Thermal Instability of Gold Thin Films

Round 1

Reviewer 1 Report

This article studies the mechanism of formation of gold nanoislands on the surface of silicon. The study was carried out both for pure silicon and the formation of nanoparticles on the surface of oxidized silicon. The work makes an important contribution to the understanding of the mechanism of formation of nano-objects that can be used in many modern technologies.

Despite the overall high level of work, some corrections and clarifications need to be made.

 

 

1. The authors used Low energy electron microscopy and Low energy electron diffraction in their work. However, the abbreviated names of these methods are given only in the description of research methods, while the abbreviations themselves are used earlier. It is necessary to eliminate this inconsistency.
2. The authors indicate the possibility of reducing the eutectic temperature in the gold-silicon system when approaching the nanolevel. It would be good to cite literary confirmations of this phenomenon specifically for the gold-silicon system, and not for other systems.
3. The paper claims that heating gold films on silicon leads to the formation of grains, which later form nanoislands. Does the grain size depend on the heating rate?
4. The atomic force microscopy method makes it possible to obtain the size distribution of surface inhomogeneities. In my opinion, it is necessary to add to the work the distribution of gold particles by size, which was probably obtained by the authors. This will make it possible to carry out a quantitative analysis of the conditions for the formation of gold nanoparticles on the surface of silicon.

Author Response

Thank you for reading the article carefully. Below are our answers to your questions. Major changes in the text are marked by red color.

1. The authors used Low energy electron microscopy and Low energy electron diffraction in their work. However, the abbreviated names of these methods are given only in the description of research methods, while the abbreviations themselves are used earlier. It is necessary to eliminate this inconsistency.

Of course you are right. We changed manuscript and we explained the abbervations

2. The authors indicate the possibility of reducing the eutectic temperature in the gold-silicon system when approaching the nanolevel. It would be good to cite literary confirmations of this phenomenon specifically for the gold-silicon system, and not for other systems.

We added citation (no 28).

3. The paper claims that heating gold films on silicon leads to the formation of grains, which later form nanoislands. Does the grain size depend on the heating rate?

According to our best hnowledge and our experiments, size and distribution of nanosilands depends on each factor: thicnkes of the initial film, time of annealing and annealing temperature. Some results we showed in eg: 

Kozioł, R.; Łapiński, M.; Syty, P.; Koszelow, D.; Sadowski, W.; Sienkiewicz, J.E.; Kościelska, B. Evolution of Ag nanostructures created from thin films: UV–vis absorption and its theoretical predictions, Beilstein J. Nanotechnol. 2020, 11, 494–507.

Kozioł, R.; Łapiński, M.; Syty, P.; et al. Experimental tuning of AuAg nanoalloy plasmon resonances assisted by machine learning method. Surf. Sci. 2021, 567, 150802.

4. The atomic force microscopy method makes it possible to obtain the size distribution of surface inhomogeneities. In my opinion, it is necessary to add to the work the distribution of gold particles by size, which was probably obtained by the authors. This will make it possible to carry out a quantitative analysis of the conditions for the formation of gold nanoparticles on the surface of silicon.

Thank you for this comment. We added histogram with distribution of nanoislands size (Fig 8b) and added comment in text.

Reviewer 2 Report

Metallic thin film, including Au, leads to formation of isolated islands,  what could be used for preparation of the plasmonic structures. In the present manuscript a study on disintegration of the gold thin film and formation of nanoislands on silicon substrates, both pure and with native silicon dioxide film was reported. The manuscript contains several important results, the paper could be published after minor correction and after certain consideration.

 

1. Are the film growth modes different on silicon and on silicon oxide?

2. Were the XP position of Au identical on Si and native oxide? Is the gold in metallic state?

 

3. For comparison it is suggested to consider the formation and behavior of Au clusters obtained on  other surfaces. (Phys. Chem. Chem. Phys. 16, (2014) 26786-26797). Cluster formation and plasmonic character of gold were observed for example on TiO₂-like materials; Au can initiate and inhibits the phase transformation of titanates."

Author Response

Thank you for reading the work carefully and for your comments. Our replies are posted below. Changes in the text are marked in red color.

