The effect of composition on structural, thermal, redox and bioactive properties of Ce-containing glasses

TheeffectofphosphateontheabilityofCe-containingbioactiveglassestoinhibitoxidativestress wasstudiedon compositions based on Hench (46.2%SiO 2 \\ 24.3%Na 2 O \\ 26.9%CaO \\ 2.6P 2 O 5 , mol%) and Kokubo (50.0%SiO 2 \\ 25.0%Na 2 O \\ 25.0%CaO) glasses. In particular, the reduction of catalase mimetic activity of Ce-containing glasses duetothepresence:i)ofP 2 O 5 intheglasscompositions,andii)ofphosphategroupsinthesolutionemployedfor catalase mimetic activity tests was explained and rationalized by combining SEM, XPS, XRD, DTA, FT-IR and UV – vis experiments with Molecular Dynamics simulations. The results suggest that the Ce ions play a different structural role in the two series of glasses. In particular, in phosphate free glasses Ce is coordinated by non-bridging oxygens (NBOs) originated from the disruption of thesilicatenetwork,whereasinphosphatecontainingglassestheNBOsaroundCeionsbelongtoorthophosphate groups. The latter groups stabilize the Ce 3+ species subtracting them from the interconversion process between Ce 3+ and Ce 4+ , which is of fundamental importance for the exhibition of the catalase mimetic activity. ©


Introduction
Some soda-lime-phosphosilicate glasses, such as Hench's Bioglass 45S5, when in contact with the body fluids, form a layer of bone-like apatite on their surfaces which allows the connection between hard tissues (bone) and soft ones (collagens and muscles) [1].The mechanism of formation of this apatite layer proposed by Hench is based on a long series of reactions occurring at the interface between the glass and the biological medium [2,3].The degree of reactivity exhibited in the biological environment is related to the chemical composition of glasses [4].In the last years, several ions have been added in the composition of bioactive glasses in order to introduce peculiar properties.For this reason, it is important to investigate with experimental and computational methods the structural role of the doped ions on the glass structure and reactivity.[5][6][7][8][9][10].
Very recently, we demonstrated that the introduction of small amounts of CeO 2 to the Hench's Bioglass® confers antioxidant properties, such as catalase mimetic activity [11].This enzyme catalyzes the decomposition of hydrogen peroxide (H 2 O 2 ) into water and oxygen, thus protecting cells from oxidative stress by reactive oxygen species (ROS) [12], such as hydrogen peroxide [13].We showed that the ability of these glasses to present catalase mimetic activity is strictly related to the simultaneous presence of Ce 3+ and Ce 4+ on the glass surface, as it happens in CeO 2 nanoparticles (CeNPs).In fact, the enzyme mimetic activity of Ce 3+ and Ce 4+ -containing materials was demonstrated for the first time by means of CeNPs [14][15][16][17][18][19].The authors showed that the amount of the Ce 3+ /Ce 4+ ratio is strictly related to the biocatalytic mimetic activity exhibited by these nanoparticles; the relative concentration of Ce 3+ and Ce 4+ is mainly affected by the dimensions of the CeNPs [19][20][21], and the kinetic of redox conversion Ce 3+ ↔ Ce 4+ depends on the environment [22].
Moreover, Singh et al. [23] showed that the interaction of phosphate groups with Ce 3+ ions present into the Ce NPs leads to the complete disappearance of Super Oxide Dismutase (SOD) activity and concomitant increase in catalase mimetic activity, concluding that a high surface concentration of Ce 3+ oxidation state favors SOD mimetic activity, while materials with a high surface concentration of Ce 4+ possess higher catalase mimetic activity.
Thus, a deep understanding of the structural role of phosphate as a component of the biomaterials and of the biological environments is required.
In our previous work [11], the catalase mimetic activity of Cecontaining bioactive glasses was investigated for a series of Ce-doped glasses based on the Hench composition in water solution.However, the antioxidant properties of these materials are strongly influenced by the composition of the solution in which they are tested, thus further investigations on their performance in the physiological medium is needed.
In particular, the aim of this work is to investigate the effect of the addition/presence of phosphate groups on the catalase mimetic activity of Ce-containing bioactive glasses with different compositions.Two series of Ce-containing potential bioactive glasses, one based on Hench's composition (where phosphate is present) and the second based on the P-free Kokubo's composition [24] have been synthesized and characterized by using Scanning Electron Microscopy (SEM), Differential Scanning Calorimetry, UV-vis, XRD, X-ray photoelectron spectroscopy (XPS) and ICP Spectroscopy and Molecular Dynamics Simulations.Bioactivity tests have been performed in simulated body fluid solution (SBF) [25] whereas the catalase mimetic activity tests have been carried out in both water and SBF.

