Controlling Oxygen Vacancies in BiFeO₃ Thin Films via Pyrolysis Temperature and O₂ Annealing

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper is quite good and may be recommended for publication, but only after clarification of certain details and responses to the questions. Main comments and question are below:

 

1. Since XPS is surface sensitive, how do you justify that the O 1s-based RIR reflects the oxygen vacancy concentration of the whole film and not only surface adsorption or contamination?
2. You report a nearly constant Fe³⁺/Fe²⁺ ratio for all films, while the O 1s analysis suggests large differences in oxygen vacancy concentration. How do you reconcile these two observations? Could the Fe²⁺ contribution be affected by surface reduction, charging, or fitting uncertainty? Please clarify how charge calibration was done (e.g., C 1s reference) and whether charge compensation was used.
3. Given the known limitations of EDS for oxygen quantification in thin films, how reliable are the reported O/Bi and O/Fe ratios?
4. Were the secondary phases (Bi₂₅FeO₃₉, Bi₂Fe₄O₉) included in the Rietveld refinement model, or was only the R3c BFO phase refined? Please clarify in the figure captions which diffraction peaks correspond to secondary phases.
5. BFO3 shows the highest dc conductivity despite being described as having the lowest oxygen vacancy concentration. How do you explain this apparent contradiction? Could microstructure, crystallinity, interfaces, or impurity phases contribute to the observed conductivity?
6. Given the presence of secondary phases, could these phases contribute to the observed hysteresis loops?
7. Lines 36-42. These sentences in the introduction need few supporting references.
8. Line 56. A reminder that oxygen vacancies can exist in three charge states is absolutely correct; however, the corresponding reference has not been provided. See, for example, the review paper: Popov, A. I., Kotomin, E. A., & Maier, J. (2010). Basic properties of the F-type centers in halides, oxides and perovskites. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, 268(19), 3084-3089.
9. The reference list includes recent papers, but could you consider adding a few 2024–2025 studies specifically focused on oxygen vacancy control in BFO thin films prepared by solution-based methods?

Author Response

Response to Reviewer 1

 

            Dear Reviewer,

Thank you for your time in reviewing our manuscript and for your comments and remarks. We revised the manuscript considering your recommendations and suggestions. The changes are highlighted in blue in the revised manuscript. We comment on your remarks as follows:

 

Reviewer's comment #1: Since XPS is surface sensitive, how do you justify that the O 1s-based RIR reflects the oxygen vacancy concentration of the whole film and not only surface adsorption or contamination?

 

Authors' comments: We agree with the referee that surface adsorption and contamination are important factors in XPS, which is a surface-sensitive technique used to determine the elemental composition, empirical formula, and chemical state of materials by measuring the kinetic energy of emitted photoelectrons to identify elements and their bonding states. Like all experimental techniques, XPS is limited by its extreme surface sensitivity (typically 10 nm), so that bulk properties are not accessible. Therefore, it is important to consider the technique's precision when analyzing experimental results and, as far as possible, to use complementary techniques to support the interpretations. Thus, as a standard procedure, the XPS data obtained in the present work reflect the surface chemical composition rather than the entire film. However, justifying that the O 1s-based Relative Intensity Ratio (RIR) in XPS reflects bulk oxygen vacancy concentration rather than surface contamination requires a combination of meticulous sample preparation, careful peak fitting, and validation via complementary techniques. Although the literature suggests that a more precise multicomponent deconvolution is needed to separate surface effects from vacancies, associating the features around 531 and 532 eV with "oxygen vacancies" can also be considered valid, provided the associated peaks are symmetrical, suggesting that additional contributions are absent or insignificant. In the present study, small asymmetries in the XPS peaks suggested possible additional O_III effects; therefore, we performed multicomponent deconvolution, as shown in Figure 3(c). The oxygen contamination and defects associated with incomplete organic removal have been discussed in the manuscript with minor modifications as follows:

 

Page 9 (lines 344-347):

“Previous works indicated that O_II is mainly associated with hydroxyl groups occupying oxygen defects in BiFeO₃ films [29,31], but is also associated with oxygen contamination, including oxidized organic groups (…)”

 

The revised manuscript also includes the following text regarding limitations of the XPS technique and a justification for possible adsorption or contamination of the surface:

 

Page 10 (lines 370-378):

“In summary, given the discussion above, it is important to note that the XPS technique is limited by its extreme surface sensitivity (typically 10 nm from the surface), so that bulk properties are not accessible. Although we did not use complementary techniques to assess surface adsorption or contamination in the studied thin films, the standard analyses indicate that the XPS data obtained in the present work reflect the surface chemical composition rather than the entire film. Therefore, justifying that the O 1s-based RIR in XPS reflects the bulk oxygen vacancy concentration rather than surface contamination requires a combination of meticulous sample preparation, careful peak fitting, and validation via complementary techniques, all to be carried out opportunistically.”

 

Reviewer's comment #2: You report a nearly constant Fe³⁺/Fe²⁺ ratio for all films, while the O 1s analysis suggests large differences in oxygen vacancy concentration. How do you reconcile these two observations?

 

Authors' comments: The observed nearly constant Fe³⁺/Fe²⁺ ratio for all films, as stated in the article (lines 311-313), in contrast to the large differences in oxygen vacancy concentration observed in the different samples from the O 1s analysis, can be reconciled by considering that oxygen vacancies do not always require a change in the average oxidation state of iron, or that the charge compensation occurs through mechanisms other than strictly reducing Fe³⁺ to Fe²⁺, as indicate by the equations  and  in the article. It is important to note that the charge compensation mechanism can also occur by non-iron defect centers. The oxygen vacancies detected by O 1s can be also compensated by electronic charge trapping or surface defect states rather than by a one-to-one conversion of Fe³⁺ to Fe²⁺, keeping the Fe³⁺/Fe²⁺ ratio relatively constant across different film preparation conditions, as observed. Therefore, the revised manuscript includes the following complementary justification:

 

Page 8 (lines 315-322):

“However, it is worth noting that charge compensation can also occur via non-iron defect centers. The oxygen vacancies detected by O 1s can be also compensated by electronic charge trapping or surface defect states rather than by a one-to-one conversion of Fe³⁺ to Fe²⁺, keeping the Fe³⁺/Fe²⁺ ratio relatively constant across different film preparation conditions, which can justify the contrast between the nearly constant Fe³⁺/Fe²⁺ ratio for all films compared to the large differences in oxygen vacancy concentration observed in the different samples from the O 1s analysis.”

