Structural evolution of CeO₂-doped alkali boroaluminosilicate glass and the correlation with physical properties based on a revised structural parameter analysis

Abstract

Alkali boroaluminosilicate glasses with composition of 56SiO₂–20B₂O₃–3Al₂O₃–8Na₂O–8K₂O–5BaO containing 0–9 mol% cerium oxide (CeO₂) were synthesized at 1500 °C using a conventional melt-quench method. Structural evolution of the as-prepared glasses was studied using infrared and Raman spectroscopy and the glasses' physical properties were characterized. The structural parameter based on the Yun, Dell and Bray model was revised by factoring multiple oxides, including CeO₂, into the computation of the ratio of alkali oxide to boron trioxide, and was used to describe the glasses' structural states divided into three categories with 3–4% as the critical content that marks the onset of drastic variations in glass structure and properties. The addition of CeO₂ below 3% are absorbed by reedmergnerite- and danburite-like groups producing one non-bridging oxygen (NBO) on the silica tetrahedrals and disconnecting tetrahedral borate [BO₄] from tetrahedral silicate [SiO₄]. Addition of CeO₂ above 3% allows the additional oxygen to combine with reedmergnerite- and danburite-like units producing two NBOs on [SiO₄] and also one or two NBOs on trigonal planar borate [BO₃] at the expense of [BO₄]. Further addition of CeO₂ beyond 5% will cause the extra NBOs to gradually combine with the disconnected boron triangles to form boron tetrahedrals in addition to continuous depolymerization of the Si–O network. Network depolymerization, enhanced linkage by charge compensation and improved compactness because of close packing are proposed as the cause of diverse variation trends in physical properties in the presence of CeO₂ below and above 3–4%, respectively. The revised structural parameter analysis is explainable by the correlation between the observed structural evolution and the physical properties, and thus, can be a useful reference for constituents' regulation of CeO₂-doped borosilicate radiation resistant glasses.

---

1. Introduction

Borosilicate glasses have been used extensively for industrial applications including heat resistant wares, glazes, photoelectron devices, immobilization of irradiative waste and so on because of their low melting temperatures, minimal thermal expansion coefficients, devitrification resistance, promising optical and mechanical properties, outstanding thermal stability and high chemical durability.¹ In particular, borosilicate possesses a superior dissolving capacity for rare-earth (RE) in the silicate system, and thus favors developing RE-containing optical glasses.^2,3

Because borosilicate glasses have always been preferentially used for radioactive waste immobilization,^4,5 aerospace facility components, for example, solar cell covers⁶ are very likely under a harsh irradiative environment that raises concerns about their radiological safety and stability; a sufficiently guaranteed irradiation resistance should be a vital prerequisite for their proper use. A considerable number of recent studies on borosilicate glasses mixed with various modifier oxides, particularly some transition metal ions, have been enthusiastically pursued with a view to improving their radiation resistance.^7–9 Cerium oxide (CeO₂) has previously been found to make glasses resistant to radiation damage,^10,11 and the mechanism for cerium to function as a radiation stabilizer is attributed to its polyvalent states namely Ce⁴⁺ and Ce³⁺ in glass via ceric–cerous redox equilibrium. They give irradiation shielding to glasses because Ce³⁺ captures holes and Ce⁴⁺ traps electrons yielded by irradiation to avoid their recombination, and thus to restrain the formation of color centers in the host glass.^12,13 Borosilicate glasses doped with CeO₂ exhibit improved irradiation resistance^14,15 and potentially have wide use where there is ambient irradiation such as aerospace, nuclear facilities, photovoltaic devices as well as medical and military fields. Addition of CeO₂ into borosilicate glass causes variations in physical properties which have been proved to be closely connected with CeO₂ induced structural changes.¹⁶

