Control of Organic Superconducting Field-Effect Transistor by Cooling Rate
Round 1
Reviewer 1 Report
In the manuscript submitted to Crystals Kawaguchi and Yamamoto suggest that they observed evidence of excitonic superconductivity in some organic compound. The experimental results, however, do not support this conclusions; hence the authors should soften their claims even more.
kappa-(BEDT-TTF)2Cu[N(CN)2]Br and kappa-(BEDT-TTF)2Cu[N(CN)2]Cl, labelled k-Br and k-Cl, are known to be extremely sensitive to smallest amount of pressure as far as their low-temperature electrical and in particular superconducting properties are concerned. About ten years ago H.M. Yamamoto and collaborators reported that thin flakes of k-Br exhibit either insulating or metallic properties depending on the substrate they are attached to, because the different thermal contraction leads to different strain. In such a case they were able to observe indications of superconducting regions by applying some electric voltage in a field effect arrangement. In the present manuscript again very thin flakes of k-Br were used, but this time the authors do not change the substrate in order to apply strain, but change the cooling rate. The materials under study are very sensitive to disorder, introduced by irradiation, but also when crossing a glassy transition around 100 K with different cooling rates.
The authors report a dependence of the low-temperature properties upon different cooling procedures, with indications of a drop of resistivity at slow cooling right at the temperature where superconductivity is observed in bulk crystals. However, no complete loss of resistance is observed nor a check of the superconducting fraction by susceptibility. The author speculate that the change in transport properties by different cooling rates is linked to the change of electronic properties by pressure. No supporting facts or arguments are presented in order to prove this crucial statement.
Similar to previous experiments, the authors then apply an electric field in an FET configuration and observe the change of transport due to the change in carrier density. Although they change the resistivity value, the superconducting state is never completely reached. Nevertheless, the authors studied the ON/OFF ratio for samples with different degrees of disorder and found that for the pure and low-disordered system the transport is p-type, while it is n-type for strong disorder.
The authors claim that superconductivity is enhanced for exactly half filling, however, this is not supported by data, but a result of speculation. The residual resistivity is in the Ohm to kOhm range.
In a final step the authors speculate that uncommon properties occur when the Fermi level is between the upper and lower Hubbard bands. For the rest of the manuscript the authors are carried away by speculations beyond any experimental ground.
In conclusion:
The authors suggest that they for the first time put into reality superconductivity mediated by electronic excitations
as predicted half a century ago. This is pure speculation and exaggeration. Hence the manuscript cannot be published in the present form.
Author Response
We appreciate the reviewers’ valuable comments. The followings are point-by-point replies associated with original comments shown in red.
Reviewer 1
In the manuscript submitted to Crystals Kawaguchi and Yamamoto suggest that they observed evidence of excitonic superconductivity in some organic compound. The experimental results, however, do not support this conclusions; hence the authors should soften their claims even more.
Reply:
We admit that our experiments are not sufficient enough to support the conclusions in the present stage. In the revised manuscript, we have removed the discussion about excitonic mechanism.
The authors report a dependence of the low-temperature properties upon different cooling procedures, with indications of a drop of resistivity at slow cooling right at the temperature where superconductivity is observed in bulk crystals. However, no complete loss of resistance is observed nor a check of the superconducting fraction by susceptibility. The author speculate that the change in transport properties by different cooling rates is linked to the change of electronic properties by pressure. No supporting facts or arguments are presented in order to prove this crucial statement.
Reply:
We do not think complete loss of resistance is always necessary to claim superconductivity, especially when the situation is reasonable to expect superconducting state [e.g. 3,19,26–28]. In the present case, it is reasonable to assume that the remaining resistance is generated by vortex motion and percolation superconductivity. We have also checked the suppression of superconductivity in magnetic field as shown in Figure S5. Note that we have already checked magnetic susceptibility in similar FETs with kappa-Br. In addition, the cooling rate dependence of superconducting fraction is confirmed in bulk crystal in literatures.
Similar to previous experiments, the authors then apply an electric field in an FET configuration and observe the change of transport due to the change in carrier density. Although they change the resistivity value, the superconducting state is never completely reached. Nevertheless, the authors studied the ON/OFF ratio for samples with different degrees of disorder and found that for the pure and low-disordered system the transport is p-type, while it is n-type for strong disorder.
The authors claim that superconductivity is enhanced for exactly half filling, however, this is not supported by data, but a result of speculation. The residual resistivity is in the Ohm to kOhm range.
Reply:
We compared the situation with those for k-type BEDT-TTF Mott-insulator in a disordered state induced by X-ray irradiation. In case of highly disordered state, on the contrary to the fast cooled state, the resistance of the system should be decreased because of Mott-gap softening. This is totally opposite behavior in our FET device and hence, we believe it is reasonable to assume that the major origin of resistance increase by fast cooling is the negative pressure effect. A discussion on this point has been added in the revised manuscript.