 

Reviewer's comment #3: Could the Fe²⁺ contribution be affected by surface reduction, charging, or fitting uncertainty? Please clarify how charge calibration was done (e.g., C1s reference) and whether charge compensation was used.

 

Authors' comments: Yes, the Fe²⁺ contribution could be affected by surface reduction, charging, or fitting uncertainty. For fittings, we have followed well-established standard procedures in the literature (background, fitting components, and optimization of all parameters). In the revised manuscript, we include the following information:

 

Page 4 (lines 166-168):

“XPS calibration was performed using 284.8 eV as the reference for adventitious carbon. Charge compensation was done by using a low-energy electron flow gun.”

 

Reviewer's comment #4: Given the known limitations of EDS for oxygen quantification in thin films, how reliable are the reported O/Bi and O/Fe ratios?

 

Authors' comments: It is well known that quantifying oxygen using EDS is difficult. For this reason, we performed a systematic analysis of image statistics from six different surface regions on the samples. Given the meticulous and properly substantiated statistical treatment, it was possible to reliably quantify the percentage of each element with its respective error bar, within the experimental error margin of the technique. Thus, we confirm that all obtained results are correct within the statistical errors. We include the following in the revised manuscript:

 

Page 4 (lines 144-148):

“The analyses were conducted at 30 kV on samples coated with a thin gold film. The quantification was performed by statistical analysis of images from six regions in each sample. The mean deviation (d) was a statistical measure used to calculate the average deviation from the mean value of the EDS data set. Thus, the obtained quantification, along with the mean deviation, provides reliability for interpreting EDS results.”

 

Reviewer's comment #5: Were the secondary phases (Bi₂₅FeO₃₉, Bi₂Fe₄O₉) included in the Rietveld refinement model, or was only the R3c BFO phase refined? Please clarify in the figure captions which diffraction peaks correspond to secondary phases.

 

Authors' comments: Yes, the secondary cubic Bi₂₅FeO₃₉ (sillenite-type structure) and orthorhombic Bi₂Fe₄O₉ (mullite-type structure) phases were included in the Rietveld refinements. Regarding this point, we include the following in the revised manuscript:

 

Page 4 (lines 151-158):

“Rietveld refinements of the XRD diffraction data were carried out using GSAS software and EXPGUI interface. Pseudo-Voigt functions were used to define the peak profiles for all refinements, while a sixth-order polynomial was used to describe the background. The occupancy parameters of the atoms were fixed at the nominal composition, but the scale factor, zero correction, background, half-width parameters, lattice parameters, positional coordinates, and isotropic thermal parameters were refined. In addition, Raman spectroscopy was also used to study the short-range structure.”

 

Page 6 (lines 220-221):

“(…) traces of the secondary Bi₂₅FeO₃₉ (sillenite-type structure) and Bi₂Fe₄O₉ (mullite-type structure) phases were observed (…)”

 

Page 6 (lines 222-230):

“For the BFO1 film, the Rietveld refinement was performed assuming a single BiFeO₃ phase with a rhombohedral structure (space group R3c). For the BFO2 and BFO3 films, the refinements assumed a coexistence model of the rhombohedral (space group R3c), cubic (space group I23), and orthorhombic (space group Pbam) structures, respectively associated with the BiFeO₃, Bi₂₅FeO₃₉, and Bi₂Fe₄O₉ phases. In addition, the cubic structure of the platinum (space group ) was also included in all refinements. Based on refinements, the amounts of both Bi₂₅FeO₃₉ and Bi₂Fe₄O₉ phases in BFO2 and BFO3 films were estimated to be less than 5%, given the fitting accuracy.”

 

Reviewer's comment #6: BFO3 shows the highest dc conductivity despite being described as having the lowest oxygen vacancy concentration. How do you explain this apparent contradiction? Could microstructure, crystallinity, interfaces, or impurity phases contribute to the observed conductivity? Given the presence of secondary phases, could these phases contribute to the observed hysteresis loops?

 

Authors' comments: For the analysis of electrical measurements, including the electrical conductivity, we consider the grain and grain boundary contributions. Based on our studies, electrical conductivity is generally dominated by grain boundaries rather than bulk grains. It is expected that conductivity is not only directly due to oxygen vacancies but also depends on crystallinity and other defects at grain boundaries. Thus, the higher crystallinity of the BFO3 film associated with the presence of secondary phases may contribute to the observed highest conductivity.

 

Reviewer's comment #7: Lines 36-42. These sentences in the introduction need few supporting references.

 

Authors' comments: Thank you for the suggestion. We included in page 1 (line 39) the new references [5,6] in the amended article, as follows:

 

[5] Meisenheimer, P.; Moore, G.; Zhou, S.; Zhang, H.; Huang, X.; Husain, S.; Chen, X.; Martin, L.W.; Persson, K.A.; Griffin, S.; Caretta, L.; Stevenson, P.; Ramesh, R. Switching the spin cycloid in BiFeO₃ with an electric field. Nat Commun. 2024, 15, 2903.

 

[6] Itoh, T.; Shigematsu, k.; Das, H.; Meisenheimer, P.; Maeda, K.; Lee, K.; Manna, M.; Reddy, S.P.; Susarla, S.; Stevenson, P.;  Ramesh, R.; Azuma, M. Electric-field-driven reversal of ferromagnetism in (110)-oriented, single phase, multiferroic Co-substituted BiFeO₃ thin films. Adv. Mater. 2025, 37, 2419580.