Structural studies on borosilicate glass modified by alkali metal oxides have provided some insights into its structure using well established models. The initial model proposed by Yun, Dell and Bray^17–19 uses two composition related structural parameters to describe structural features of the simple silicon dioxide–sodium oxide–boron trioxide (SiO₂–Na₂O–B₂O₃) ternary system. This provides a valuable approach to determine the possible structural evolution upon addition of alkali metal oxide although it had received much criticism and had consistently been revised and developed from then on.^20,21 Structure studies relating to borosilicate glass modified by metal oxides such as cadmium oxide,⁷ Bi₂O₃,⁸ SrO,²² PbO⁹ for radiation resistance purposes have been widely conducted, whereas CeO₂ is sometimes involved just as a non-radioactive surrogate to simulate the presence of minor actinides in nuclear glasses.^23,24 Even so, understanding the structure-property correlations of cerium-containing borosilicate glass especially for complex multi-component system remains limited, for example, recent reviews on borate and borosilicate glasses have comprehensively summarized the effects of different additives but unfortunately without mentioning cerium.^25,26

More importantly, the role cerium plays in glass structure so far remains debatable. Studies of the effect of CeO₂ on aluminosilicate glass,²⁷ borosilicate glass,¹⁶ lead silicate glass,²⁸ borate glass^29,30 and phosphate glass^31,32 generally accepted that the role of cerium was to act as a network modifier. Nevertheless, some of the investigations on the structure of borate,^33,34 and phosphate glasses³⁵ containing CeO₂, instead suggested that in these glasses cerium ions play a dual role, as both a network former and a network modifier depending on its content. In contrast, the structure of alkali boroaluminosilicate glasses is more complex because of the presence of multiple types of oxides, and doping CeO₂ in such a system is expected to arouse intricate structural speciation leading to macroscopic property variation. Furthermore, additional oxygen introduced by CeO₂ may raise the problems on how to reassess the structural parameters and how to interpret the related structural evolution based on the Yun, Dell and Bray model. In this work, the aim was to extend the applicability of this model by factoring multiple oxides into the computation of the R value to illustrate the structural changes of boroaluminosilicate glass induced by CeO₂. The correlations between structural evolution and variations in physical properties were tentatively elaborated based on structure parameter analysis combined with semi-quantitative spectroscopic characterization with the aim of guiding the tailoring of the performances of CeO₂ doped boroaluminosilicate glasses used in ambient irradiation.

2. Experimental

2.1 Glass preparation

The base glass with a fixed composition 56SiO₂–20B₂O₃–3Al₂O₃–8Na₂O–8K₂O–5BaO (molar ratio) was synthesized using a conventional melt-quench method using analytically pure reagents SiO₂, aluminium hydroxide, B₂O₃, sodium carbonate, potassium carbonate, barium carbonate, and CeO₂ as starting materials. Similarly CeO₂ doped glasses were fabricated by just adding a certain percentage of CeO₂ powder (mol%) over the base glass (keeping the ratio of base components constant). Prior to melting, batches of mixtures had been ball milled for 2 h. The batched mixture was heated to 1500 °C within 75 min and kept melting for 5 h for complete fining in an open platinum crucible under an air atmosphere, and afterwards the melt was poured onto a preheated steel plate and was pressed to form a disk and was then annealed in a muffle furnace at 550 °C for 1 h to eliminate the internal stress. The samples were ultrasonically cleaned in acetone and then in anhydrous ethanol successively. Glass samples were referred to by their doped molar percentage of CeO₂ relative to the base glass (in moles), i.e., 0–9% with a step of 1 mol%, respectively.

2.2 Spectroscopic analysis

Infrared (IR) spectra measurements of the glasses from 400 to 4000 cm⁻¹ were performed at room temperature with resolution of 2 cm⁻¹ using a Fourier transform infrared spectrometer (FTIR, Nicolet iS10) using the standard potassium bromide (KBr) pellet method. The recorded spectrum of the pulverized sample pellet was subtracted from that of the matrix KBr. Raman spectra in the range of 200–2000 cm⁻¹ were recorded at room temperature on a LabRAM HR800 micro-spectrometer (Horiba Jobin Yvon) using the 514.5 nm line of an Ar⁺ laser with the measured power of 5 mW, and the resolution of the system was less than 1 cm⁻¹.

