Al₂₀⁺does melt, albeit above the bulk melting temperature of aluminium

Abstract

Employing first principles parallel tempering molecular dynamics in the microcanonical ensemble, we report the presence of a clear solid-liquid-like melting transition in Al20(+) clusters, not found in experiments. The phase transition temperature obtained from the multiple histogram method is 993 K, 60 K above the melting point of aluminium. Root mean squared bond length fluctuation, the velocity auto-correlation function and the corresponding power spectrum further confirm the phase transition from a solid-like to liquid-like phase. Atoms-In-Molecules analysis shows a strong charge segregation between the internal and surface atoms, with negatively charged internal atoms and positive charge at the surface. Analysis of the calculated diffusion coefficients indicates different mobilities of the internal and surface atoms in the solid-like phase, and the differences between the environment of the internal atoms in these clusters with that of the bulk atoms suggest a physical picture for the origin of greater-than-bulk melting temperatures.

Characterizing the Greater-Than-Bulk Melting Behavior of Ga–Al Nanoalloys

“…Moreover, among the charged surface atoms it is the gallium atom which becomes negatively charged and aluminium which becomes positively charged in both cases. A similar picture, where an electrostatic cage confines the internal atoms was also observed for the monometallic Ga 20 + and Al 20 + clusters [38].…”

“…Strong charge segregation between the internal and surface sites plays an important role in the greater-than-bulk melting behaviour as also reflected by the diffusion coefficients (Fig. 4) of the representative clusters [38].…”

Building machine learning force fields for nanoclusters

“…[1][2][3][4][5] Because of the variety of isomers accessible at finite temperatures and the lack of translational symmetry, implying a non-trivial interplay between their geometric and electronic structure, a comprehensive understanding of metallic nanoclusters remains challenging, despite their potential use in many advanced applications. [6][7][8][9][10][11][12][13][14][15][16] Density Functional Theory (DFT) is the most common framework to investigate the static and dynamical properties of nanoclusters of few tens of atoms, for which the standard classical force fields cannot systematically be relied upon to provide sufficient accuracy. However, DFTbased calculations are very expensive, and probing limited time scales in first principles dynamical simulations may lead to poor sampling of the nanoclusters' configuration space.…”

Building machine learning force fields for nanoclusters

“…Metallic nanoparticles have fascinating chemophysical properties, different from those of their individual atomic constituents and their bulk counterparts [1][2][3][4][5] . Because of the variety of isomers accessible at finite temperatures and the lack of translational symmetry, implying a non-trivial interplay between their geometric and electronic structure, a comprehensive understanding of metallic nanoclusters remains challenging, despite their potential use in many advanced applications [6][7][8][9][10][11][12][13][14][15][16] . Density Functional Theory (DFT) is the most common framework to investigate the static and dynamical properties of nanoclusters of few tens of atoms, for which standard classical force fields cannot systematically be relied upon to provide sufficient accuracy.…”