Synthesis of the glass samples
Two series of bioactive glasses doped with variable percentages of cerium oxide have been synthesized via the melting method [11].The parent glass of the first series, hereafter identified as H series, is the composition of the well-known 45S5 Bioglass®, whereas the parent glass of the second one (hereafter referred as K series) is the phosphatefree glass proposed by Kokubo et al. [24].The molar compositions of the samples discussed in this paper are reported in Table 1.The samples were prepared by mixing reagent grade SiO 2 , Na 2 CO 3 , CaCO 3 , Na 3 PO 4 •12H 2 O (only for H glass series), and CeO 2 in an agate mortar.Then each batch was put into a Pt crucible, and melted in an electric oven.The heating ramp was set to 15 °C/min up to 1000 °C and 8 °C/min up to 1350 °C.Samples were maintained at this temperature for 2 h, to ensure optimal melting and mixing of all the oxides, and finally quenched at room temperature on a graphite plate.The obtained glasses were homogeneous and transparent.
Attempts to synthesize K_5.3 (series K with 5.3% of CeO 2 ) and H_7.2 (series H with 7.2% of CeO 2 ) systems yield heterogeneous and opaque samples characteristic of partial crystallized materials.The glasses were milled in order to have diameters in the range 500-250 μm.The powder dimension was chosen in order to compare the present results with that obtained previously [11].
We heated each of the samples containing cerium to their crystallization temperature for 2 h, and then we performed the X-ray diffraction (XRD) analyses over the crystallized samples.

Characterization of the glasses
The powder glass samples were characterized before and after soaking in solution.The sample before the soaking was indicated "as quenched (AQ)".

UV-vis spectroscopy
UV-vis absorption spectra on AQ and soaked powders were collected to gain qualitative information about the oxidation state of cerium in the glasses.We used the HP8452 UV-vis spectrometer equipped with the diffuse reflectance sphere, and we collected all the spectra in the range 200-850 nm, using BaSO 4 as reference.

Differential Thermal Analysis (DTA)
Differential thermal analyses were performed over the AQ glasses to obtain the values of their melting points and their crystallization temperatures (T c ).We weighted ~30 mg of powders in a platinum crucible and performed the analysis in air.We set the scan rate at 10 °C/min, for a range between 20°and 1200 °C.

X-ray diffraction (XRD)
XRD measurements were carried out over the AQ and the soaked glasses.We analyzed all the specimens in the (2 θ) 10°-55°range employing an X'Pert PRO − PANAnalytical apparatus, equipped with Ni-filtered Cu Kα radiation (λ = 1.54060Å).

Infrared spectroscopy (IR)
The IR spectra were collected for the AQ and the soaked glasses in the range 400-4000 cm − 1 by using a FT-IR Perkin Elmer 1600 spectrometer.We performed the measurements over pellets prepared with 1 mg of glass and 100 mg of KBr.

X-ray photoelectron spectroscopy (XPS)
XPS was used to obtain quantitative information on the relative concentration of Ce 3+ and Ce 4+ ions on the surface of the AQ samples.The spectra were measured using Al K α photons as the exciting probe.The Ce 3d photoemission spectra were fitted using a Shirley-type background and five spin-orbit split doublets, three related to Ce 4+ ionic species, and two related to Ce 3+ ionic species, following previous works [26,27].The free parameters in the fitting were the areas of the peaks, while the shifts in binding energy position and the Lorentzian/ Gaussian width of the individual components were fixed to the values used in reference [11].The Ce 3+ and Ce 4+ relative concentrations were obtained from the relative weight of the area of the peaks related to each ionic species.The absolute values of the Ce 3+ and Ce 4+ relative concentrations evaluated in this way are approximate; however, their variation among the glass series may be considered reliable.