2.3 Measurement of physical properties

Density was measured in accordance with the Archimedes principle using xylene as an immersion liquid at room temperature. Measurements were repeated three times and the densities (d) were determined by applying eqn (1) and then averaged. In eqn (1) ρ_b is the density of the buoyant substance, W_a and W_b are the sample weights in air and the buoyant substance, respectively.

d = ρ_bW_a/(W_a − W_b) (1)

The molar volume (V_m) was calculated using eqn (2):

where x_i is the molar fraction of the individual oxide in the glass composition and M_i is the molecular weight of the oxide.

Measurements of mechanical properties were carried out using an indentation test with a Nanotest system (Micro Materials Limited).^36,37 Hardness (H) and elastic modulus (E) of the glass were first determined using a low-load indentation test setting the maximum load to 500 mN and the loading/unloading rate to 5 mN s⁻¹, and the results were calculated using the software included with the instrument, using eqn (3) and (4) which were based on the Oliver and Pharr method,³⁸ where P is the maximum load, A is the projected contact area, and dP/dh represents the slope of the initial portion of the unloading curve. Subsequently a high-load indentation test was conducted setting the load to 5 N and the loading/unloading rate to 50 mN s⁻¹. The flawing impression was captured in situ to determine the fracture toughness (K_IC) using eqn (5),³⁹ where a is the length measured from the center of the indentation to the crack tip, l is the radial length of the crack and c is the length measured from the center of contact to the end of the corner radial crack.

Spectrophotometric analysis using a T6 UV-visible (UV-Vis) spectrophotometer (Persee, Beijing, China) was carried out to measure the transmittance of the prepared glass as a function of wavelength from 190 to 1100 nm with a resolution of 1 nm and then the absorption coefficient was obtained according to the Beer–Lambert law.

3. Results

3.1 Glass structure

3.1.1 FTIR analysis. On the whole, the FTIR spectra of the as-prepared glasses shown in Fig. 1 resemble each other in shape. A peak at 1620 cm⁻¹ arises from the vibrations of water molecules or hydroxide related groups in the samples⁴⁰ and a shoulder at about 1460 cm⁻¹ is assigned to carbonate groups which result from the interaction between carbon dioxide and silicate melts at high temperatures.⁴¹ A broad convolved peak in the range of 900–1200 cm⁻¹ consists of several constituent bands associated with the stretching of tetrahedral silicate ([SiO₄]) with various non-bridging oxygens (NBOs).⁴² Specifically, the intense band located at 1020 cm⁻¹ can be attributed to the combined asymmetric stretching of Si–O–Si and B–O–B network built of [SiO₄] and tetrahedral borate ([BO₄]) groups.^9,22,43 The shoulder at 1178 cm⁻¹ is ascribed to a fully bonding [SiO₄] tetrahedral denoted as Q⁴ (silica tetrahedral with four bridging oxygens), which appears to decrease in intensity with increasing CeO₂ content, suggesting the depolymerization of the glass network because of the CeO₂ addition. The structural features within this range, as described by Raman spectrometry, are discussed in the following section.

Fig. 1 IR spectra of the base and doped glasses with various content of CeO₂.

Furthermore, the bands occurred at 1400 cm⁻¹ because of the asymmetric stretching relaxation of the B–O bond of trigonal borate ([BO₃]) units^25,44,45 and a shoulder at 880 cm⁻¹ is associated with B–O stretching in [BO₄].^46,47 With the increase of CeO₂ below 5%, the [BO₃] tend to intensify and concurrently the [BO₄] gradually reduces, which is likely to be related to the dominant conversion of [BO₄] to [BO₃] in the presence of additional oxygen upon doping with CeO₂. Progressively increasing CeO₂ over 5% appears to stop this trend and this implies suppressed conversion of [BO₄] to [BO₃]. A weak band at 781 cm⁻¹ corresponds to the bending of the Al–O bond in the tetrahedral aluminate ([AlO₄]),^48,49 which appears to be unaffected by CeO₂ addition because aluminum has already existed in tetrahedral form in preference to boron because of the more alkaline oxide than aluminium oxide (Al₂O₃) in the base glass. In addition, there is a band located at about 710 cm⁻¹ arising from the bending of B–O–B^25,50 and another one at 450 cm⁻¹ because of the bending of the Si–O–Si bond,^43,51 however this gives less clear information about the structural variation.

