Electronic Structure of Oxygen-Deficient SrTiO3 and Sr2TiO4
Round 1
Reviewer 1 Report
The authors applied DFT to characterize different single and extended oxygen deficiency defect states in SrTiO3 (STO) and Sr2TiO4. It was established rather shallow and narrow in-gap defect state for single oxygen vacancy and a small (~0.6 eV) shift of core (p, s) levels. In contrast, the row oxygen vacancies along the (010) axis displays a broad defect band in the bandgap and a large energy shift of core states ~2.4 eV. They discussed the importance of the results obtained for conductive properties of STO as well as for the “visibility” of defect states for local probe techniques (STM), basing on the anisotropic nature of defect states. Indeed, the oxygen defects states of STO and their influence on the conductivity is extremely important for studies of 2D electron gas at LAO/STO interfaces as well as for growth of oxide films as the STO is a prominent substrate. In this sense the manuscript provides a valuable information on the development of such states.
However, some questions should be answered before the manuscript could be published in Crystals: 1) I recommend to improve English as some sentences look rather long and some words are missing (e.g. the second sentence in the abstract); 2) In Fig. 2a) one can see the reduced bandgap for STO Eg~2.3 eV. Why? All other band structure calculations result in a normal bulk-like bandgap ~3 eV. The effect of vacancies on the bandgap of STO is known, but I guess this should be not the case for a single oxygen vacancy; 3) The broadening of the defect level structure in (010) and (110) rows of defects in comparison with the single vacancy is clearly seen. The authors should comment on that; and 4) How the discussed defect states impact the evolution of the conductivity and eventually the insulator/metal transition when their concentration increases? Which defects are playing a dominate role: the single and homogeneously distributed effects or the extended?
Summarizing, I recommend the revision of the manuscript along the above listed points. After revision it could be suitable for publication.
Author Response
The authors applied DFT to characterize different single and extended oxygen deficiency defect states in SrTiO3 (STO) and Sr2TiO4. It was established rather shallow and narrow in-gap defect state for single oxygen vacancy and a small (~0.6 eV) shift of core (p, s) levels. In contrast, the row oxygen vacancies along the (010) axis displays a broad defect band in the bandgap and a large energy shift of core states ~2.4 eV. They discussed the importance of the results obtained for conductive properties of STO as well as for the “visibility” of defect states for local probe techniques (STM), basing on the anisotropic nature of defect states. Indeed, the oxygen defects states of STO and their influence on the conductivity is extremely important for studies of 2D electron gas at LAO/STO interfaces as well as for growth of oxide films as the STO is a prominent substrate. In this sense the manuscript provides a valuable information on the development of such states.
We are happy to read that the referee shares our opinion about the importance of the subject and are grateful for the positive evaluation.
However, some questions should be answered before the manuscript could be published in Crystals:
1) I recommend to improve English as some sentences look rather long and some words are missing (e.g. the second sentence in the abstract);
Long sentences have been shortened and typos were corrected (indicated in blue in the revised manuscript).
2) In Fig. 2a) one can see the reduced bandgap for STO Eg~2.3 eV. Why? All other band structure calculations result in a normal bulk-like bandgap ~3 eV. The effect of vacancies on the bandgap of STO is known, but I guess this should be not the case for a single oxygen vacancy;
This reduction of the band-gap is a unique feature of the TiO2 terminates (001) surface. Additional states arising from the top TiO2 layer at the top of the valence band are responsible for this band-gap narrowing (Ref. [35] of the new manuscript). This is now mentioned around line 109 in the text.
3) The broadening of the defect level structure in (010) and (110) rows of defects in comparison with the single vacancy is clearly seen. The authors should comment on that; and
The observed broadening correlates with the spatial overlap of the defect wave-functions as can be expected from a simple LCAO-MO model. We comment on this observation in lines 148-150 of the manuscript.
4) How the discussed defect states impact the evolution of the conductivity and eventually the insulator/metal transition when their concentration increases? Which defects are playing a dominate role: the single and homogeneously distributed effects or the extended?
Indeed an interesting point: assuming a random distribution of isolated defects and taking the Mott criterion as guideline about 4x10^18/ccm free charge carriers are necessary to initiate a I/M transition. Assuming that they are thermally activated from a shallow donor level, a rather high defect concentration is necessary (about 10^20/ccm as observed in other perovskites, new Ref. [43]). The fact that in reduced SrTiO3 the I/M transition is observed at much lower defect concentration hints at a locally enhanced defect density. Whether the simulated defect rows are a good model for the actual situation is hard to say from the ab initio side, but they feature some signatures that are sometimes observed in conductive SrTiO3 (like the strong core-level shifts). We added a short discussion about this aspects at the end of the discussion.
