Submit to this Journal Review for this Journal Propose a Special Issue

Article Menu

Share Help Cite Discuss in SciProfiles

Open AccessArticle

Peer-Review Record

Optimizing the PECVD Process for Stress-Controlled Silicon Nitride Films: Enhancement of Tensile Stress via UV Curing and Layered Deposition

Coatings 2025, 15(6), 708; https://doi.org/10.3390/coatings15060708

by Jianping Ning^1,2, Chunjie Niu², Zhen Tang², Yue Sun², Hao Yan² and Dayu Zhou^1,*

Reviewer 1:

Kamal Kayed

Reviewer 2: Anonymous

Reviewer 3:

Aomar Hadjadj

Reviewer 4: Anonymous

Coatings 2025, 15(6), 708; https://doi.org/10.3390/coatings15060708

Submission received: 31 March 2025 / Revised: 6 June 2025 / Accepted: 8 June 2025 / Published: 12 June 2025

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The review report

The manuscript title: Optimizing PECVD Process for Stress-Controlled Silicon Nitride Films: Enhancement of Tensile Stress via UV Curing and Layered Deposition.

In this manuscript, the authors present the results of an experimental study on the effects of silane gas (SiH₄) flow rate during the deposition of silicon nitride films using the PECVD method. The study also investigated the effects of UV curing and layer deposition on the tensile stress of the prepared silicon nitride films. The main conclusion of the research is that combining optimization of PECVD deposition parameters with UV curing and layer deposition provides a strategy for engineering mechanical properties for advanced applications.

The authors have used references relevant to the topic of their article, and the conclusions presented are consistent with the evidence and arguments presented, and address the main question raised. However, this article can be accepted for publication after taking into account the following points:

1- Silane gas (SiH₄) should be added to the keywords.

2- In line 39, the authors cite the following statement from Reference 39: "Hydrogen incorporation during the PECVD process plays a key role in the formation of compressive stress. Hydrogen from primary gases, such as silane (SiH₄) and ammonia (NH₃), forms Si-H and N-H bonds, increasing internal stress and reducing film stability." This result is not a result of reference 39, so authors should consult the appropriate reference.

3- What are the similarities and differences between this research and the research included in Reference 21?

4- I suggest replacing the phrase "the number of deposition cycles" with "the number of deposited layers".

5- In Figure 11, how does the ratio N-H/Si-H represent the hydrogen ratio (represented on the vertical axis)?

6- In Figure 11, the hydrogen ratio does not appear to increase significantly with increasing deposition cycles in the pre-UV case, as the authors reported. It can be seen that the changes in the content are very small and negligible. On the other hand, what is the experimental evidence that deviating the hydrogen ratio more than 1:1 leads to a decrease in tensile stress?

7- In general, consideration must be given to interpreting all changes in the behavior of curves in all Figs.

8- The effect of UV treatment of silicon nitride films on both hydrogen content and tensile stress has been reported in several previous works (regardless of the films synthesis method), so the unique results obtained should be highlighted and compared with those of previous research.

Author Response

Reviewer # 1

Response: Thank you! We have made the corresponding modifications (see the point-to-point responses shown below).

Comment #R1.1: Silane gas (SiH₄) should be added to the keywords.

Response: Thank you for suggesting this improvement. Revised.

Modifications:

Keywords: PECVD, Silane Gas (SiH₄), Silicon Nitride Films, Tensile Stress, UV Curing, Layered Deposition

Comment #R1.2: In line 39, the authors cite the following statement from Reference 39: "Hydrogen incorporation during the PECVD process plays a key role in the formation of compressive stress. Hydrogen from primary gases, such as silane (SiH₄) and ammonia (NH₃), forms Si-H and N-H bonds, increasing internal stress and reducing film stability." This result is not a result of reference 39, so authors should consult the appropriate reference.

Response: Thank you for pointing this. Revised.

Modifications:

Hydrogen from precursor gases such as silane (SiH₄) and ammonia (NH₃) forms Si-H and N-H bonds, increasing internal stress and reducing film stability [1,11,16].

Comment #R1.3: What are the similarities and differences between this research and the research included in Reference 21

Response: Thank you for suggesting this improvement. We would like to emphasize that although both studies investigate PECVD-deposited SiNx:H films and employ similar analytical techniques (e.g., FTIR for H-bond analysis, stress measurements via wafer curvature), they differ substantially in terms of research goals, methodologies, and scientific contributions, as detailed below:

Previous work (Electronics, 2024) focused on establishing quantitative correlations between H-related chemical bonds (Si–H, N–H) and key physical properties of SiNx:H films—such as refractive index (RI), breakdown field (EB), stress, and wet etch rate. The goal was to elucidate underlying mechanisms of hydrogen incorporation under various deposition parameters (e.g., RF power, temperature, pressure, gas ratio) for 28 nm node passivation applications. Current work is centered on process innovation to achieve high tensile stress in PECVD SiNx films. It introduces UV curing and layered deposition as two sequential strategies to actively modify film bonding structure and maximize tensile stress. The emphasis is on engineering an optimized fabrication sequence rather than mapping passive correlations.
The previous study demonstrated that increasing Si–H/N–H ratio increases RI but reduces EB; total hydrogen correlates more strongly with wet etching rate than with stress. Importantly, it concluded that film stress is not linearly dependent on hydrogen content alone. The current study takes that insight forward: by actively reducing hydrogen bonds through UV treatment (primarily N–H), the film network becomes more crosslinked (Si–N–Si), and tensile stress is effectively increased. By controlling hydrogen desorption per layer, a cumulative densification effect is achieved, resulting in stress >1500 MPa, which was not addressed in the earlier work.
In summary, the prior work is a scientific correlation study exploring intrinsic H-bond/property relationships under varied PECVD conditions, while the current work is a process development study establishing a new route to engineer high-stress SiN films through UV-layered strategies. Both studies are complementary but distinctly focused—one on understanding mechanisms, the other on applying them for advanced stress control.

Comment #R1.4: I suggest replacing the phrase "the number of deposition cycles" with "the number of deposited layers".

Response: Thank you for pointing this. Based on your suggestions, we have carefully revised the manuscript to correct this inappropriate phrasing in all relevant sections.

Modifications:

Figure 9 illustrates the effect of the number of deposition layer on the stress and RI of SiN films before and after UV curing. In this study, the total thickness of the SiN films is maintained at 60 nm for both single and multiple deposition layers. Figure 9(a) presents the relationship between initial stress and the number of deposition layers before UV curing. The results indicate that as the layer number increases from 1 to 6, the initial stress rises from 725 MPa to 920 MPa, representing an increase of approximately 30%.
Figure 9(b) shows the variations in post-stress and loss-stress as a function of the number of deposition layers. Notably, after three PECVD deposition layers followed by UV curing, the SiN films exhibit the highest tensile stress, reaching approximately 1630 MPa—nearly twice that of the as-deposited film without UV curing.
Additionally, Figure 9(b) illustrates the trend of loss-stress with increasing number of deposition layers.
Figure 10 presents the variation in , , and as a function of the number of deposition layer.
This suggests that during multilayer deposition, thermal effects play a dominant role in enhancing tensile stress.
Figure 11 illustrates the variation of the N-H/Si-H ratio as a function of the number of deposition layers. As shown in Figure 11, the N-H/Si-H ratio increases with increasing deposition layers, with a more pronounced increase observed after UV curing. Notably, when the number of deposition layers is less than two, the in SiN films is lower than the . However, as the number of deposition layers increases, the N-H/Si-H ratio undergoes a reversal.

Comment #R1.5: In Figure 11, how does the ratio N-H/Si-H represent the hydrogen ratio (represented on the vertical axis)?

Response: Thank you for pointing this. This expression was inappropriate. In Equation (4), we have defined the physical meaning of the H_ratio. Based on this definition, we have revised Figures 11, 5, and 8 accordingly. Additionally, all instances of the term “H ratio” have been removed from the manuscript.

Modifications:

Figure 5. (a) Variation of the N-H/Si-H ratio as a function of SiH₄ flow rate before and after UV curing, and (b) Δ H concentration as a function of SiH₄ flow rate. The plasma power is 175 W, NH₃ flow rate to 880 sccm, and UV power to 85 W.

Figure 8. Variation of the N-H/Si-H ratio as a function of UV power. The plasma power is 175 W, SiH₄ flow rate to 91 sccm, the SiN films thickness is 60 nm, and NH₃ flow rate to 880 sccm.

Figure 11. Variation of the N-H/Si-H ratio as a function of the number of deposition layer cycles. The plasma power is 175 W, UV curing power is 85 W, SiH₄ flow rate to 91 sccm, the SiN films thickness is 60 nm, and NH₃ flow rate to 880 sccm.

Comment #R1.6: In Figure 11, the hydrogen ratio does not appear to increase significantly with increasing deposition cycles in the pre-UV case, as the authors reported. It can be seen that the changes in the content are very small and negligible. On the other hand, what is the experimental evidence that deviating the hydrogen ratio more than 1:1 leads to a decrease in tensile stress?

