Solute Diffusion in Metals: Larger Atoms Can Move Faster

Abstract

First-principles calculations for the diffusion of transition metal solutes in nickel challenge the commonly accepted description of solute diffusion rates in metals. The traditional view is that larger atoms move slower than smaller atoms. Our calculation shows the opposite: larger atoms, in fact, can move much faster than smaller atoms. Conventional mechanisms involving the effect of misfit strain or the solute-vacancy binding interactions cannot explain this counterintuitive diffusion trend. Instead, the origin of this behavior stems from the bonding characteristics of the d electrons of solute atoms, suggesting that a similar diffusion trend also occurs in other types of host lattices.

Effects of Alloying Elements on Elastic, Stacking Fault and Diffusion Properties of Fcc Ni from First-principles: Implications for Tailoring the Creep Rate of Ni-base Superalloys

“…It can be seen that the shear modulus decreases with increasing equilibrium volume, agreeing with the general trend observed for elastic properties relating moduli and equilibrium volume [9]. The shear modulus values of Ni 31 X alloys are affected greatly by the properties of the respective alloying element, such as, the larger shear moduli of Ni 31 Ru and Ni 31 Os are mainly caused by the larger shear moduli of pure elements Ru and Os compared to Ni [9]. Among all of the 26 alloying elements, the largest decrease of elastic properties of Ni is due to element Y, followed by Zr and Sc.…”

“…Figure 4 plots activation energy calculated for some of the alloying elements by Janotti et al [31] versus activation energies calculated in the present work, sorted by increasing atomic number for the 3d, 4d, and 5d transition elements. Other than the elements Y, Zn, Sc, Al, and Si, which were not previously calculated by Janotti et al [31], excellent correlation with the present work for all elements with the exception of Zr, and to a lesser extent some of the 4d solutes.…”

“…Figure 4 plots activation energy calculated for some of the alloying elements by Janotti et al [31] versus activation energies calculated in the present work, sorted by increasing atomic number for the 3d, 4d, and 5d transition elements. Other than the elements Y, Zn, Sc, Al, and Si, which were not previously calculated by Janotti et al [31], excellent correlation with the present work for all elements with the exception of Zr, and to a lesser extent some of the 4d solutes. This can be attributed to the work of Janotti et al [31] using the ultrasoft pseudopotential method [15], because the PAW [16] method used in the present work reconstructs the exact valence wave function with smaller core radii than the ultrasoft potentials, making the PAW a little more computationally expensive, but also more accurate.…”

“…For the sake of simplicity and efficiency, D 0 can be estimated by treating the vibrational degrees of freedom near the saddle point as nearly simple harmonic oscillators [32]. This value was found to be on the order of 10 -5 m 2 /s for the dilute diffusion of transition metals in Ni [2,31]. Thus, 10 -5 m 2 /s will be used for the diffusion prefactor in this work as an approximation, a value that well represents the experimental diffusion prefactor of pure Ni (8.5×10 -5 -17.7×10 -5 m 2 /s [33][34][35]).…”

“…In the present work, the temperature dependence of the correlation factor, ݂ ଶ and the activation energy for diffusion is calculated as [6,31] ܳ ൌ ‫ܧ∆‬ ‫ܧ߂−‬ + ‫ܧ߂‬…”