Review Reports
- Juno Bae1,2,
- Yoojin Lim2 and
- Jong Min Park1,2
- et al.
Reviewer 1: Anonymous Reviewer 2: Anonymous Reviewer 3: Anonymous
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsIn this paper, Bae et al. investigate the effects of rapid thermal annealing (RTA) in nitrogen ambience on Al0.8Sc0.2N ferroelectric capacitors. AFM, XRD, and XPS analyses reveal improved crystallinity, reduced oxygen content, and tensile strain. Leakage current is reduced by more than 50%, endurance improves from 1000 to 5000 cycles, and the dielectric constant shows a modest increase. However, several issues should be addressed before the manuscript can be considered for publication. Below are a few points for the authors to consider:
- The fabrication description is clear, but more detail is needed regarding the choice of annealing times (3 min vs. 13 min). Was there a specific rationale for these durations?
- The endurance improvement to 5000 cycles is promising. How does this compare to the state-of-the-art endurance performance of AlScN-based MFM capacitors? Can the authors provide a benchmark table and comment on this?
- Even after annealing, the leakage current density (~70 mA/cm2 at 3 MV/cm) is still relatively high, especially compared to HfO2-based devices. Can the authors comment on this?
- The increase of the dielectric constant is attributed to Pt diffusion into AlScN. However, the evidence is indirect. Have the authors conducted other characterizations like SEM-EDX/ TEM? At minimum, please acknowledge the limitation and clarify the degree of certainty.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe manuscript addresses rapid thermal annealing (RTA) of AlScN-based ferroelectric capacitors, focusing on structural, surface, and electrical properties. While the manuscript on AlScN thin films is relevant. However, the mechanism of the annealed AlScN thin films is unclear and rather confusing, with no proper explanation. The contribution to advancing knowledge is minimal, and the experimental evidence is insufficiently robust to support the strong claims. In the current form, the manuscript should not be accepted and should be considered rejected.
Major Comments
(1) The authors claimed a slight increase in surface roughness upon RTA annealing, in the atomic force microscopy (AFM) images, attributing it to the thermally driven grain growth. However, it could be seen that the grain size was nearly similar with no evident changes. Authors need to further verify their claim with additional measurements like scanning electron microscopy (SEM).
(2) Based on the X-ray diffraction (XRD) analyses, the authors claimed a shift in the (0002) Bragg peaks from lower to higher angles for the as-deposited AlScN film to the 13-minute RTA AlScN film, respectively. They attributed this shift to a reduction in the c-axis lattice constant due to in-plane tensile strain arising from the lattice mismatch between the bottom Pt electrode and AlScN film. The authors need to calculate such tensile stress between the film and the electrode. Moreover, tensile or compressive stress normally arises during the growth process rather than the annealing process [Advanced Science 11, 2408784 (2024)]. Can the authors explain it? Did the authors consider thermal stress between the film layer and the Pt layer as mentioned in Nano Energy 55, 182–192 (2019)?
(3) A slight shift in the Pt (111) diffraction peaks was also noticed, which the authors attributed to the inherent lattice mismatch between the electrode stack and Si substrate. First, the shift in the Pt (111) is relatively larger than the shift noticed in the AlScN films upon annealing. Authors should correct this claim. Secondly, did the authors confirm the same peak position of the Si (004) in the XRD results? For fair comparison, the authors should plot the enlarged XRD pattern where the Si peak should be aligned and then present their arguments accordingly.
(4) In the abstract and the discussion part, XRD analysis reveals enhanced crystalline and improved c-axis orientation due to an increase in peak intensities upon annealing temperature, which mitigates grain-boundary defects and suppresses leakage pathways. However, this argument is superficial. To demonstrate this argument clearly, authors should perform the rocking curves of the corresponding films, extracting the full width at half maximum (FWHM). If the FWHM values decrease with increasing annealing in RTA, then the authors can claim such arguments. Moreover, for better crystalline quality, the surface roughness obtained in AFM should also be lower, as revealed in J. Korean Phys. Soc. 80, 628-633 (2022). Authors should further explain their argument in support of or against the mentioned reference.
(5) In the X-ray photoelectron microscopy (XPS) depth profiles of the AlScN films, the authors claimed a reduction in oxygen signals, as well as a portion of Al appears to react with the underlying platinum to form a Pt2Al3 intermetallic layer, which lowers the absolute Al atomic fraction detected by XPS. This is interesting, but it seems that the thickness of the films was not identical and rather different, where the Pt signals appear to vanish at 225 etch time (s) in the first two cases AlScN films, while the AlScN films annealed for 13 films showed that the Pt signals started at 225 etch time and ended around 300 etch time (s). Authors need to clarify these points by performing the cross-section SEM measurements.
(6) The authors mentioned no change in the polarization electric field (P-E) hysteresis loops, only a reduction in the coercive field of the corresponding P-E loops, and a reduction in the leakage current occurred with the annealing time in RTA. In my opinion, such changes noticed in the XPS results could be linked with a reduction in oxygen signals, leading to crystalline quality of the films and linked electrical properties.