Response: Thank you for suggesting this improvement. The citation of Reference [21] in this section is incorrect. The corresponding experimental results should instead be discussed with reference to Figures 3(b) and 5(a). As shown in Figure 5(a), with increasing SiH₄ flow rate, the N–H/Si–H ratio first gradually approaches 1:1 and then shifts away from this ratio. Correspondingly, Figure 3(b) reveals that as the N–H/Si–H ratio increases from below 1:1 toward approximately 1:1, the tensile stress of the UV-treated SiN films exhibits a slight increasing trend. Notably, the tensile stress reaches its maximum when the N–H/Si–H ratio is close to 1:1. Beyond this point, when the N–H/Si–H ratio exceeds 1:1, a significant decrease in tensile stress is observed.

Modifications:

With further deposition cycles, the N–H/Si–H ratio deviates increasingly from 1:1, leading to a noticeable reduction in tensile stress. This trend is consistent with the experimental results shown in Figures 3(b) and 5(a), which indicate that maximum tensile stress occurs when the N–H/Si–H ratio is close to 1:1, while stress decreases significantly as the ratio exceeds this threshold.

Comment #R1.7: In general, consideration must be given to interpreting all changes in the behavior of curves in all Figs.

Response: Thank you for this constructive feedback. In accordance with your suggestion, we have added the missing description of the curve trends in the manuscript, as detailed in the underlined sentence.

Modifications:

It can be seen that when the SiH₄ flow rate is less than approximately 93 sccm, the N–H/Si–H ratio is greater than 1:1. As the SiH₄ flow rate increases, the N–H/Si–H ratio gradually approaches 1:1. Furthermore, after UV curing, the N–H/Si–H ratio shifts further away from 1:1 compared to the SiN films without UV treatment.
As shown in Figure 8, the N–H/Si–H ratio remains approximately 1:1 in the absence of UV treatment. When the UV power increases from 55 W to 70 W, the H ratio remains relatively constant.
Figure 11 illustrates the variation of the N-H/Si-H ratio as a function of the number of deposition layers. As shown in Figure 11, the N-H/Si-H ratio increases with increasing deposition layers, with a more pronounced increase observed after UV curing. Notably, when the number of deposition layers is less than two, the in SiN films is lower than the . However, as the number of deposition layers increases, the N-H/Si-H ratio undergoes a reversal. With further deposition cycles, the N–H/Si–H ratio deviates increasingly from 1:1, leading to a noticeable reduction in tensile stress. This trend is consistent with the experimental results shown in Figures 3(b) and 5(a), which indicate that maximum tensile stress occurs when the N–H/Si–H ratio is close to 1:1, while stress decreases significantly as the ratio exceeds this threshold.

Comment #R1.8: The effect of UV treatment of silicon nitride films on both hydrogen content and tensile stress has been reported in several previous works (regardless of the films synthesis method), so the unique results obtained should be highlighted and compared with those of previous research.

Response: Thank you for your valuable feedback. While the effect of UV treatment on SiN films has been widely reported, our study makes a unique contribution by focusing on the impact of the N-H/Si-H ratio on tensile stress. We systematically investigate how changes in this ratio influence film stress, offering new insights into the relationship between bonding configurations and mechanical properties. Additionally, our work explores the optimal parameters for combining layered deposition with UV curing, a method that leads to a cumulative enhancement of tensile stress. This approach surpasses previous studies, where UV curing was applied to single-layer films, and provides a novel strategy for achieving significantly higher tensile stress in SiN films.

Modifications:

The layered deposition approach introduced in this study provides an alternative strategy for enhancing the tensile stress of SiN films. By repeatedly depositing thin SiN layers followed by UV curing, a cumulative stress effect is observed in the films. Each deposition-curing cycle contributes to SiN film densification and cross-linking, resulting in a progressive increase in tensile stress [32]. The fundamental mechanism of the layered deposition approach lies in the controlled formation of SiN films, where each layer undergoes UV curing to improve its mechanical properties. This method also focuses on the impact of the N-H/Si-H ratio on tensile stress, a key factor that differentiates this study from previous research. It is observed that when the N-H/Si-H ratio approaches 1:1, tensile stress reaches its maximum, demonstrating a clear link between bonding configuration and mechanical properties.

The thickness of individual deposition layers plays a crucial role in determining the overall mechanical properties of the film. During the deposition process, the accumulation of hydrogen and other defects often leads to higher intrinsic stress in thicker layers. In contrast, thinner layers provide more opportunities for effective curing and densification, thereby improving mechanical performance. In this study, the use of thin layers (~20 nm) in the layered deposition approach resulted in films with better uniformity and significantly higher tensile stress. The optimized combination of layer thickness and deposition cycle number provides a means to tailor the mechanical properties of SiN films for specific applications. By adjusting these parameters, it is possible to achieve films with a wide range of tensile stresses, from moderate to highly stressed films, depending on device requirements.

The enhancement of tensile stress in SiN films via UV curing and layered deposition, as demonstrated in this study, offers distinct advantages over conventional stress engineering methods. Traditional approaches for stress modulation in PECVD-deposited SiN films include thermal annealing, doping, deposition parameter optimization, and plasma post-treatment. This study contextualizes the findings within the broader landscape of stress engineering techniques and highlights the unique benefits of the UV curing-based approach, especially in terms of achieving high tensile stress through controlled hydrogen desorption and cross-linking, a strategy not commonly explored in previous works.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Comments and Suggestions for Authors

Optimizing PECVD Process for Stress-Controlled Silicon Nitride Films: Enhancement of Tensile Stress via UV Curing and Layered Deposition

The manuscript focuses on optimizing the PECVD process to control the tensile stress in silicon nitride films. It specifically investigates how varying the silane flow rate during PECVD deposition, along with the application of UV curing and multilayer deposition, can be used as strategies to enhance the tensile stress of the resulting SiN films. In general, the presentation is clear and well-organized, and the study provides a thorough examination of the factors affecting tensile stress in SiN films. However, some critical aspects must be considered for the rejection of its acceptance for publication:

Regarding the relevance and originality of the current paper I consider it irrelevant for the field, the proposed strategy for the optimization of tensile stress in the silicon nitride thin films by combining PECVD method with UV post-deposition treatment has been extensively discussed in the literature, the multilayered deposition approach has also been explored by other authors. Considering these aspects, I do not recommend the publication of this paper. Moreover, the following aspects further support the decision to reject the publication.

In Section 2.3, which focuses on the performed characterization methods, the authors specified that the structural properties of the SiN films were investigated using several techniques. Specifically, FTIR spectroscopy was employed to analyze the bonding configurations within the SiN films and monitor changes in bond intensities after UV curing, while XPS spectroscopy was used to evaluate the hydrogen content in the resulting silicon nitride thin films. However, neither the FTIR nor the XPS spectra are presented or discussed in the main body of the article, which significantly limits the validity of the structural analysis and weakens the accuracy of the study

Furthermore, it should be noted that the authors do not master the XPS technique, and they should know that the hydrogen content cannot be directly quantified using conventional XPS due to the technique's limitations in detecting the hydrogen atoms. In contrast, XPS can provide valuable information about the bonding states, especially the presence and relative concentration of Si-N bonds, which, however, were not presented in the article, The same situation applies to the FTIR investigation: the authors conclude that the mechanism of hydrogen degassing and bond rearrangement is confirmed by FTIR analysis, citing a significant reduction in the intensities of the Si–H and N–H peaks after UV curing. However, they do not present any FTIR spectra to support these conclusions, mentioning only two references to other publications, [23] and [25], which compromises the rigor of their study.

Another critical aspect regarding the consistency of their study is the ellipsometry investigation. Although it was mentioned to investigate the optical properties of the SiN thin films, the authors only provide, in Figure 3, the dependency of the refractive index on the SiH4 flow rate. These values are presented without essential details regarding the optical model used to analyze the data, the dispersion model applied, the fitting method, or any assumptions involved in extracting the refractive index and film thickness. As a result, the reliability and reproducibility of the reported optical constants cannot be verified.

Regarding the bibliography, I believe the study is based on a limited number of references, with only four representing current research in the field. As a result, the study lacks in-depth scientific support from the specialized literature. In this context, the introduction section is inadequate and insufficient, both for the study itself and for the scientific community

Recommendation: By addressing the critical aspects outlined above, I consider the publication inappropriate in its current form.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

Could be improved.