Molecular dynamic (MD) simulations
Since the H series was already characterized in a previous work [28] we have here generated structural models of the K series, only.Classical MD simulations of the K, K_1.2 and K_3.6 glasses were performed by means of the DL_POLY® 2.14 package [29] employing a wellestablished melt-quench computational protocol [30].
In this approach, an initial random configuration containing about 10,000 atoms enclosed in a periodic cubic box is melted at 3200 K for 100 ps and then cooled down to 300 K at a nominal cooling rate of 5 K ps −1 .The resulting glass structures have been subjected to a final NVT trajectory of 70 ps; the structural analysis was performed on 501 configurations sampled at regular intervals during the last 50 ps of MD trajectory.The box sides (reported in Table S1 of the ESI) were set accordingly to densities calculated with the Priven empirical method [31] (Table S1, ESI), which is encoded in the Sciglass package [32].
Because of the employment of non-experimental densities, the final glass models were relaxed for 0.2 ns, at 300 K and atmospheric pressure using the Berendsen NPT ensemble to check whether significant volume variations occurred after the relaxations.All boxes slightly expanded and for all of them the volume variations were comprised between 0.72 and 1.65%.The Ce 3+ /Ce 4+ ratio of each glass composition was fixed accordingly to experimental findings.The interatomic forces acting between the constituting ions have been described by using a shell model force field able to accurately reproduce the short and medium range structure of multicomponent oxide glasses [28,[33][34][35][36][37][38][39].In this model, cations are represented by rigid points bearing full formal charges (Si 4+ , Ca 2+ , Ce 3+ , Ce 4+ , Na + ), whereas anions are split in a shell of charge -Y and a massive core of charge Z (Z + Y is the total atomic charge; −2 for O ions), which are coupled adiabatically by a harmonic spring potential [24].The oxygen shell interacts with the Si 4+ , Ce 3+ , Ce 4+ , Ca 2+ and Na + cations through a Buckingham term, and Coulomb forces act between all species bearing a formal charge, in the Born Mayer model.Three-body screened harmonic potentials have been employed to control the O\ \Si\ \O angle, to guide tetrahedral coordination.Therefore, in the shell model, the polarizability of oxygen is straightforwardly included, and it is also environment dependent due to the forces acting on the oxygen shell by the other ions.This allows for a better glass structure relaxation during quenching and leads to glasses with improved medium-range structures with respect to those generated by using the rigid ion model [40].A detailed description of the force-field functional forms as well as the complete list of parameters is reported in refs.[28,33,34].

Catalase mimetic activity tests
The samples were soaked (under continuous stirring) in a 1 M solution of H 2 O 2 for24 h, 96 h, and 168 h.A constant glass mass/volume of solution ratio of 5 mg/mL was kept.The same tests were also performed in a solution 1 M of hydrogen peroxide in simulated body fluid (SBF).The SBF solution was prepared through the method explained in Ref. [25].After the soaking, we filtered the solutions, we measured the pH value and we determined the residual H 2 O 2 through titration with KMnO 4 .The reacted powders were dried in oven at 60 °C overnight, and characterized again after the treatment.

ICP-OES analyses
To quantify the release of the different ions in solution during the reaction we analyzed the SBF solutions before and after the contact with the glasses with the ICP-OES Perkin Elmer Optima 4200DV, equipped with ultrasonic nebulizer CETAC.

Bioactivity tests
The bioactivity tests in SBF were performed with the method explained in Ref. [25].We soaked 150 mg of powder samples in 100 mL of SBF for 7, 14, 21 and 28 days and then, we filtered the solutions, washed the glass powders with acetone and analyzed the solids with the FT-IR and the XRD to evaluate the formation of hydroxyl-apatite on the surface of the samples.

Environmental Scanning Electron Microscopy-Energy Dispersive Spectroscopy (ESEM-EDS)
After the soaking in SBF, we collected images of the reacted glasses to gain a morphological and elemental characterization of the surfaces and to determine if there were detectable apatite-like areas.The morphological and elemental analyses of the solid products were carried out by means of Environmental Scanning Electron Microscopy-Energy Dispersive Spectroscopy using a FEI Quanta 200 instrument (Fei Company, The Netherlands), equipped with an INCA 350 EDS apparatus (Oxford Instruments, UK); EDS analysis was performed over different areas of the sample and the maximum value of σ was 0.5%.

