2023 AIChE Annual Meeting
(197bw) Force Fields and Initial Conformations Optimization of Amorphous Polymers in Multi-Scale (atomistic and coarse-grained) Simulations.
Authors
Sahar Zenoozi - Presenter, University of South Florida
Peter J. Ludovice, Georgia Institute of Technology
Clifford Henderson, University of South Florida
Polymers are one of the most diverse classes of materials thanks to the nearly infinite number of combinations of structures possible due to a variety of characteristics of polymer synthesis including the ability to utilize numerous different polymerization approaches available today, the wide array of monomer chemistries and architectures that are easily synthesized, the numerous polymer architectures beyond linear homopolymers that are available through various techniques such as the use of branching chemistries, and the relative ease of enchaining various combinations of different monomers along a polymer chain. Although this flexibility to exquisitely control and tailor polymer structure makes this class of materials well-suited to solve a variety of engineering challenges across a variety of application areas, the size of the polymer design space also presents a significant challenge when tasked with determining the best polymer structure to produce that can satisfy the property and processing requirements for a particular problem. With the dramatic gains in computational capability in recent years, molecular modeling has finally transitioned from a tool that can be used for predictive design and optimization of polymer structure as opposed to most past use cases where it was best suited to explore either general polymer behavior or to semi-quantitatively match and explain experimental data for real polymer systems in hindsight. Thus, the ability to develop of reliable predictive computational models for the behavior and properties of real polymer systems is of great importance for the current and future design and optimization of new materials. In this work, a methodology is presented which can be used to develop improved atomistic and coarse-grained force fields for molecular dynamics simulation of real polymers. Furthermore, an efficient method for producing initial conformational states in dense linear polymer simulations is also presented.
The first challenge when creating polymer molecular dynamics simulations that are expected to quantitatively match, and better yet predict, polymer properties is the selection of a force field to represent the interactions of the atoms and/or coarse-grained structural elements of the polymer. There is in general always a need for the ability to conduct accurate atomistic MD simulation of polymers. In some cases, such atomistic simulations can be useful themselves and in other cases they can be useful for parameterizing and training coarse-grained models that can allow for larger time and spatial scale simulations. In this work, an improved force-field is developed for the simulation of linear hydrocarbon polymers based on initial use of the OPLS force-field with various refinement and optimization approaches. The end result, as will be demonstrated from the results presented for several exemplar linear homopolymers including polystyrene (PS), polypropylene (PP), and polyisoprene (PI), is a force-field which can reproduce the physiochemical properties of such linear hydrocarbon polymers with significantly improved accuracy. Once such improved atomistic force-fields and simulation were available, these capabilities were then utilized to develop improved coarse-grained models for such linear hydrocarbon polymer systems. In particular, models that allow for higher levels of coarse graining while retaining the ability to predict important polymer structural and physical properties were explored and developed. The general methodology used for this coarse-grained model development will be presented and the specific coarse-grained models developed for PS, PP, and PI will be presented along with a discussion of the ability of such coarse-grained models to accurately reproduce the behavior and properties of these polymer systems.
An often underappreciated challenge in such polymer modeling problems is the ability to generate realistic polymer conformations for polymers at their experimental density. This paper will also discuss the importance of such polymer conformations with respect to the general ability to produce accurate molecular dynamics models for linear polymer systems that can reproduce the properties of well characterized polymer systems (i.e. PS, PP, and PI). Improved methodologies for generating initial conformations for use in atomistically detailed simulations of dense polymer systems will be discussed and benchmarked in detail.
The end result of this work is a set of methodologies and both atomistic and coarse-grained polymer simulation models that can reproduce a wide variety of the physiochemical properties of linear hydrocarbon polymers. The future potential to extend this work to both other known and as of yet to be synthesized polymer systems will be discussed.
The first challenge when creating polymer molecular dynamics simulations that are expected to quantitatively match, and better yet predict, polymer properties is the selection of a force field to represent the interactions of the atoms and/or coarse-grained structural elements of the polymer. There is in general always a need for the ability to conduct accurate atomistic MD simulation of polymers. In some cases, such atomistic simulations can be useful themselves and in other cases they can be useful for parameterizing and training coarse-grained models that can allow for larger time and spatial scale simulations. In this work, an improved force-field is developed for the simulation of linear hydrocarbon polymers based on initial use of the OPLS force-field with various refinement and optimization approaches. The end result, as will be demonstrated from the results presented for several exemplar linear homopolymers including polystyrene (PS), polypropylene (PP), and polyisoprene (PI), is a force-field which can reproduce the physiochemical properties of such linear hydrocarbon polymers with significantly improved accuracy. Once such improved atomistic force-fields and simulation were available, these capabilities were then utilized to develop improved coarse-grained models for such linear hydrocarbon polymer systems. In particular, models that allow for higher levels of coarse graining while retaining the ability to predict important polymer structural and physical properties were explored and developed. The general methodology used for this coarse-grained model development will be presented and the specific coarse-grained models developed for PS, PP, and PI will be presented along with a discussion of the ability of such coarse-grained models to accurately reproduce the behavior and properties of these polymer systems.
An often underappreciated challenge in such polymer modeling problems is the ability to generate realistic polymer conformations for polymers at their experimental density. This paper will also discuss the importance of such polymer conformations with respect to the general ability to produce accurate molecular dynamics models for linear polymer systems that can reproduce the properties of well characterized polymer systems (i.e. PS, PP, and PI). Improved methodologies for generating initial conformations for use in atomistically detailed simulations of dense polymer systems will be discussed and benchmarked in detail.
The end result of this work is a set of methodologies and both atomistic and coarse-grained polymer simulation models that can reproduce a wide variety of the physiochemical properties of linear hydrocarbon polymers. The future potential to extend this work to both other known and as of yet to be synthesized polymer systems will be discussed.