Summarizing, I recommend the revision of the manuscript along the above listed points. After revision it could be suitable for publication.
We hope that the revised version fulfills now the referees criteria.
Reviewer 2 Report
Al-Zubi et. al present an interesting theoretical investigation of the electronic structure signature of oxygen vacancies near the surface of SrTiO3 bulk material, with the aim that future spectroscopic experiments may identify such defects. The paper it well written and suitable for publication. I have a few suggestions only, mentioned in the following.
Section 2.1 basically describes previous studies on DFT calculations performed on bulk SrTiO3, however is listed under the Results section. I would suggest to reorganize this part such as it is part of the introduction. Line 69. “… with a formal charge +1…” Which scheme was used to analyze the partial charges? Mulliken, Lowdin, Bader? Line 79. “…our calculation gives…” - Line 119. Typo: “oeiented” Line 176. Typo: delete extra “the”. Authors comment the problems of using GGA functionals. Good description of lattice parameters but poor description of electronic properties. What if instead of GGA another type of functional is used, such as meta-GGA of hybrid methods? Perhaps a comment can be included in the methods section. This is slightly described in the discussion of results, but I think it can be made more specific. What is a typical sensitive of the spectroscopic methods proposed in the paper to detect such defects? Can be numbers included? For some of the arrangements of defects that were calculated, changes in the electronic structure were small. Can be better describe which one are suitable for experimental detection?
Author Response
Al-Zubi et. al present an interesting theoretical investigation of the electronic structure signature of oxygen vacancies near the surface of SrTiO3 bulk material, with the aim that future spectroscopic experiments may identify such defects. The paper it well written and suitable for publication. I have a few suggestions only, mentioned in the following.
We thank the referee for his positive comments and address the suggestions in the following:
Section 2.1 basically describes previous studies on DFT calculations performed on bulk SrTiO3, however is listed under the Results section. I would suggest to reorganize this part such as it is part of the introduction.
We agree with the referee and included the information in the introduction, adding also a new reference [26] containing a further study from the literature that illustrates the diversity of the ab initio results.
Line 69. “… with a formal charge +1…” Which scheme was used to analyze the partial charges? Mulliken, Lowdin, Bader?
To derive the formal charge, the two electrons left behind at the vacancy were distributed on the nearest neighbor Ti atoms. The actual increase of local charge was calculated by integrating the charge density in a 2A sphere around the Ti atoms and comparing to Ti sites far from the vacancy. This is now mentioned in the text (line 78).
Line 79. “…our calculation gives…” -
Line 119. Typo: “oeiented”
Line 176. Typo: delete extra “the”.
The typos were corrected.
Authors comment the problems of using GGA functionals. Good description of lattice parameters but poor description of electronic properties. What if instead of GGA another type of functional is used, such as meta-GGA of hybrid methods? Perhaps a comment can be included in the methods section. This is slightly described in the discussion of results, but I think it can be made more specific.
Indeed, as shown in Ref. [27], hybrid functionals like HSE can also give a good account of the structure and the band gap. Computationally, they are more costly than DFT+U methods which affects in particular the calculation of large supercells. For a neutral isolated oxygen vacancy, the results are rather similar to ours. Another hybrid method based on B3PW was applied in Ref. [21], again giving a good lattice constant with a 0.3 eV overestimation of the band gap. Consequently, the neutral defect state is more split from the conduction band. We included this information in the methods section (after line 201).
What is a typical sensitive of the spectroscopic methods proposed in the paper to detect such defects? Can be numbers included?
Unfortunately, the detection limits depend on different parameters, peak width and separation, background, etc. Typically in XPS detection limits for different atoms are about 0.1% [Shard, Surf. Interf. Anal. 46, 175 (2014)]. For Ti+4/Ti+3 a limit of 0.5% was proposed recently in a study of TiO2 [14]. We discuss this shortly in the introduction, lines 37-40.
For some of the arrangements of defects that were calculated, changes in the electronic structure were small. Can be better describe which one are suitable for experimental detection?
Being not experimentalists, we just can give some general arguments. The larger the ratio between peak separation and peak width (typically 1.5 eV FWHM for Ti 2p states), the easier the detection will get. Therefore small core level shifts of 0.5 eV will be only visible if defect concentrations get in the order of a few percent. In addition, the surface sensitivity of most spectroscopic methods and the possibility of defect accumulation at the surface has to be taken into account when analysing XPS data. We added some corresponding comments at the end of section 2.1.2.