Results
A full characterization of the K series of glasses, here investigated for the first time, is reported.For the H series of glasses the catalase mimetic activity tests in H 2 O 2 + SBF and the characterizations performed after soaking are carried out and discussed for the first time, whereas the results of bioactivity, [41] and catalase mimetic activity in water [11], have been reported previously and are here considered for comparison purposes.

Characterization of the K series of glasses 3.1.1. UV-vis spectroscopy
Fig. 1a reports the UV-vis spectra of the K series of AQ glasses, together with the spectra of CeO 2 and Ce(NO 3 ) 3 •6H 2 O, as reference for Ce 4+ and Ce 3+ respectively.The K sample shows a weak band around 300 nm, consistent with the UV-vis signal of the soda-lime-silicate glasses [42].The doped samples show different spectra, with a strong absorption band between 350 and 400 nm, and a second weak band around 220 nm.Following the literature [43,44], we could assign the first signal to the charge transfer (CT) O 2 − 2p \ \Ce 4+ 4p transition and the second one to the electronic transition 4f 1 -5d 1 of Ce 3+ .In addition, the shoulder between 500 and 600 nm can be assigned to the CT Ce 3+ -Ce 4+ transition [45].Fig. 1b reports the spectra collected for the AQ samples of H_3.6 and K_3.6.It is possible to notice that at high λ (above 400 nm), the absorption is slightly higher for the K_3.6 sample.As previously proposed [28,40] this effect could be attributed to an increment in the content of Ce 4+ with respect to Ce 3+ .Thus, we can assert that in the K_3.6 glass there is a major content of Ce 4+ with respect to the H_3.6 glass, even if the content of CeO 2 is the same.The difference in the UV region (200-400 nm) between the H_3.6 and K_3.6 glasses is probably due to the phosphate groups present only in the H_3.6 sample.In fact, the band at around 300 nm is due to Si\ \O\ \(Si, Ca) absorption characteristic of silicate glasses, whereas the absorption at 250 nm is due to the P\ \O-(P, Ca) groups [46].In order to confirm this finding, the UV-vis absorption spectra of the H and the K glasses have been reported in Figure S1 of the ESI.The figure shows that the band centered at 250 nm in the H glass is not observed in the K glass.

X-ray photoelectron spectroscopy (XPS)
Fig. 2 reports the Ce 3d XPS peaks of the H_3.6 [11] and K_3.6 samples and the fittings obtained using the procedure described in the experimental section.In spite of the low signal to noise ratio, due to the low Ce concentration in the samples, some differences in the spectra of the two samples can be appreciated.In particular, the intensity of the band around 917 eV, related to Ce 4+ ions only, is much higher in the K sample spectrum.The Ce 4+ relative concentrations, evaluated from the fitting of the Ce 3d peaks, with the procedures outlined in the experimental section, are 26 ± 5% [11] for the H_3.6 and 46 ± 5% for the K_3.6 sample, and confirm the higher relative Ce 4+ concentration in K samples observed qualitatively by UV-vis spectroscopy.

Differential Thermal Analysis (DTA)
Fig. 3a reports the results of the DTA analysis performed over the K_1.2 and the K_3.6.We found that T c increases as the Ce content increases, indeed for the K sample the T c is 661 °C, as reported in Ref. [47], for K_1.2 it is 672 °C and for K_3.6 it is 685 °C.A similar trend of T c was also detected for the H series, in fact the value increase from 724 °C for the H glass to 735°, 749 °C for H_1.2 and H_3.6, respectively.However, the glass with the maximum CeO 2 content, H_5.3, showed a slightly decrease of T c to 738 °C (Fig. 3b).

