Nonisothermal nucleation in the gas phase is driven by cool subcritical clusters

Latent-to-sensible heat conversion kinetics during nanoparticle coalescence

Coagulational growth in an aerosol is a multistep process; first particles collide, and then they coalesce with one another. Coalescence kinetics have been investigated in numerous prior studies, largely through atomistic simulations of nanoclusters (102-104 atoms). However, with a few exceptions, they have either assumed the process is completely isothermal or is a constant energy process. During coalescence, there is the formation of new bonds, decreasing potential energy, and correspondingly increasing internal kinetic (thermal) energy. Internal kinetic energy evolution is dependent not only on coalescence kinetics but also on heat transfer to the surrounding gas. Here, we develop and test a model of internal kinetic energy evolution in collisionally formed nanoclusters in the presence of a background gas. We find that internal kinetic energy dynamics hinge upon a power law relationship describing latent-to-sensible heat release as well as a modified thermal accommodation coefficient. The model is tested against atomistic models of 1.5-3.0 nm embedded-atom gold nanocluster sintering in argon and helium environments. The model results are in excellent agreement with the simulation results for all tested conditions. Results show that nanocluster effective temperatures can increase by hundreds of Kelvin due to coalescence, but that the rise and re-equilibration of the internal kinetic energy is strongly dependent on the background gas environment. Interestingly, internal kinetic energy change kinetics are also found to be distinct from surface area change kinetics, suggesting that modeling coalescence heat release solely due to surface area change is inaccurate.

Multiscale approach to the theory of nonisothermal homogeneous nucleation

Molecular dynamics (MD) of the Lennard-Jones cluster in the environment of supersaturated vapor at fixed temperature and density is used for the investigation of nonisothermal nucleation. The results allow one to single out different processes occurring at different time scales, the Ornstein-Uhlenbeck fluctuations at the short time scale and a combination of slow diffusion and drift of the fluctuation packet that represents a cluster, at the long time scale. The multiscale approach is developed, in which a separate treatment of different time scales makes it possible to consider strongly correlated cluster size and temperature. This reduces the nonisothermal cluster evolution to a one-dimensional problem. The fluctuation packet drift velocity and diffusivity are calculated based on the cluster microscopic thermophysical parameters determined in this work from MD data for isothermal clusters. The proposed approach is consistent with the results of our MD simulation.

Understanding why constant energy or constant temperature may affect nucleation behavior in MD simulations: A study of gas hydrate nucleation

Molecular dynamics simulations have been widely used in exploring the nucleation behavior of many systems, including gas hydrates. Gas hydrates are ice-like solids in which gas molecules are trapped in water cages. During hydrate formation, a considerable amount of heat is released, and previous work has reported that the choice of temperature control scheme may affect the behavior of hydrate formation. The origins of this effect have remained an open question. To address this question, extensive NVE simulations and thermostatted (NPT and NVT) simulations with different temperature coupling strengths have been performed and compared for systems where a water nanodroplet is immersed in a H2S liquid. Detailed analysis of the hydrate structures and their mechanisms of formation has been carried out. Slower nucleation rates in NVE simulations in comparison to NPT simulations have been observed in agreement with previous studies. Probability distributions for various temperature measures along with their spatial distributions have been examined. Interestingly, a comparison of these temperature distributions reveals a small yet noticeable difference in the widths of the distributions for water. The somewhat reduced fluctuations in the temperature for the water species in the NVE simulations appear to be responsible for reducing the hydrate nucleation rate. We further conjecture that the NVE-impeded nucleation rate may be the result of the finite size of the surroundings (here the liquid H2S portion of the system). Additionally, a local spatial temperature gradient arising from the heat released during hydrate formation could not be detected.

Further cautionary tales on thermostatting in molecular dynamics: Energy equipartitioning and non-equilibrium processes in gas-phase simulations

Molecular dynamics (MD) simulations of gas-phase chemical reactions are typically carried out on a small number of molecules near thermal equilibrium by means of various thermostatting algorithms. Correct equipartitioning of kinetic energy among translations, rotations, and vibrations of the simulated reactants is critical for many processes occurring in the gas phase. As thermalizing collisions are infrequent in gas-phase simulations, the thermostat has to efficiently reach equipartitioning in the system during equilibration and maintain it throughout the actual simulation. Furthermore, in non-equilibrium simulations where heat is released locally, the action of the thermostat should not lead to unphysical changes in the overall dynamics of the system. Here, we explore issues related to both obtaining and maintaining thermal equilibrium in MD simulations of an exemplary ion-molecule dimerization reaction. We first compare the efficiency of global (Nosé-Hoover and Canonical Sampling through Velocity Rescaling) and local (Langevin) thermostats for equilibrating a system of flexible compounds and find that of these three only the Langevin thermostat achieves equipartition in a reasonable simulation time. We then study the effect of the unphysical removal of latent heat released during simulations involving multiple dimerization events. As the Langevin thermostat does not produce the correct dynamics in the free molecular regime, we only consider the commonly used Nosé-Hoover thermostat, which is shown to effectively cool down the reactants, leading to an overestimation of the dimerization rate. Our findings underscore the importance of thermostatting for the proper thermal initialization of gas-phase systems and the consequences of global thermostatting in non-equilibrium simulations.