An unexpected phase transformation of ceria nanoparticles in aqueous media

Abstract

Cerium oxide Nanoparticles (CNPs) are of significant interest to the scientific community due to their wide spread applications in a variety of fields. It is proposed that size dependent variations in the extent of Ce3+ and Ce4+ oxidation states of cerium in CNPs determines the performance of CNPs in application environments. To obtain greater molecular and structural understanding of chemical state transformations previously reported for ceria ≈ 3 nm nanoparticles (CNPs) in response to changing ambient conditions, microXRD and Raman measurements were carried out for various solution conditions. The particles were observed to undergo a reversible transformation from a defective ceria structure to a non-ceria amorphous oxy-hydroxide/peroxide phase in response to the addition of 30% hydrogen peroxide. For CNPs made up of ~8 nm crystallites, a partial transformation was observed and no transformation was observed for CNPs made up of ~ 40 nm crystallites. This observation of differences in size dependent transition behavior may help explain the benefits of using smaller CNPs in applications requiring regenerative behavior.

Introduction:

Cerium oxide nanoparticles (CNPs) are widely used in a variety of applications ^{1, 2} including those associated with catalysis ^{3, 4}, fuel cells⁵, solar energy ^{6, 7} and protecting human health^{4, 8, 9}. However, there are also concerns about the fate and possible deleterious impacts of CNPs in the environment and potentially toxic effects on human exposure ^{10, 11}. Relevant to many applications is the ability of CNPs to transform between Ce³⁺ and Ce⁴⁺ oxidation states, alter structure and change shape ^12-14. Greater knowledge of the nature of the transformations of CNPs is important to understand the desirable and undesirable properties of CNPs and the observed inconsistencies in biological and chemical behaviors of CNPs in the literature ^{10, 11, 15, 16}.

Bulk cerium oxide has two well-known stable stoichiometries: the cubic fluorite-type dioxide (CeO₂) and the hexagonal cerium sesquioxide (Ce₂O₃). The close thermodynamic stability of these stoichiometries leads to a continuous range of partially reduced CeO_2-x phases, where oxygen vacancies can be rapidly formed or eliminated¹⁷. The oxygen capture and release capability is important in a number of catalytic processes and is responsible for some of the useful biomedical applications. Several research groups have studied the physical, chemical and structural changes in cerium oxide nanoparticles (CNPs) as a function of their size and reported intriguing observations and identified a variety of reasons for size dependent behavior ^18-21. These studies report chemical transformations of CNPs in different environments ²² and biological systems^{13, 16, 23, 24}. The dopant and size dependence of the amounts of Ce³⁺ in CNPs have been studied and higher concentrations of Ce³⁺ have been frequently reported in smaller size particles ²⁵. Colvin²⁶ and Seal ²⁷ research groups have shown the influence of particle size on the behavior of CNPs with reference to their applicability in biomedical applications. In a recent review examining the observed variation in CNP behaviors, Gagnon et al. ¹⁰ found that biologically protective effects were more commonly, but not always, found for CNPs with diameter < 20 nm.

Dynamic properties, specifically the regenerative oxidation state switching, of CNPs have been found to contribute to its biomedical applications. The switching behavior has been observed to be moderated by the presence of ligands and adsorbates in both engineered and natural environmental and biological systems ^{13, 14}. In previous work, we reported a reversible transformation in chemical state of CNPs made up of ~3 nm crystallites that agglomerated to particles of ≈ 12 nm ¹² in response to the addition of H₂O₂ to an aqueous solution. Biological studies of particles made by this process were found to have beneficial antioxidant behaviors ^28-32. The detailed nature of environmentally driven transformations of CNPs is difficult to fully characterize by ex situ methods because environmental changes can alter the particles ³³. Some of the useful in situ techniques cannot provide the level of direct molecular information important to understanding details of the processes involved. Such challenges highlight the importance of semi in situ, close to in situ conditions but not absolutely in situ, characterization methods ^34-36 and a combination of techniques is often required to understand the mechanisms driving the regenerative transformations and to minimize unintended changes in particles during analysis ^{16, 18, 37}.

Colleagues and reviewers of our earlier observations of chemical state transformations ¹², challenged us to learn in more detail the nature of the transformations observed. Although changes in the chemical state of CNPs have been observed to be size dependent, other experimental and theoretical studies have found that the presence of surface water can also alter the state of the particles ³⁸. Our observations of the influence of changes in an aqueous solution environment potentially adds an additional variable influencing the chemical state of CNPs. In this paper we report a more detailed examination of the nature of this peroxide driven transformation using (semi) in situ characterization. Structural and chemical changes during interaction of CNPs in solutions in response to the addition of H₂O₂ were monitored using microX-ray diffraction and Raman spectroscopy.

