Deciphering the incipient stages of phase change in superheated fluids – and their subsequent evolution – remains a major challenge, as it requires bridging microscopic nucleation with macroscopic bubble dynamics and heat transfer. This gap has long hindered predictive boiling models and limited progress in emerging thermal technologies. Here, we leverage large-scale simulations of fluctuating hydrodynamics, combined with a rare-event technique, to rationalise liquid–vapour transformation from nucleation to bubble growth. We quantify the statistics of nucleated bubbles and hydrodynamic fields, onset temperatures, and non-trivial wettability effects on the nucleation pathways. Despite operating at the mesoscale, the simulations recover microscopic nucleation signatures observed in atomistic studies, including the dependence of both the boiling onset temperature and the crossover between homogeneous and heterogeneous nucleation pathways on surface wettability. At larger scales, they reproduce the experimentally observed transition from an early exponential regime of bubble-contact areas to a post-critical power-law scaling. These findings establish a quantitative bridge between fluctuation-driven nucleation and emergent boiling dynamics across scales.

Identifying the incipient conditions of liquid-vapour transformation, the number of bubbles formed, their spatiotemporal scales, and the role of inertia remains a major challenge, reflecting how elusive the early stages of phase change are. Here, we present a theoretical framework that combines large deviation theory, multiphase fluctuating hydrodynamics and real fluid thermodynamics to compute the most probable nucleation pathways in metastable liquids. We identify the optimal trajectories connecting metastable and stable states and determine the full spatiotemporal structure of the nucleation process. Our results reveal that nucleation is not solely governed by thermodynamic forces, but is also shaped by hydrodynamic phenomena such as wave propagation and inertial effects. The approach predicts boiling thresholds for water, nitrogen, and helium, in agreement with experiments. It provides a unified, predictive description of phase-change kinetics linking microscopic fluctuations to macroscopic hydrodynamic observables, opening routes to prediction and control of phase change.

We show that the three-point skewness of concentration fluctuations is non-vanishing in free liquid diffusion, even in the limit of vanishingly small mean concentration gradients. We exploit a high-Schmidt reduction of nonlinear Landau-Lifshitz hydrodynamics for a binary fluid, both analytically and by a massively parallel Lagrangian Monte Carlo simulation. Non-Gaussian statistics result from nonlinear coupling of concentration fluctuations to thermal velocity fluctuations, analogous to the turbulent advection of a passive scalar. Concentration fluctuations obey no central limit theorem, counter to the predictions of macroscopic fluctuation theory for generic diffusive systems.

Within the framework of the E-Nucl project, and in collaboration with the group of G. Eyink at Johns Hopkins,  we investigated the emergence of non-equilibrium long-range correlations in concentration fluctuations during the free diffusion of a solute in a solvent. While such correlations are well established experimentally, their dynamical origin has remained unclear. To address this problem, we adopted the model proposed by Donev, Fai, and Vanden-Eijnden (DFV), derived from the high-Schmidt-number limit of the Landau–Lifshitz fluctuating hydrodynamics for binary mixtures. Starting from an initially sharp concentration interface in an unbounded domain, we combined analytical arguments and numerical simulations using tools from turbulence theory. The results reveal the spontaneous emergence of a quasi-steady, self-similar regime in which concentration correlations exhibit distinct spatial scalings at short and large distances. Specifically, the well-known regime of “giant concentration fluctuations” is complemented by a newly identified regime characterised by algebraic decay at large separations. The study provides fresh insight into the dynamic mechanisms underlying non-equilibrium correlations and offers quantitative predictions that can be directly tested in experiments.

Within the E-Nucl project, we uncovered complex transition pathways for vapour bubble nucleation in metastable liquids, under both homogeneous and heterogeneous conditions. The study, published in the Journal of Fluid Mechanics (Gallo et al., “Complex transition pathways in boiling and cavitation”), combines Navier–Stokes–Korteweg dynamics with rare-event techniques to resolve transition mechanisms and nucleation times. The results show that nucleation pathways deviate significantly from classical theory: bubble volume alone is not a sufficient reaction coordinate, as the process is driven by long-wavelength fluctuations with densities close to the metastable liquid. We further propose a hydrodynamics-based strategy to infer nucleation times, validated against state-of-the-art theories. The analysis reveals non-trivial effects of surface wettability on heterogeneous nucleation, including homogeneous nucleation persisting at moderate hydrophilic wettabilities, while hydrophobic walls anticipate the spinodal. Although demonstrated for a prototypical fluid, the approach is general and can be extended to complex geometries and real fluids, paving the way to predictive modelling of nucleation in engineering systems.