Large scale molecular dynamics simulations of homogeneous nucleation

Area of Science:

• Physical Chemistry
• Computational Physics
• Materials Science

Background:

• Homogeneous vapor-to-liquid nucleation is a fundamental process in nature and technology.
• Accurate theoretical models are crucial for predicting nucleation phenomena.
• Previous models showed significant discrepancies with experimental and simulation data.

Purpose of the Study:

• To perform large-scale molecular dynamics (MD) simulations of homogeneous vapor-to-liquid nucleation.
• To compare simulation results with experimental data and existing nucleation theories.
• To develop a new empirical nucleation model.

Main Methods:

• Large-scale molecular dynamics (MD) simulations using Lennard-Jones (LJ) atoms (1-8 billion atoms, up to 1.2 μs).
• Simulations covered wide ranges of supersaturation ratios (S ≃ 1.55-10^4) and temperatures (kT = 0.3-1.0ε).
• Analysis of nucleation rates, critical cluster sizes, and growth rates per encounter.

Main Results:

• Achieved high accuracy in simulating nucleation rates as low as 10^17 cm⁻³s⁻¹ and critical cluster sizes up to 100 atoms.
• Demonstrated very good agreement between MD simulations and recent argon nucleation experiments.
• Identified limitations of classical and semi-phenomenological nucleation models, with discrepancies up to 9 orders of magnitude.
• Growth rates per encounter in simulations ranged from 0.05 to 0.24.

Conclusions:

• MD simulations provide a reliable tool for studying nucleation, showing excellent agreement with experimental data.
• Classical nucleation theory underestimates rates at low temperatures and overestimates them at high temperatures.
• A new empirical nucleation model, based on free energy functions from subcritical clusters, accurately predicts nucleation rates.