Comparative Analysis of DFT+U, ACBN0, and Hybrid Functionals on the Spin Density of YTiO3 and SrRuO3
Round 1
Reviewer 1 Report
This paper straight forward analysis of the theoretical spin density map of two ferromagnetic perovskites. The paper is clearly written and easy to read but to me it remain unclear if there is any "new science" included.
Minor suggestions:
Figures 1 & 4 are hard to read because the text is too small.
The main point of the paper could be presented more clearly.
Author Response
We thank the referee for the suggestion. We are not ambitious enough to
state that "DFT is straying away from the path to the exact functional"
(Science 355, 49 (2017)) where the authors compared quantitatively a large
set of experimental charge densities to DFT (LDA, GGA, mGGA and higher
rungs in the Jacob's ladder) and showed that better total energies are
not linked to better charge densities.
There exists other published works on the similar line:
* 10.1021/acs.jctc.0c00440: how to compare quantitatively charge density in solids
* 10.1021/acs.jctc.7b00550: compare DFT charge density to CCSD(T) of molecules
* 10.1063/1.4792436: compare DFT charge density of solids to electron diffraction data
Therefore we believe that comparing DFT charge/spin density to the
experimental one is a strong enough motivation. We hope that in the
future other groups will embark on measuring more spin densities of
molecules and solids. Clearly there must be a quantitative way to compare
densities and QTAIM is an unbiased one that can be applied both to theory
and experiments.
We added this sentences and references to the introduction.
We've placed figures 1 and 4 on a full landscape page.
Reviewer 2 Report
The authors have performed the theoretical study of the electronic structure and the spin density of YTiO3 and SrRuO3 with the density functional theory. They employed several exchange-correlation functionals for the systematic study. The simulations have been reasonably performed, and the provided data will be useful for the community. Hence, I can recommend the publication of this manuscript after addressing the following points.
1) Large U for oxygen
The authors mentioned, “Interestingly, the ACBN0 method yielded U(Ti-3d)=0.26 eV and a large value on oxygen, U(O-2p)=8.31”. Here, why does the U for oxygen become so large?
2) Comparison with HSE
Around the line of 91, the authors mention, “… the manifold of occupied oxygen bands was found 6 eV below the bottom of the conduction band as in the ACBN0 case”. However, the reported oxygen bands in Ref. [30] is more similar to the DFT+U result in the present manuscript rather than the ACBN0. Here, ACBN0 seems to provide too deeply bound oxygen bands. This point has to be clarified.
3) Comparison with previous theoretical results
The authors compared their spin density with the previous experiments in Ref. [17,19]. However, in those references, the DFT results are also provided. Hence, the authors should also mention the previous theoretical results and address the consistency/difference/novelty of the present work.
4) Typo
At the end of the line of 90, the author wrote, “It the same paper...”. However, this should seem to be “In the same paper”.
Author Response
We thank the referee for the positive comments. Here is our reply:
1) Large U for oxygen
The large U for oxygen is found by ACBN0 in a number of perovskites.
This is reported for instance in PRB 101, 165117 (Fig.1). In that paper
the author state that "according to cRPA calcualtions and spectroscopy
results" the magnitude of U(2p) is expected to be of the same order
of U(3d/4d).
Based on our past experience on BaBiO3 (unpublished) we noted that when
the states at the Fermi level have a large O(2p) character, applying U
on oxygen is more effective than applying it on the transition metal.
The results is a slightly more ionic metal-oxygen bond which allows
charge disproportionation in BaBiO3 (i.e. Ba2 Bi(III) Bi(V) O6).
As a corollary, because ACBN0 acts on a large set of atomic orbitals,
the Hubbard value of the transition metal is diminished. By changing
the set of active orbitals (i.e. including O(2s) and metal(4s/5s))
we expect changes in the U values. Indeed by removing oxygen from the
active orbitals, the U value on the metal is expected to be large.
We added this paragraph to the discussion.
2) Comparison with HSE
We agree with the referee observation. Overall the HSE DOS of Ref.[30] is
more similar to the case U=5 eV, except that the the oxygen states are
-5 eV below the left shoulder of the "Upper Hubbard Band". In ACBN0 the
distance between the oxygen states and the UHB is 6 eV, similarly to HSE.
Unfortunate ACBN0 is not able to split the LHB from the UHB.
We added a comment in the manuscript.
3) Comparison with previous theoretical results
Refs.[17,19] used QTAIM to integrate the spin density in the Bader atomic
basins. They used only the PBE0 screened functional and localized basis code
(CRYSTAL). We stated it only in the caption of Fig.1 and we added a full
sentence in the text.
Regarding SrRuO3, the DFT calculations reported in Ref.[19] where done with
the PBE functional and the spin density was integrated inside atomic spheres.
4) Typo
Fixed!