The exact half –filling state is apparent from the R-Vg plot shown in Fig. 2a. The resistance plateau observed at negative Vg range clearly shows that the Fermi level is in the gapped region.
In a final step the authors speculate that uncommon properties occur when the Fermi level is between the upper and lower Hubbard bands. For the rest of the manuscript the authors are carried away by speculations beyond any experimental ground.
In conclusion:
The authors suggest that they for the first time put into reality superconductivity mediated by electronic excitations as predicted half a century ago. This is pure speculation and exaggeration. Hence the manuscript cannot be published in the present form.
Reply:
We have revised the title and discussions by removing our previous claim on the excitonic mechanism.
Reviewer 2 Report
This is a reasonably well written report on an innovative setup involving charge transfer molecular conductors. The setup exploits the differential expansion rate of the substrate and device material to produce an effective tensile stress on the device, that can be adjusted according to the cooling rate of the structure.
The experimental data show clear cooling-rate dependence of the resistivity and drastically different gate voltage variations, depending on the regime of slow and fast cooling rates.
The authors motivate their results in terms of layer differential expansion, that effectively maps the entire device into layers with effectively different U/W values due to the strain, essentially producing a phase separated system with different behavior. Although such explanation is tantalizing, I am not sure it is the only possibility.
For example, one can think that the fast cooling rate produces inhomogeneous freezeout of the ethylene groups throughout the sample, in such a way that regions with different charge density are produced. In addition, fast cooling may produce large crystal defects (dislocation, twinning, and even larger “cracks”) that disconnect some of the metallic regions, creating disorder and electronic traps. These effects may produce significant resistivity at low temperature, especially if metallic regions are below the percolation threshold. This scenario does not require phase-separated layers, and yet could qualitatively explain the behavior. In other words, the mechanism invoked may not be needed to understand the data. Perhaps the authors have done structural characterization of the device, to explore what are the physical changes induced by their clever method and could include such data. A related question is whether they have explored the thickness dependence of the results, as thinner samples would likely be more homogeneous. Can they identify the systematics of the thickness dependence?
I would recommend that the authors include a discussion on possible alternative mechanisms for the observed behavior, and then discuss why these cannot explain the data in its entirety. At this point, I would moreover request that the authors tone down their claims implicit in the title and abstract or provide additional evidence for the validity of their scenario and main conclusions.
Author Response
Response to reviewers’ comments:
We appreciate the reviewers’ valuable comments. The followings are point-by-point replies associated with original comments shown in red.
Reviewer 2
This is a reasonably well written report on an innovative setup involving charge transfer molecular conductors. The setup exploits the differential expansion rate of the substrate and device material to produce an effective tensile stress on the device, that can be adjusted according to the cooling rate of the structure.
Reply:
Thank you again for your valuable comments.
The experimental data show clear cooling-rate dependence of the resistivity and drastically different gate voltage variations, depending on the regime of slow and fast cooling rates.
The authors motivate their results in terms of layer differential expansion, that effectively maps the entire device into layers with effectively different U/W values due to the strain, essentially producing a phase separated system with different behavior. Although such explanation is tantalizing, I am not sure it is the only possibility.
For example, one can think that the fast cooling rate produces inhomogeneous freezeout of the ethylene groups throughout the sample, in such a way that regions with different charge density are produced. In addition, fast cooling may produce large crystal defects (dislocation, twinning, and even larger “cracks”) that disconnect some of the metallic regions, creating disorder and electronic traps. These effects may produce significant resistivity at low temperature, especially if metallic regions are below the percolation threshold. This scenario does not require phase-separated layers, and yet could qualitatively explain the behavior. In other words, the mechanism invoked may not be needed to understand the data. Perhaps the authors have done structural characterization of the device, to explore what are the physical changes induced by their clever method and could include such data. A related question is whether they have explored the thickness dependence of the results, as thinner samples would likely be more homogeneous. Can they identify the systematics of the thickness dependence?
Reply:
Thank you very much for kind suggestions. Other scenarios which do not require phase-separated layers such as enhanced Umklapp scattering at half-filling may be discussed in addition to the suggested story. However, as pointed out by both of the reviewers, those stories are still speculative as far as the experimental data are limited in the present form. Therefore, we refrain from proposing and discussing those possibilities in this article and keep the chance in future occasion with further supporting data.
I would recommend that the authors include a discussion on possible alternative mechanisms for the observed behavior, and then discuss why these cannot explain the data in its entirety. At this point, I would moreover request that the authors tone down their claims implicit in the title and abstract or provide additional evidence for the validity of their scenario and main conclusions.
Reply:
As described above, we have made the discussion simple in the new version. We hope our revised manuscript can meet your criteria for publication.
Round 2
Reviewer 1 Report
The authors followed the advice to soften the claims and eliminate unsupported speculations. It is an interesting paper that can be published in the present version.
Reviewer 2 Report
I believe the authors have taken to heart the comments from the referees and adjusted their description accordingly. I am glad to recommend this version of the manuscript for publication.
