Review Reports - Metal-to-Insulator Transition in Ultrathin Manganite Heterostructures

Round 1

Reviewer 1 Report

This manuscript (MS) reports a review on the thickness-driven metal-to-insulator (dead layer here) transition (TDMIT) in the half metal La2/3Sr1/3MnO3 (LSMO) thin films, which is very helpful to many readers. In detail, the TDMIT concentrates on the formation of the dead layer regarded as the non-metallic phase (or the semiconducting phase or layers), because the dead layer can be fatal on fabrication of a nanolevel spintronics device. The analysis’s of the TDMIT are reasonable. In particular, the distinguished part of the MS is to review the mechanism of the TDMIT. The MIT research is one of the most important hot field in the condensed matter physics. The authors view the MIT researches through the LSMO. I feel this can be very good paper if explanations on fundamental MIT mechanisms are added. I have searched many papers and studied them for authors. My comments are given below.

Comments

I think the authors know the Mott transition, but I would like to mention it one more. In brief, as a representative MIT (metal-to-insulator) theory, the Mott MIT can be considered. It is the indirect transition and is the formation (or breakdown) of the critical Coulomb interaction between carriers from a metal to the Mott insulator, (vice versa). Mott derived the Mott criterion, n_c^1/3a_H » 0.23, from a correlated metal. In materials, n_c» 3 x 10¹⁸ cm^-3 is obtained. Here, the n_c is the density of not impurity or doped carriers but carriers in the correlated metal because the Mott theory started from metal. According to Mott’s logic, when the number of free carriers in metal is reduced to n_c, the MIT occurs. By the way, fortunately, the n_c is closely consistent with impurity concentration inducing the insulator-to-metal transition (IMT).

Moreover, when the number of carriers is reduced, the Coulomb interaction between carriers decreases. This is a problem in Mott’s calculations. Many papers have suggested the MIT is induced by impurity, oxygen deficiency, pressure, chemical doping, lattice strains, etc, as shown in this MS. The MIT occurs even in very high quality single crystals. However, in pure materials, the IMT does not occur. All materials undergoing the MIT or the IMT show the characteristic of the extrinsic semiconductor. Although they are not Mott insulator, all extrinsic semiconductors undergo the IMT or the MIT. We know that all extrinsic semiconductors with both the semiconductor direct gap and the impurity indirect gap undergo the MIT or the IMT. When the indirect gap disappears, the direct gap also disappears. That is, when a small energy amounting to the indirect gap is applied to the materials, the direct large band gap is broken. This is the characteristic of the MIT.

All compound materials have intrinsically a stable electronic structure corresponding to insulator. The metal structure is unstable. We make an unstable metal material through doping. If the doping condition is released, the material necessarily return to the stable insulator (or dead layer).

In the case of LSMO (x=1/3), the material was doped and became an unstable metal by doping. By the way, when the thickness decreases, doping is not enough to be metal. Then the metal is converted to the dead layer. This is nature. We can make a very thin metal layer by reducing lattice match and using high-doped target. However, we cannot perfectly remove the dead layer, because the metal exists on the CDW unstable point (CDW instability) (https://journals.aps.org/prb/abstract/10.1103/PhysRevB.54.90) or on the diverging point (https://aip.scitation.org/doi/abs/10.1063/1.4926860?journalCode=jap).

I think quality of the manuscript will be enhanced and be very good paper if authors digest and reflect the above comments.

Extra error, in page 10, in line 23, a verb is not shown. Please check it.

Dec. 2, 2018

Author Response

Reply to the Reviewer 1

This manuscript (MS) reports a review on the thickness-driven metal-to-insulator (dead layer here) transition (TDMIT) in the half metal La_2/3Sr_1/3MnO₃ (LSMO) thin films, which is very helpful to many readers. In detail, the TDMIT concentrates on the formation of the dead layer regarded as the non-metallic phase (or the semiconducting phase or layers), because the dead layer can be fatal on fabrication of a nanolevel spintronics device. The analysis’s of the TDMIT are reasonable. In particular, the distinguished part of the MS is to review the mechanism of the TDMIT. The MIT research is one of the most important hot field in the condensed matter physics. The authors view the MIT researches through the LSMO. I feel this can be very good paper if explanations on fundamental MIT mechanisms are added. I have searched many papers and studied them for authors. My comments are given below.