3.1.2 Raman analysis. The Raman spectra of glasses studied with their intensity normalized are depicted in Fig. 2. The band centered at about 490 cm⁻¹ is consistent with the bending vibrations of Si–O–Si bonds.⁵² A broad band at about 770 cm⁻¹ related to four coordinated boron bonds in diborate^52,53 appears to first drop off with the increase of CeO₂, and correspondingly the band at about 1450 cm⁻¹, which is ascribed to B–O⁻ vibrations of borate triangles⁵⁴ from pyroborate units, is observed to intensify. However, further increasing the CeO₂ content above 5% tends to reverse this trend. This is in line with the IR spectra results, which suggest the conversion of [BO₄] into [BO₃] when CeO₂ is added stepwise below 5% and a subsequent hampered variation trend above that.

Fig. 2 Raman spectra of the glasses studied containing varying amounts of CeO₂.

Assignment of band 635 cm⁻¹ remains controversial⁵³ because it can be attributed to either the breathing mode of B–O–Si(B) bonds from danburite- and reedmergnerite-like rings consisting of [SiO₄] and [BO₄] tetrahedrals^52,53 in a mixed borosilicate phase or the bending of ring-type metaborate groups mainly composed of boron triangles.^1,26 Considering this band weakens with rising CeO₂ content, it tends to be attributed to the former because the weakening band at 635 cm⁻¹ should signify intensifying breakage of B–O–Si(B) bonds in borosilicate rings, which are built of silicon and boron tetrahedrals rather than boron triangles. Otherwise, its trend in variation may conflict with those of other bands as well as the previously mentioned IR results.

The bands across the range 800–1200 cm⁻¹ associated with the stretching mode of terminal groups of the silicon–oxygen tetrahedrons Qⁿ (n is the number of bridging oxygen atoms) are of much more significance for the analysis of structural changes,⁵⁵ and thus are specifically elaborated in Fig. 3. As this broad envelope comprises component peaks situated amid 780–890 cm⁻¹, 900–940 cm⁻¹, 950–1020 cm⁻¹, 1060–1130 cm⁻¹ and 1150–1200 cm⁻¹ assigned to the Si–O stretching of Qⁿ with n being 0, 1, 2, 3 and 4, respectively,^52,54–56 peak deconvolution using several Gaussians are employed to fit multipeaks at these bands. In this study, satisfactory fitting results with negligible errors in peak center and area were acquired with accuracy which conformed to the appropriate fitting parameters (reduced chi-square < 10⁻⁵, coefficient of determination R² > 0.999). The total area of the fitting peaks was used to normalize the individual deconvolved peak generating the proportional area of every constituent peak that represents the relative intensity of symmetric stretching of Qⁿ. The band intensity is assumed to be proportional to the structural species' area fraction that can be used to roughly evaluate the relative Qⁿ fraction in the glass structure.

Fig. 3 Deconvolution of the Raman spectra of base and doped glasses with various contents of CeO₂.