Author Response

Reviewer # 2

Response: Thank you for your thoughtful and constructive feedback. We truly appreciate your time and effort in reviewing our manuscript. We would like to address the points you raised regarding the relevance and originality of the study, the XPS technique, and the ellipsometry data. Kindly allow us to explain the updates made in the manuscript:

Relevance and Originality:

While we acknowledge that UV curing and multilayer deposition have been explored in previous studies, our work introduces a novel combination of layered deposition and UV curing. Specifically, we focus on the N-H/Si-H ratio and its effect on tensile stress, a unique aspect not thoroughly investigated in prior work. Additionally, this study systematically optimizes the combination of deposition cycles and UV curing parameters, resulting in a cumulative enhancement of tensile stress. The combination of these factors leads to the formation of highly cross-linked SiN films with superior mechanical properties. We believe this approach presents significant advancements over existing methods and offers a robust strategy for tailoring SiN film stress for advanced semiconductor applications. We have already discussed this in the manuscript, as detailed below:

This method yields films with greater uniformity and reduced defect density compared to single-layer films. Additionally, we systematically optimized the parameters for UV curing and layered deposition to achieve the highest tensile stress. The cumulative effect of multiple deposition-curing cycles facilitates the formation of highly cross-linked, dense SiN films, ultimately enhancing tensile stress. The results demonstrate that tensile stress increases significantly with the number of deposition-curing cycles, reaching optimal stress levels after three to five cycles. Notably, the optimized combination of layer thickness (~20 nm) and deposition cycles ensures the highest tensile stress without compromising film quality.The thickness of individual deposition layers plays a crucial role in determining the overall mechanical properties of the film. During the deposition process, the accumulation of hydrogen and other defects often leads to higher intrinsic stress in thicker layers. In contrast, thinner layers provide more opportunities for effective curing and densification, thereby improving mechanical performance. In this study, the use of thin layers (~20 nm) in the layered deposition approach resulted in films with better uniformity and significantly higher tensile stress. The optimized combination of layer thickness and deposition cycle number provides a means to tailor the mechanical properties of SiN films for specific applications. By adjusting these parameters, it is possible to achieve films with a wide range of tensile stresses, from moderate to highly stressed films, depending on device requirements.

XPS Technique:

We sincerely apologize for the error regarding the use of XPS to quantify hydrogen content. This issue was identified by the academic editor during the initial submission process, and we have since corrected the manuscript to accurately reflect the XPS technique’s role in analyzing bonding states, not hydrogen quantification. The revised version of the manuscript, which was corrected prior to resubmission, properly addresses this and removes any erroneous mention of XPS for hydrogen quantification. Unfortunately, it appears the version you received was the earlier, unmodified version. We apologize for this confusion and have ensured that the current manuscript now adheres to proper technical accuracy.

Ellipsometry Data:

Regarding the use of ellipsometry to investigate the optical properties of SiN films, we have expanded on the explanation in Section 2.3 of the revised manuscript. This ensures full transparency and clarity regarding how the refractive index (RI) and film thickness were determined. We hope these added details address your concern about the ellipsometry data analysis and its accuracy.

We kindly ask for your reconsideration of the manuscript based on these revisions, which have clarified the points raised and enhanced the rigor of the study. We are confident that these changes address all of the concerns and would greatly appreciate the opportunity to have the manuscript re-evaluated for publication.

Comment #R2.1: In Section 2.3, which focuses on the performed characterization methods, the authors specified that the structural properties of the SiN films were investigated using several techniques. Specifically, FTIR spectroscopy was employed to analyze the bonding configurations within the SiN films and monitor changes in bond intensities after UV curing, while XPS spectroscopy was used to evaluate the hydrogen content in the resulting silicon nitride thin films. However, neither the FTIR nor the XPS spectra are presented or discussed in the main body of the article, which significantly limits the validity of the structural analysis and weakens the accuracy of the study.

Response: We apologize for the incorrect usage of XPS in the previous version. Regarding FTIR analysis, we apologize for not providing the FTIR spectra to support our conclusions in the previous submission. However, we have updated Section 2.3 with a more comprehensive calculation method for evaluating the bonding configurations in SiN films. We did include references to previous studies that have conducted FTIR analysis under similar conditions, and we have expanded the section to include theoretical interpretations based on existing literature, such as the reduction in Si-H and N-H peak intensities after UV curing. It is important to note that our detection of Si-H and N-H bonds and the corresponding changes are fully consistent with our previous work, as discussed in reference [26].

Modifications:

2.3 Characterization analysis

Fourier-transform infrared spectroscopy (FT-IR) spectra were collected with a Nicolet 6700 Flex (Thermo Fisher, Massachusetts USA), which provide the structural properties and bonding configuration of SiN films before and after UV curing. The FT-IR spectra was measured among the wavelength range from 4000 cm⁻¹ to 400 cm⁻¹ with a step width of 2 cm⁻¹. The presence of Si-H, N-H, and Si-N bonds were determined with the location of the characteristic peaks. Meanwhile, the atomic percentage of H atom ( ) in Si-H or N-H was calculated with equation 1 based on the bond density ( ) of Si-H or N-H (equation 2). And the total atomic percentage of H atom in Si-H and N-H ( ) was obtained with equation 3, the bond density ratio of N-H and Si-H (H ratio) was measured by equation 4 accordingly [26].

where is the absorption peak area of the Si-H or N-H in the SiN film, is the Si-H or N-H bond cross section, and are 7.4 10-18 and 5.3 10-18 cm², and is the SiN flim thickness.

The curvature radius of the wafer was measured via T910 (Skyverse Technology, China). The test mode was a line scan with 51 points. Mechanical stress of SiN film was measures by substrate bow methods, in which the stress of the SiN film ( ) is determined by the Stoney’s equation [26].

where is the radius of curvature of the bare substrate and is the radius of curvature with the film on the substrate. and are the Young’s modulus and Poisson ratio of the substrate, respectively, and and are the thickness of the substrate and the film, respectively.

Comment #R2.2: Another critical aspect regarding the consistency of their study is the ellipsometry investigation. Although it was mentioned to investigate the optical properties of the SiN thin films, the authors only provide, in Figure 3, the dependency of the refractive index on the SiH4 flow rate. These values are presented without essential details regarding the optical model used to analyze the data, the dispersion model applied, the fitting method, or any assumptions involved in extracting the refractive index and film thickness. As a result, the reliability and reproducibility of the reported optical constants cannot be verified.

Response: Thank you for pointing this. In accordance with your suggestion, further details have been added in Section 2.3.

Modifications:

The refractive index (RI) of the SiNx:H film was measured via the film metrology system with a wavenumber range of 200 to 800 cm⁻¹ (Tensor 8500, KLA, Milpitas, CA, USA). The RI data were collected from five wafers, and 49 points were measured for each wafer. A systematic study of the optical properties of PECVD-prepared silicon nitride films was conducted using the Tauc-Lorentz fitting model. The film thickness, refractive index, and fitting errors were calculated. Nonlinear regression fitting was performed using the Levenberg-Marquardt algorithm, with the minimum square error (MSE) used to evaluate the fitting quality. The Tauc-Lorentz model, specifically designed for silicon nitride and amorphous semiconductor films, was applied to fit the silicon nitride films, and the fitting values showed good agreement with the measured data.

Comment #R2.3: Regarding the bibliography, I believe the study is based on a limited number of references, with only four representing current research in the field. As a result, the study lacks in-depth scientific support from the specialized literature. In this context, the introduction section is inadequate and insufficient, both for the study itself and for the scientific community

Response: Thank you for your valuable feedback. We understand your concern regarding the limited number of references in the initial manuscript. In response to your suggestion, we have further supplemented the bibliography, particularly in the Introduction section. Additional references from recent studies in the field have been incorporated to provide a more comprehensive scientific foundation and to strengthen the contextual relevance of our work. This expansion ensures that the study is well-supported by current literature and provides a more in-depth background for the scientific community. We have revised the Introduction accordingly to ensure it fully addresses the research context, highlighting recent developments and placing our work within the broader landscape of tensile stress optimization in SiN films.

Modifications:

The formation of internal stress in thin films is closely related to the film fabrication process, deposition conditions, and the microstructure of the films. Studies have shown that there are certain regularities in the mechanisms underlying the formation of internal stress in thin films. E. Chason et al. [13] examined stress evolution during the electrodeposition of Ni on Si substrates, proposing a model where stress is attributed to competing processes at growing grain boundaries, with the stress depending on the rate of grain boundary formation. W. D. Nix et al. [14] studied the stress during crystallite coalescence in polycrystalline films, suggesting that the balance between surface area reduction, grain boundary creation, and film strain leads to large tensile stresses in the films. G. Guisbiers et al. [15] developed a model for intrinsic residual stress in thin films, based on nanograin size-dependent phase transitions, and showed that stress arises from volume changes in the nanograins, considering relaxation processes and experimental comparisons with Ta, Mo, Pd, and Al films. Seel [16] suggested that after island-like deposition, atoms diffuse on the island surfaces or enter strain regions at grain boundaries, leading to the relaxation of tensile stress. P. Chaudhari [17], however, argued that grain growth causes the relaxation and reduction of compressive stress. Additionally, it is believed that the formation of internal stress in thin films may result from impurity atoms or capillary stress. Impurity effects refer to the incorporation of substrate atoms or impurity atoms, such as inert gases or oxygen, into the film during growth, which generates compressive stress due to the surface size effect. The capillary stress mechanism suggests that as the film is deposited, the microcrystalline lattice constant increases, but due to the constraints of the substrate, the lattice constant cannot continue to increase, resulting in compressive stress inside the film.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

This paper deals with the effects of UV curing on stress in PECVD deposited SiN thin films. The paper contains a number of interesting results. The presentation needs serious improvement before it can be considered for publication.

Introduction:

In the introduction the authors state that "Plasma-enhanced chemical vapor deposition (PECVD) is the preferred method for SiN deposition due to its lower processing temperature, which enables compatibility with temperature-sensitive substrates." This sentence needs to be revised as a deposition temperature of around 400°C (Ref 26) is already too high for deposition compatible with flexible substrates.