1. Are the film growth modes different on silicon and on silicon oxide?

Thank you for this comment. Yes, modes are various, due to accelerate process by interface diffusion in case of pure silicon. We added a comment on that fact in discussion on the end of results paragraph.

2. Were the XP position of Au identical on Si and native oxide? Is the gold in metallic state?

In our opinion the main peaks are coresponding to metallic gold. Hawever asymmetry on the high energy side can be noticed. It could be related to additional gold-silicide compound that appears on a surface.

3. For comparison it is suggested to consider the formation and behavior of Au clusters obtained on  other surfaces. (Phys. Chem. Chem. Phys. 16, (2014) 26786-26797). Cluster formation and plasmonic character of gold were observed for example on TiO₂-like materials; Au can initiate and inhibits the phase transformation of titanates."

It is a very good comment. Thank you. We referred to these results in the discussion.

Reviewer 3 Report

I have a good impression about the article. It is clearly written, all figures are well designed and scientifically informative, and the results are reasonably interpreted.

The main thing that I did not understand is a motivaton and originality of the current research. What are the new and important results compared to dozens of cited references and previous works?

Formation of gold droplets is a key process to obtain good SERS substrates, such substrates are commercially available, so I believe that process of their formation is very well studied (e.g. https://www.nature.com/articles/srep14790). Thus, in introduction you have to explain in better detail why the current research is important and what are the key findings of the current work.

There are also several technical remarks:

1) Figure 1. The thickness of 16 ML gold film is 3 nm. The lattice constant of gold is 0.4 nm. Thus, I can expect that 16 ML are 16*0.4 nm = 6.4 nm, not 3.

2) Before line 156, you are writing about 16 ML film. Later, all figure captions indicate that you are dealing with 3 ML film. Why did you change the thickness?

3) Figures 6 (right) and 8 seem to be collected from the same (or identical) samples (600 K). As you write, AFM gives droplets size of ~100 nm. However, droplets are much smaller on SEM (~20 nm). Why is that?

4) Figure 9. You are writing about "asymmetry on the high energy side" for spectra c and d. What do you mean? I do not see any difference between the shapes of Au4f5/2 peaks. There are, probably, additional peaks slightly above 90 eV. Are these additional peaks responsible for a gold-silicide compound?

Author Response

We kindly thank you for reading the work and your comments. Below we include detailed responses to your comments. All changes in the text are marked in red.

The main thing that I did not understand is a motivaton and originality of the current research. What are the new and important results compared to dozens of cited references and previous works?

Formation of gold droplets is a key process to obtain good SERS substrates, such substrates are commercially available, so I believe that process of their formation is very well studied (e.g. https://www.nature.com/articles/srep14790). Thus, in introduction you have to explain in better detail why the current research is important and what are the key findings of the current work.

You are absolutelly right. We we emphasized the novelty of this work in discussion on the end of results paragraph. We also referred to the suggested work.

 

There are also several technical remarks:

1) Figure 1. The thickness of 16 ML gold film is 3 nm. The lattice constant of gold is 0.4 nm. Thus, I can expect that 16 ML are 16*0.4 nm = 6.4 nm, not 3.

In our system, the Au layer grows with the (111), not (001) plane. From this it follows that the thickness of 1ML Au is the distance between the (111) Au planes, a/sqrt(3) = 2.35A. 16 ML gives ca 3.6 nm thickness. Our layer is a little thinner (3 nm from AFM measurements) than we assumed. Later calibrations of the Au evaporator showed that it slowed down a bit.

2) Before line 156, you are writing about 16 ML film. Later, all figure captions indicate that you are dealing with 3 ML film. Why did you change the thickness?

Of course you are right. It is our mistake. It should be a 16 ML or 3 nm. We corrected text.

3) Figures 6 (right) and 8 seem to be collected from the same (or identical) samples (600 K). As you write, AFM gives droplets size of ~100 nm. However, droplets are much smaller on SEM (~20 nm). Why is that?

Yes, size of structures is slightly different for AFM and SEM/LEEM imaging. It is due to various technique. AFM method in not a direct method, but it measures forces. It would make a differences.

4) Figure 9. You are writing about "asymmetry on the high energy side" for spectra c and d. What do you mean? I do not see any difference between the shapes of Au4f5/2 peaks. There are, probably, additional peaks slightly above 90 eV. Are these additional peaks responsible for a gold-silicide compound?