Area fraction subtotals of various Qⁿ species as a function of CeO₂ content are displayed in Fig. 4(a) where the fitted curves are shown as a visual guide. As revealed in Fig. 4(a), the base glass builds its structure mainly using higher level tetrahedra (Q⁴ and Q³) whose fractions decrease remarkably while the Q¹ and Q² fractions grow nonlinearly with CeO₂ additions below 5%, and up to and after 3% within this range, different rates of change are exhibited. Q¹ and Q² fractions growing nonlinearly upon CeO₂ additions below 5%, and up to and after 3% within this range different changing pace are exhibited. Addition of CeO₂ exceeding 5% allows a gradual drop of Q² but a rise of Q³, which roughly conforms to the transformation relationship: 2Q² → Q³ + Q¹.⁵⁷ To avoid the possible uncertainty on assigning a few wavering bands located between different Qⁿ species, the similar items specified as higher levelled Qⁿ (Q³ and Q⁴) and lower levelled ones (Q⁰, Q¹ and Q²) are merged separately into two classes. Fig. 4(b) reveals the gradual drop of the summed area fraction of Q³ and Q⁴ compared to the rising total of Q⁰, Q¹ and Q². There is an evident turning point close to a level of 3% of CeO₂ signifying the nonlinear evolution of silicate structural units upon addition of CeO₂. For the glasses investigated, this could be attributed to the deepening depolymerization and reconfiguration of the network starting at about 3–4% CeO₂ that can be viewed as a critical content of structural evolution.

Fig. 4 Variation of the Qⁿ area fraction as a function of CeO₂ content: (a) individual Qⁿ species with n being 0, 1, 2, 3 or 4; (b) higher levelled Qⁿ species (Q³ and Q⁴) compared to lower levelled ones (Q⁰ and Q¹ and Q²) (fitting curves shown as guide for the changing trends).

3.2 Physical properties

3.2.1 Density. As shown in Fig. 5, glass density can be thought to increase as a function of CeO₂ content with a drastic rise above 3% until it reaches a decline at about 5% upon which the density starts a gradual increase from the lower value. Correspondingly, the molar volume estimated by eqn (2) increases slightly for CeO₂ levels below 3% when it immediately starts to decline with further increasing CeO₂ levels up to 5%, thereafter it roughly tends to level off. Variation of the density is ruled by the combination of mass fluctuation and the resultant network compactness produced by CeO₂. The molar volume, a measure to assess structural looseness, is essentially dependent on network connectivity and structural compactness with the former reflecting polymerization and the latter, the atomic packing density. It seems that weight gain by taking in heavy cerium ions accounts predominantly for the increase in density, whereas incremental CeO₂ may make structural compactness the dominant factor governing the fluctuation of molar volume. Given the depolymerization of the network upon CeO₂ addition, structural compactness appears to outperform the network looseness to varying degrees in different doping ranges that determine the diverse structural states of the glass.

Fig. 5 Glass density and molar volume as a function of CeO₂ content.

3.2.2 Mechanical properties. As shown in Fig. 6, doping CeO₂ with a content lower than 3% reduces the hardness and elastic modulus, whereas further increasing the concentration over 3% will cause a reversal of a rise in hardness and elastic modulus until relatively stationary values for CeO₂ in excess of 5% are reached. Nevertheless, the fracture toughness increases gradually to an extrema at a level of 3% CeO₂ and then drops to a lower level with the continuous increase of CeO₂ level above 5% as displayed in Fig. 7. These results echo the structural features suggesting that doping CeO₂ at a level below 3% leads to a network with dominant looseness over compactness which is detrimental to hardness and modulus but favors of fracture toughness and that further doping CeO₂ beyond a 3% level results in an inversely affected structure imposing the opposite effect on the previously mentioned mechanical properties. Doping over 5% leads to a relatively relaxed but multifactor-balanced network structure linked to lower mechanical properties with little fluctuation.

Fig. 6 Glass hardness and elastic modulus measured by indention as a function of CeO₂ content.

Fig. 7 Glass fracture toughness measured by indention as a function of CeO₂ content.

