Comment on "Oxygen Vacancy Ordering and Electron Localization in CeO2: Hybrid Functional Study"

GANDUGLIA-PIROVANO, M. VERÓNICA; MURGIDA, GUSTAVO E.; FERRARI, VALERIA; LLOIS, ANA MARÍA

JOURNAL OF PHYSICAL CHEMISTRY C

AMER CHEMICAL SOC

Año: 2017 vol. 121 p. 21080 - 21083

1932-7447

Han et al. have recently published an article1 on theordering of oxygen vacancies and electron localization inbulk CeO2 using density functional theory (DFT) with theHeyd−Scuseria−Ernzerhof (HSE06) functional.2−4 On thebasis of their results, they concluded that oxygen vacancies tendto linearly order in the [111] CeO2 direction with a weakenedexcess charge localization compared with the case of a singlevacancy. Moreover, distinct vacancy-induced lattice relaxationswere found to be crucial for the interpretation of their results.This Comment is written to prevent misconceptions regardingthe localization of the excess charge and the associated latticerelaxations discussed in the work of Han et al.1that misses anycitation to previous recent work5−7 on related subjects usingthe DFT+U approach with the Perdew−Becke−Ernzerhof(PBE) functional.8 As an example, Han et al.1 describe the caseof an isolated bulk vacancy and claim that the two excesselectrons left behind upon vacancy formation localize on twonearest-neighbor Ce4+ cations that are reduced to Ce3+.Moreover, for the cases of a third-neighbor vacancy pair(VV3) and vacancy line along the [111] direction, their resultsindicate a homogeneous distribution of the Ce3+ ions onnearest-neighbor sites to the vacancies; namely, each vacancyhas two first cationic Ce3+ neighbors. It is astonishing that theyget such results because with DFT(PBE)+U the secondneighborcationic sites to an isolated vacancy are preferred,5−7and the distribution of first-neighboring Ce3+ to vacancy pairs ispredicted to be inhomogeneous.7 The consistency betweenboth the DFT(PBE)+U and DFT(HSE06) approaches fordescribing the electronic structure of partially reduced ceria andfor predicting energy differences between different Ce3+distributions is well documented.7,9 Here we used exactly thesame DFT(HSE06)-based methodology10 and computationalcode (VASP)11 as Han et al.1 and reconsidered the cases ofisolated vacancies, VV3 vacancy pairs, and lines along the [111]direction in bulk CeO2. From our point of view, their studysuffers from a number of critical flaws. The main issues are (I)not having investigated in detail many possible configurationsof the Ce3+ ions and (II) the reproducibility and correctness oftheir calculated defect structures. These aspects are vital for theinterpretation of their results, as outlined in the followingsections.