Results and discussion :

For the sake of clarity, the results will be presented in two parts: The first will be devoted to the first case in Figure 2 (Low tensile stress) and the second to the second case in Figure 2 (High tensile stress).

The procedure for normalizing the hydrogen content described in the following sentence is not very clear and I don't see the point: "To illustrate the magnitude of variation in H_N-H + H_Si-H and H_total rather than merely their relative content changes, each of these components is normalized by the sum of all three". For example, Figures 4a and 4b can be omitted as the information on the evolution of the different forms of bonded hydrogen (Si-H and N-H) is already included in Figures 5a and 5b. Similarly, Figure 7b and Figures 10a and 10b can be omitted. A figure (11b) showing the evolution of ΔH as a function of the number of cycles would be welcome in addition to the current Figure 11.

As the hydrogen content is measured ex situ, the authors should discuss the role that air oxidation could play in the evolution of stress in the SiN film layer.

Author Response

We sincerely thank the Reviews for taking efforts for timely review of the manuscript. We also thank the respected reviewers for their insightful observations and useful suggestions. The preciseness and brevity in the comments reflected the amount of efforts and time they have invested for reviewing the manuscript and we are sincerely grateful to them. We must admit that the inclusion of the suggestions in the revised manuscript has enriched the quality of the work.

Point by point response is provided below to earnestly clarify all the concerns and authors hope that it matches expectations of the respected reviewers. Changes are also underlined in the revised manuscript.

Reviewer # 1

Response: Thank you! We have made the corresponding modifications (see the point-to-point responses shown below).

Comment #R1.1: In the introduction the authors state that "Plasma-enhanced chemical vapor deposition (PECVD) is the preferred method for SiN deposition due to its lower processing temperature, which enables compatibility with temperature-sensitive substrates." This sentence needs to be revised as a deposition temperature of around 400°C (Ref 26) is already too high for deposition compatible with flexible substrates.

Response: Thank you for suggesting this improvement. Revised.

Modifications: Plasma-enhanced chemical vapor deposition (PECVD) is widely used for SiN deposition due to its relatively low processing temperature compared to other CVD techniques, which allows integration with a variety of substrates, although compatibility with highly temperature-sensitive or flexible substrates may still be limited.

Comment #R1.2: Results and discussion : For the sake of clarity, the results will be presented in two parts: The first will be devoted to the first case in Figure 2 (Low tensile stress) and the second to the second case in Figure 2 (High tensile stress).

Response: We thank the reviewer for the valuable suggestion. To enhance clarity, we have reorganized the "Results and Discussion" section to more explicitly separate the two cases illustrated in Figure 2—namely, the low-tensile-stress condition (prior to UV curing and layering) and the high-tensile-stress condition (after UV curing and layered deposition). While the overall structure of the section remains intact, we have added clear transitional statements and subtitles to distinguish the two regimes for better readability, as recommended.

Modifications: For clarity, the experimental results are discussed in two distinct regimes. The first focuses on low-tensile-stress conditions derived from standard PECVD deposition (Figure 2, upper process), and the second emphasizes enhanced tensile stress achieved through UV curing and layered deposition (Figure 2, lower process).

3.1 Low-tensile-stress regime: Effect of PECVD process parameters

3.2 Low-tensile-stress regime: Effect of UV Curing parameters

3.3 High-tensile-stress regime: Layered deposition and UV curing

Comment #R1.3: The procedure for normalizing the hydrogen content described in the following sentence is not very clear and I don't see the point: "To illustrate the magnitude of variation in HN-H + HSi-H and Htotal rather than merely their relative content changes, each of these components is normalized by the sum of all three". For example, Figures 4a and 4b can be omitted as the information on the evolution of the different forms of bonded hydrogen (Si-H and N-H) is already included in Figures 5a and 5b. Similarly, Figure 7b and Figures 10a and 10b can be omitted. A figure (11b) showing the evolution of ΔH as a function of the number of cycles would be welcome in addition to the current Figure 11.

Response: We appreciate the reviewer’s constructive feedback regarding the clarity and relevance of the hydrogen normalization procedure and the associated figures. In response, we have removed the sentence describing the normalization of hydrogen content, as well as Figures 4(a), 4(b), 7(b), and 10(a)(b), in accordance with the reviewer’s suggestion. These deletions help to streamline the presentation and avoid redundancy with the information already provided in Figures 5(a) and 5(b). In addition, we have added a new Figure 8(b), which shows the evolution of ΔH as a function of the number of deposition-curing cycles, as requested. The manuscript structure has been adjusted accordingly to maintain logical consistency and improve overall clarity.

Modifications:

Figure 8 (a) Variation of the N-H/Si-H ratio, and (b) Δ H concentration as a function of SiH₄ flow rate.. The plasma power is 175 W, UV curing power is 85 W, SiH₄ flow rate to 91 sccm, the SiN films thickness is 60 nm, and NH₃ flow rate to 880 sccm.

Comment #R1.4: As the hydrogen content is measured ex situ, the authors should discuss the role that air oxidation could play in the evolution of stress in the SiN film layer.

Response: Thank you for pointing this. We acknowledge that ex situ measurements may allow for limited surface oxidation of the SiN films, which could potentially influence both hydrogen content and stress behavior. To address this concern, we have added a discussion on the possible role of air oxidation in stress evolution in the revised manuscript (Section 3.2, paragraph 1). Although the short exposure time and ambient storage conditions in our study are unlikely to cause substantial oxidation, the potential for surface modification and its minor contribution to stress relaxation have now been explicitly noted.

Modifications: It should also be noted that, as the hydrogen content was measured ex situ, there exists a possibility of limited surface oxidation occurring upon air exposure prior to characterization. While the short ambient exposure time in this study minimizes such effects, minor oxidation at the film surface could potentially contribute to slight stress relaxation or variations in hydrogen bond concentrations. However, given the consistent measurement trends and the dominant influence of UV curing observed, the contribution of air oxidation is considered to be secondary. o further modulate the stress state of the as-deposited low-tensile-stress SiN films, UV curing was introduced as a post-treatment method. The following section explores how variations in UV curing parameters, particularly curing timed and UV power, influence hydrogen desorption behavior and tensile stress development.

---------------------------------------------------------------------------------------------------------

Thank you, once again!

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The topic is potentially interesting and relevant for Coatings. However the manuscript is very badly organized and difficult to follow. It is thus difficult to assess its overall scientific soundness.

The presentation of the data is often unclear and several contradicting statements have been reported in the text (see below)

The lack of attention is evident even from the fact that in some figure captions the size of the characters are not the same (see e.g. the caption of fig. 4 and of fig. 7).

A detailed list of comments is appended below.

1) Explain what is the species corresponding to the term 6.1 10 ^22 in the denominator of eq. 1 and of eq. 3.
2) I think that or in the paper or in the supporting information sample IR spectra before and after UV curing or selected values of the flows must be shown.
3) I thin that also the absolute and not only the relative change of the N_H and Si-H intensities must be shown in the manuscript.
4) The authors write that “The primary purpose of UV curing was to break the Si-H and N-H bonds present in the film, facilitating the outgassing of hydrogen and the subsequent formation of Si-N-Si crosslinks, which are responsible for the increase in tensile stress.”
The intensities recorded in Fig.4 b (post UV curing) corresponds to another subsequent exposure to plasma after UV curing or just to the intensities recorded after UV curing without any further exposure to plasma?
In the former case how does the relative intensity change immediately after UV curing ? In the latter case how is it possible that the total number of H bonds increases after UV curing as shown in Fig. 4?

5) The sentence on line 232: “Although the relative intensities of 𝐻𝑁−𝐻 and 𝐻𝑆𝑖−𝐻 remain largely unaffected by the SiH₄ flow rate before and after UV curing (as shown in Figure 4), the total H retention within the film decreases significantly after UV curing” clearly contradicts the statement reported before at line 205-208 where it is stated that:
“It can be seen that when the SiH₄ flow rate is less than approximately 93 sccm, the N–H/Si–H ratio is greater than 1:1. As the SiH₄ flow rate increases, the N–H/Si–H ratio gradually approaches 1:1. Furthermore, after UV curing, the N–H/Si–H ratio shifts further away from 1:1 compared to the SiN films without UV treatment.”

