Thermomechanical Stresses in Silicon Chips for Optoelectronic Devices

Round 1

Reviewer 1 Report

There are a number of comments on the content of the manuscript.

1) Justify the choice of substrate materials. For the production of electronics, organic-based composite materials (polymers) are very often used as substrate materials. Why didn't you choose them?

2) Have you used Raman spectroscopy before to determine stresses?

3) Fig. 12. How do you explain the effect of pressure during soldering of the transition layer AuSn on the magnitude of stresses in silicon?

4) Why is there no similar Fig. 12 graphs for silicon samples on a ceramic substrate?

5) The Conclusion is very long. Please make it concise.

6) Section "References". 1 link missing. And the number of references (23) is most likely not enough for publication in the journal Applied Science.

Author Response

We thank reviewer 1 for comments and suggestions on our paper. All suggested corrections, listed below, have been highlighted in yellow in the text.

1) Justify the choice of substrate materials. For the production of electronics, organic-based composite materials (polymers) are very often used as substrate materials. Why didn't you choose them?

“Selecting the right substrate materials for the production of microelectronics is a key process. The substrate must be highly reliable, in configurations precisely chosen to meet the needs of a specific application. Several substrate materials are used, which are carefully chosen not only on the base of the performances to reach but also following other criteria like safety, environmental impact, economic costs, process complexity. Organ-ic-based composite materials in polymeric form have attracted considerable attention because they offer the opportunity to produce substrates with multiple customizable functionalities, large areas, flexibility, lightweight, and at relative low cost [Tzu-Hsuan Cheng, Kenji Nishiguchi, Yoshi Fukawa, B. Jayant Baliga, Subhashish Bhattacharya, Douglas C. Hopkins, Thermal and Reliability Characterization of an Epoxy Resin-Based Double-Side Cooled Power Module, Journal of Microelectronics and Electronic Packaging (2021) 18 (3): 123–136, https://doi.org/10.4071/imaps.1427774. Oksana Ostro-verkhova, Handbook of Organic Materials for Electronic and Photonic Devices, 2019, ISBN: 978-0-08-102284-9, https://doi.org/10.1016/C2016-0-05254-3 X. Zhao, Y. Jiang, B. Gao, K. Nishiguchi, Y. Fukawa, and D.C. Hopkins, Novel polymer substrate-based 1.2kV/40A double-side intelligent power module, 2017 IEEE 67th Electronic Components and Technology Conference (ECTC), pp. 1461–1467, Orlando, FL, 2017. Y. Kaji, Y. Hatanaka, S. Hiramatsu, S. Kondo, S. Asada, and Y. Otsubo, “Novel IGBT modules with epoxy resin encapsulation and insulating metal baseplate,” 2016 28th International Symposium on Power Semiconductor Devices and ICs (ISPSD), pp. 475–478, Prague, 2016.] Immense scope is expected in the future of flexible organic electronics. However, the current performance and lifetimes of electronic devices based on organic substrates is still lower than the ones of traditional ceramic or other inorganic materials [N. Chasserio, S. Guillemet-Fritsch, T. Lebey, and S. Dagdag, Ceramic substrates for high-temperature electronic integration, Journal of Electronic Materials, Vol. 38, pp. 164–174, 2009.], like aluminum nitride or copper. In the current paper, ceramic and copper are used as substrate materials. Ceramic substrates are well known for high-temperature electronic integration. Copper substrates in microelectronics are widely used because of the advantageous properties of copper, like low electrical resistivity, high electromigration resistance, high thermal conductivity, excellent chemical and thermodynamic characteristics, antimicrobial properties and low-cost. A problem with Cu is the tendency to form oxides on the surface. The authors have previously used copper substrates to prepare assemblies with LEDs instead of Si. Moreover, they have suggested to use formic acid or formate salts to prevent the formation of oxides [Omid Mokhtari, Fosca Conti, Rodolfo Saccon, Sri Krishna Bhogaraju and Gordon Elger, Formic acid and formate salts for chemical vapor deposition of copper on glass substrates under atmospheric pressure, New J. Chemistry 2021, 45(43), pp. 20133-20139, Doi: 10.1039/d1nj02476k].”

Explanations about the substrate materials have been added in the introduction.

 

2) Have you used Raman spectroscopy before to determine stresses?

