In scientific research papers, several face-centered cubic metals such as Ag, Cu, Au, Ni, and Pt have presented an upward curvature in the Arrhenius plot of the atomic self-diffusion coefficient close to the melting point. This phenomenon has stimulated hypothesis around three potential mechanisms: (1) Different types of defects like monovacancies and divacancies contributing to the diffusion process with differing activation energies; (2) An intrinsic temperature dependence of enthalpy and entropy changes associated with the formation and migration of monovacancies; (3) Contributions from double jumps of monovacancies. This particular discussion will primarily focus on the role of correlated multiple jumps and their temperature dependence of the diffusion coefficient.

Diffusion of vacancies in solids close to melting had already been known to undergo correlated jumps, but a statistically significant measurement of this effect had not been reported until the active research undertaken by our team. We undertook molecular dynamic simulations in gold, aiming to characterize qualitatively and quantitively the effects of correlated multiple jumps on the temperature dependence of the diffusion coefficient. Our results, based on a large number of event analyses, show that multiple jumps are present and highly effective in bending the Arrhenius plot near the melting point.

To integrate this new data into the diffusion theory that explicitly accounts for dynamical effects, a microscopic interpretation of the multi-jump mechanisms was needed. The simulations utilized for this research didn't account for mechanisms related to defect types or enthalpy and entropy changes.

It is particularly important to note our methodology in conducting the simulations. We utilized a simulation system consisting of 255 particles with periodic boundary conditions, interacting via a well-developed "glue" many-body potential. It was carefully chosen as suitable for our study as it mimics, particularly well, the melting temperature and thermal expansion of Au, as well as the T= 0 energetics of a monovacancy.

We then proceeded to analyzing all the vacancy jumps in simulations performed at twelve different temperatures, between 1000K and 1550K, at steps of 50K. Melting, in this model, occurs near 1350K, meaning that runs at 1400K, 1450K, 1500K and 1550K refer to an overheated crystal which, in the absence of free surfaces, remains stable within the simulation time scale.

In conclusion, the migration of just monovacancies within our simulations produces a measurable curvature. The curve bends just below Tm and continues into the superheated region. For the atomic migration contribution to the diffusion coefficient, we see an upward trend beginning at the melting point, further highlighting the correlation between melting point and diffusion mechanisms. This is a significant contribution to metallic atomic diffusion knowledge and opens avenues for further research.