The presentation of the data should be largely re-organized showing
a) Sample IR spectra before and after UV curing.
b) Showing the absolute intensity of N-H and Si-H as a function of flow before and after curing.
c) Only after having shown a and b it might make sense to show the ratios reported in fig.4 and in fig.5
6) At line 216 it is written that: “A higher ammonia concentration promotes the formation of Si-N bonds, which are 216 stronger and more stable than Si-H bonds [18,29,31]”Is there any spectroscopic evidence of Si-N bonds in the IR spectra ?
7) At line 223 it is written that: “Moreover, the results indicate that when the concentration of Si-H bonds in the film is significantly higher than that of N-H bonds, as shown in Figure 5(a), the N-H/Si-H ratio continues to decrease after UV curing. This suggests that the thermally generated H atoms from Si-H bonds may also interact with dangling N bonds, forming additional N-H bonds.”
As already written above, the absolute N-H and Si-H intensities before and after UV curing should be shown so that the formation of additional N-H bond should be immediately evident and not tentatively deduced from the ratio of the intensities.
8) I do not understand fig. 6b. How there can be different points corresponding to different UV treatment time before irradiation ? ! There should be one value pre-UV and then different one after UV. It would be much simpler to show the ratio as a function of UV time starting from 0 (which would obviously correspond to the pre-UV point).
9) At line 259 it is written that: “This suggests that UV curing preferentially disrupts N-H bonds, as illustrated in the inset of Figure 6(b).” At line 212 it is written “UV curing leads to a further reduction of Si-H bonds, significantly increasing the N-H/Si-H ratio”
By the way, how is it possible that UV curing preferentially disrupts N-H bonds if they are more strong than Si-H bonds ? At line 213 it is written indeed that “Studies have shown that in the ground state, the bond energies of N-H and Si-H bonds are 4.8 eV and 4.0 eV”

10) Why the error bars in fig. 7 a are much larger for the first three points than for the other ones? With the error bars for the first three points the statement reported at line 263 " post-stress initially decreases 263
slightly"is not supported by the data.

In conclusion I think that the presentation of the data must be largely re-organized and the readability of the manuscript substantially improved.

Only then it will be possible to properly assess the manuscript.

Author Response

Reviewer # 2

The topic is potentially interesting and relevant for Coatings. However the manuscript is very badly organized and difficult to follow. It is thus difficult to assess its overall scientific soundness. The presentation of the data is often unclear and several contradicting statements have been reported in the text (see below)

Response: We sincerely thank the reviewer for the critical evaluation and constructive feedback. In response to the concerns regarding the organization and clarity of the manuscript, we have thoroughly revised the entire structure to improve logical flow and readability. In addition, we carefully reviewed the entire manuscript for internal consistency and have corrected all previously identified contradictions in both text and data interpretation. We hope that the revised version presents a more coherent and scientifically sound discussion of our results.

Comment #R2.1: The lack of attention is evident even from the fact that in some figure captions the size of the characters are not the same (see e.g. the caption of fig. 4 and of fig. 7).

Response: We thank the reviewer for the observation regarding inconsistencies in the figure caption formatting. In response to concerns raised by other reviewers regarding data presentation, Figures 4 and 7 have been removed from the revised manuscript. We believe this adjustment improves the overall clarity and focus of the data and also addresses the formatting issue noted here.

Comment #R2.2: 1) Explain what is the species corresponding to the term 6.1 10 ^22 in the denominator of eq. 1 and of eq. 3.

Response: We thank the reviewer for pointing out the need to clarify the constant term in the denominator of Equations 1 and 3. The term 6.1 × 10²² is a proportionality factor used to convert integrated absorbance values from FTIR measurements into hydrogen atomic concentration (at.%), based on the Lanford–Rand method (William A. Lanford and M. J. Rand, J. Appl. Phys., 49, 1978, 2473–2477). This method establishes a correlation between FTIR absorbance and hydrogen concentration by calibrating with absolute hydrogen content measured using NRA-RBS techniques. In this model, the species corresponding to this factor are hydrogen atoms bonded in N–H and Si–H configurations within the SiNx:H matrix. It is widely used in hydrogen quantification of PECVD SiNx:H films, as also discussed in technical documentation by Thermo Fisher Scientific (Application Note AN53081, 2019), and reflects a commonly accepted calibration for films measured by FTIR. A clarification of this constant and its origin has been added to the manuscript following Equations 1 and 3.

Modifications: The constant 6.1 × 10²² in the denominator is a proportionality factor derived from the Lanford–Rand method, which correlates FTIR absorbance with hydrogen atomic concentration. It has been validated against RBS/NRA measurements, as referenced in Thermo Fisher Application Note AN53081 (2019).

Comment #R2.3: 2) I think that or in the paper or in the supporting information sample IR spectra before and after UV curing or selected values of the flows must be shown.

Response: Thank you for your valuable feedback. In response, we have added representative FTIR spectra before and after UV curing to the revised manuscript to illustrate the bonding configuration changes. In addition, Figure 8(b) has been included to show the variation in absolute hydrogen content as a function of UV curing power, providing quantitative insight into hydrogen desorption. These additions help clarify the effects of UV treatment on the chemical structure of the SiN films and address the reviewer’s request for more direct experimental evidence.

Modifications: Figure 8(b) presents the variation in hydrogen content within the SiN films as a function of the number of layered deposition cycles. It can be observed that, in the absence of UV treatment, the hydrogen concentration remains nearly constant regardless of the number of deposition layers. However, when UV curing is combined with the layered deposition process, a substantial reduction in hydrogen content is evident, reaching a minimum at three deposition-curing cycles. Further increasing the number of cycles does not lead to additional hydrogen removal, which is consistent with the observed trend in tensile stress. Correspondingly, Figure 9 shows the FTIR spectral evolution under various conditions: (a) varying SiH₄ flow rates, (b) different UV curing times, (c) UV curing power levels, and (d) layered deposition with UV curing. The H ratio values were directly derived from the integrated intensities of the corresponding FTIR absorption peaks (According to Equations (1)–(4)).

Figure 9. FTIR spectra under different conditions: (a) variation in SiH₄ flow rate; (b) UV curing time; (c) UV curing power; and (d) layered deposition with UV curing.

Comment #R2.4: 4) The authors write that “The primary purpose of UV curing was to break the Si-H and N-H bonds present in the film, facilitating the outgassing of hydrogen and the subsequent formation of Si-N-Si crosslinks, which are responsible for the increase in tensile stress.” The intensities recorded in Fig.4 b (post UV curing) corresponds to another subsequent exposure to plasma after UV curing or just to the intensities recorded after UV curing without any further exposure to plasma? In the former case how does the relative intensity change immediately after UV curing ? In the latter case how is it possible that the total number of H bonds increases after UV curing as shown in Fig. 4?

Response: We thank the reviewer for the insightful question regarding the hydrogen bonding intensities presented in Figure 4(b). As noted, the data were obtained immediately after UV curing, without any subsequent plasma exposure. The observed increase in total hydrogen bonds at higher SiH₄ flow rates is primarily attributed to the increased incorporation of hydrogen species during deposition, which results from the elevated SiH₄ concentration. This effect reflects the initial chemical composition of the film rather than the formation of new hydrogen bonds during UV treatment.

In response to related suggestions from other reviewers concerning data presentation clarity, we have removed Figure 4 from the revised manuscript to avoid potential misinterpretation. The discussion of hydrogen bonding and UV curing effects has been consolidated and clarified accordingly in the text.

Modifications: Received

Comment #R2.5: 5) The sentence on line 232: “Although the relative intensities of ??−? and ???−? remain largely unaffected by the SiH₄ flow rate before and after UV curing (as shown in Figure 4), the total H retention within the film decreases significantly after UV curing” clearly contradicts the statement reported before at line 205-208 where it is stated that:
“It can be seen that when the SiH₄ flow rate is less than approximately 93 sccm, the N–H/Si–H ratio is greater than 1:1. As the SiH₄ flow rate increases, the N–H/Si–H ratio gradually approaches 1:1. Furthermore, after UV curing, the N–H/Si–H ratio shifts further away from 1:1 compared to the SiN films without UV treatment.”The presentation of the data should be largely re-organized showing. a) Sample IR spectra before and after UV curing.
b) Showing the absolute intensity of N-H and Si-H as a function of flow before and after curing. c) Only after having shown a and b it might make sense to show the ratios reported in fig.4 and in fig.5

Response: Thank you for your valuable feedback. As mentioned in our response to Comment 2.4, Figure 4 and the corresponding text have been removed from the revised manuscript. This decision was made to address both the potential contradiction noted by the reviewer and similar concerns raised by other reviewers regarding the interpretation of relative hydrogen bond intensities. In response to the reviewer’s recommendation, we have added representative FTIR spectra before and after UV curing, which are now presented in Figure 9. These spectra illustrate the characteristic changes in N–H and Si–H bond absorption bands as a result of UV treatment.

Modifications:

Figure 9. FTIR spectra under different conditions: (a) variation in SiH₄ flow rate; (b) UV curing time; (c) UV curing power; and (d) layered deposition with UV curing.

Comment #R2.6: 6) At line 216 it is written that: “A higher ammonia concentration promotes the formation of Si-N bonds, which are 216 stronger and more stable than Si-H bonds [18,29,31]”Is there any spectroscopic evidence of Si-N bonds in the IR spectra?

Response: We thank the reviewer for the important question. Yes, spectroscopic evidence of Si–N bond formation is present in our FTIR data. We have included representative FTIR spectra (see Figure) that clearly show the characteristic Si–N stretching absorption band around ~830–890 cm⁻¹. The intensity of this peak increases progressively with UV curing time, power, and the number of deposition-curing cycles, indicating an enhancement in Si–N bond formation during post-treatment.

Figure. FTIR spectra under different conditions: (a) UV curing time and (b) UV curing power.