 

Reviewer's comment #8: Line 56. A reminder that oxygen vacancies can exist in three charge states is absolutely correct; however, the corresponding reference has not been provided. See, for example, the review paper: Popov, A. I., Kotomin, E. A., & Maier, J. (2010). Basic properties of the F-type centers in halides, oxides and perovskites. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, 268(19), 3084-3089.

 

Authors' comments: Thank you for the suggestion. We have included the new suggested reference [13] in the modified article on page 2 (line 56), as follows:

 

[13] Popov, A.I.; Kotomin, E.A.; Maier, J. Basic properties of the F-type centers in halides, oxides and perovskites. is Nucl. Instrum. Methods Phys. Res. B. 2010, 268, 3084–3089.

 

Reviewer's comment #9: The reference list includes recent papers, but could you consider adding a few 2024–2025 studies specifically focused on oxygen vacancy control in BFO thin films prepared by solution-based methods?

 

Authors' comments: Thank you again for the suggestion. However, we decided to keep the references as they are, since they reflect the article as conceived.

 

In summary, we hope to have answered the Reviewer's questions and improved the manuscript quality for readers. We would like to thank you again for your comments and for the opportunity to review the manuscript carefully, and for providing a constructive discussion.

Reviewer 2 Report

Comments and Suggestions for Authors

Reviewer’s comments

In this manuscript, the authors synthesized BiFeO₃ thin films under different pyrolysis and crystallization conditions to evaluate and control oxygen vacancy formation. Higher pyrolysis temperature (360 °C) and crystallization in an O₂ atmosphere significantly reduced oxygen vacancies, whereas films processed at lower temperature in air showed high defect concentrations that could not be fully corrected by post-annealing. The presence of oxygen vacancies markedly degraded the dielectric, electrical, photovoltaic, and magnetic properties, underscoring the importance of oxygen control during synthesis.

Overall, the manuscript demonstrates many experimental results but requires more conceptual analysis and a clear presentation of its results to reach the level expected for a competitive publication in the field of multiferroic oxide thin films. At the current state, the mechanism of the manuscript is not clear and rather confusing. Until all the raised concerns are answered properly, the manuscript can not be considered for publication.

 

Major Comments

1. First, the novelty of the work is not sufficiently clarified. BiFeO₃ thin films have been extensively studied in relation to strain-driven phase transitions, polarization enhancement, domain engineering, defect chemistry, and magnetoelectric coupling. The manuscript does not clearly explain how the present films differ fundamentally from previously reported BiFeO₃ systems. It remains unclear whether the key contribution lies in a new growth strategy, making it difficult to assess progress beyond incremental gains.
2. For BF1, BF2, and BF3 films, the authors have used various growth parameters, such as different pyrolysis and crystallization temperatures, post-annealing temperature, times for these conditions, and then provided different (air and oxygen) environments without identifying a clear link. Considering various conditions without a clear explanation of the role of each parameter and the linked structural, electrical, and optical properties makes this manuscript very puzzling and confusing. The authors should restructure the manuscript by clarifying each parameter separately; otherwise, the manuscript will remain unclear.
3. The alkali metals are very sensitive to growth temperature and time at high temperatures [M. Sheeraz et al., ACS Appl. Electron. Mater. 6, 11, 8328–8338 (2024)]. Can the author explain why there are no Bi vacancies present at high crystallization and post-annealing temperatures? The authors should highlight this point in the manuscript based on the suggested reference.   
4. Weighted-profile R-factor (R_wp) represents the weighted average of the differences between the observed diffraction intensities and the calculated intensities, taking measurement errors into account. A lower R_wp value indicates better agreement between the calculated and experimental patterns. Generally, values below 15% are considered indicative of a well-refined result [G. Sim, et al., J. Electr. Electron. Mater., Vol. 38, No. 5, pp. 481-498 (2025)]. In this work, all the Rwp and Rp values are over 25 and 20%, respectively, which demonstrates a poor fit rather than the best fit. The authors should briefly explain the reason based on the mentioned tutorial reference on the refinement.
5. Based on the XPS data, if the oxygen vacancies are higher in BFO1 and BFO2 than in BFO3, then why is the oxygen content higher in BFO1 and BFO2 than in BFO3 as observed in EDS data? Similarly, the Bi and Fe content is also not consistent.
6. The authors should provide references to some sentences. One of the examples is “since Raman frequency shift is susceptible to internal and external factors associated with structural and physical properties, such as crystallite size, crystallinity, phase composition, oxygen vacancies, defects, strain, and compressive/tensile stresses.”. There are similar kinds of authors claimed where the references should be provided.
7. In comparison to the shift in the Raman spectral peaks at 144 and 177, a clear shift was noticed in the A1 modes at 430 and 425 for BFO1 and BFO3 films, respectively. The authors should clearly explain this point compared to the other peaks with a little variation.

Author Response

Response to Reviewer 2

 

            Dear Reviewer,

Thank you very much for your comments and remarks on our manuscripts. We revised it considering your comments and recommendations. The changes are highlighted in blue in the revised manuscript. We comment your remarks as follows:

 

Overall Reviewer's comment: Overall, the manuscript demonstrates many experimental results but requires more conceptual analysis and a clear presentation of its results to reach the level expected for a competitive publication in the field of multiferroic oxide thin films. At the current state, the mechanism of the manuscript is not clear and rather confusing. Until all the raised concerns are answered properly, the manuscript can not be considered for publication.

 

Authors' comments: Thank you very much for your comments. However, we regret that our work is viewed as a competitive endeavor within the scientific community. In our opinion, scientific research is not a business. In any case, thank you very much for your time spent reviewing our work and for your comments.

