posted on 2023-11-03, 11:37authored byYue Fang, Hanyu Gao
Polymer materials design is essential in a wide range
of industries,
and the simulation of polymerization reactions provides a promising
approach to the efficient design of synthesis recipes for desired
property targets. The kinetic Monte Carlo (KMC) algorithm has been
widely used in the simulation of polymerization reactions due to its
ability to track the most delicate details of individual molecular
sequences. However, the disadvantage of high computational cost has
limited its broader applications, primarily caused by the large number
of molecules that must be sampled in KMC simulations. This work proposes
a novel scaling acceleration algorithm (SAA) specifically designed
for hybrid KMC simulations of linear radical polymerization, which
can significantly reduce the number of molecules in KMC simulations
of radical polymerization, thus greatly accelerating the simulations.
The algorithm can be used in radical reaction systems (currently limited
to simple kinetic schemes) with species concentrations on significantly
different scales, thus presenting a novel method for simulating polymerization
reactions for efficient polymer materials design. To be more specific,
this was achieved by solving a continuum model in parallel with the
KMC simulation to provide an accurate value of radical concentrations
for calculating a scaling factor (SF), which was then used to recover
the correct results in KMC simulations when using an insufficient
number of molecules. The algorithm was tested on three case studies:
(1) free-radical polymerization, copolymerization, (2) atom-transfer
radical polymerization (ATRP), homopolymerization, and (3) ATRP, copolymerization.
The results showed that SAA simulates the system with at least 100
times fewer molecules than what would have been required for an accurate
KMC simulation while achieving acceptable accuracy in the key characteristics
of the polymerization system (including conversion, molecular weight
distribution, and sequence distribution). In addition, sensitivity
analysis of the ATRP model showed that the algorithm can be adequately
applied to a wide range of cases with extreme rate constants. While
SAA offers a speedier approach to simulate radical polymerization
reactions, it is crucial to note that this acceleration process incurs
a minor compromise in accuracy, particularly in the calculation of z-average molecular weight (Mz). The algorithm currently accounts for basic polymerization
reaction schemes, assuming rapid ordinary differential equation solving
and omitting chain-length dependencies.