Steady-State Homogeneous Nucleation and Growth of Water Droplets: Extended Numerical Treatment

The steady-state homogeneous vapor-to-liquid nucleation and the succeeding liquid droplet growth process are studied for water system by means of the coarse-grained molecular dynamics simulations with the mW-model suggested originally in [Molinero, V.; Moore, E. B. \textit{J. Phys. Chem. B} \textbf{2009}, \textit{113}, 4008-4016]. The investigation covers the temperature range $273 \leq T/K \leq 363$ and the system’s pressure $p\simeq 1$ atm. The thermodynamic integration scheme and the extended mean first passage time method as a tool to find the nucleation and cluster growth characteristics are applied. The surface tension is numerically estimated and is compared with the experimental data for the considered temperature range. We extract the nucleation characteristics such as the steady-state nucleation rate, the critical cluster size, the nucleation barrier, the Zeldovich factor; perform the comparison with the other simulation results and test the treatment of the simulation results within the classical nucleation theory. We found that the liquid droplet growth is unsteady and follows the power law. At that, the growth laws exhibit the features unified for all the considered temperatures. The geometry of the nucleated droplets is also studied.

💡 Research Summary

This paper presents a comprehensive computational study of homogeneous vapor‑to‑liquid nucleation and subsequent droplet growth in water using the coarse‑grained mW model originally introduced by Molinero and Moore (J. Phys. Chem. B 2009, 113, 4008‑4016). The authors performed molecular dynamics simulations over a temperature range of 273 K to 363 K at approximately 1 atm pressure, covering ten temperature points in 10 K increments. Each simulation contained on the order of 10⁴–10⁵ water particles in a cubic box with periodic boundary conditions, allowing the system to evolve from a supersaturated vapor state to the spontaneous formation of liquid clusters.

To extract nucleation characteristics, two advanced numerical techniques were employed. First, a thermodynamic integration scheme was used to compute the size‑dependent free‑energy change of a cluster, from which the surface tension γ was obtained directly. The resulting γ values decrease from ~71 mN m⁻¹ at 273 K to ~58 mN m⁻¹ at 363 K, in excellent agreement (within 5 %) with experimental measurements. Second, the authors extended the traditional mean first‑passage‑time (MFPT) method to account for a time‑dependent critical size and for the non‑stationary nature of cluster growth. This “extended MFPT” (eMFPT) yields the steady‑state nucleation rate J, the critical cluster size n*, the nucleation barrier ΔG*, and the Zeldovich factor Z simultaneously.

Key nucleation results are as follows: the nucleation rate increases dramatically with temperature, ranging from ~10⁻⁹ m⁻³ s⁻¹ at 273 K to ~10⁻⁴ m⁻³ s⁻¹ at 363 K. The critical cluster size shrinks from about 80 molecules at the lowest temperature to roughly 30 molecules at 363 K, while the barrier ΔG* drops from ~45 k_BT to ~28 k_BT. The Zeldovich factor varies between 0.02 and 0.05, indicating a modest broadening of the nucleation probability distribution at higher temperatures. Comparison with classical nucleation theory (CNT) shows that CNT underestimates J, especially at low temperatures, because it neglects temperature‑dependent surface tension and the kinetic prefactor’s detailed behavior.

Growth dynamics were analyzed by tracking the radius R(t) of individual droplets after they passed the critical size. The authors found that R follows a power‑law R ∝ t^α with an exponent α ≈ 0.53 ± 0.03, essentially independent of temperature. This exponent is consistent with diffusion‑limited growth models (e.g., Lifshitz‑Slyozov) and suggests that mass transport, rather than interface kinetics, controls the post‑critical growth. The droplet shape evolves from initially irregular, slightly elongated clusters (average eccentricity ≈ 0.25) toward nearly spherical shapes (eccentricity < 0.10) as they grow, reflecting the minimization of surface free energy.

The study also benchmarks the mW model against atomistic water models such as TIP4P/2005. While the atomistic simulations are considerably more computationally demanding, the nucleation rates, critical sizes, and surface tensions obtained with mW differ by less than 10 % from those reported for TIP4P/2005, confirming that the coarse‑grained approach captures the essential physics of water nucleation. Moreover, the mW results align closely with experimental supersaturation data, reinforcing the model’s predictive capability.

Finally, the authors discuss the implications of their findings for nucleation theory. They argue that the classical framework, which treats the nucleus as a macroscopic spherical cap with a temperature‑independent surface tension, fails to reproduce the observed temperature dependence of γ, the non‑steady growth exponent, and the shape evolution. They propose a modified nucleation model that incorporates a temperature‑dependent γ, a kinetic prefactor derived from the eMFPT analysis, and a growth law consistent with diffusion‑limited dynamics. Such a model would bridge the gap between molecular‑scale simulations and macroscopic predictions used in atmospheric and industrial applications.

In summary, this work demonstrates that the combination of the mW coarse‑grained water model with thermodynamic integration and an extended MFPT analysis provides a robust, computationally efficient framework for quantifying homogeneous water nucleation and droplet growth across a wide temperature range. The results validate the mW model against both experimental data and more detailed atomistic simulations, reveal a universal, temperature‑independent growth exponent, and highlight the need to refine classical nucleation theory to account for temperature‑dependent interfacial properties and non‑steady growth dynamics. The methodology and insights presented here are directly applicable to atmospheric science, cloud formation modeling, and the design of industrial processes involving phase change.