(7) The endurance measurements reported in the manuscript extend only up to 5,000 cycles, which is far below what is required for non-volatile ferroelectric memory applications (typically >106 cycles). Similarly, the retention measurements are limited to a very short timescale (~103 seconds). These values are insufficient to demonstrate practical reliability. The authors need to further extend their tests to more realistic conditions or tone down their claims regarding memory applicability.
(8) There are many typos and grammatical errors that should be carefully revised.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsPoint 1:
How many samples were used in each group in this project? With very few data points how to ensure the performance improvement is real but not due to sample variations?
Point 2:
Is it possible to obtain more data points in Figure 5? The improvements are not obvious in endurance, retention, polarization, polarization loss as shown in (a)-(d) in Figure 5. For example, Figure 5 (a), positive polarization of as-deposited becomes larger than that of RTA-3min, RTA-13min starting 10s, while most of the negative polarization values are in between that of RTA-3 min, RTA-13min. (b) and (c) do not show clear tendency of how RTA time affects polarization and polarization loss. (d) successfully shows the sudden failure of as-deposited and RTA-13min at different cycle numbers. Can the authors discuss more about these figures and explain the observations to support the statements?
Point 3:
Please rewrite the conclusion. From Line 233-239, the conclusion seems draw from comparison of only 3 samples, which are not adequate number of samples. For Figure 4 (c) and (d), it seems like in negative E, the leakage is larger for RTA samples, have the authors look into the cause of this?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors have addressed my previous concerns. The manuscript can be accepted in its current form now.
Author Response
Dear Reviewer,
We would like to express our sincere gratitude for positive review.
Regards,
Reviewer 2 Report
Comments and Suggestions for AuthorsI read the response of the authors. However, it seems that the authors did not answer my raised concerns clearly. I raised total 8 comments along with some suggestions while the authors response incude only 4 comments. Did the authors uploaded wrong file?
Also, the in the revised manuscript, the authors added some new reference. The reference 37 and 38 are wrong. The authors should cross-check these reference and cite properly along with their DOI. The title of one of the reference is "On the surface oxidation and band alignment of ferroelectric Sc0.18Al0.82N/GaN heterostructures" and another is "Protonation-Driven Polarization Retention Failure in Nano-Columnar Lead-Free Ferroelectric Thin Films".
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsPoint 1:
The justification of data representativeness should be put in text, not only in response to the reviewer. The figures should also be included.
Point 2:
Figure 5 does not contain box plots and make data less convincing. It appears to be measurements on 3 samples instead of what the authors explain in the reply. Should explain in details about the improvement in endurance and why no obvious trend in retention and polarization. The 3 min RTA almost has the same endurance behavior as as-deposit one.
Point 3:
The conclusion is not well-written. The conclusion does not mention at all about the almost-no-influence in retention and polarization. The significance of RTA method should be further justified.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 3
Reviewer 2 Report
Comments and Suggestions for AuthorsAccept
Author Response
Dear Reviewer,
We would like to express our sincere gratitude for positive review.
Regards,
Reviewer 3 Report
Comments and Suggestions for AuthorsPoint 1:
What is the definition this study pick for endurance failure? It is a bit confusing by looking at Figure 5 (d). In some of the readings, when Pr/Pr0<0.5 (a criteria that people pick, can be other numbers), it considers fail. However, by looking at Figure 5 (d), it appears that as the cycles increase, Pr/Pr0 is larger and larger, which is counter-intuitive.
Point 2:
Thanks for adding Figure 6, which provide more statistics to this study.
By looking at Figure 6 (e), even the only improvement that the authors claim in this study is questionable… Figure 6 (e) shows the RTA method instead of what theoretically it should “enhanced crystallinity and grain growth” and therefore improving endurance, it actually introduces extra randomness. Instead of what line 261-263 states “The lim-261 ited thermal budget likely resulted in partial grain growth and incomplete defect relaxa-262 tion, leading to only modest endurance enhancement compared to the as-deposited film”, by looking at Figure 6(e) and compare as-deposited & RTA-3min, the mean decreases from 1000 to 700, the 25%-75% range increases rom 0 to 466, which means RTA-3min endurance results are even worse than that of as-deposited samples.
Point 3:
Also Figure 6 (e): How to explain the much worse endurance across the samples as RTA time increases? By looking at this large increase in variation, how to justify RTA is an effective method to improve endurance? Hard to believe this large behavioral fluctuation is acceptable.
Point 4:
Figure 6 s not sufficiently discussed. The only mention of Figure 6 in text is in line 270-272, with no discussion.
Point 5:
The endurance results from Figure 6 (e) and Figure 5 (d) have conflicts: how did the authors get the 5000 as RTA-13min results and put in Table 3?
Point 6:
The leakage current density and cycles of this work seems worst among the listed ferroelectric in Table 1. How to justify the significance of this study?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf