Proton-assisted creation of controllable volumetric oxygen vacancies in ultrathin CeO2−x for pseudocapacitive energy storage applications

Two-dimensional metal oxide pseudocapacitors are promising candidates for size-sensitive applications. However, they exhibit limited energy densities and inferior power densities. Here, we present an electrodeposition technique by which ultrathin CeO2−x films with controllable volumetric oxygen vacancy concentrations can be produced. This technique offers a layer-by-layer fabrication route for ultrathin CeO2−x films that render Ce3+ concentrations as high as ~60 at% and a volumetric capacitance of 1873 F cm−3, which is among the highest reported to the best of our knowledge. This exceptional behaviour originates from both volumetric oxygen vacancies, which enhance electron conduction, and intercrystallite water, which promotes proton conduction. Consequently, simultaneous charging on the surface and in the bulk occur, leading to the observation of redox pseudocapacitive behaviour in CeO2−x. Thermodynamic investigations reveal that the energy required for oxygen vacancy formation can be reduced significantly by proton-assisted reactions. This cyclic deposition technique represents an efficient method to fabricate metal oxides of precisely controlled defect concentrations and thicknesses.


Response:
The authors thank the reviewer for providing the references, which show high capacity of cerium oxide materials as supercapacitors. Accordingly, the authors have consulted the references cited by the reviewer 8, 9 . Owing to the ultrathin nature of the films in the present work, all of the capacitances reported are on the basis of volume and area of each electrode (i.e., volumetric and areal capacitances). These data by nature are not comparable to those reported by others for the gravimetric capacitances of bulk samples 10 , as reported in the two suggested papers. By convention, the capacitances of ultrathin films of very low masses (ng-g scale mass) are reported on the basis of volume or area 11,12,13,14 . This is due to the fact that the low masses of the ultrathin films greatly inflate the specific capacitance 10 , which does not reflect the actual capacitance. In contrast, the capacitances of nanoparticles (mg scale mass) are reported on the basis of mass (i.e., gravimetric capacitance). Nonetheless, the authors believe that the very high capacitance reported in the suggested references is an excellent example of why CeO 2 materials should be explored further for energy storage materials. The high values reported for gravimetric capacitance draw attention to the excellent performance of CeO 2 , regardless of the method of measurement, and accordingly the following text has been added to the Introduction of the main paper (Page 1):

Textual Modification:
Main paper (Page 1): Although intercalation pseudocapacitance has not been observed to date in CeO 2 , very high specific capacitances of this material have been reported 17,18 , indicating its potential for energy storage applications. Fig.4c, I think it is hard to believe their data. I suggest the authors to recalculate their results with different sizes.

Response:
In regard to the reviewer's drawing attention to the importance of particle size reported in suggested reference 2 , which is cited in the revised version of the manuscript, some points of clarification concerning the computational methodology are merited. Also, the reviewer's comment has been addressed by modifying the manuscript to include textual changes, clarifications, and new numerical data have been generated.
In the present work, the main aim of the density functional theory (DFT) calculations is to provide physical insights into how the structural and electronic properties of bulk ceria change in the presence of oxygen vacancies. The DFT energy band gaps (E g ) reported in Figure 4c are for bulk stoichiometric and non-stochiometric systems, so both systems must be periodic along the three Cartesian directions. Since the system boundaries are placed at infinity, by definition, nanosize effects cannot be considered in the simulations.
Further, electronic band structure calculations incorporating nano-size effects cannot be done using first-principles DFT because, for the electronic band structure properties of a system to be well defined, the system must be periodic along at least one direction. If this is not the case, then the Bloch theorem, on which the theory of electronic band structure of crystals is based, is not valid. Therefore, the electronic band structures of nanoparticles in principle are not well defined since they are zero-dimensional and hence lack infinite periodicity. Further, in the present work, faceted ceria nanoparticles were not produced; instead ceria thin films (i.e., periodic along two Cartesian directions) of ultrafine crystallites with randomly oriented faceting were synthesised.
However, in order to probe the role of nano-size effects on the DFT simulations, bearing in mind the preceding comments, new DFT simulations designed to replicate the features indicated in

Section 9. Effect of crystallite size on calculated band gap energy
It is possible that there may be nano-size effects on the estimation of the E g in ceria, as suggested by atomic packing considerations of Ce-O bond lengths of nanoparticles 19 . However, such size effects cannot be simulated directly by DFT since the electronic band structure of a system requires periodicity in at least one dimension for the Bloch theorem to be valid and nanoparticles are zerodimensional and hence lack periodicity. However, potential size effects on the E g have been examined by the DFT calculations by considering 2% expansion or 2% contraction of the Ce-O bond lengths that correspond to analogous calculations based on the presence and absence of ligands in the growth environment 19 . The resultant DFT calculations for stoichiometric CeO 2 ( Figure 12) reveal that 2% expansion or contraction in the Ce-O bond lengths leads to an almost negligible 2% reduction or 3% increase in the E g , respectively. For non-stoichiometric CeO 2-x , 2% expansion or contraction results in only a 3% decrease or 2% increase in the E g , respectively. In effect, if there is any size effect on the E g corresponding to the preceding conditions 19 , it is not significant.

Reviewer 2
General Comment: This manuscript reports a kind of ultrathin CeO 2 material with oxygen vacancies. This material was prepared by using an electrodeposition technique and exhibited very high capacitance. The topic of the manuscript is helpful for the energy storages in the future and the electrochemical performance of the material is very good. However, there are some problems in the current manuscript and it must be revised before being accepted.

Response:
The authors thank the reviewer for his/her constructive comments. The response to each of the comments are given in below: So, the authors should revise the statements in the manuscript.

Response:
The authors appreciate this correction.
All references to "first observation for redox pseudocapacitive behaviour" have been removed.

Comment 2:
What does the prepared ultrathin CeO 2-x material look like? Like membranes which cover the FTO substrate layer by layer? Please provide some SEM images of the morphologies of the material.

Response:
Low-and high-resolution SEM images of the surfaces of coated and uncoated FTO substrates (scan rates 3000, 1000, 300, 50 mVs -1 ) have been added to Supplementary Section 3 as Supplementary

Comment 3:
What is the loading of the material? If the loading is very low, the electrochemical performance would be very high while the material is not promising actually.

Response:
The loads of the ultrathin films were in the range 10-50 gcm -2 , which covers a wide range from low to high masses compared to a great number of ultrathin films reported previously, which were in the range 0.3-80 gcm - 2 11, 12, 13, 14, 15, 16, 17 . In fact, the issue of how the capacitance for materials of different classifications (dimensionality and mass) should be evaluated and subsequently reported is an important controversial concept, which was the focus of a recent paper in Science 10 . The method of capacitance measurement is critical, especially when working with ultrathin films with very limited mass; this also is the case with hybrid materials containing graphene. As the reviewer noted, for these materials, it is recommended not to report the capacitance based on the mass of the film but, instead, it should be reported based on the volume and/or area of the electrode materials.
Therefore, owing to the potential to imply artificially high gravimetric capacitance, which is unrealistic relative to the real electrochemical performance of the films, all of the capacitance values reported in the present work are based on volume and surface area of the CeO 2-x electrode, in agreement with previous reports 13,18,19,20 .

Comment 4:
In Fig2, the authors identified the peaks (Ox1, Ox2, Ox3, Re1, Re2 and Re3) in the CV curves, but the evidence is not enough. Please write the chemical reaction equations for each peak and confirm them. If the authors insist on the explanations of the peaks, please calculate the areas of every peak and discuss the variation trend of the areas.

Response:
As the reviewer requested, all the peaks have been identified and the corresponding chemical reactions are given (Supplementary Section 6; Tables 5-7). Additionally, all of the peak areas, which correspond to the charge contribution to the reaction, have been deconvoluted using the Gaussian method and subsequently discussed (Supplementary Section 6; Table 8 4 redissolves as soluble Ce 3+ during reduction (Re 1 ), which is confirmed by the lower current density of Re 1 relative to that of Ox 1 ((anodic peak current)/(cathodic peak current) or I pa /I pc <1). These reactions and corresponding calculations are given in Table 5.  Figure 3(d)). As given in Table 6, the reduction of as-prepared CeO 2 occurs at E = -0.13 V vs Ag/AgCl. This is followed by partial oxidation (annihilation of oxygen vacancies) by the following oxidation reaction (Ox 2 ) occurring at E = +0.07 V vs Ag/AgCl. This is further confirmed in Supplementary Section 7. Owing to the formation of oxygen vacancies within the bulk of the deposited CeO 2-x , the insertion/disinsertion of protons commences during the reduction (Re 3 ) and oxidation (Ox 3 )