 

Reviewer's comment #1: First, the novelty of the work is not sufficiently clarified. BiFeO3 thin films have been extensively studied in relation to strain-driven phase transitions, polarization enhancement, domain engineering, defect chemistry, and magnetoelectric coupling. The manuscript does not clearly explain how the present films differ fundamentally from previously reported BiFeO3 systems. It remains unclear whether the key contribution lies in a new growth strategy, making it difficult to assess progress beyond incremental gains.

 

Authors' comments: We regret that the article was not written clearly enough to convey the research's novelty. In our opinion, the last paragraph in the introduction, as in the submitted manuscript, described clearly step by step the purpose and challenges of the study. However, we modified some phases in the revised manuscript, as follow:

 

Page 3 (lines 94-107):

"This work investigates the effects of processing on the control of oxygen vacancies in BiFeO3 thin films prepared by the CSD method and the consequent impact on their structural, dielectric, electrical, optical, photovoltaic, and magnetic properties. For this purpose, we produced three BiFeO3 thin films under different preparation conditions to determine how oxygen vacancies are linked to their crystallization kinetics and physical properties. Two films were prepared in an open-atmosphere electric furnace under the same pyrolysis and crystallization conditions. Nucleation at low-temperature pyrolysis induces oxy-gen vacancies in both films, but one of them was subjected to a long period of subsequent post-annealing in an O2 atmosphere to correct the imbalance. A third film was prepared under an O2 atmosphere at higher pyrolysis and crystallization temperatures. The main objective of the study is to make clear that once oxygen vacancies are established during synthesis, there are practically no alternatives to correct the imbalance in subsequent steps, reinforcing the importance of controlling point defects during nucleation and crystallization."

 

In the revised article, we reinforce the novelty of the study in the new paragraph:

 

Page 3 (lines 108-115):

"The study aims to elucidate how the oxygen atmosphere and pyrolysis temperature influence the structural evolution of BiFeO₃ thin films and their physical properties. The novelty of the present investigation lies in the variation of pyrolysis temperature and oxy-gen environment during crystallization and post-annealing of spin-coated BiFeO₃ thin films, followed by a comparative analysis of their effects on physical properties. Rather than focusing on the material itself, the study emphasizes how different thermal treatments in oxygen and air environments affect the microstructural development and overall film quality of BiFeO₃ thin films."

 

Reviewer's comment #2: For BF1, BF2, and BF3 films, the authors have used various growth parameters, such as different pyrolysis and crystallization temperatures, post-annealing temperature, times for these conditions, and then provided different (air and oxygen) environments without identifying a clear link. Considering various conditions without a clear explanation of the role of each parameter and the linked structural, electrical, and optical properties makes this manuscript very puzzling and confusing. The authors should restructure the manuscript by clarifying each parameter separately; otherwise, the manuscript will remain unclear.

 

Authors' comments: Thank you for your comments. To clarify the preparation conditions and improve reader understanding, redundant text has been removed and the paragraph was amended as follows:

 

Page 6 (lines 219-330):

“A single-phase BiFeO₃ is observed in Figure 2(a) in the BFO1 film. On the other hand, traces of the secondary Bi₂₅FeO₃₉ (sillenite-type structure) and Bi₂Fe₄O₉ (mullite-type structure) phases were observed in coexistence with the BiFeO₃ phase in the BFO2 and BFO3 films. For the BFO1 film, the Rietveld refinement was performed assuming a single BiFeO₃ phase with a rhombohedral structure (space group R3c). For the BFO2 and BFO3 films, the refinements assumed a coexistence model of the rhombohedral (space group R3c), cubic (space group I23), and orthorhombic (space group Pbam) structures, respectively associated with the BiFeO₃, Bi₂₅FeO₃₉, and Bi₂Fe₄O₉ phases. In addition, the cubic structure of the platinum (space group ) was also included in all refinements. Based on refinements, the amounts of both Bi₂₅FeO₃₉ and Bi₂Fe₄O₉ phases in BFO2 and BFO3 films were estimated to be less than 5%, given the fitting accuracy.”

 

In addition, details of sample preparation have been introduced in the Materials and Methods section as follows:

 

Page 3 (lines 130-141):

"For film preparation, the chemical solution was initially deposited onto Pt/TiO₂/SiO₂/Si(100) substrates by spin coating at 5000 rpm for 30 seconds, followed by thermal annealing in an open electric furnace (open atmosphere) for the pyrolysis (to remove water and organics) at 300°C for 60 min for the BFO1 and BFO2 films, and at 360°C for 60 min for the BFO3 film. For each film, the second and subsequent layers were sequentially deposited on the previously pyrolyzed film following the same protocol to increase the film thickness. After pyrolysis and completing eight depositions, the BFO1 film was crystallized in air (open atmosphere) at 600 °C for 40 min, the BFO2 was also crystallized in air at 600 °C for 40 min, and after completing the crystallization, it was post-annealed at 600 °C for 5 hours under an O₂ atmosphere at 18.5 psi. Finally, the BFO3 film was crystallized in an O₂ atmosphere at 18.5 psi at 640 °C for 40 min, without additional post-annealing."

 

 

Reviewer's comment #3: The alkali metals are very sensitive to growth temperature and time at high temperatures [M. Sheeraz et al., ACS Appl. Electron. Mater. 6, 11, 8328–8338 (2024)]. Can the author explain why there are no Bi vacancies present at high crystallization and post-annealing temperatures? The authors should highlight this point in the manuscript based on the suggested reference.  