X-ray diffraction (XRD)
The results of XRD analyses are shown in Fig. 4. As already reported by Lusvardi et al. [47] in a previous investigation, the XRD analysis performed after thermal treatment at 661 °C of the K glass reveals that the Na 2 CaSi 2 O 6 phase crystallizes from the glass.The presence of cerium does not affect the crystallization process; in fact, the Na 2 CaSi 2 O 6 cyclosilicate (JCPDS 77-2189) is the principal crystal phase observed in both the K_1.2 and K_3.6 crystallized glasses.For sake of clarity, only the three principal peaks: 2θ = 33.5°,2θ = 34.0°and2θ = 48.5°arehighlighted in Fig. 4. In addition, a weak signal at 28.5°characteristic of CeO 2 (JCPDS 34-0394) can be observed in the K_3.6 sample.The same signal is not present in the spectra of K_1.2 probably because of the small amount of cerium, and it is present together with the others two most intense peaks of the CeO 2 in the AQ K_5.3 glasses.The same analysis performed on the H series [11] revealed the presence of CeO 2 in the H_1.2, and H_3.6 systems; the intensity of the peak attributed to CeO 2 is more intense with respect to that detected for K_3.6 system.This experimental evidence highlights that in the P-containing systems (H series) the presence of phosphorous enhances the Ce\ \Ce clustering inside the glass network.However, the presence of PO 4 groups favors the crystallization in H_3.6 and H_5.3 systems of the CePO 4 phase.This suggests that in the H series the Ce ions are also surrounded by phosphate groups and form Ce-phosphate rich zones in the glass network.

Molecular dynamics simulations
The structural models generated by means of classical molecular dynamics simulations were analyzed in details in order to shed some lights on the short and medium range order of the K glass series as a function of the CeO 2 content and to compare them with those obtained for the H series. Table 2 reports the average X-O coordination numbers (where X = Si, Ce 3+/4+ , Ca and Na).As expected, silicon ions are fourcoordinated, and their coordination numbers are not influenced by the addition of Ce ions.
Ce 3+ ions are coordinated by about 6.3-6.4NBOs whereas Ce 4+ ions by 6.7-6.8NBOs and their coordination numbers (CNs) do not significantly change with CeO 2 content.These values are slightly smaller than those found in Ce-containing bioactive glasses of the Hench's family of about 0.2-0.3NBOs but are still in good agreement with the values obtained in previous MD investigations [28,48,49] and EXAFS and X-ray diffraction measurements [50][51][52] on phosphate glasses.These findings together with the observation that only NBOs are coordinated to Ce ions suggest that Ce first coordination shell does not drastically change with glass composition if a proper number of NBOs is present in the glass structure.As for the Na and Ca ions, their CN are in agreement with previous MD and NMR investigations on silicate glasses [53][54][55][56].The CN of Na slightly increases from 5.6 to 5.8, whereas that of Ca increases from 5.5 to 5.7 with the addition of CeO 2 .It is interesting to note that the number of NBOs and bridging oxygen (BOs) atoms coordinated to Ca remains essentially constant in the four compositions analyzed, whereas a depletion of BO and an increment of NBO are observed around Na cations upon CeO 2 addition.These are the same trends previously observed in the H series [11].
To investigate the effect of CeO 2 on the glass polymerization, the Q n distributions (Q n stands for 4-coordinated network former ion bonded to n bridging oxygens) of Si for the K, K_1.2, K_3.6, and K_5.3 glasses are listed in Table 3.
The un-doped K glass presents a trinomial Q n (Si) distribution dominated by Q 2 species (55.7%) with a similar population of Q 1 and Q 3 species (21.2%).By adding CeO 2, the population of Q 2 species drastically decreases to 43.6% (for K_5.3Ce glass) whereas Q 1 increases to about 40% and Q 3 decreases to 10.8%.
Therefore, the addition of CeO 2 leads to a more fragmented network with the network connectivity (NC) monotonically decreasing from 2.4 for the K glass to 1.6 for K_5.3 glass.At constant content of CeO 2 , the glass polymerization is more pronounced for the K series than the H one for which the NC decreases from 2.1 (H glass) to 1.7 (H_5.3 glass) [11].
The more marked influence exerted by Ce ions in the NC of K glass series is due to the absence of phosphate groups which are able to donate NBO to Ce ions, indeed, in the K series the number of NBOs necessary to stabilize the first coordination sphere of Ce ions can only be supplied from the silicate depolymerization.
To gain insights on the medium range order and on the cations distributions in these glasses the cation − cation coordination numbers (CN X-Y ) have been computed by integrating the first peak of the corresponding pair distribution functions, and reported in Table 4.
In previous investigation [11] we observed that the addition of CeO 2 oxide in the H glass series caused an increment in the number of Na ions in the second coordination sphere of Si: from 4.5, for H glass, to 5.9 for H_5.3.At the same time, the number of Na surrounding P decreases from 6.0, for H, to about 5.4-5.7 for Ce-containing compositions, whereas the number of Ca ions around Si and P cations was observed to be less dependent on the addition of CeO 2 .These data revealed that Ce ions tend to satisfy their coordination requirements by using NBO coming from the orthophosphate units.Table 4 shows that a different  phenomenon occurs in the K series.In fact, the number of Na ions in the second coordination sphere of the Si ions decreases from 6.7 (K glass) to 6.3 (K_5.3 glass), whereas the number of Ca ions around Si decreases from 3.5 to 3.3.
At the same time the number of Si ions around Ce(III) and Ce(IV) ions decreases from 7.0 to 6.9 and from 7.8 to 7.1, respectively.The number of Ca ions around Ce(III) increases from 2.2 to 2.5, while it remains constant around Ce(IV).The number of Na ions around Ce(III) ions decreases from 6.8 to 6.4, whereas it increases from 6.3 to 6.5 around Ce(IV).
This reveals that by adding CeO 2 , Na ions tend to migrate towards Ce(IV), whereas Ca ions towards Ce(III).
To investigate further which modifier among Ca and Na surrounds preferentially silicon and cerium ions we have computed the ratio between the number of Ca and Na in the second coordination sphere of Si, Ce(III) and Ce(IV) ions and compared it to the Ca/Na ratio in the nominal glass composition (which is 0.5).If the ratio is higher than 0.5 the cation is preferentially surrounded by Ca ions, whereas the contrary is true when the ratio is smaller than 0.5.The values reported in the Table 5 reveal that silicon has similar preference towards Ca and Na, whereas Ce ions tend to be surrounded by Na ions forming zones richer in Ce and Na.