We have used Raman spectroscopy for the determination of stresses starting from GaN samples, see papers reported in Reffs 6,7,13,20,21,22, 28 and 29, and in the following articles, not included in the text as they are less relevant to the stress problems of the Si:

- R. Signorini, D. Pedron, F. Conti, A. Hanss, S. K. Bhogaraju, G. Elger, in 24th IEEE/Therminic 2018, 1.

- E. Liu, F. Conti, R. Signorini, E. Brugnolotto, S. K. Bhogaraju, G. Elger, in 20th EuroSimE 2019.
- R. Signorini, F. Conti, E. Brugnolotto, D. Pedron, E. Liu, S. K. Bhogaraju, G. Elger, in SPIE-Optics and Optoelectronics 2019, 11031, 1.

- E. Brugnolotto, S.K. Bhogaraju, E Liu, F. Conti, D. Pedron, R. Signorini, G. Elger, in 25th IEEE/Therminic 2019,

 

3) Fig. 12. How do you explain the effect of pressure during soldering of the transition layer AuSn on the magnitude of stresses in silicon? “Figure 12 shows that the magnitude of stress in the sample prepared under pressure is much lower than in the sample prepared in the simplest method, i.e. without the use of an additional pressure of 5 N. A hasty conclusion would be that the effect of pressure during soldering of the transition AuSn layer could be helpful because for the assembly without pressure the Raman signal of the Si peak changes strongly, in comparison to the signal of unmounted Si. Further measurements to check the interconnections were done during this investigation. In particular, cross-section analyses of the assemblies have revealed that the Si chip was not properly soldered to the substrate, i.e. the use of the pressure of 5 N was a disadvantage and studies in this direction are in progress. It is worth to note that in this case the accuracy and precision of the Raman results need to be confirmed by other experimental analysis to avoid any miss interpretation of the data.”

Explanations have been added close to the figure.

4) Why is there no similar Fig. 12 graphs for silicon samples on a ceramic substrate? Since the silicon sample on the ceramic substrate do not show appreciable stress, close to the unassembled Si_unmounted, its characterization has been performed only at room temperature not also at low and high temperatures.

5) The Conclusion is very long. Please make it concise. Conclusion has been modified as suggested.

6) Section "References". 1 link missing.  Reference 1 has been corrected. And the number of references (23) is most likely not enough for publication in the journal Applied Science. As suggested references have been added to the paper.

Reviewer 2 Report

1.      Rewriting the abstract is necessary. Doesn't meet the criteria for a scientific abstract in its current form. Additionally, it must to include the primary outcomes.

2.      The keyword list could be enhanced to address the core purpose and research as it consists of generic terms.

3.      The key references and resources should be incorporated with all the material that is offered in the article, notably in the introduction.

4.      Although the phrase "silicon sample" is frequently used, it should use a more thorough definition.

5.      The present study's preparation techniques for silicon chips on multiple substrates are utterly unclear. As a result, in the updated edition, the writers must include specifics of the preparation techniques.

6.      How was the Raman spectrum employed by the author for optical analysis?

7.      It is important to integrate all Raman spectral peak observations with appropriate references.

 

8.      The conclusion should be condensed since it is too long and doesn't adequately cover the key findings and their justifications.

Author Response

We thank reviewer 2 for comments and suggestions on our paper. The following are the replies to the comments made.

1. Rewriting the abstract is necessary. Doesn't meet the criteria for a scientific abstract in its current form. Additionally, it must to include the primary outcomes. Abstract has been revised; primary outcomes have been included.
2. The keyword list could be enhanced to address the core purpose and research as it consists of generic terms. The keyword list has been modified according to the suggestion.
3. The key references and resources should be incorporated with all the material that is offered in the article, notably in the introduction. Key references and resources have been added in the introduction.
4. Although the phrase "silicon sample" is frequently used, it should use a more thorough definition. More appropriate definitions have been used in text to indicate the correct meaning of "silicon sample".
5. The present study's preparation techniques for silicon chips on multiple substrates are utterly unclear. As a result, in the updated edition, the writers must include specifics of the preparation techniques.

“All Si-chips were soldered to the substrate using a vacuum-assisted closed reflow oven (Unitemp, RSS-160). For an efficient soldering, surface oxides and undesired residuals were removed using formic acid vapor, which has strong reducing effects. The assemblies were prepared at 320 °C, using a bonding time of 80 s (for the samples without additional pressure) and 30 s (for the samples with additional pressure). Several assemblies were prepared to confirm the reproducibility of the experimental results. Fig. 3 shows the temperature profiles in the reflow oven. Before to proceed with the Raman measurements, the final assemblies were optically inspected by a 3D digital microscope (Keyence VHX-900F).”