Comment #R2.7: 7) At line 223 it is written that: “Moreover, the results indicate that when the concentration of Si-H bonds in the film is significantly higher than that of N-H bonds, as shown in Figure 5(a), the N-H/Si-H ratio continues to decrease after UV curing. This suggests that the thermally generated H atoms from Si-H bonds may also interact with dangling N bonds, forming additional N-H bonds.” As already written above, the absolute N-H and Si-H intensities before and after UV curing should be shown so that the formation of additional N-H bond should be immediately evident and not tentatively deduced from the ratio of the intensities.

Response: Thank you for your valuable feedback. While direct FTIR peak intensities of N–H and Si–H were not plotted individually, we would like to clarify that Figure 5(b) already presents the total hydrogen concentration (H_total) after UV curing as a function of SiH₄ flow rate. When this total hydrogen concentration is combined with the N–H/Si–H ratio (H_ratio) shown in Figure 5(a), it is straightforward to derive the absolute concentrations of N–H and Si–H bonds. This approach provides a reliable estimation of how each bonding species evolves and supports our discussion regarding potential bond reconfiguration during UV curing.

Comment #R2.8: 8) I do not understand fig. 6b. How there can be different points corresponding to different UV treatment time before irradiation? There should be one value pre-UV and then different one after UV. It would be much simpler to show the ratio as a function of UV time starting from 0 (which would obviously correspond to the pre-UV point).

Response: Thank you for your valuable feedback. We agree that displaying multiple pre-UV data points may cause confusion, as there should be a single reference state prior to UV treatment. In response, we have revised Figure 6(b) so that the pre-UV data is now represented as a single reference line.

Modifications:

Figure 5 (a) Post-stress and loss-stress as a function of UV curing time, and (b) variation of the H ratio as a function of UV curing time. The plasma power is 175 W, SiH₄ flow rate to 91 sccm, NH₃ flow rate to 880 sccm, and UV power to 85 W.

Comment #R2.9: 9) At line 259 it is written that: “This suggests that UV curing preferentially disrupts N-H bonds, as illustrated in the inset of Figure 6(b).” At line 212 it is written “UV curing leads to a further reduction of Si-H bonds, significantly increasing the N-H/Si-H ratio”. By the way, how is it possible that UV curing preferentially disrupts N-H bonds if they are more strong than Si-H bonds? At line 213 it is written indeed that “Studies have shown that in the ground state, the bond energies of N-H and Si-H bonds are 4.8 eV and 4.0 eV”

Response: We thank the reviewer for this insightful observation. It is indeed correct that, in the ground state, N–H bonds are stronger than Si–H bonds, with reported bond energies of 4.8 eV and 4.0 eV, respectively. We fully agree that this would typically suggest that Si–H bonds should dissociate more readily under thermal or photonic energy input.

However, modeling and experimental studies (see: V. Zubkov, M. Balseanu, and L. Q. Xia, Materials Research Society, 2006, 910: 0910-A19-04) have shown that under UV excitation, the situation differs. In the excited state, N–H bonds in excited states may become more prone to dissociation due to the formation of different bonding configurations, such as ring-like or chain-like structures.

Furthermore, the UV curing process involves non-equilibrium dehydrogenation dynamics, where interstitial hydrogen atoms generated from Si–H or other sources may collide with N dangling bonds, contributing to secondary bond breaking or reconfiguration. This could explain the relatively faster decrease in N–H bond concentration in some experimental cases, as reflected in our FTIR data. We have revised the discussion in the manuscript to clarify this mechanism and avoid apparent contradictions in interpretation.

Modifications: Before UV curing, the N-H/Si-H ratio in SiN films is approximately 1:1. As the UV curing time increases from 100 s to 600 s, the N-H/Si-H ratio remains nearly constant at 0.9:1, excluding measurement errors. This suggests that UV curing preferentially disrupts N-H bonds, as illustrated in the inset of Figure 5(b). Studies have shown that in the ground state, the bond energies of N–H and Si–H bonds are approximately 4.8 eV and 4.0 eV, respectively [28], suggesting that Si–H bonds should be more easily dissociated thermally. However, under ultraviolet irradiation, the dissociation dynamics are influenced by photo-excitation and the resulting non-equilibrium processes. Overall, UV curing time has minimal impact on the N-H/Si-H ratio. The effect of UV power on SiN films stress and the N-H/Si-H ratio has been investigated.

Comment #R2.10: 10) Why the error bars in fig. 7 a are much larger for the first three points than for the other ones? With the error bars for the first three points the statement reported at line 263 " post-stress initially decreases 263 slightly" is not supported by the data.

Response: We thank the reviewer for pointing out the inconsistency in the error bars shown in Figure 7(a). In the original version, the larger error bars associated with the first three data points were due to limited sampling and higher variability in the initial experimental batches. To address this issue and improve the statistical reliability of the results, we have conducted additional measurements under the same conditions to increase the sample size for those data points. In the revised version of the manuscript, the updated Figure 7(a) reflects reduced error bars based on the expanded dataset.

To reduce statistical uncertainty, we applied the following approach:

Five independent wafers were measured under each UV power condition;
Stress values were calculated from 49 measurement points per wafer using a wafer curvature tool;
The mean value and standard deviation were computed and plotted as the central value and error bar, respectively.

These measures significantly improved the consistency of the data, and we have revised the relevant description in the manuscript accordingly.

Modifications:

Figure 6 (a) Post-stress and loss-stress as a function of UV curing power, and (b) variation of the N-H/Si-H ratio as a function of UV power. The plasma power is 175 W, SiH₄ flow rate to 91 sccm, and NH₃ flow rate to 880 sccm.

---------------------------------------------------------------------------------------------------------

Thank you, once again!

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have completed the required revisions, so I suggest accepting the article for publication.

Author Response

We sincerely thank the reviewer for the positive evaluation and recommendation for acceptance. We greatly appreciate the valuable feedback provided throughout the review process, which helped improve the quality of our manuscript.

Reviewer 3 Report

Comments and Suggestions for Authors

A final simplification, which I mentioned in my previous report, is to replace equation (3) with H_total = H_Si-H + H_N-H

Author Response

Reviewer # 1

Comment #R1.1 A final simplification, which I mentioned in my previous report, is to replace equation (3) with H_total = H_Si-H + H_N-H

Response: Thank you for suggesting this improvement. Revised.

Modifications: H_total = H_Si-H + H_{N-H （3）}

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

Despite the revision the quality of the presentation remains quite poor.

The discussion is difficult to follow and in my opinion contraddictory.

The topic remains potentially interesting but it is impossible to assess whether the conclusions are supported by experimental results. The authors added a figure with infrared spectra but without showing how the integrated adsorbances of both N-H and Si-H peak change with flux, power, time lenght of UV curing and number of cycles it is impossible to assess whether the conclusions are supported by the results.

Fig. 9 should be presented immediately after fig. 3 and a new figure showing adsorbances should be shown as well.

In conclusion the manuscript must be substantially revised improving data presentation before it can be be given further consideration.

A few more specific comments highlighting the most puzzling concerns are reported below.

1) If the H ratio is defined as the ratio of the N-H to the Si-H adsorbance it is apparent from fig,.9 that for fluxes lower than 93 scm (brown and orange traces) the intensity of the N-H peak is lower than the one of the Si-H peak resulting in a H-ratio lower than 1 and not greater than 1 as claimed in the text and even clearly in fig. 4.

2) According to fig. 9b with increasing exposure to UV the intensity of the Si-H peak decreases so that the ratio between N-H and Si-H (defined as H ratio increases as stated in the text). Fig. 4 a is thus confusing: what is represented here? What do the three colours in Fig 4a correspond to ? In the text at line 215 it is stated that the H ratio continues to decrease after UV curing thus contraddicting both fig 9 and the discussions at lines 202-203.

By the way fig 9 (from which all these informations are obtained !) should be placed before fig. 4 and not at the end of the manuscript.

3) After showing that the flow rate affects the N-H/Si-H ratio how is it possible to write that "the relative intensities of H in N-H and H in Si-H remain largely unaffected by the SiH₄ flow rate before and after UV curing" ?

As suggested above, the presentation of the data is still totally unsatisfactory: fig. 9 should come before showing how the H ratio depends on the different parametes, then a new figure showing the intensity of the N- and Si-H peak as a function of flow, length of exposure, power and layer numbers should be introduced and only then it will make sense to go on with the subsequent figures and their discussion !

4) What are the peaks in the high wavenumber region in fig. 9? They are present in b and c but not in a. Morever they appear as increased adsorbance in b and c and decreased adsorbance in d. This should be discussed in the paper.

Author Response

Reviewer # 2

Despite the revision the quality of the presentation remains quite poor.

The discussion is difficult to follow and in my opinion contraddictory.

Fig. 9 should be presented immediately after fig. 3 and a new figure showing adsorbances should be shown as well.

In conclusion the manuscript must be substantially revised improving data presentation before it can be be given further consideration.

A few more specific comments highlighting the most puzzling concerns are reported below.

Response: Thank you for your valuable feedback. We have made significant revisions based on your comments. Specifically, we have clarified the discussion to address inconsistencies and improve clarity. Additionally, we have added new data showing the changes in the integrated absorbances of the N-H and Si-H peaks under varying conditions (e.g., flow, power, UV curing time, and number of cycles), which directly support our conclusions. Following your suggestion, we have placed Fig. 9 immediately after Fig. 3 and included a new figure to clearly display the absorbance changes. We believe these revisions have greatly improved the manuscript and hope that the updated version meets your expectations.