H 2 O 2 degradation
Fig. 5 reports the results of the titrations of the solutions after the soaking of the glasses for different times.It is possible to notice that for all the solutions the degradation of hydrogen peroxide is ceriumdependent, increasing as cerium content increases.Furthermore, the degradation of H 2 O 2 changes with the environment, being higher in water solution than in SBF solution.In particular, we found that for both the series the most doped samples (H_5.3 and K_3.6, respectively) degraded 35% of H 2 O 2 after 168 h of soaking in the first solution, and only ca.25% of H 2 O 2 after the same time in SBF.

pH measurements
When the reactions take place in hydrogen peroxide, an important increment in the values of pH during time (from 5.45 to 9.76) is observed.This increment is due to the exchanging reactions between the alkali and the alkali-earth ions of the glass and the protons of the solution [1,57].Instead, fluctuations in pH values are very limited (from 7 to 8.5) for both the series when the samples are soaked in hydrogen peroxide and SBF.This is an expected behavior when working with buffered solutions, as in this case.

UV-vis spectroscopy
Fig. 6 shows the results of the UV-vis analyses over the reacted samples in H 2 O 2 + SBF, as well as those of the AQ glasses and of Ce 3+ and Ce 4+ species reported as reference.It is possible to notice that for all the samples, regardless from the considered series, the absorption band shifts to higher λ after 24 h of the soaking time, and then it remains unchanged for longer reaction time.This confirms, in qualitative manner, that the conversion of Ce 3+ to Ce 4+ takes place during the first 24 h of reaction with hydrogen peroxide for all samples (both for H and K series).[58,59].

XPS analysis
Ce 3d XPS spectra of the three H samples measured after different soaking times in H 2 O 2 + SBF solution are shown in Fig. 7.While sample H_1.2 does not show significant changes of Ce 3d peak shape as a function of soaking time, the spectra of the H_3.6 and H_5.3 samples are modified by soaking in the solution.In particular, as shown in Fig. 7 panel d), the Ce 3+ relative concentration shows a significant decrease after 4 h soaking and a subsequent increase back to the original level.The decrease is more pronounced for the H_5.3 sample than for the H_3.6 one, i.e. when the cerium oxide concentration is higher.For the K series of glasses after soaking in SBF solution the Ce 3d XPS signal intensity is very small, probably due to the precipitation on the surface of insoluble phases.This hindered the possibility of obtaining reliable estimates of the Ce 3 + and Ce 4 + relative concentrations for these samples.