We observed that in response to the environmental changes – adding H₂O₂ to the CNPs, as in the case of suspensions – the particles made up of ~ 3 nm crystallites fully transformed from a ceria structure to an amorphous complex and returned to the crystalline cerium oxide phase upon solution aging. For comparison, after the addition of a H₂O₂ containing solution ~ 40 nm ceria nanoparticles were not observed to undergo this transformation and particles made up of crystallites of ~ 8 nm appeared to partially transform during the time frame of our experiments. Observations from both microXRD and Raman studies indicate that CNPs of small size can transfer fully between a peroxide phase consisting of Ce, O, H when in the Ce⁴⁺ state and a highly defected cerium oxide when in Ce³⁺ state. Larger sized CNPs either do not readily undergo such transformation, partially transform or transform at a much slower rate.

Materials and Methods:

Cerium oxide nanoparticles:

CNPs primarily used in this study were synthesized in the laboratory while larger particles used for comparison were purchased from different vendors. The sizes reported throughout the discussion will be the average crystallite size as it is an appropriate representation of the individual crystallite and particle reactivity. Therefore after this section we will refer to the particles as the 3nm, 8 nm and 40 nm particles, regardless of other measurements of particle size.

The CNPs with ~3nm crystallite size were synthesized by dissolving Ce(NO₃)₃. 6H₂O, in DI water (18.2 M.Ω.) at room temperature, and a stoichiometric amount of 30% H₂O₂ was added drop-wise to yield a final 30 mM CNPs suspension as reported previously¹². The aging of this solution over a period of 15-20 days yields CNPs with predominantly Ce³⁺ oxidation state Nanoparticles in as-synthesized or as-exposed to hydrogen peroxide in solution had a strong yellow/orange-red color due to cerium in Ce⁺⁴ oxidation state, while crystallites aged in-solution were slightly yellow or clear due to the presence of cerium in predominantly Ce³⁺ state. The chemical state transformation of particles produced by this process were reported in some detail previously¹².

The ~ 10-20 nm particles were purchased as suspension from Alfa Aesar (20% in solution) and used as is without any further purification or treatment. These particles were received as an acetate ion stabilized suspension at pH 3.0. The average crystallite size of these nominally 10-20nm particles was found to be 8nm

Larger ~50nm particles were purchased from Nanostructured and Amorphous Materials Inc. as powder. The crystallite sizes of all three nanoparticles were determined using the Scherer equation from the XRD measurements. The crystallite size of these particles was observed to be ~ 40 nm.

Preparation of particles for microXRD measurements:

Suspensions of all the three sizes of Nanoparticles were filtered using a 20 nm filter A small agglomerate/chunk/portion of wet material was loaded into a capillary, carefully removed from the filter paper with a laboratory spatula (as indicated in the insert of Figure 3) for the microXRD measurements. Particles loaded in the capillary were aged, allowed to dry inside the capillary at laboratory conditions. Diffraction patterns collected for as loaded nanoparticles, after adding 30% hydrogen peroxide and also after allowing further aging. To compare the structural changes that take place upon reaction with H₂O₂ the same amount of 30% hydrogen peroxide was added to the dried particles in capillary and allowed to react for 1 hour.

Figure 3.

MicroXRD data of cerium oxide samples (~3nm crystallite size) after aging (high Ce³⁺) and after the addition of H₂O₂. The inserts show digital images of the nanoparticle and the color change with time as captured on the microXRD instrument, a broken capillary is also clearly visible in the images. The plots are overlaid for clarity and the intensities are not on the same scale and hence should not be compared quantitatively.

Preparation of particles for Raman measurements:

For Raman studies the nanoparticles were collected over the filter paper before and after reaction with hydrogen peroxide. These particles were used for a variety of further studies looking at impacts of aging, drying and peroxide additions. Some of the 3 nm particles were collected immediately after synthesis and, as observed in previous work, were equivalent to aged particles to which H₂O₂ was added.

X-ray Diffraction:

A Rigaku D/MAX RAPID II microdiffractometer with a curved imaging plate and a rotating Cr anode operating at 35 kV and 25 mA was used for the x-ray diffraction measurements. An optical prefix consisting of horizontal and vertical mirrors specific to Cr Kα radiation was employed in the incident beam to provide focused x-rays at the sample position. At the end of the incident beam path, a 0.3 mm collimator was used to optimize the resolution and signal to noise ratio of the diffraction patterns. The incident beam path from x-ray generator to the collimator is protected by O-ring sealed tubing and it is purged with He gas to improve the x-ray flux at the sample position. For the diffraction measurements of dried samples, thin film on Si substrate was mounted on a stainless steel sample-holder. JADE 8.5 from Materials Data Inc. and PDF4+ data base from ICSD were used to analyze x-ray diffraction data.

Raman Spectroscopy:

Raman spectra were acquired on wet and dry powder samples directly from the filter paper in a backscattering configuration using a Spex (Edison, NJ) Model 1877 Raman spectrometer. The 488 nm line of a Coherent (Santa Clara, CA) Innova 307 Ar+ ion laser was used for excitation and the detector was a Princeton Instruments (Trenton, NJ) LN/CCD detector. Laser power was attenuated to <10 mW at the sample to minimize decomposition. The slit width was 400 μm and the exposure time was 100 s for all samples. Spectral acquisition and data analysis were performed using Princeton Instruments Winspec and Galactic Industries (Salem, NH) Grams/32AI software, respectively. The estimated uncertainty of the peak frequencies is ±1 cm⁻¹. As described by Patil et al ³⁹ “Cerium oxide has a cubic fluorite-type structure and belongs to the Oh 5 Fm3m space group. This structure has six optical-phonon branches, which yield three zone-center frequencies. These frequencies are 272, 465, and 595 cm⁻¹, corresponding to the doubly degenerate TO mode, the triply degenerate Raman-active mode, and the nondegenerate LO mode, respectively. The triply degenerate Raman-active mode frequency can be directly detected by Raman measurements, whereas the TO and LO frequencies are given indirectly by fits to infrared reflectivity”.

Results and Discussion

In our earlier work, as summarized in Figure 1, it was shown that the CNPs prepared by wet chemical methods using hydrogen peroxide as an oxidizing agent initially produced, within about a day, CNPs with cerium in predominantly +4 oxidation state ¹². Aging CNPs in solution at room temperature leads to a slow transition of cerium from +4 to +3 oxidation state. Detailed UV-Visible spectrophotometry, transmission electron microscopy and X-ray photo electron spectroscopy analyses of the as-synthesized and aged CNPs was reported previously ^{12, 40-42}. This cycle of oxidation and reduction of the CNPs sets the framework for the experiments described in this paper. As reported in earlier studies, TEM images of particles extracted from solution at different times and with different oxidation states did not appear to differ³⁸. The Raman and XRD experiments described above were undertaken to learn more about what was happening to the particles during the transitions and in presence of H₂O₂.

Figure 1.

Schematic representation of the regenerative capability and oxidation state switching of ceria NPs in an aqueous environment. The pictures of bottles containing the NP suspension in DI water indicate the color changes after different aging periods. Adapted from ¹² and used by permission.

Raman spectroscopy data collected during CNP synthesis were collected at stages consistent with Fig 1, during synthesis, after particle aging and upon the addition of H₂O₂ solution Fig. 2. The discussion of the Raman spectra for each of the crystallite size focuses on the relative signal strengths of two Raman lines: i) ~460-470 cm⁻¹ (cerium oxide) and ii) ~ 845 cm⁻¹(cerium oxyhydroxide/peroxide )³⁹. A third Raman line at ~ 877 cm⁻¹ (surface or unreacted peroxide) could be observed soon after the addition of H₂O₂ solution to dried or aged particles, but is not essential to this discussion as it disappears within a few hours after adding the H₂O₂ solution to the particles. The presence, absence and relative strengths of peaks i) and ii) provide information about the transformations.

Figure 2.

Raman analysis of 3 nm CNPs at different stages of their oxidation reduction cycle. Raman spectroscopy data collected immediately after synthesis, after particle aging and upon the addition of H₂O₂ solution to crystalline CNPs and aged, after addition of H₂O₂, at room temperature for 3-4 days. 2-5 units variations in the wave numbers and change in width of peaks may be ascribed to defects in the particles.

As synthesized 3 nm CNPs in the +4 oxidation state did not show any peak corresponding to cerium oxide (~460-470 cm⁻¹), however a peak at ~ 845 cm⁻¹ was observed. After aging at room temperature for 15 days, the 845 cm⁻¹ peak intensity decreased and a peak at ~ 450-460 cm⁻¹ evolved indicating the transformation to nanocrystalline cerium oxide. For similar aging periods where this peak appeared, optical and XPS measurements indicated the presence of Ce ³⁺ in CNPs [31], i.e. higher oxygen vacancy concentration ³⁵. When these aged 3 nm CNPs were treated with H₂O₂ the peak at 450-460 cm⁻¹ disappeared and the peak at 845 cm⁻¹, as observed for the freshly synthesized 3 nm particles, reappeared. At roughly 3-4 days after the addition of H₂O₂ solution, the peak at 845 cm⁻¹ started to disappear and the peak corresponding to crystalline cerium oxide reappeared. To test for reproducibility of the non-ceria peak, a subset of the studies were repeated at a later time on a different Raman system by a researcher not familiar with the earlier measurements. These measurements were consistent with the observations, particularly observing the presence of the ≈ 845cm⁻¹ peak upon the addition of H₂O₂ solution.

These cyclic observations of disappearance of the cerium oxide Raman peak upon addition of hydrogen peroxide and its reappearance upon aging in air suggests that the cycling of cerium oxide proceeds with a transformation between a defected ceria state and an oxyhydroxide state. This state is metastable and decomposes as the solution ages at RT and in ambient air to generate crystals of cerium oxide nanoparticles.

The presence of the CeO₂ peak at a lower wave number (450-460 cm⁻¹) than characteristic of bulk ceria (465 cm⁻¹) corroborates well with the observations by Scholes and others ⁴³. A shift to the lower wave numbers for nanosized particles is attributed to contributions from phonon confinement, strain, size distribution, defects and variations in phonon relaxation with particle size.

The presence of a Raman peak characteristic of an oxy-hydroxide, ~845 cm⁻¹ peak, immediately raises questions of particle structure. Therefore microXRD data also were collected on aged particles and after the addition of H₂O₂. Consistent with the cerium oxide peaks observed by Raman, aged (dry) 3 nm crystallites show clear diffraction peaks corresponding to the fluorite lattice of cerium oxide as labeled in Fig. 3 in red. A lower intensity broad peak corresponding to a 2θ value of ~ 31° was also observed in the diffraction pattern along with fluorite structured cerium oxide peaks. The broad peak could not be matched with any known peaks of cerium oxide or related structures from the literature or the data base. Addition of few drops of 30% H₂O₂ to this sample caused the fluorite structure to disappear and the relative intensity of the broad peak corresponding to a 2θ value ~ 31° increased. The inset digital images in Figure 3 show the change from yellow colored CNPs for the aged sample (Ce³⁺) to orange red (Ce⁴⁺) after the addition of H₂O₂. In this transition stage it is not possible ascertain whether these particles are pure cerium oxide Nanoparticles or some other intermediate phase is not possible without further supporting data.

These results add a new dimension to previous work on the interaction of cerium with H₂O₂ which has been of significant interest to the researchers working in catalysis, conversion coatings, and biomedical applications. The cerium-H₂O₂ interaction resulted in the formation of cerium oxyhydroxide/peroxide; Ce(H₂O₂)³⁺ with further reactions expected to produce a O₂⁻² containing species Ce^IV (O₂)(OH)₂. Such peroxide species were reported to result in Raman peaks around 840-850 cm⁻¹. In an earlier study ⁴³ on conversion coatings, it was shown that the formation of peroxide species is enhanced with an increase in the amount of hydrogen peroxide. It was also demonstrated with the help of Raman, XPS and XRD that the crystallite size of cerium oxide in the coatings was smaller with higher amount of peroxide⁴³. While the Raman spectroscopy data reported previously indicated the formation of peroxide phase during the synthesis of cerium oxide, our study shows that CNPs can undergo this reactive transformation between cerium oxide and cerium oxy-hydroxide/peroxide even after the formation of CNPs.

Because of previous studies suggesting size dependent variations in the behaviors of CNPs, we undertook additional micro-XRD and Raman measurements on 8nm and 40 nm CNPs. Comparison of the Raman data for 8 nm and 40 nm CNPs is shown in Fig. 4. The 3 nm particle data shown demonstrates, Fig. 3, the full cycle including spectra from as synthesized, aged, after the addition of H₂O₂ and further aging after the addition. The spectra for the 8 nm and 40 nm CNPs are shown only up to 2 hours after the addition of H₂O₂. At 5 minutes, the spectra show very low intensity peak of crystalline cerium oxide, which is attributed to the masking effect of the residual hydrogen peroxide on the particles. Spectra collected 2 hours after the addition of hydrogen peroxide, show that both the ~843cm−¹, oxy-hydroxide/peroxide, peak and the ~460-470 cm⁻¹ cerium oxide are present for the 8 nm particles, but only a 466 cerium oxide peak is present for the 40 nm CNPs. It should be worth noting that along with the peroxide peak at ~ 860-87 cm⁻¹, the oxyhydroxide/peroxide peak is also present in case of 8nm particles whereas in case of 40 nm particles the characteristic peak of oxy-hydroxide(peroxide) peak is missing. This further strengthens observation that 8 nm particle is partially transformed by the addition of H₂O₂ resulting in the intermediate phase but the 40 nm CNPs show no evidence of a transformation. The peaks at 860-870nm are attributed to peroxide peaks due to presence of residual hydrogen peroxide.

Figure 4.

Raman spectra from 8 nm and 40 nm CNPs. The plots for the 8 and 40 nm CNPs show data of before addition of H2O2, 5 minutes and 2 hours after the addition of H₂O₂ solution showing the absence of a oxy-hydroxide/peroxide peak for the 40 nm particles and the presence of both the ceria and oxy-hydroxide peak for the 8 nm particles. To ascertain the consistency and uniformity of information, data was collected from multiple locations. The peaks in the graph near 860-870 nm are the peroxide peaks – due to residual hydrogen peroxide left on the particles. Intensities of the Raman peaks shall not be used for quantitative comparison but for qualitative understanding alone.

The microXRD data for the 8 nm and 40 nm particles indicate a similar behavior as shown in Fig. 5. The diffraction peaks confirm mostly the structure of CeO₂ after the addition of H₂O₂ with the presence of a low intensity peak at the a 2θ value of ~ 31° for the 8 nm particles and no evidence was observed changes in crystalline cerium oxide for the 40 nm particles. As might be expected, no major color changes were observed in 8 and 40 nm CNPs upon addition of H₂O₂ (data not shown).

Figure 5.

MicroXRD data from 8nm(left) and 40nm (right) CNPs before and after(inset) addition of H₂O₂ solution to the particles. The presence of an amorphous phase, as observed for the 3 nm particles is not observed for the 40 nm particles and present to a small degree for the 8 nm particles.

In the case of reaction between hydrogen peroxide and 3 nm/8 nm ceria crystallites; microXRD data confirm the presence of an amorphous like phase (at a 2θ value of ~ 31°) whose intensity varies with the CNP size. Hence, it is proposed that the amount of peroxide phase (amorphous in nature) is dependent on the extent of reaction or transformation of CNPs and decreases with increased crystallite size of CNPs. The presence of the low intensity peak at 2θ value of ~ 31°in case of 3 nm crystallites before the addition of H₂O₂ can be attributed to the presence of this phase from the synthesis method adopted by the authors involving H₂O₂. It is important to mention that irrespective of the method of synthesis, addition of H₂O₂ can produce the 31° peak, as evidenced in the case of 8 nm crystallites.

With aging or drying at room temperature after the addition of hydrogen peroxide, the crystalline cerium oxide peaks reappeared over time. Corresponding Raman and XRD data confirmed the time and crystallite size dependence of the transformation. Relatively smaller nanosize ceria particles with larger surface energy and defect concentration result in visible chemical transformation from crystalline cerium oxide to an amorphous peroxide phase in presence of H₂O₂ as confirmed by the broadening and shift of the Raman peak^{39, 43-46}. Absence of cerium-peroxide phases in the spectra obtained for 40 nm in presence of H₂O₂ indicates greater stability of these CNPs (relative to 3 and 8 nm crystallites) at 40 nm size. The transformation tendency of the 8 nm CNPs is observed to be in-between the highly reactive 3nm CNPs and relatively stable 40 nm CNPs.

As the peroxide phase is observed to be dependent on the extent of transformation or the extent of re-transformation, we propose that the regenerative oxidation state switching in CNPs with crystallite sizes below 10 nm may involve formation of molecular structures that do not appear to have a flourite structure of crystalline cerium oxide. The environment induced switching of Ce between the +3 and +4 oxidation state in solution¹² and in biological systems ^{13, 23} has been observed by many research teams. However, the identification of a reversible phase with a non-ceria structure using optical absorption spectra was not reported and was likely observable by the methods used. The use of semi in situ studies to monitor the transformation of nanoparticles with microXRD and Raman studies enabled the identification of the formation of a peroxide phase for small crystallites in a peroxide rich environment.

The observed ability of surface species to induce phase transformations in smaller sized nanoparticles is consistent with other studies. For example the adsorption of water on 3 nm ZnS altered both the surface and bulk structure ⁴⁷. The authors of that study noted that ‘the structure and reactivity of nanoparticles will depend on both particle size and the nature of surrounding molecules⁴⁷. As other examples, varying of surface ligands was found to control the crystal structure of ~ 2 nm CdSe NPs ⁴⁸ and ligands were observed to change the structure of nano-sized Au square sheets from hcp to fcc ⁴⁹. In the current study it is proposed that the peroxide species react with surface cerium ions in 3nm CNPs to transform the crystallites from a cerium oxide phase to a cerium oxyhydrdoxide/peroxide because of the relatively high number of surface cerium atoms and a generally higher number of structural defects due to particle instabilities ²⁵. As the crystallite size grows, the reaction still takes place, but there is a lower relative number of surface cerium ions that undergo transformation to a cerium oxyhydroxide/peroxide as compared to the bulk crystalline cerium oxide and the larger particles have fewer defects and greater stability. At even higher crystallite sizes, the reaction is negligible compared to the bulk cerium ions and thus the particles do not show a reactive transformation from crystalline cerium oxide to a cerium oxyhydrdoxide/peroxide and its regeneration as highly defective cerium oxide.

The measurements reported primarily focused on observing transformations on a previously reported type of CNP, but also involved the use of two additions, particles of different particle sizes from different sources. Although we have called out the importance of size, it is known that both size ^{10, 26, 27} and history ^{10, 11} can influence particle behaviors, but size, associated defective structure in smaller size, seems to be significant. The identification of a non-ceria transition phase for some types and sizes of crystallites may assist the efforts to understand the differences in biological impact of CNPs of different sizes and from different sources.

Conclusion:

The authors started these experiments with the intention of obtaining greater molecular and structural understanding of the transformations previously observed for 3 nm CNPs. For these particles a reversible transformation from a ceria to a non-ceria phase during the reaction of CNPs with hydrogen peroxide was found. For CNPs with crystallite size of 8 nm, a partial transformation was observed and no transformation was observed for 40 nm particles. These observations are consistent with the frequently reported importance of surface ligands on CNP transformations ^{2, 13, 16} and the apparent benefits of small particle size on the regenerative behaviors of CNPs¹⁰. We note that during the synthesis process for the 3-5 nm crystallite CNPs, the peroxide Ce⁴⁺ phase is intermediate between the initial solution and the final highly defected Ce³⁺ phase with a ceria structure. However, the smaller particles have the ability to transform between these two phases.

Considering cerium oxide’s ability to scavenge radicals and regenerative oxidation state switching ability, it appears that the ability of CNPs to transform between Ce⁴⁺ and Ce³⁺ rich phases is facilitated by small size, but is not always constrained to be a transformation between defected and non-defected ceria phases. Likely the tendency of small particles to have many vacancies in the ceria structure ²⁵, leading to the formation of the highly defected Ce³⁺ phase with a ceria structure facilitates the transformation to other phases. The specific nature of the phases formed likely depends on the solution environment and the potential ligands contained ^{13, 16}. We note that in phosphate containing environments Ce phosphates can be formed¹ and that the presence of polymers can alter the extent and rate of Ce +3 to +4 transformation ⁴¹.

It is also evident from this study that it is important to understand the size range in which nanoparticles will have the desired materials properties such as higher reactivity or stability and the ability to perform in various application environments.^{10, 35} Size dependent reactive transitions, along with any influence of the synthesis methods used, may impact some of the desired behaviors and may be the reason behind the contradictory reports of CNP properties that appear in the literature ^10-12. It is essential to consolidate the observations reported in this article with additional studies including first principle modelling to ascertain the exact nature of the intermediate cerium-peroxide species and the type of molecular bonding involved for the systems and environments involved. Such studies should be aimed at unequivocally establishing the chemistry and structure of this cerium-based-peroxide intermediate phase (and potentially other similar phases in different environments) and provide the necessary impetus to move closer to practical use of cerium oxide nanoparticles in biomedical applications including cancer treatment.