Comment #R2.1: 1) If the H ratio is defined as the ratio of the N-H to the Si-H adsorbance it is apparent from fig,.9 that for fluxes lower than 93 sccm (brown and orange traces) the intensity of the N-H peak is lower than the one of the Si-H peak resulting in a H-ratio lower than 1 and not greater than 1 as claimed in the text and even clearly in fig. 4.

Response: We thank the reviewer for pointing this out. We acknowledge that this is a labeling error (the curve labels and corresponding colors were reversed). As the SiH₄ flow increases, the intensity of the Si-H peak increases, leading to a decrease in the H-ratio. We have corrected this error in Figure 4(a).

Modifications:

Figure 4 (a) FTIR spectra and (b) Variation of and with SiH₄ flow rate. The plasma power is 175 W, NH₃ flow rate to 880 sccm, and UV power to 85 W.

Comment #R2.2: 2) According to fig. 9b with increasing exposure to UV the intensity of the Si-H peak decreases so that the ratio between N-H and Si-H (defined as H ratio increases as stated in the text). Fig. 4 a is thus confusing: what is represented here? What do the three colours in Fig 4a correspond to? In the text at line 215 it is stated that the H ratio continues to decrease after UV curing thus contraddicting both fig 9 and the discussions at lines 202-203. By the way fig 9 (from which all these informations are obtained!) should be placed before fig. 4 and not at the end of the manuscript.

Response: We appreciate the reviewer’s suggestion. Fig. 4a represents the relationship between SiH₄ flow and the H-ratio, with the three curves corresponding to: the blue line—pre-UV, the black line—post-UV, and the green line—a reference line (which was poorly labeled, and we have corrected this). However, since Fig. 9b corresponds to the effect of UV treatment time, it should be related to the changes shown in Fig. 5b, rather than Fig. 4a. The position of Fig. 9 has been moved before Fig. 4.

Modifications:

Figure 5 (a) Variation of the N-H/Si-H ratio as a function of SiH₄ flow rate before and after UV curing, and (b) Δ H concentration as a function of SiH₄ flow rate. The plasma power is 175 W, NH₃ flow rate to 880 sccm, and UV power to 85 W.

Comment #R2.3: 3) After showing that the flow rate affects the N-H/Si-H ratio how is it possible to write that "the relative intensities of H in N-H and H in Si-H remain largely unaffected by the SiH₄ flow rate before and after UV curing"? As suggested above, the presentation of the data is still totally unsatisfactory: fig. 9 should come before showing how the H ratio depends on the different parametes, then a new figure showing the intensity of the N- and Si-H peak as a function of flow, length of exposure, power and layer numbers should be introduced and only then it will make sense to go on with the subsequent figures and their discussion!

Response: We appreciate your constructive suggestions. The sentence in question indeed contains some issues, and we have revised it accordingly. Furthermore, we have reorganized and added the data presentation as per your recommendation. The revised figures can be found in Figs. 4, 6, 8, and 11.

Modifications:

Furthermore, the total H retention within the film decreases significantly after UV curing.
Figure 4(a) shows the variation of the Si-H and N-H peaks in the FTIR spectra as a function of SiH₄ flow rate, while Figure 4(b) illustrates the changes in and with varying SiH₄ It is evident that as the SiH₄ flow rate increases, the N-H bond decreases from 9.5% to 4.7%, while the Si-H bond increases from 4.9% to 11.3%. Based on equations (1)-(4), the change in the N-H/Si-H ratio ( ) was further calculated.

Figure 4 (a) FTIR spectra and (b) Variation of and with SiH₄ flow rate. The plasma power is 175 W, NH₃ flow rate to 880 sccm, and UV power to 85 W.

Figure 6(a) shows the variation of the Si-H and N-H peaks in the FTIR spectra as a function of UV treatment time, while Figure 6(b) presents the changes in and with increasing UV treatment time. It can be observed that as the UV treatment time increases from 100s to 600s, both the N-H and Si-H bonds remain relatively stable, at approximately 8% and 9%, respectively, with a slight decreasing trend.

Figure 6 (a) FTIR spectra, and (b) Variation of and with UV treatment time. The plasma power is 175 W, SiH₄ flow rate to 91 sccm, NH₃ flow rate to 880 sccm, and UV power to 85 W.

Figure 8(a) presents the variation of the Si-H and N-H peaks in the FTIR spectra as a function of UV power, while Figure 8(b) illustrates the changes in and with increasing UV power. It can be observed that as the UV power increases from 55 W to 70 W, both the N-H and Si-H bonds remain relatively stable, at approximately 8% and 9%, respectively. When the UV power increases to 75 W, the N-H bond decreases significantly, while the Si-H bond increases. Based on the results averaged over three measurements, when the UV power exceeds 75 W, the concentrations of Si-H and N-H stabilize and no further significant changes are observed.

Figure 8 (a) FTIR spectra, and (b) Variation of and with UV power. The plasma power is 175 W, SiH₄ flow rate to 91 sccm, and NH₃ flow rate to 880 sccm.

Figure 11(a) shows the variation of the Si-H and N-H peaks in the FTIR spectra as a function of deposition cycles, while Figure 11(b) presents the changes in and with increasing deposition cycles. It can be observed that as the number of deposition cycles increases from 1 to 6, both the N-H and Si-H bonds decrease in parallel, from approximately 8% to 5%.

Figure 11 (a) FTIR spectra, and (b) Variation of and with the number of deposition layers. The plasma power is 175 W, UV curing power is 85 W, SiH₄ flow rate to 91 sccm, the SiN films thickness is 60 nm, and NH₃ flow rate to 880 sccm.

Comment #R2.4: 4) What are the peaks in the high wavenumber region in fig. 9? They are present in b and c but not in a. Morever they appear as increased adsorbance in b and c and decreased adsorbance in d. This should be discussed in the paper.

Response: The peaks observed in the high wavenumber region (~3600–3800 cm⁻¹) are influenced by atmospheric interference. Since the experiment was not completed within a single day, environmental factors contributed to variations in this region. However, as there are no characteristic species in this wavenumber range, these peaks have been classified as noise. We have clarified this point in the manuscript.

Modifications: It is worth noting that, as observed in Figs. 4a, 6a, 8a, and 11a, different forms of noise signals appear in the ~3600-3800 cm^-1 range. This is primarily attributed to variations in the background atmosphere during each FTIR measurement. However, since no characteristic species are formed in this wavenumber range, the impact on the results is considered negligible.

---------------------------------------------------------------------------------------------------------

Thank you, once again!

Author Response File: Author Response.pdf

Round 3

Reviewer 4 Report

Comments and Suggestions for Authors

I appreciate the efforts of the authors to revise the manuscript as I suggested in my last report. They added the intensity of N-H and Si-H feature as suggested. This allowed me to check the consistency of the analysis and the conclusions in larger detail. Unfortunately this more transparent analysis indicates that at least in the case of Fig. 6 and of Fig. 11 the analysis is not correct. Consequently several conclusions are not supported by the data.

Specifically:

1) I am not convinced about the correctness of this analysis of fig. 6.
The baseline for the peaks around 2300 cm^-1 (Si-H) are the same but the intensity after 100 s UV (green trace) is clearly higher than the one after 500 s UV (brown trace). How can the si-H intensity be nearly constant as reported fig 6 b? !

2) Fig. 7. It is difficult to distinguish traces. I suggest to change the colours avoiding the use of he same colours for different powers. I agree that N-H intensity decreases but it is not as obvious that the Si-H increases.

3) The statement at lines 325-326 confirms my doubts about the analysis of fig. 6. The authors say here that there is a significant reduction of the Si-H and N-H peaks after UV curing while fig. 6 indicates they are constant.
As already stated in my comment above the analysis of fig. 6 is not correct, and with it the conclusions of fig. 7 and what follows.

4) According to fig 11b the N-H intensity should reduce by almost a factor of two with cycle numbers.This is clearly not the case (see pdf enlarging fig. 11 a,). The areas under the green and blue traces are nearly the same ! Also the area under the brown trace is not evidently lower. So in my opinion the areas shown in fig. 11b are not correct and the conclusions are nor supported by the data.

In conclusion I think the analysis of IR spectra presents important inconsistencies. In my opinion the conclusions are not properly supported by the data. I then can only recommend to reject the manuscript.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

We would like to sincerely thank you for your valuable and constructive feedback. We greatly appreciate your efforts in reviewing our revised manuscript and pointing out the important issues with the data analysis. We acknowledge that the previous data processing was not sufficient to support the conclusions, and we recognize that time constraints during the previous revision limited our ability to address the FTIR data issues properly.