ICP-OES analyses
Table 6 lists the content of Si, Ca and P ions of the H 2 O 2 + SBF solutions, before and after soaking of the H and K series.It is possible to notice that the trends for Si and Ca are the same for all the glasses: their concentrations increases with time.Instead, P behaves differently in the two series: it increases with time soaking in the H samples, while it decreases in the K series.The data collected with the ICP-OES suggest that the H series is a little bit more durable than the K series.
In particular, glasses of the K series release more Si with respect to the H series (Table 6).However, for both of them, the durability increases as the content of cerium increases.Cerium ions in solutions are under  the detection limit for this element, thus their concentrations are assumed to be below 1 ppm in each solution.
3.3.Bioactivity tests 3.3.1.Infrared spectroscopy Fig. 8 shows the spectra of the FT-IR analyses for the K series after 28 days of soaking in SBF solution.The aim of this analysis is to investigate the range of the IR spectra between 700 and 400 cm −1 in order to verify the presence of the bands at 605 and 565 cm −1 that is usually assigned to apatite-like crystalline phase [60].The K_1.2 and K_3.6 sample after 28 days of soaking in SBF show the presence of two band at 605 and 654 cm −1 confirming the crystallization of an apatite-like phase.However, for the sample K_3.6 in the range characteristic of PO 4 vibration another band at around 540 cm −1 is present.This band is attributed to the CePO 4 crystal phase (see the spectra of pure CePO 4 reported in Fig. 8) confirming the tendency of Ce-containing glasses to form insoluble phosphate on the glass surface during the bioactivity test.In fact, in a previous paper [41] it was demonstrated that the introduction of Ce in the H series decreases the bioactivity because the released cerium ions caused the precipitation of a Ce-containing phosphate phase.

ESEM-EDS analyses
All the studied glasses of the K series have reacted surface characterized by agglomerates with different morphologies (Fig. 9).
In particular, for the K and K_1.2 samples (Fig. 9 sections a, b, c and d), the darker areas correspond to the degraded glass which are free of the more mobile Na ions, whereas the bright ones are relative to an apatitic layer; the Ca/P molar ratio ranges from 1.5-1.7 and it is near to the stoichiometric ratio of hydroxyl-apatite.
In the sample K_3.6 (Fig. 9 sections e and f) two different morphologies in the superimposed agglomerates are observed: i) particles of spherical shape formed by Si, Ca, P and O with a Ca/P molar ratio ranges from 1.5-1.6;ii) particles of linear edges slightly enriched of Ce and markedly impoverished of Ca with respect to the spherical ones.The EDS analysis performed on these last agglomerates show a Ce/P and Ca/P molar ratios of 0.6 and 0.2-0.3,respectively.
These morphological and compositional studies indicate that all the glasses are bioactive, the presence of Ce in the bright areas, could be related to the formation of a cerium phosphate or mixed calcium-cerium phosphate phases, as previously hypothesized by IR spectra, and as subsequently demonstrated by XRD analysis.

XRD characterization
Fig. 10 shows the XRD spectrum of the samples K_3.6, K_1.2 and K after 28 days of soaking in SBF solution, which is the most meaningful spectrum collected.
All the XRD plots show the presence of peaks characteristic of hydroxyl-apatite (hkl = 211, 012 and 002, JCPDS 09-0432), confirming the bioactivity of these glasses.In the sample K_3.6 peaks attributed to CePO 4 (hkl = 120 and 20-2, JCPDS 32-0199) are also found.These results confirm the ability of the K-series glasses to form hydroxylapatite layers important for bioactivity.However, of Cerium, present in high %, interacts with the phosphate ions of the SBF solution giving rise to an insoluble phosphate phase (CePO 4 ), that has been detected in this study for the first time.