3.2.3 Optical band gap. UV-Vis light absorbed by glass excites the oxide ions to higher energy levels, and the threshold energy required to activate such an intrinsic excitation can be measured by the optical band gap, which generally refers to the energy difference (in electron volts) between the top of the valence band and the bottom of the conduction band. The optical band gap carries the structural information of the glass network and can be calculated using the UV-Vis spectrum.^58,59 For direct and indirect optical transitions, a model proposed by Mott and Davis quantitatively correlates the linear absorption coefficient (k, cm⁻¹) to the incident photon energy E expressed using eqn (6):

kE = B(E − E_opt)ⁿ (6)

where E can be calculated by ch/λ (here c is the light velocity, λ is wavelength and h is Planck's constant), B is a constant and E_opt denotes the optical band gap. The exponent n = 1/2, 3/2, 2 and 3 is for an allowed direct, forbidden direct, allowed indirect and forbidden indirect transition, respectively. Transformation of eqn (6) yields a linear relationship between (kE)^1/n and E, the interception of the linear region extrapolated against the zero axis gives the value of E_opt. As demonstrated in Fig. 8, (kE)^2/3 as a function of photon energy E (n = 3/2) shows an extended linear region signifying the induced transitions for the glass studied are direct forbidden which is similar to the results obtained by the processing method adopted by Abdel Rehim.⁵⁹ The E_opt of the studied glasses obtained as a function of CeO₂ concentration are displayed in Fig. 8(b).

Fig. 8 (a) A demonstration of the plots for the computation of the optical band gap using UV-Vis spectra; (b) E_opt variation as a function of CeO₂ content (the dashed line is shown as a guide).

As exhibited in Fig. 8(b), initially E_opt decreases gradually and then declines steeply with increasing CeO₂ content below 5%. As stated in the previous section, the doped CeO₂ in glass acting as a network modifier creates additional NBOs by breaking up the Si–O network. However, cerium ions with a high atomic number will greatly augment the electron population density distribution in the network. The weakened covalence bonds stemming from the depolymerized network in combination with the increased electron density distribution will contribute to the shrinkage of E_opt. Furthermore, further increasing the CeO₂ content beyond 3% seemingly intensifies most the two influences that allow a drastic drop of E_opt. Nevertheless, decrease in E_opt apparently slows down with CeO₂ doping ≥5%, indicating that the newly introduced NBOs are reused for bridging the structural groups such as conversion of some [BO₃] to [BO₄] as revealed by IR and Raman spectrometry analysis.

4. Discussion

Composition of the glass studied in this research comprises multiple network formers, namely, Si, B and Al, and it is very likely that these formers allow the mixing of the borate and aluminosilicate network because of the appropriate constituent relationship. According to the model proposed by Yun, Dell and Bray, structural parameters R = M₂O/B₂O₃ and K = SiO₂/B₂O₃ (in molar ratio) govern the possible structural speciation upon addition of alkali oxides (M₂O) in a simple ternary system, for example, Na₂O–B₂O₃–SiO₂ glasses, as follows.^1,19 For R < 0.5, the borate present in a triangular structure together with those which have a tetrahedral structure, because of the conversion of [BO_3/2] to [BO_4/2]⁻, are compensated by M⁺ for electric neutrality. For 0.5 < R < R_max (0.5 + K/16), formation of reedmergnerite- and danburite-like structural units occurs. Whereas for R_max < R < R_d1 (0.5 + K/4), additional modifiers start to cause depolymerization of the network yielding NBOs in silica tetrahedrons. For R_d1 < R < R_d3 (2 + K), the additional modifier is assumed to produce NBOs on a trigonal borate at the expense of tetrahedral units as well as two NBOs on [SiO₄].

Analogously, CeO₂ as a modifier is expected to yield additional oxygens leading to either depolymerization of the silicate network or variation in the coordination number of formers with cerium ions acting for charge balance.²⁷ Here, multiple oxides including CeO₂ are factored into the computation of the R value. In view of the polyvalence of cerium, the average additional oxygen of one mole of cerium ions can be equivalent to x mole alkali oxide expressed as a function of [Ce³⁺]/[Ce⁴⁺] given in eqn (7), where items in square brackets represent the ingredient's content. Here [Ce³⁺]/[Ce⁴⁺] strongly relies on the CeO₂ concentration as well as the redox conditions of glass synthesis. Apparently, the value of x ranges from 1.5 to 2 irrespective of the scale of [Ce³⁺]/[Ce⁴⁺]. Correspondingly, the structural parameter R can be rewritten as eqn (8). In essence, the amended R represents the relative amount of modifiers' content for charge compensation.