We greatly appreciate the reviewer for positive comments and the suggestion to improve further the manuscript.

I think the authors know the Mott transition, but I would like to mention it one more. In brief, as a representative MIT (metal-to-insulator) theory, the Mott MIT can be considered. It is the indirect transition and is the formation (or breakdown) of the critical Coulomb interaction between carriers from a metal to the Mott insulator, (vice versa). Mott derived the Mott criterion, n_ca_H>>0.23, from a correlated metal. In materials n_c>>3×10¹⁸ cm^-3 is obtained. Here, the n_c is the density of not impurity or doped carriers but carriers in the correlated metal because the Mott theory started from metal. According to Mott’s logic, when the number of free carriers in metal is reduced to n_c, the MIT occurs. By the way, fortunately, the n_c is closely consistent with impurity concentration inducing the insulator-to-metal transition (IMT).

In the case of LSMO (x=1/3), the material was doped and became an unstable metal by doping. By the way, when the thickness decreases, doping is not enough to be metal. Then the metal is converted to the dead layer. This is nature. We can make a very thin metal layer by reducing lattice match and using high-doped target. However, we cannot perfectly remove the dead layer, because the metal exists on the CDW unstable point (CDW instability).(https://journals.aps.org/prb/abstract/10.1103/PhysRevB.54.90) or on the diverging point (https://aip.scitation.org/doi/abs/10.1063/1.4926860?journalCode=jap). I think quality of the manuscript will be enhanced and be very good paper if authors digest and reflect the above comments.

Extra error, in page 10, in line 23, a verb is not shown. Please check it.

The MIT in LSMO is not likely Mott MIT. The metallicity in LSMO is related to delocalization of electron due to double exchange mechanism modified by doping. The evidence from lattice distortion, chemical compositions change can all cause the MIT, rather than electron-electron correlation. The Mott criteria can be applied to Mott MIT system such as VO2, but we don’t think it is the scenario for LSMO.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper reports review-like on developments in perovskite thin films. It is dealing with layers in the perovskite films, in which the bulk magnetic and electric properties are destroyed (dead layers). The underlying mechanism and the thickness of such layers are in the focus of the paper. Chapter 4 is nicely summarizing the state-of-the-art knowledge about the mechanism.

In general the paper is very well written and is very informative. Nevertheless, some further improvement is required. Thus, the authors should consider the following points in a revised version of the paper:

1. Sketch in Fig. 2b: Why is the color of the outer three layers different from that of the lower two layers? Shall the green part represent the STO layer, on which the LSMO is grown? This can be assumed from the text, but it should be more clearly said.

2. Fig. 3c: Why is the growth of a LSMO layer slower at higher oxygen pressure? in a review the growth mechanism should be explained shortly.

3. Fig. 6: Is it possible that the surface dead layer results from stoichiometry differences of the LMSO at the surface (besides of structural distortion)? Possibly, just the oxygen content might be higher. I suggest to carry out X-ray photoelectron spectroscopy (XPS) analysis to learn more about the chemical surface composition. In this respect, what do the authors mean with “surface off-stoichiometry” (line 192)? How is this term justified by data?

4. Chapter 4.5: It would be good to specify the “extrinsic ingredients” more. Also the discussion of site ration for A and B cations should be supported by data.

5. Chapter 4 compares the different suggestions for mechanism. However, in a review paper I expect to see also some summarizing comments, highlighting the opinion of the authors about the different mechanisms. An overall picture about the different contributions to the mechanism must be given. Such a summary is rudimentary given in lines 456 ff, but it should be extended in a summarizing (sub-)chapter.