Supplementary
reactions, as given in Table 7. Although these phenomena have not been observed previously for The growth in the peak area of can be ascribed to the total increase in oxygen vacancy concentration formed within bulk of the ultrathin film. Consistent with thickening of the CeO 2 film and increasing the amount of oxygen vacancies, the volume of hydrogen insertion/disinsertion reactions increase, resulting in having the highest areal contribution in the cyclic voltammogram.
Supplementary Table 8. Charge contribution and area calculation for identified peaks.

Comment 5:
The thicknesses of the films are obtained by TEM technique (in Supplementary Fig.12 and Fig. 15). However, the field of view of TEM is a very narrow. Moreover, the CeO 2-x films are not uniform (for example, Supplementary Fig.12 b is a typical image). The thicknesses of films should be measured by using other suitable techniques.

Response:
Following the reviewer's request, the thicknesses of the films deposited were characterised by two additional methods: 1) AFM analyses were used to measure the thicknesses of the ultrathin films by scanning the cross-sectional area (21 m 2 ) between film and substrate.
2) Time-of-flight secondary ion mass spectrometry (TOFSIMS) was used to confirm the thicknesses (scanning surface area 900 m 2 ). In order to highlight these data, 3-dimensional imaging was

Comment 6:
In Table 5 of Supplementary Section 5, the 2nd and 3rd reactions are unreasonable.
Why is OHthe final product in an acidic surrounding? If the reaction equations are not true, the calculated results should be changed.

Response:
The authors thank the reviewer for highlighting the necessity of clarifying the apparent contradiction whereby OHis a product under acidic conditions. In order to confirm and clarify the effect of pH, differentiating calculations and experiments were done. This additional work is discussed in Supplementary Section 7, as given below: Textual Modification:

Supplementary Material (Pages 15-16):
Since the formation of OHin an acidic environment might seem counter-intuitive, we have conducted further thermodynamic calculation. As shown in Table 10, oxygen vacancy formation in a relatively strong acidic environment (e.g., pH 5.5) rich in hydrogen ions leads to the local formation of water 17 . In such a condition, the thermodynamic calculations (Table 10)  In contrast, as shown in Table 11, oxygen vacancy formation in a weakly acidic environment (e.g., pH 6) less rich in hydrogen ions leads to the formation of OHions, thus making the local environment essentially basic. In the presence of CeO 2-x , the amount of H + in basic solutions is much lower than acidic solutions, so the extent of reaction between as-produced O 2with two H + , leading to H 2 O formation, is reduced significantly. Therefore, the formation of OHis favoured over that of H 2 O 17 , as confirmed by the calculations given in Table 11. This is consistent with the cyclic voltammetry results shown in Figure 8 (Supplementary) and Figure 2 (main text), where CeO 2 film deposition occurred at pH = 6.0 and was followed by the oxygen vacancy formation peak at E 1/2 = ~-0.03 eV vs.
Ag/AgCl. Cycling at less acidic (i.e., more basic) pH values of 6.5 and 7.0 also showed the presence of an oxygen vacancy formation peak. In effect, a minimal pH of 6.0 is required in order to form an oxygen vacancy at accessible energies of approximately -0.1 eV vs. Ag/AgCl, depending on pH of the environment. i), where the significant differences in peak areas are clear. Further, using the thickness of the FTO (~700 nm), the volumetric capacitances of the FTO substrate were calculated to be a factor of ≤5 x 10 -10 less than that of the ultrathin CeO 2-x film (~24 nm). These capacitance values for both the FTO and ultrathin CeO 2-x films are given in Table 1.  Owing to the determined negligible impact of the FTO on the total capacitance of the CeO 2-x , this contribution can be disregarded. However, in order to illustrate the precision of the calculations, a note has been added to the main text as given below.