Calculation of the R of the base glass without CeO₂ gives the value of 0.9, which falls in the range between R_max (0.675) and R_d1 (1.2), meaning that there is the presence of mixed of 3- and 4-fold coordinated boron and silicate tetrahedral with varying NBOs (Qⁿ) configured partly in reedmergnerite- and danburite-like groups according to the above model. Additional alkalis (K⁺, Na⁺, Ba²⁺) in excess of R_max are absorbed by reedmergnerite- or danburite-like groups producing NBOs on silicate in the base glass. An Si–O–B mixed network constructed of reedmergnerite- and danburite-like groups is verified by characteristic band in the Raman spectrum at 630 cm⁻¹ in Fig. 2. The typical three-dimensional structural models of these two groups are illustrated in Fig. 9(a) for reedmergnerite and Fig. 9(b) for danburite, which shows that the significant difference between them is that double boron tetrahedrals are included in the danburite-like group whereas there is only one connecting to four surrounding silica tetrahedrals in the reedmergnerite-like group.

Fig. 9 Structural modeling graphics of (a) reedmergnerite-like unit and (b) danburite-like unit assumed to be present in the base glass (red: O, blue: Si, yellow: B, grey: alkali metal).

For CeO₂-doped glasses, CeO₂ as a modifier behaving similarly to additional alkali or alkali earth metallic oxides will readily be absorbed to a vitreous structure causing NBOs in silica network and degrading the connective dimensionality of boron in excess of the rated doping amount. This is shown by the Raman and IR spectra in the previous sections that revealed the progressive rupture of the silicate network as well as part conversion of borate tetrahedrons to triangles upon CeO₂ addition. The incremental modifier above R of base glass because of the CeO₂ addition is dependent on the term on the far right of eqn (8). Given that x equals 1.5–2 and [B₂O₃] is 20 mol%, R can be computed, using eqn (8), to be 1.2 in the presence of 3% CeO₂, a value close to R_d1. That is to say, 3% is seems to be the theoretical critical CeO₂ concentration that just divides the structural parameter R into two distinct categories, i.e., R_max < R < R_d1 and R > R_d1 which correspond to the two respective structural states for the studied glass composition containing CeO₂ below 5%. Furthermore, by raising the CeO₂ concentration it may give rise to R until approaching a new boundary value R_d3 (4.8) roughly corresponding to the glass containing 5% CeO₂. Further increasing the CeO₂ content over 5% allows for R > R_d3 that has been rarely considered by the Yun, Dell and Bray model.

For the first structural state with CeO₂ <3% (R_max < R < R_d1), the added CeO₂ in trivalent and tetravalent states are absorbed by reedmergnerite- and danburite-like groups producing one NBO on the silica tetrahedrals and disconnecting [BO₄] from [SiO₄] at one corner as shown in Fig. 10(a) and (b). In this case, the Qⁿ units gradually degrade to lower level ones and boron triangles continuously yield upon adding CeO₂. However, with increasing CeO₂ from 3% to 5% in a stepwise manner, the glass structure runs into the second state (R_d1 < R < R_d3), and in return allows the additional oxygen to combine with reedmergnerite- and danburite-like units progressively rupturing the Si(B)–O skeletal network with two NBOs on [SiO₄] and also one or two NBOs on [BO₃] at expense of [BO₄] as illustrated in Fig. 10(a′) and (b′), respectively. These steps continue until mostly [BO₃] groups with two NBOs form when increasing the CeO₂ content. When further adding CeO₂ in excess of 5% with R > R_d3, the extra NBOs yielded will gradually combine with the disconnected boron triangles from the [SiO₄] network forming boron tetrahedrals in addition to continuous depolymerization of the Si–O network. In this case, the B–O network achieved enhancement to some extent despite loosening the Si–O skeletal network, and consequently the network can maintain relative stability. This seems to explain the structural features arrived at in the forgoing sections. Thus, a level of 3–4% should be viewed as the approximate critical content that marks the onset of most of the extreme variations in properties as well.

