Thermodynamic and kinetic origins of micro-explosions in Al-Mg alloy nanoparticles via machine learning molecular dynamics

Aluminum-magnesium (Al-Mg) alloy nanoparticles are promising high-energy fuels for advanced flight vehicles. While alloying enhances combustion performance, the atomistic mechanisms driving accelerated ignition and micro-explosions remain poorly understood. Existing models attribute micro-explosions to preferential Mg vaporization, yet the atomic-scale origins have not been resolved, specifically regarding how Mg disrupts the host lattice, reshapes transport pathways, and modulates the vaporization energy landscape. In this study, a high-precision machine learning potential for the Al/Mg/O system was developed to investigate these mechanisms via molecular dynamics simulations. Beyond the conventional preferential-vaporization mechanism, the results reveal that the intrinsic size mismatch of Mg atoms introduces local topological perturbations, accelerating disordering and expanding free volume. This lowers spatial mass-transfer barriers, promoting the surface segregation and vaporization of Mg, while inward oxygen permeation establishes a bidirectional mass-transport network. Quantitative kinetic analyses reveal an exponential decay of the apparent activation energy for vaporization (from 1.25 eV for Al to 0.75 eV for the 60 wt% Mg alloy), a departure from linear mixing rules, indicating a cooperative lattice-destabilization effect. This barrier reduction promotes internal vapor pressure accumulation, causing particles to breach the oxide shell yield limit and trigger micro-explosions, ultimately transforming the reaction from localized surface passivation to global deep oxidation. By resolving the thermodynamic-kinetic coupling governing this transition, this work provides an atomistic, composition-dependent framework for the combustion design of metallic fuels.