Thank you for this comment. For us asymmetry means, that there is probably a very weak peak, close to 90 eV. We added position of notice asymmetry (additional peak).

Round 2

Reviewer 3 Report

Though the article was improved and the answers to my questions are quite reasonable, I'm still not fully satisfied by the answers to points 1 and 3.

1) Why are you writing about (111) gold? Your film is amorphous, as you state in lines 137-139 and as seen in Fig 4b. You just have to define somewhere (I would suggest line 85) that you calculate monolayer in (111) direction to avoid confusion.

3) AFM and SEM can not give such huge difference. I have good experience in both techniques, and the most recent sample I characterized was an oxide film with the grain size ~ 100 nm. AFM and SEM images were identical. AFM also produces very nice pictures with finer grains. Based on your AFM image, I can estimate an average film thickness as ~3 nm, so this image do not have huge artifacts. Maybe you have incorrect scale bar in Fig 6 (it should be 100 nm instead of 500)? Or maybe AFM imaged sample was annealed for longer time than 15 minutes? Or maybe you miss something else?

Author Response

Thank you again for a very thorough reading of the work and your comments. Our correctons are marked in green.

1) Why are you writing about (111) gold? Your film is amorphous, as you state in lines 137-139 and as seen in Fig 4b. You just have to define somewhere (I would suggest line 85) that you calculate monolayer in (111) direction to avoid confusion.

First of all, the sentence quoted by the reviewer (137-139 lines) refers to the Au layer deposited on amorphous native Si oxide, where the Au layer also reveals an amorphous structure, as we wrote. In such case, we agree with the reviewer, that the considerations about orientation of the growing layer is obviously pointless. However in the case of Au layer deposited on clean Si substrate the character of growth is completely different and needs to be described in more details to clarify any doubts. The diffraction pattern presented in the Fig. 4b (mentioned in the review) was collected after deposition of Au layer on clean Si substrate, not an oxide. As we said, presented image shows weakly ordered low-buckled silicene, which is formed on the top of Au layer. It is worth to mark that the energy of electron beam was relatively low to optimize the intensity of the low-buckled silicene spots. Unfortunately, for such low energy the diffraction spots corresponding to the Au layer are located beyond Ewald's sphere (field of view). Therefore, in order to study the atomic structure of the Au layer, a diffraction pattern for higher energy (31.7 eV) was collected (see picture below). The diffraction image shows two main rotated in-plane structural domains characterized by six-fold symmetry marked with two hexagons. Additional weak ordered elongated spots observed between two main domains indicate other rotated domains with smaller crystallite sizes. The calculated distance between (0,0) and (0,1) diffraction spots corresponds to the value equal to 2.46 Å in the real space roughly corresponding to the distance between atomic rows perpendicular to the [11-2] direction within (111) plane of fcc Au (2.50 Å). Thus, both the observed symmetry and the calculated lattice parameter clearly support our statement that (111) plane of Au layer deposited on clean Si(111) substrate is parallel to the surface of the sample which also remains consistent with the literature e.g. [1]. The outer most weak ordered ring is associated with low-buckled silicene.

[1]       R. Daudin, T. Nogaret, T. U. Schülli, N. Jakse, A. Pasturel, and G. Renaud, Epitaxial Orientation Changes in a Dewetting Gold Film on Si(111), Phys Rev B 86, 094103 (2012).

 

3) AFM and SEM can not give such huge difference. I have good experience in both techniques, and the most recent sample I characterized was an oxide film with the grain size ~ 100 nm. AFM and SEM images were identical. AFM also produces very nice pictures with finer grains. Based on your AFM image, I can estimate an average film thickness as ~3 nm, so this image do not have huge artifacts. Maybe you have incorrect scale bar in Fig 6 (it should be 100 nm instead of 500)? Or maybe AFM imaged sample was annealed for longer time than 15 minutes? Or maybe you miss something else?

Thank you very much for pointing out the error. We made a mistake in the figure description. The AFM image shows the structure after annealing in the LEEM/LEED chamber. So annealing took much longer than 15 minutes. It resulted in the growth of bigger structures. We have corrected the description with a comment. Once again, we apologize for this error and thank you for the thorough correction.

Author Response File: Author Response.docx