Explanations about the assemblies preparation are added in “Materials and Sample Preparation”.

6. How was the Raman spectrum employed by the author for optical analysis? As rightly noted, no explanation of how the Raman data were processed to obtain the Raman shift values ​​shown in the various tables has been reported. The description has been added in the text, at the end of section 2.2 Methods.

“All the collected Raman spectra have been interpolated with a Lorentzian function to determine their parameters, like area, intensity, amplitude and peak position.”

7. It is important to integrate all Raman spectral peak observations with appropriate references. The appropriete REFERENCE HAS BEEN ADDED
8. The conclusion should be condensed since it is too long and doesn't adequately cover the key findings and their justifications. Conclusions have been revised accordingly suggestions.

Reviewer 3 Report

The authors analyzed the thermomechanical stresses generated during the assembly process of silicon devices based on various assembly parameters (material, temperature and pressure, etc.), since internal stresses can deform and sometimes destroy the microstructure of the device. Until now, soldering using AuSn alloy has been common in many manufacturing processes, and many studies on analyzing thermomechanical stress using Raman spectrum have been conducted, but this study is very interesting in terms of the practical use temperature range of MEMS devices.

In Fig. 13(c), Fig. 14(c) and Fig. 15(b), the stress distribution is unstable. Did the author observe any mechanical defects after soldering in this study?

 

Minor

1. Some pictures are cut off or the color legend overlaps the data. please correct

2. The figure number in the text is corrected, and the figure number mentioned in the text must also be corrected (i.e. Figure 9~ Figure 12).

3. Re-edit the text, including Table 7, to conform to the MDPI style.

4. Clearly correct the sample names mentioned in the figure caption.

5. Number 1 in the reference list is missing. Please check.

Author Response

We thank reviewer 3 for comments and suggestions on our paper. All suggested corrections, listed below, have been highlighted in yellow in the text.

In Fig. 13(c), Fig. 14(c) and Fig. 15(b), the stress distribution is unstable. Did the author observe any mechanical defects after soldering in this study?

Fig. 13(c), Fig. 14(c) and Fig. 15(b) show an unstable stress distribution. This can be correlated to mechanical defects after the soldering process. Cross-sections of the samples confirmed the experimental results that the samples had cracks within the solder joint. Alternative method to test the solder joint integrity is offered by the transient thermal testing, which is a non-destructive tool [G.Elger, S.V.Kandaswamy, R.Derix, F.Conti, Detection of solder joint cracking of high power LEDs on Al-IMS during temperature shock test by transient thermal analysis, In: Proceeding of the 20^th IEEE International Workshop on Thermal Investigations of IC's and Systems (THERMINIC), Greenwich, London, UK, September 24th–26th, 2014.]. It is worth to remind that also samples with cracks, detected by Raman spectroscopy and confirmed by transient thermal analysis and cross section, can still have functioning electric contacts, as demonstrated with high power LEDs, which passed the light-on test [G. Elger, S. V. Kandaswamy, E. Liu, A. Hanss, M. Schmid, R. Derix, F. Conti, Analysis of solder joint reliability of high power LEDs by transient thermal testing and transient finite element simulations, Microelectronics Journal, 2015, 46, 1230-1238. http://dx.doi.org/10.1016/j.mejo.2015.08.007.]. The Raman results of the current study suggest the high potentiality, accuracy and precision of this optical spectroscopy to estimate the quality of samples.

Explanations about the stress distributions in Fig. 13c, Fig. 14c and Fig. 15b are added in the text.

 

Minor

1. Some pictures are cut off or the color legend overlaps the data. please correct The figures have been corrected and a pdf version of the paper has been attached, since the overlap seems to be due to a variation in the layout of the article when it was loaded.
2. The figure number in the text is corrected, and the figure number mentioned in the text must also be corrected (i.e. Figure 9~ Figure 12). The number of the figures have been corrected.
3. Re-edit the text, including Table 7, to conform to the MDPI style. The style of the Tables has been corrected conforming to the MDPI style.
4. Clearly correct the sample names mentioned in the figure caption. Sample names in the figure captions have been corrected.
5. Number 1 in the reference list is missing. Please check. It has been corrected.

Round 2

Reviewer 1 Report

The manuscript can be recommended for publication.

Reviewer 2 Report

The authors well addressed the reply to the reviewer’s comments and revised their work accordingly. Therefore, I recommend the publication of the manuscript in its present form in Applied Sciences.