 

Authors' comments: Bismuth volatilization during preparation of BiFeO₃ thin films is a relatively well-known issue in the literature and mainly occurs during high-temperature processing, such that several physical and chemical factors contribute to this phenomenon. Consequently, these Bi vacancies mainly arise from Bi loss during high-temperature processing and from defect charge compensation mechanisms in the crystal lattice. Several researchers have, in the recent past, adopted the practice of including an excess of bismuth to compensate for the supposed loss of bismuth during synthesis. However, we have strong evidence (results not yet published) that this procedure is a mistake and stems from inadequate control of the precursors. In our case, the rigorous control of precursors in an environment with controlled humidity and temperature allows us to prepare more stable films and, therefore, less prone to Bi volatilization at relatively low temperature ranges (below 650 °C). In addition, it is important to mention that the controlled oxygen-rich atmospheres, as used in the present work, is favorable to stabilize Bi oxide phases and reduce volatilization. When people talk about evidence of Bi vacancies in BiFeO₃ films, especially when investigated by XPS, they are usually referring to indirect but consistent spectroscopic fingerprints that arise because removing Bi changes the local chemistry of Fe and O and alters charge compensation. However, this is not a "direct visualization of vacancies," but rather of chemical shifts, changes in peak shape, and variations in relative intensity that correlate with non-stoichiometry. In our case, there is no conclusive evidence for bismuth vacancies.

 

Reviewer's comment #4: Weighted-profile R-factor (Rwp) represents the weighted average of the differences between the observed diffraction intensities and the calculated intensities, taking measurement errors into account. A lower Rwp value indicates better agreement between the calculated and experimental patterns. Generally, values below 15% are considered indicative of a well-refined result [G. Sim, et al., J. Electr. Electron. Mater., Vol. 38, No. 5, pp. 481-498 (2025)]. In this work, all the Rwp and Rp values are over 25 and 20%, respectively, which demonstrates a poor fit rather than the best fit. The authors should briefly explain the reason based on the mentioned tutorial reference on the refinement.

 

Authors' comments: We agree with the Reviewer that this issue should be clarified, and additional information should be included in the revised article so that the reader can better understand the refinement complexity and infer possible reasons for the values ​​obtained for parameters associated with the credibility of the refinements. There is no doubt that the above-cited reference is a good tutorial reference on the Rietveld refinement, especially for beginners. However, some specific points must be considered beyond this tutorial:

 

1. It is important to note that the XRD data refers to thin films, not to single crystals, bulk ceramics or powders, and were obtained from a conventional X Ray diffractometer. Although the measurements were performed in step mode (4 seconds) to increase counts, the intensity obtained was still low (ceramic thin films naturally produce lower intensities) compared to other high-intensity sources, such as neutron diffraction or synchrotron radiation.
2. High R_wp and R_p values often arise from low-intensity data, high background noise, or significant peak broadening. High R-values (R_wp > 25% and R_p > 20%) combined with a low (GoF) around 1.9-2.0, as in the present work, often indicate a good structural model fit hidden behind poor counting statistics or sometimes a background difficult to model. All these points are possible considering thin films XRD data obtained from conventional X Ray diffraction.
3. Although the examination of the R_wp, R_p, c² parameters associated with the integrated visual inspection of the obtained theoretical patterns can provide indicators of good refinements and adequate models, it is important to highlight the inherent difficulties of Rietveld refinements when analyzing multi-phase systems. To ensure the results are reliable in these cases, it is often helpful to validate them using additional characterization techniques beyond Rietveld refinements, as used in the present work.
4. Beyond the inspection of the R_wp, R_p, c² parameters, the most critical evaluation is checking the difference plot between observed and calculated patterns. If the difference curve is relatively flat and does not show significant systematic peaks (e.g., peak shapes, positions not matching), the structural model is likely good. In the present work, the difference curves follow the requisite for good fits.

 

In summary, considering the above explanations, our Rietveld refinements are of good quality. The following texts were included in the revised manuscript:

 

Page 4 (lines 151-158):

"Rietveld refinements of the XRD diffraction data were carried out using GSAS software and EXPGUI interface. Pseudo-Voigt functions were used to define the peak profiles for all refinements, while a sixth-order polynomial was used to describe the background. The occupancy parameters of the atoms were fixed at the nominal composition, but the scale factor, zero correction, background, half-width parameters, lattice parameters, positional coordinates, and isotropic thermal parameters were refined.

 

Page 6 (lines 220-222):

“(…) traces of the secondary Bi₂₅FeO₃₉ (sillenite-type structure) and Bi₂Fe₄O₉ (mullite-type structure) phases were observed (…)”

 

Page 6 (lines 222-230):

"For the BFO1 film, the Rietveld refinement was performed assuming a single BiFeO₃ phase with a rhombohedral structure (space group R3c). For the BFO2 and BFO3 films, the refinements assumed a coexistence model of the rhombohedral (space group R3c), cubic (space group I23), and orthorhombic (space group Pbam) structures, respectively associated with the BiFeO₃, Bi₂₅FeO₃₉, and Bi₂Fe₄O₉ phases. In addition, the cubic structure of the platinum (space group ) was also included in all refinements. Based on refinements, the amounts of both Bi₂₅FeO₃₉ and Bi₂Fe₄O₉ phases in BFO2 and BFO3 films were estimated to be less than 5%, given the fitting accuracy."

 

Reviewer's comment #5: Based on the XPS data, if the oxygen vacancies are higher in BFO1 and BFO2 than in BFO3, then why is the oxygen content higher in BFO1 and BFO2 than in BFO3 as observed in EDS data? Similarly, the Bi and Fe content is also not consistent.

 

Authors' comments: Like all experimental techniques, XPS is limited by its extreme surface sensitivity (typically 10 nm), so that bulk properties are not accessible. For XPS analysis, several points must be considered to quantify oxygen vacancies and iron states from fits, including difficulties in quantifying these elements and the potential for surface contamination, which requires a combination of meticulous sample preparation, careful peak fitting, and validation via complementary techniques. In principle, the observed nearly constant Fe³⁺/Fe²⁺ ratio for all films, as stated in the article (lines 311-313), in contrast to the large differences in oxygen vacancy concentration observed in the different samples from the O 1s analysis, can be reconciled by considering that oxygen vacancies do not always require a change in the average oxidation state of iron, or that the charge compensation occurs through mechanisms other than strictly reducing Fe³⁺ to Fe²⁺, as indicate by the equations  and  in the article. It is important to note that charge compensation can also occur via non-iron defect centers. The oxygen vacancies detected by O 1s can also be compensated by electronic charge trapping or surface defect states rather than by a one-to-one conversion of Fe³⁺ to Fe²⁺, keeping the Fe³⁺/Fe²⁺ ratio relatively constant across different film preparation conditions, as observed.