Textual Modification (Page 14, Electrochemical Measurements section)
In order to confirm that the total capacitances obtained can be attributed to the ultrathin CeO 2-x films, the volumetric capacitance of bare FTO was determined to be insignificant (≤5 x 10 -8 % of that of the CeO 2-x capacitance).

Comment 2:
Based on the SEM images, it could be observed that the deposited CeO 2 covers the particles of the FTO substrate and the gaps between the particles are still there. Moreover, the average size of the particles is about 50 nm, which is close to the thickness of the deposited CeO 2 "film". How can the authors obtain the thickness of the deposited "film" with AFM?

Response:
The authors thank the reviewer for this perceptive comment.
The original text provided data for the film thicknesses over only the limited area of TEM images.
Consequently, in order to confirm the accuracy and precision of these data, AFM imaging was done over a large area of ~210 μm 2 at the stepped interfaces between FTO substrates (uncoated darker area on the right side of the AFM image in Figure 2) and the ultrathin CeO 2-x films (coated brighter area on the left side of the AFM image in Figure 2). The reliability of the data was increased further by effectively determining areal scan using 128 scan lines to generate the mean profiles rather than the single linear-step technique. The consistency of the thicknesses of the ultrathin CeO 2-x films is illustrated using representative raw AFM data accompanied by the corresponding height profile, as shown in Figure 2. To ensure the accuracy of the AFM measurement, the calibration of the AFM scanner was checked using a standard 180nm calibration grid and the height sensor measurement was precise and within 1% variation. The mean thicknesses of the films were evaluated by measuring the overall height difference between the area covered by the thin film and the FTO substrate. Further, TOF-SIMS imaging (Supplementary Fig. 9 and 10), which is based on elemental mapping over a large area of 900 μm 2 , confirmed the reliability of the AFM thickness measurements. Going one step further, the optical reflectance method of ellipsometry also was done over the very large area of 2.0 mm 2 , which indicated the thickness homogeneity for individual films. All of the new data are given in Table 2. Since the ellipsometry data provide a fourth confirmatory set of thickness measurements, these data have not been included to the main drafts. Nonetheless, details of the process are given as follows: The measurement was done using a Nanocalc-2000-UV/VIS/NIR (Ocean Optics, FL, USA) instrument using deuterium light with wavelengths in the range 250-450 nm. The light-covered area of the sample surfaces was 2.0 mm 2 . In order to simulate the experimental spectrum, NanoCalc software was used.
Prior to measurement, the instrument was calibrated using a bare FTO substrate as the reference sample. The thicknesses of both the functional layer (CeO 2-x ) and the sublayer (FTO) were calculated by taking the measurements in reflection mode at three random points on the surface of each sample. Table 5 is wrong. The header is "oxidation of Ce 3+ ", however, the chemical equation describes the change of Ce 4+ to Ce 3+ . Please correct it and recalculate the energy.

Response:
The authors thank the reviewer for noting the mistake in the Table heading. As for the calculation, Supplementary Table 5 shows thermodynamic calculations and results for the energy required for the (Ox 1 -Re 1 ) redox reaction, leading to the relevant potential (E 1/2 ). This redox reaction includes both Ox 1 , which is attributed to the oxidation of Ce 3+ to Ce 4+ , and Re 1 , which is attributed to the reduction of Ce 4+ to Ce 3+ . Thus, the theoretical potential (standard cell potential) can be obtained according to the Nernst equation to generate the theoretical E 1/2 by calculation of the total Gibbs free energy using either the reduction or oxidation reaction. Experimentally, the standard typical method to determine the required energy and the corresponding potential is done by averaging the peak potentials of the redox reaction (Ox 1 and Re 1 ), hence E 1/2 = ( E Ox +E Re 2 ); this is given in Supplementary Table 5.
Consequently, the calculations are sound but we have clarified the issue by correcting headers of Supplementary Tables 5, 6, and 7.

Comment 4:
In Methods part of "synthesis of ultrathin film", the pH of the solution should be listed.

Response:
As requested, the pH value was included in the "synthesis of ultrathin film" section.