6. Chapter 6: The two sentences in lines 476-479 have the same content; I guess one of them is enough.

7. Chapter 6, lines 483 ff: Some more distinct practical examples for (daily) applications of the MIT would be good, to attract reading also for non-physicists. Maybe they should already be given in the Introduction.

8. The whole paper requires a careful language check, in best case by a native speaker. There is a number of small mistakes (e.g. lines 27, 29, 68, 147, 148, 157, 397, 410, and many more) which must be avoided. Also the layout of the text should be checked in order to avoid inappropriate line-breaks like e.g. in lines 38 or 43.

Author Response

It is a general view of a specific ABO₃ perovskite film on an oxide perovskite substrate. So the bottom green layer is A'B'O₃ perovskite film and top layer is a specific film. This has been specified in revised version.

2. Fig. 3c: Why is the growth of a LSMO layer slower at higher oxygen pressure? in a review the growth mechanism should be explained shortly.

In higher background gas, the plume will be more scattering by the background gas. Therefore, the amount of the forwarding species which can arrive on substrate decreases, resulting lower growth rate. In order to address reviewer’s comment, new sentences have been added to revised version in Page 3

The observed slower growth rate in higher oxygen pressure is due to the side scattering of the plume by background gas which reduces the amount of arriving species on substrates.

3. Fig. 6: Is it possible that the surface dead layer results from stoichiometry differences of the LSMO at the surface (besides of structural distortion)? Possibly, just the oxygen content might be higher. I suggest to carry out X-ray photoelectron spectroscopy (XPS) analysis to learn more about the chemical surface composition. In this respect, what do the authors mean with “surface off-stoichiometry” (line 192)? How is this term justified by data?

The surface off-stoichiometry can be the variation of oxygen content or La/Sr from bulk value. In Fig. 6, the (La,Sr) content becomes smaller near surface for LSMO films without capping while the La/Sr is more uniform through the whole films from interface to top surface in STO capped LSMO.

4. Chapter 4.5: It would be good to specify the “extrinsic ingredients” more. Also the discussion of site ration for A and B cations should be supported by data.

Extrinsic ingredients include oxygen vacancies, structure defects, off-stoichiometry such as loss to A or B site due to the unoptimal growth condition. This has been specified in revised version. The data to supported A and B ratio is shown in new Fig. 14e-h in revised version.

In revised manuscript, we added a sub-chapter to summarize the previous research on mechanism of the MIT.

6. Chapter 6: The two sentences in lines 476-479 have the same content; I guess one of them is enough.

To fully understand the origin of dead layer still requires more effort. The key to understand properties of ultrathin films will be the advanced techniques which can probe the materials at atomic scale. Recently developed techniques which can probe the materials at atomic scale may eventually lead to the solution. These advanced techniques include STEM and PNR, RXR, which are powerful techniques to provide very local lattice, spin and electronic structure

Is changed to

The key to understand properties of ultrathin films will be the advanced techniques which can probe the materials at atomic scale. Recently developed techniques such as STEM and PNR, RXR, which can provide very local lattice, spin, orbital configuration and electronic structure may eventually reveals of the origin of thickness driven metal to insulator transition and the dead layer.

The understanding and control of dead layer are also very attractive for technological application

Is changed to

The understanding and control of dead layer are also very attractive for technological applications, such as transistor, memory device, and sensor.

We thoroughly went through the manuscript and polished the English.

Reviewer 3 Report

The review article is very informative and, in general, well written and well structured. The English language usage has a recurring omission of the article "the", which needs correction. Other language suggestions are given in the annotated copy. The acronym LSMO is used first for the specific La2/3Sr1/3MnO3 and later for a generic La1-xSrxMnO3. I recommend to have different acronyms to avoid ambiguity, perhaps LSMO-1/3 can be used to refer to the specific case x=1/3. It would be good to specify at what thickness the films are not to be considered "ultrathin" anymore. This could impact what references may have been omitted from the review. Depending on this definition, a few additional references may need to be included for completeness. A few of these are mentioned below.

