Response to letter “Electron correlation and relativity of the 5f electrons in the U[sbnd]Zr alloy system”

Abstract

In the Letter [Söderlind et al., J. Nucl. Mater. 444, 356 (2014)], Söderlind et al. state their interpretation that 1) we view electron correlation to be strong and including spin-orbit coupling (SOC) to be necessary for U metal and UZr alloy in our article [Xiong et al., J. Nucl. Mater. 443, 331 (2013)]. Further, they argue that 2) density functional theory (DFT) without adding the Hubbard U potential, especially when solved using all electron methods, already models U and UZr accurately, and 3) adding the Hubbard U potential to DFT in DFT + U models U and UZr worse than DFT according to volume, bulk modulus, and magnetic moments predicted from their calculations of the γU phase of elemental U metal. With respect to Söderlind et al.’s interpretation 1), we clarify that our opinions are that U and UZr are not strongly, but weakly to moderately correlated and that including SOC is beneficial but not necessary for modeling most ground state properties of U and UZr. With respect to Söderlind et al.’s argument 2) we demonstrate that previously neglected and very recent experimental data suggest that DFT in Söderlind's full-potential linear muffin-tin orbital calculations [Söderlind, Phys. Rev. B 66, 085113 (2002)] in fact models the bulk modulus and elastic constants of αU with errors considerably larger than other related elements, e.g., most transition metals. With respect to Söderlind et al.’s argument 3) we argue that they have inappropriately focused on just one phase (the BCC γU phase of U metal), neglecting the other phases which represent the majority of our evidence, and made overgeneralizations based on results at only one Ueff value of 2 eV. We therefore maintain our original conclusion that the accuracy of DFT for modeling U and UZr has room for improvement and DFT + U can be of value for this purpose on at least some ground state properties.