On the other hand, it is well known that quantifying oxygen by EDS is difficult and has limitations. Regarding EDS, we performed a systematic analysis of image statistics from six different surface regions on the samples. Given the meticulous, properly substantiated statistical treatment, it was possible to reliably quantify the percentage of each element, with its respective error bar, within the experimental error margin of the technique. Thus, we confirm that all obtained results are correct within the statistical errors. So, we believe that apparent inconsistencies can be understood considering the possibilities described above.

 

The oxygen contamination and defects associated with incomplete organic removal have been discussed in the manuscript with minor modifications as follows:

 

Page 9 (lines 344-347):

"Previous works indicated that O_II is mainly associated with hydroxyl groups occupying oxygen defects in BiFeO₃ films [29,31], but is also associated with oxygen contamination, including oxidized organic groups (…)"

 

The revised manuscript also includes the following text regarding limitations of the XPS technique and a justification for possible adsorption or contamination of the surface:

 

Page 10 (lines 370-378):

"In summary, given the discussion above, it is important to note that the XPS technique is limited by its extreme surface sensitivity (typically 10 nm from the surface), so that bulk properties are not accessible. Although we did not use complementary techniques to assess surface adsorption or contamination in the studied thin films, the standard analyses indicate that the XPS data obtained in the present work reflect the surface chemical composition rather than the entire film. Therefore, justifying that the O 1s-based RIR in XPS reflects the bulk oxygen vacancy concentration rather than surface contamination requires a combination of meticulous sample preparation, careful peak fitting, and validation via complementary techniques, all to be carried out opportunistically."

 

The revised manuscript includes the following complementary justification on the possible charge compensation via non-iron defect centers:

 

Page 8 (lines 315-322):

"However, it is worth noting that charge compensation can also occur via non-iron defect centers. The oxygen vacancies detected by O 1s can be also compensated by electronic charge trapping or surface defect states rather than by a one-to-one conversion of Fe³⁺ to Fe²⁺, keeping the Fe³⁺/Fe²⁺ ratio relatively constant across different film preparation conditions, which can justify the contrast between the nearly constant Fe³⁺/Fe²⁺ ratio for all films compared to the large differences in oxygen vacancy concentration observed in the different samples from the O 1s analysis."

 

Regarding the EDS analysis, we include the following in the revised manuscript:

 

Page 4 (lines 144-148):

"The analyses were conducted at 30 kV on samples coated with a thin gold film. The quantification was performed by statistical analysis of images from six regions in each sample. The mean deviation (d) was a statistical measure used to calculate the average deviation from the mean value of the EDS data set. Thus, the obtained quantification, along with the mean deviation, provides reliability for interpreting EDS results."

 

Reviewer's comment #6: The authors should provide references to some sentences. One of the examples is "since Raman frequency shift is susceptible to internal and external factors associated with structural and physical properties, such as crystallite size, crystallinity, phase composition, oxygen vacancies, defects, strain, and compressive/tensile stresses.". There are similar kinds of authors claimed where the references should be provided.

 

Authors' comments: Following your suggestion, we include the new references [32], [33], [34], and [35] as follows:

 

Page 8 (lines 274-275):

“(…) such as crystallite size and crystallinity [32], phase composition and oxygen vacancies [33], defects and strains [34], and compressive/tensile stresses [35].”

 

Reviewer's comment #7: In comparison to the shift in the Raman spectral peaks at 144 and 177, a clear shift was noticed in the A1 modes at 430 and 425 for BFO1 and BFO3 films, respectively. The authors should clearly explain this point compared to the other peaks with a little variation.

 

Authors' comments: In the article, as submitted, we discussed the 144 cm^-1 and 177 cm^-1 modes as in the below transcribed paragraph, as well as redshift in the BFO2 film, and we conclude that this one is possibly associated not only with oxygen vacancies.

 

Page 7 (lines 275-282):

"The Raman shift of the A1(LO) modes in the BFO2 film shifts to lower frequencies or longer wavelengths (red shift), in comparison to the BFO1 film modes around 144 cm^-1 and 177 cm^-1, as shown in Figure 2(d). Despite the red shift often being attributed to improved crystallinity in the literature, this does not appear to be the case, as BFO1 and BFO2 exhibit similar crystallinity, whereas BFO3 shows higher crystallinity. The redshift in the BFO2 film indicates a direct effect of the synthesis conditions, possibly associated not only with oxygen vacancies, as will be discussed below, but also with other defects that could not be explicitly identified."

 

In summary, we hope to have satisfactorily answered the Reviewer's questions and improved the manuscript quality for readers. We would like to thank you again for your contribution to our work.

Reviewer 3 Report

Comments and Suggestions for Authors

This paper reports on the research of BiFeO3 in multiferroic applications. Under the same annealing conditions, BiFeO3 exhibits the best performance after extra O2 annealing, as evidenced by its lowest leakage current and highest photocurrent.

This study is detailed and structurally rigorous.

Please explain why the 𝜀′(f) is the lowest and why the dielectric loss 𝜀′′(𝑓) is the highest for BiFeO3 under O2 annealing condition. The 𝜀′(f) is due to the polarization that may be decreased by oxygen vacancies, but improved material quality by O2 annealing should give lower dielectric loss 𝜀′′(𝑓).