Carrier tuning the metal-insulator transition of epitaxial La0.67Sr0.33MnO3 thin film

on Nb doped SrTiO3 substrate

J. M. Zhan, P. G. Li, H. Liu, S. L. Tao, H. Ma, J. Q. Shen, M. J. Pan, Z. J. Zhang, S. L. Wang, and G. L. Yuan

LSMO thin films with high metal–insulator transition temperature onbuffered SOI substrates for uncooled microbolometersˇS. Chromika,∗, V.ˇStrbíka, E. Dobroˇckaa, T. Rochb, A. Rosováa, M.ˇSpankováa, T. Lalinsk´ya,G. Vankoa, P. Lobotkaa, M. Ralbovsk´yc, P. Choleva

CHARACTERIZATION OF ELECTRICAL TRANSPORT

IN LSMO WITH ENHANCED TEMPERATURE

OF METAL–INSULATOR TRANSITION

Vladim´ır ˇStrb´ık — ˇStefan Chromik

Overall, after minor corrections, the article is worthy of publication.

Comments for author File: Comments.pdf

Author Response

The review article is very informative and, in general, well written and well structured. The English language usage has a recurring omission of the article "the", which needs correction. Other language suggestions are given in the annotated copy. The acronym LSMO is used first for the specific La_2/3Sr_1/3MnO₃ and later for a generic La_1-xSr_xMnO₃. I recommend to have different acronyms to avoid ambiguity, perhaps LSMO-1/3 can be used to refer to the specific case x=1/3. It would be good to specify at what thickness the films are not to be considered "ultrathin" anymore. This could impact what references may have been omitted from the review. Depending on this definition, a few additional references may need to be included for completeness. A few of these are mentioned below.

Carrier tuning the metal-insulator transition of epitaxial La_0.67Sr_0.33MnO₃ thin film on Nb doped SrTiO₃ substrate，J. M. Zhan, P. G. Li, H. Liu, S. L. Tao, H. Ma, J. Q. Shen, M. J. Pan, Z. J. Zhang, S. L. Wang, and G. L. Yuan

LSMO thin films with high metal–insulator transition temperature onbuffered SOI substrates for uncooled microbolometers, Š. Chromik, E. Dobročka, T. Roch, A.Rosová, M. Španková, T. Lalinský, G. Vanko, P. Lobotka, M. Ralbovský, P. Choleva

CHARACTERIZATION OF ELECTRICAL TRANSPORT IN LSMO WITH ENHANCED TEMPERATURE OF METAL–INSULATOR TRANSITION, Vladimír Štrbík and Štefan Chromik

Overall, after minor corrections, the article is worthy of publication

In revised version, we will avoid ambiguity by defining only La_2/3Sr_1/3MnO₃ as LSMO and for other component, we will use La_1-xSr_xMnO₃. Some research groups also use La_0.7Sr_0.3MnO₃ formula instead, we still the same acronym LSMO since it actually refers to same compound.

Regarding to the terminology “ultrathin”, so far there is not consensual exact boundary between “ultrathin film” and normal thin film, but it should be in a region where dimensionality, interface or surface plays a strong role and thus the thickness is normally just few nanometers. We will give a brief discussion about it in revised manuscript by reforming the sentences in line 31

Of particularly crucial is when the thickness of the layer is down to several unit cell (u.c.) or said few nanometers

Is changed to

Particularly crucial is when the layers enter into ultrathin region where the thickness only several unit cell (u.c.) or said few nanometers.

We greatly appreciate the reviewer for providing references. The first reference which discusses the MIT in LSMO controlled by depleted layer at LSMO/Nb-STO junction provides deep insight into the charge effect on MIT and should be useful for understanding the MIT in LSMO. While the other two are related to temperature driven MIT in very thick LSMO, maybe not quite relevant. In revised manuscript, we will cite the first reference as Ref. 75