Uncertainty quantification in MD simulations of concentration driven ionic flow through a silica nanopore. II....

Area of Science:

• Computational chemistry
• Materials science
• Physical chemistry

Background:

• Molecular dynamics (MD) simulations are crucial for understanding ionic flow in nanopores.
• Uncertainty quantification (UQ) is essential for reliable MD predictions.
• Previous work established UQ for ionic flow; this study refines it for specific parameters.

Purpose of the Study:

• To quantify the sensitivity of ionic flow in silica nanopores to Lennard-Jones potential parameters (ɛ(Na(+)) and ɛ(Cl(-))).
• To map parameter uncertainty to MD predictions of ionic fluxes.
• To determine the most suitable regression model for noisy MD data.

Main Methods:

• Forward propagation analysis to link parameter uncertainty to MD predictions.
• Polynomial chaos expansions and Bayesian inference for UQ.
• Bayes factor analysis to select regression models.
• Green-Kubo time correlations for bulk transport coefficients.

Main Results:

• The Na(+) conductance response surface is effectively inferred despite significant noise.
• Underlying trends in Cl(-) conductance are partially obscured by noise.
• Correlations between ionic conductances and bulk transport coefficients (viscosity, diffusivities) are analyzed.

Conclusions:

• Parametric uncertainty in Lennard-Jones potentials significantly impacts MD predictions of ionic flow.
• The methodology allows for robust UQ in complex nanoscale systems.
• Understanding these sensitivities is key for designing nanoporous materials for specific ionic transport applications.