Discussion
The comprehensive study carried out on Ce-containing potential bioactive glasses confirms that in both phosphate-free and phosphate containing glasses the cerium ions are present in both oxidation states (3+ and 4+).UV-vis spectra and XPS results clearly indicate that the relative concentration of Ce 4+ is higher in the phosphate-free ones, conferring to the K-series of glasses higher bioactivity and higher catalase mimetic activity with respect to the H-series.This may be due to the absence of phosphate groups that are able to entrap and stabilize the Ce 3+ ions.
In fact, both computational and experimental (XRD on heat-treated of the phosphate-containing glasses) results highlight the formation inside the glass structure of zones rich in Ce and PO 4 units.These zones, characterized by low solubility, increase the chemical durability, delay the degradation glass process, and cause a decrement of bioactivity with respect to the pristine (H) glass.
The analysis of the Ce\ \Ce clustering performed on the MD-derived structural models [61,62] (Fig. 11) carried out on the phosphate-free glasses show that the K_5.3 glass presents a higher amount of Ce clusters with 2-and 4-members, with respect to H_5.3 system.These Ce-rich zones can be considered seed for CeO 2 crystallization.Indeed, XRD measurements on heat-treated Ce-containing glasses belonging to K series detects the presence of CeO 2 crystal phases.These results highlight, for the first time, a strict correlation between the behavior of Ce NPs and glasses containing Ce 3+ /Ce 4+ .In fact, a recent paper [63] reported on the ability of phosphate groups to inhibit the catalytic activity of Ce NPs by stabilizing the Ce 3+ species preventing their oxidation to Ce 4+ .
Another interesting result is that the degradation of H 2 O 2 is lower in SBF solutions with respect to H 2 O solutions; this can be explained by the XPS results obtained in the present and previous works.For the sake of clarity, some interesting data regarding the behavior of H_5.3 after soaking in H 2 O 2 and H 2 O 2 + SBF solutions are summarized in Table 7.
In H 2 O 2 water solution a drastic decrease of the Ce 3+ /Ce 4+ ratio is observed after 1 h of soaking, while 4 h of soaking are necessary to reach the same value of the ratio in H 2 O 2 + SBF (Table 7).Thus, the phosphatic component of the SBF solution stabilizes the Ce 3+ ions present on the glass surface, preventing their conversion into the oxidized species, and delaying the depletion of H 2 O 2 as shown in Fig. 5 (sections a and b).The same phenomena are observed, albeit in a smaller extent, for the H_1.Furthermore, also the presence of the phosphate units, which constitute the structure of the H series of glasses, increases the stability of Ce 3+ ions subtracting it from the dynamic Ce 3+ ↔ Ce 4+ equilibrium fundamental for catalase mimetic activity.This explains the superior catalase mimetic activity of the K series with respect to the H series, comparing glasses with the same CeO 2 content.

Conclusions
The results of this study allow the elucidation of structuralbioactivity relationships of Hench and Kokubo derived-glasses doped with Ce.
In particular, the structural role of Ce and the simultaneous presence of phosphate and cerium in the bulk or on the surface of the glasses have been correlated to the chemical durability (bioactivity) and the interconversion between Ce 3+ ↔ Ce 4+ (catalase mimetic activity).
The influence exerted by the Ce ions on the glass connectivity and solubility of the K-derived glasses with respect to the H ones is due to the absence/presence of phosphate groups.In fact, phosphate groups donate NBOs to Ce ions in the H-derived glasses, whereas in the K ones the NBOs necessary to stabilize the first coordination sphere of the Ce ions are supplied only by the de-polymerization of the silicate network.Moreover, phosphate groups stabilize the Ce 3+ species, forming CePO 4 insoluble phases; this prevents the interconversion process between Ce 3+ and Ce 4+ , which is a redox reaction of fundamental importance for the exhibition of catalase mimetic activity.Therefore, the presence of phosphate groups both in the intimate glass structure and/or in the environment (SBF solution vs. pure water) causes a lower catalase mimetic activity.

Fig. 5 .
Fig. 5. (a) and (c) Degradation of H 2 O 2 1 M after soaking of the glasses doped with CeO 2 ; (b) and (d) degradation of H 2 O 2 1 M + SBF after soaking of the glasses doped with CeO 2 .K and H are reported as control.The lines serve only to guide the eyes.

Fig. 7 .
Fig. 7. Ce 3d XPS spectra of (a) H_1.2,(b) H_3.6, (c) H_5.3 samples after different soaking times in H 2 O 2 + SBF solution.A Shirley type background was subtracted from the data.The fits of the spectra are also shown as a dashed line.(d) Ce 3+ relative concentration evaluated from the fitting of Ce 3d spectra.

Table 1
Nominal composition in %mol of the studied systems.

Table 5
Average Ca/Na ratio in the glass and Ca/Na ratio around Si, Ce(III) and Ce(IV) ions.

Table 6
Amount of Si, Ca and P (expressed in ppm ± std.dev.%)released at different times of reaction.