Author Response

Response to Reviewer 3

 

            Dear Reviewer,

Thank you for your time in reviewing our manuscript and for your comments. We revised the manuscript considering your remarks. The changes are highlighted in blue in the revised manuscript. We comment your remarks as follows:

 

Reviewer's comment #1: This paper reports on the research of BiFeO3 in multiferroic applications. Under the same annealing conditions, BiFeO3 exhibits the best performance after extra O2 annealing, as evidenced by its lowest leakage current and highest photocurrent.

 

Authors' comments: Thank you for your comments.

 

Reviewer's comment #2: Please explain why the ?′(f) is the lowest and why the dielectric loss ?′′(?) is the highest for BiFeO3 under O2 annealing condition. The ?′(f) is due to the polarization that may be decreased by oxygen vacancies, but improved material quality by O2 annealing should give lower dielectric loss ?′′(?).

 

Authors' comments: A realistic description of dielectric permittivity, as well as electrical conductivity, involves contributions from the grains and their boundaries, which means that the dielectric response is a combined effect of volume and interfaces. In Fig. 4(a), the lower real dielectric permittivity () of the BFO3 film is more realistic, with dominance of the grain responses, than for the BFO1 and BFO2 films, whose higher dielectric permittivities result from a more effective contribution from the grain boundaries. Something similar occurs with the conductivity, which reflects the imaginary part of the dielectric permittivity (). However, in general, ac conductivity in most materials is dominated by grain boundaries. At very low frequencies or in highly conductive materials, free charge carriers can accumulate at interfaces. So, the increase of  at low frequencies for BFO3 is indicative of Maxwell-Wagner polarization, probably a combination effect of high crystallinity and defects at the interfaces. We described in detail the grain and grain boundary contributions to the real and imaginary dielectric permittivities in the cited references [27] and [28] in the manuscript, especially in the supplementary information of Ref. [27].

 

In summary, we hope to have answered the Reviewer's question. We would like to thank you once again for your contribution to our work.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

After a successful revision, this paper can be recommended for publication.

Author Response

We appreciate the Reviewer's support for the publication of our work.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have made efforts to revise the manuscript and provide responses to the reviewer comments; however, the explanations provided are still not sufficiently convincing from a scientific perspective. In its current form, I am not fully satisfied with the responses, and therefore, I am not in a position to recommend acceptance of the manuscript. Several important issues remain unresolved and should be addressed more clearly in the revised version. Addressing the concerns mentioned below would significantly improve the scientific rigor and clarity of the manuscript.

 

Major and Minor comments

1) First, the central question regarding the advancement of the present work in comparison to other published work remains unclear. The main concern was to understand how the current BiFeO₃ thin films differ from or improve upon previously reported BiFeO₃ systems in terms of growth strategy, defect control, or functional properties. Although the authors rewrote part of the introduction, the response still does not clearly demonstrate how this work advances beyond existing studies. The authors should provide a clear comparison with previously reported BiFeO₃ thin films, preferably in the form of a comparison table summarizing processing conditions, defect control strategies, and key physical properties. This would help the readers and reviewers better evaluate the scientific contribution of the present work.

2) Another important concern relates to the similarity index of the manuscript. According to the iThenticate report, the overall similarity is approximately 35%, which is relatively high and generally not acceptable for publication in Nanomaterials without careful revision. The authors should thoroughly revise the manuscript, rephrase overlapping sections, and ensure that the text is written in a more original and concise manner.

3) Regarding the discussion on Bi vacancies, the authors argue that strict control of the precursor environment prevents Bi volatilization and therefore suppresses the formation of Bi vacancies. However, this explanation remains largely speculative because no direct experimental evidence has been provided. The authors mention that they have strong supporting results that are not yet published, but such statements cannot be considered sufficient scientific justification in a peer-reviewed manuscript. If such evidence exists, the authors should include the relevant data or characterization results to support their claim in their response letter rather to the reviewer and not in the manuscript.

4) Furthermore, the authors should more clearly discuss this issue in relation to the previously suggested reference and explain why Bi vacancies are not observed in their system while previous studies report a clear trend of Bi loss during high-temperature processing [M. Sheeraz et al., ACS Appl. Electron. Mater. 6, 11, 8328–8338 (2024) and M. Abazari, et al., Appl. Phys. Lett. 96 (8), 082903 (2010)].

5) With respect to the Rietveld refinement discussion, the explanation provided by the authors regarding the challenges associated with thin-film XRD data is reasonable. However, this justification should be clearly incorporated into the manuscript itself rather than remaining only in the response to reviewers, as general readers do not know the exact reasons. A brief explanation based on the literature should be included in the manuscript to clarify why relatively high refinement parameters may still be observed in thin-film measurements and how the authors ensured the reliability of their structural analysis [G. Sim, et al., J. Electr. Electron. Mater., Vol. 38, No. 5, pp. 481-498 (2025)].

Author Response

Reviewer question: 1) First, the central question regarding the advancement of the present work in comparison to other published work remains unclear. The main concern was to understand how the current BiFeO₃ thin films differ from or improve upon previously reported BiFeO₃ systems in terms of growth strategy, defect control, or functional properties. Although the authors rewrote part of the introduction, the response still does not clearly demonstrate how this work advances beyond existing studies. The authors should provide a clear comparison with previously reported BiFeO₃ thin films, preferably in the form of a comparison table summarizing processing conditions, defect control strategies, and key physical properties. This would help the readers and reviewers better evaluate the scientific contribution of the present work.

Authors’ response: We thank the reviewer for their concern. We want to point out that the focus of the work was not to improve or differentiate our films from other ones reported on the literature. Our focus was to emphasize how critical are the first steps of the synthesis of the films on the final materials, including their properties, mostly by changing or not the oxygen vacancies. While most of the studies focus on controlling defects with doping, very few have devoted to understanding the link with earlier synthesis steps. As such, in lines 93 to 101, we added a brief sentence explaining this specific mention to stress the intended outcome of our work, reproduced below.

