Session: 03-05-02: Integrated Computational Materials Engineering

Paper Number: 121934

121934 - Optimal System Size for Molecular Dynamics Simulation of Epoxy Materials 

Abstract:

Molecular dynamics (MD) simulation plays an important role in the academia and industrial research to describe the molecular behavior of materials at the nanoscale. Modeling of the amorphous material are subjected to statistical fluctuation which will lead to the increase of the uncertainties in the results. To address this issue, multiple replicates are created to improve the sampling performance and precision in the predicted properties. Unfortunately, there is not enough studies in the literature regarding the relationship between MD model size and the resulting precision in predicted mechanical properties. In this study, different MD model size epoxy resin was generated to obtain the optimal MD model size to balance efficiency and precision.

Introduction: 

Accurate MD modeling methods are valuable for understanding material responses in situations where experiments are challenging [1]. Perhaps the main issue with MD modeling is the computational cost required to simulate realistic material systems, which limits the size and time scales to nanometers and nanoseconds, respectively. Since polymer materials are amorphous in nature and having material periodicity characteristics that are orders of magnitude above the nanometer length scale, relatively large MD simulation periodic boxes are required to accurately predict nano-scale properties that statistically resemble those observed on the laboratory scale. Consequently, MD simulations of polymer materials can suffer from unintended size effects[2,3]. This study predicts the optimal system size that balances both predictive precision and simulation time, which is necessary for the rapid design of next-generation composite materials through ICME approaches.

Material and Method:

DGEBF and DETDA monomers were mixed with the ratio of 2:1 in periodic simulation boxes and reproduced in each direction of the simulation box to create six different system sizes ranging between 5265 to 36,855 atoms using LAMMPS. Five replicates were independently built for each of the six system sizes for statistical sampling purposes. In the next step, each independent replicate for each system size was densified to target density of 1.20 g/cc, polymerized and equilibrated at 300K and 1 atm. The shear and bulk modulus were predicted and the simulation times for these simulations were recorded for each replicate and model size. The Young's modulus and Poisson's ratio values were determined using the standard equations of linear elasticity for isotropic materials [4].

Results:

For the range of model sizes considered, the results indicate that the predicted mass density precision is not significantly affected by the MD model size. However, the precision of the predicted elastic properties and strength are substantially affected by model size. As expected, simulation times are significantly affected by model size. Taken together, the results show that epoxy MD model sizes of 15,000 atoms are an excellent balance between precision and simulation time. That is, larger models do not provide significant increases in precision in predicted mechanical properties, and smaller models result in clear reductions in precision.

Conclusion:

This study investigates the trade-off between computational efficiency and prediction accuracy in modeling epoxy systems using molecular dynamics (MD) simulations. Larger MD simulations yield greater precision but require significantly more time compared to smaller simulations. The impact of model size on predictive accuracy varies depending on the physical and mechanical properties being predicted. These findings are valuable for employing MD as a tool for designing composite materials in ICME research.

References:

1.         Li, C.; Strachan, A. Molecular Scale simulations on Thermoset Polymers: A review. Journal of Polymer Science Part B: Polymer Physics 2015, 53, 103-122.

2.         El Hage, K.; Hedin, F.; Gupta, P.K.; Meuwly, M.; Karplus, M. Valid Molecular Dynamics Simulations of Human Hemoglobin Require a Surprisingly Large Box Size. Elife 2018, 7, e35560.

3.         Wan, S.; Sinclair, R.C.; Coveney, P.V. Uncertainty Quantification in Classical Molecular Dynamics. Philosophical Transactions of the Royal Society A 2021, 379, 20200082.

4.         Odegard, G.M.; Patil, S.U.; Deshpande, P.P.; Kanhaiya, K.; Winetrout, J.J.; Heinz, H.; Shah, S.P.; Maiaru, M. Molecular Dynamics Modeling of Epoxy Resins Using the Reactive Interface Force Field. Macromolecules 2021, 54, 9815-9824.

Presenting Author: Khatereh Kashmari michigan technological university

Presenting Author Biography: I am currently pursuing a Ph.D. at Michigan Tech, with a specialization in Molecular Dynamics (MD) and Computational Material Science. my research mainly focus on molecular dynamics modeling of polymer material their properties. I also play piano and tennis in my spare time. I love hiking and camping during summer.