Upon reviewing your comments in detail, we identified several critical points that required attention. The main issues include the following:

Baseline Calibration: The initial FTIR data was not baseline corrected (i.e., set to zero and aligned as a straight line), which led to the data appearing less consistent and clear visually. This likely contributed to the issues you raised. To address this, we have now performed baseline calibration on all the raw data (as shown in Figs. 4, 6, 8, and 11). Since the H-content was determined based on the integration of the Si-H and N-H peaks, we recalculated the H-content and H-ratio after baseline correction. This revision has led to some minor differences in the data (e.g., Figs. 5(b), 7(b), 9(b), and 12(a)), but these are only slight variations and do not affect the overall conclusions.
Effect of Film Thickness: The intensity of FTIR peaks is significantly influenced by the thickness of the SiN films, which can lead to some FTIR data points appearing anomalous and affecting the analysis. However, the final H-content values are relative results normalized to the film thickness, so this does not impact the experimental conclusions. Nevertheless, to improve the consistency of the FTIR data, we repeated measurements for some of the anomalous data points and re-plotted the FTIR results accordingly.

In addition to addressing the points you raised, we have carefully reviewed the entire manuscript once again. We have also included the raw FTIR data for your review, allowing for a more transparent evaluation of our results.

We hope that these comprehensive revisions will help clarify the issues and that the changes will allow you to reassess our manuscript more favorably. Once again, we would like to thank you for your thoughtful and constructive suggestions, which have greatly contributed to improving the quality of this work.

We look forward to your feedback and hope for a positive outcome.

Sincerely,

Dayu Zhou

Reviewer # 2

Comment #R2.1: 1) I am not convinced about the correctness of this analysis of fig. 6.
The baseline for the peaks around 2300 cm^-1 (Si-H) are the same but the intensity after 100 s UV (green trace) is clearly higher than the one after 500 s UV (brown trace). How can the si-H intensity be nearly constant as reported fig 6 b? !

Response: Thank you for your insightful comment. As mentioned in our previous response, we have conducted a thorough revision of the manuscript. In addition to addressing the issues you raised, we have also included the FTIR results for the Pre-UV condition in Fig. 4, as a point of comparison. This should provide further clarity on the baseline and intensity variations observed in the Si-H peaks after UV treatment. We hope that these updates help clarify the concerns and allow for a more accurate assessment of the data.

Modifications:

Figure 4 (a) Pre-UV FTIR spectra, (b) Post-UV FTIR spectra, and (c) Variation of and with SiH₄ flow rate. The plasma power is 175 W, the NH₃ flow rate is 880 sccm, and UV power is 85 W.

Figures 4(a) and (b) show the variation in the Si-H and N-H peaks in the FTIR spectra as a function of SiH₄ flow rate, while Figure 4(c) illustrates the changes in and with varying SiH₄ flow. A comparison between Figure 4(a) and Figure 4(b) reveals a significant decrease in both Si-H and N-H concentrations following UV curing treatment. As the SiH₄ flow rate increases, the Si-H bond concentration gradually increases, while the N-H bond concentration decreases. Specifically, as shown in Figure 4(c), before UV treatment, as the SiH₄ flow rate increases from 85 sccm to 107 sccm, the N-H bond concentration decreases from ~15% to 11%, while the Si-H bond concentration in the SiN:H film increases from 11.5% to 17%. After UV curing treatment, with increasing SiH₄ flow, the N-H bond concentration in the SiN film decreases from 9.5% to 5.2%, while the Si-H bond concentration increases from 5% to 12.5%. Based on equations (1)-(4), the change in the N-H/Si-H ratio ( ) was subsequently calculated.

Figure 6 (a) Pre-UV and Post-UV FTIR spectra, and (b) Variation of and with UV treatment time. The plasma power is 175 W, the SiH₄ flow rate is 92.5 sccm, the NH₃ flow rate is 880 sccm, and the UV power is 85 W.

Figure 6(a) shows the variation of the Si-H and N-H peaks in the FTIR spectra as a function of UV treatment time, while Figure 6(b) presents the changes in and with increasing UV treatment time. As shown in Figure 6, before UV treatment, both N-H and Si-H bond concentrations are approximately 13.8%, with an of 1:1. During the increase in UV treatment time from 100 s to 400 s, both N-H and Si-H bond concentrations exhibit a slight decrease. However, when the UV treatment time reaches 600s, the concentrations of N-H and Si-H bonds in the SiN film stabilize, at approximately 8% and 9%, respectively.

Comment #R2.2: 2) Fig. 7. It is difficult to distinguish traces. I suggest to change the colours avoiding the use of he same colours for different powers. I agree that N-H intensity decreases but it is not as obvious that the Si-H increases.

Response: Thank you for your valuable suggestion. I believe you may be referring to Fig. 8 rather than Fig. 7. To improve the clarity of the data presentation, we have not only revised Fig. 8 but also optimized all of the figures, with a particular focus on the FTIR data (see Figs. 4, 6, 8, and 11). In response to your comment, we have modified the color scheme and plotting style to avoid using the same colors for different powers, making the traces easier to distinguish.

Modifications:

Figure 8 (a) Pre-UV and Post-UV FTIR spectra, and (b) Variation of and with UV power. The plasma power is 175 W, SiH₄ flow rate is 91 sccm, and NH₃ flow rate is 880 sccm.

The effect of UV curing power on the stress of SiN films and the N-H/Si-H ratio has been investigated. Figure 8(a) shows the variation of the Si-H and N-H peaks in the FTIR spectra as a function of UV curing power, while Figure 8(b) illustrates the changes in and with increasing UV power. It can be observed that as the UV curing power increases from 55 W to 70 W, both the N-H and Si-H bond concentrations remain relatively stable, at approximately 8% and 9%, respectively. When the UV power increases to 75 W, the N-H bond concentration decreases significantly, while the Si-H bond concentration increases. Based on the results averaged over three measurements, when the UV power exceeds 75 W, the concentrations of Si-H and N-H stabilize, and no further significant changes are observed.

Comment #R2.3: 3) The statement at lines 325-326 confirms my doubts about the analysis of fig. 6. The authors say here that there is a significant reduction of the Si-H and N-H peaks after UV curing while fig. 6 indicates they are constant.

As already stated in my comment above the analysis of fig. 6 is not correct, and with it the conclusions of fig. 7 and what follows.

Response: Thank you for your insightful comment. To address your concern, we have added the FTIR results for each condition before UV curing for comparative analysis (see Figs. 4, 6, 8, and 11). Upon reviewing these updated results, it is evident that both the Si-H and N-H bond concentrations show a significant decrease after UV curing, as indicated by the new data.

Comment #R2.4: 4) According to fig 11b the N-H intensity should reduce by almost a factor of two with cycle numbers. This is clearly not the case (see pdf enlarging fig. 11 a,). The areas under the green and blue traces are nearly the same ! Also the area under the brown trace is not evidently lower. So in my opinion the areas shown in fig. 11b are not correct and the conclusions are nor supported by the data.

Response: Thank you for your thoughtful comment. As you pointed out, the baseline calibration and plotting style issues previously mentioned led to difficulties in data analysis, which affected the accuracy of the results. In response to this, we have carefully revised Fig. 11 to address these concerns. The updated figure now presents clearer data, and we have attached the raw data for your review to ensure transparency and allow for a more detailed evaluation. Additionally, in Fig. 11a, we have included the FTIR results for the Pre-UV condition to ensure the completeness of the data.

We hope these revisions resolve the issues you raised and provide a more accurate representation of the data.

Modifications:

Figure 11 (a) Pre-UV FTIR spectra, (b) Post-UV FTIR spectra, and (c) Variation of and with the number of deposition layers. The plasma power is 175 W, UV curing power is 85 W, the SiH₄ flow rate to 92.5 sccm, the SiN films thickness is 60 nm, and NH₃ flow rate is 880 sccm.

Figures 11(a) and (b) show the variation in the Si-H and N-H peaks in the FTIR spectra as a function of deposition cycles, while Figure 11(c) illustrates the changes in and with increasing deposition cycles. As shown in Figure 11(a), before UV treatment, the concentrations of Si-H and N-H bonds in the SiN film remain largely unchanged with an increase in the number of deposition cycles, exhibiting only a slight decreasing trend. As illustrated in Figure 11(c), when the number of deposition cycles increases from 1 to 6, both Si-H and N-H bond concentrations decrease from ~13.5% to 12.5%. When both layer-by-layer deposition and UV curing are applied, the concentrations of Si-H and N-H bonds in the SiN film continue to decrease. As shown in Figure 11(c), when the number of deposition cycles is less than 3, the concentration of N-H bonds is slightly lower than that of Si-H bonds. However, when the number of deposition cycles exceeds 3, the concentration of N-H bonds becomes slightly higher than that of Si-H bonds. This suggests that, even with a thinner film, further UV curing does not promote the release of hydrogen from the film. It is worth noting that, as observed in Figures 4(a,b), 6(a), 8(a), and 11(a,b), different forms of noise signals appear in the ~3600-3800 cm^-1 range. This is primarily attributed to variations in the background atmosphere during each FTIR measurement. However, since no characteristic species are formed in this wavenumber range, the impact on the results is considered negligible.

---------------------------------------------------------------------------------------------------------

Thank you, once again!

Author Response File: Author Response.pdf

Article Menu

Optimizing the PECVD Process for Stress-Controlled Silicon Nitride Films: Enhancement of Tensile Stress via UV Curing and Layered Deposition

Further Information

Guidelines

MDPI Initiatives

Follow MDPI