“Given that context, it turns clear that controlling defects in bismuth ferrite-based materials is critical to their application. While this has been an antique subject of interest, most studies focus on manipulation of defects and properties through doping [27,28,29], while only few works [30,31] devoted to demonstrating the crucial role of the first thin film synthesis steps on the defects and consequent properties. Still, even though some studies in this topic may be found, a throughout connection between the synthesis, structure and multiple functional properties in BiFeO₃ films is missing.

To embody the current literature in this relevant topic, we herein address the effects of processing on the control of oxygen vacancies in BiFeO₃ thin films prepared by the CSD method and the consequent impact on their structural, dielectric, electrical, optical, photovoltaic, and magnetic properties.”

 

Reviewer question: 2) Another important concern relates to the similarity index of the manuscript. According to the iThenticate report, the overall similarity is approximately 35%, which is relatively high and generally not acceptable for publication in Nanomaterials without careful revision. The authors should thoroughly revise the manuscript, rephrase overlapping sections, and ensure that the text is written in a more original and concise manner.

Authors’ response: We thank the reviewer for their concern. While we agree that plagiarism and extensive AI are getting increasingly more common, we humbly also want to point out that focusing merely in a software derived metric, that may be reduced through rewording and paraphrasing, may lead to other problems, as nicely discussed in doi.org/10.3390/publications13010001. To produce our text, we adhered to the strict writing guidelines and field specific terms and definitions to sustain scientific clarity and rigor. Still, while the number itself may seem large at first, it is common and perfectly fine in research communication. Having that said, we don’t believe that artificially changing the text for the sake of reducing a similarity index is productive. Hence, we respectfully prefer to keep the present writing the way it is.

 

Reviewer question: 3) Regarding the discussion on Bi vacancies, the authors argue that strict control of the precursor environment prevents Bi volatilization and therefore suppresses the formation of Bi vacancies. However, this explanation remains largely speculative because no direct experimental evidence has been provided. The authors mention that they have strong supporting results that are not yet published, but such statements cannot be considered sufficient scientific justification in a peer-reviewed manuscript. If such evidence exists, the authors should include the relevant data or characterization results to support their claim in their response letter rather to the reviewer and not in the manuscript.

AND

4) Furthermore, the authors should more clearly discuss this issue in relation to the previously suggested reference and explain why Bi vacancies are not observed in their system while previous studies report a clear trend of Bi loss during high-temperature processing [M. Sheeraz et al., ACS Appl. Electron. Mater. 6, 11, 8328–8338 (2024) and M. Abazari, et al., Appl. Phys. Lett. 96 (8), 082903 (2010)].

Authors’ response: Since the two questions point to the same context, we respond to both at once. We thank the reviewer for sending the two references. Even though these studies are not in BiFeO₃-based films, we understand that bismuth volatilization is commonly pointed as a source of stoichiometric deviations in any Bi-containing materials. While mentioning this eventual phenomenon, we didn’t find compelling evidence of elemental volatilization being shown on these works. Accordingly, the references only cite this information as a possibility, but do not demonstrate the phenomenon. We do not state it with the intention of derogating the two works. In contrast, we acknowledge their appeals for interested readership. What we mean is that the bismuth volatilization is almost a universal claim for most Bi-containing materials, including BiFeO₃ (and we have pointed out references in this matter in the introduction), but we truly feel that it hasn’t being indubitably proved yet. With that said, we do not discharge that Bi vacancies can be formed in general and that they were not formed in our films. However, since Bi-related signals in our XPS data have not indicated any difference between the films, there is reason to believe that Bi vacancies should not be the reason why the films have marked different properties. Also, if extensive volatilization had occurred, we would have observed the pronounced formation of Fe-rich phases. Instead, when secondary phases were formed, both Fe-rich and Bi-rich phases were present, suggesting that nominal stoichiometry remains balanced between iron and bismuth. In this case, the presence of the two phases should be due to kinetic reasons, as we have addressed in a recent study (reference [25] of the manuscript). Lastly, the films were kept for rather short periods of time at temperatures way below the melting point of Bi-based phases such as Bi₂O₃ and BiFeO₃, indicating that the tendency of Bi loss during the thermal treatments should be negligible. Having this context stated, we included a brief discussion (lines 364 to 370) to emphasize our views on the eventual effects of bismuth vacancies in our films, as reproduced below.

“It is important to mention that bismuth volatilization is largely claimed to occur in Bi-containing oxide materials, leading to vacancies and stoichiometric deviations. Even though we do not discard that it may happen, the fact that bismuth-related XPS data are the same for the three films and that XRD reveals that no Fe-rich phases (e.g. Bi₂Fe₄O₉ and Fe₂O₃) are formed alone strongly suggest that bismuth defects should not be the main reason for markedly different properties, as shown in the latter.”

 

Reviewer question: 5) With respect to the Rietveld refinement discussion, the explanation provided by the authors regarding the challenges associated with thin-film XRD data is reasonable. However, this justification should be clearly incorporated into the manuscript itself rather than remaining only in the response to reviewers, as general readers do not know the exact reasons. A brief explanation based on the literature should be included in the manuscript to clarify why relatively high refinement parameters may still be observed in thin-film measurements and how the authors ensured the reliability of their structural analysis [G. Sim, et al., J. Electr. Electron. Mater., Vol. 38, No. 5, pp. 481-498 (2025)].

Authors’ response: We added the following brief comment (lines 294 to 298) in the caption of Table 3 to ensure the best clarification.

“Table 3. Summary of structural parameters obtained from the Rietveld refinement of the XRD patterns of the studied BiFeO₃ thin films. Despite rather high R_wp and R_p values, the values are a good indication of reliable fits, as evidenced by the visual inspection of data in Figure 2(a). Reasons for suboptimal R_wp and R_p values may include the small interaction volume due to data collected in thin films instead of bulk crystals, inherent incoherence and strains in the substrate/film interface, and multiphase refinements due to the presence of Pt signals, among others.”