Fig. 10 Assumed structural variations upon adding CeO₂ (a) reedmergnerite-like, CeO₂ < 3%, (a′) reedmergnerite-like, CeO₂ > 3%, (b) danburite-like, CeO₂ < 3%, (b′) danburite-like, CeO₂ > 3%.

Under these circumstances, cerium ions can occupy network interstitial sites with Ce³⁺ associated with three neighboring Q³ tetrahedrals and Ce⁴⁺ with four for charge compensation. Several possible configuration modes but not limited to these by virtue of electrostatic interaction are depicted in Fig. 10 and correspond to diverse structural states. Apparently, the scope of such functionality extends drastically when double NBOs are produced on silicon tetrahedrals in case of CeO₂ levels beyond 3%. It is supposed that cerium ions are much more tightly bonded into the skeletal network providing stronger network linkages and immobilizing the relative displacement of loosened chains. This structural modification enhances to some extent the network interconnectivity despite local scission of the Si(B)–O bond. This postulation is essentially in agreement, but from a different perspective, with the notion reported elsewhere regarding the CeO₂-doped borate glass that the role of CeO₂ has been considered as a network former by forming [CeO₄] units at the expense of its role as a network modifier.^33,34 Alternatively, cerium ions of a large size and huge mass situated at the interstitial site will occupy much free volume of the network where they are wedged into the space with Na⁺, K⁺, Ba²⁺ seeking optimal close packing, and thus improve structural compactness.

On the whole, doping with CeO₂ has three influences on the structure of the studied glass, namely, depolymerization of the borosilicate network, enhanced linkage by charge compensation and improved compactness. The latter two characteristics progressively increase to play the dominant role over the former until settling in the structural state of R > R_d1 corresponding to CeO₂ > 3%. For doping with CeO₂ levels ranging from 5% to 9%, the relative balance of network connectivity and compactness allows for a mild variation in network rigidness and glass properties. Improved linkage and compactness combined with sharp weight gain cause the glass density to increase upon addition of CeO₂. Correspondingly, the molar volume appears to initially increase slightly and then decrease when the CeO₂ level exceeds about 3% and tends to stabilize in the range of 5–9%. This structural evolution upon CeO₂ addition has also been reflected in mechanical and optical properties that present diverse changing trends under lower and higher CeO₂ contents, respectively, because of the inversely fluctuating impact induced by CeO₂. There is an intrinsic link between the glass properties and the evolved structural states depending on the content of CeO₂ relative to host network formers, and it provides some guidance for developing CeO₂-doped boroaluminosilicate glass materials with tailored performances.

5. Conclusion

By factoring multiple oxides including CeO₂ into the computation of structural parameter R (the ratio of alkali oxide to B₂O₃) based on the Yun, Dell and Bray model, the structural states of alkali boroaluminosilicate glass (56SiO₂–20B₂O₃–3Al₂O₃–8Na₂O–8K₂O–5BaO) containing 0–9 mol% CeO₂ are divided into three categories with 3–4% as the critical content that marks the onset of drastic variations in glass structure and properties. When the added CeO₂ levels are below 3%, they are absorbed by reedmergnerite- and danburite-like groups producing one NBO on silica tetrahedrals and disconnecting [BO₄] from [SiO₄]. Addition of a CeO₂ level above 3% allows the additional oxygen to combine with reedmergnerite- and danburite-like units producing two NBOs on [SiO₄] and also one or two on [BO₃] at the expense of [BO₄]. Further adding CeO₂ at a level in excess of 5% will make the extra NBOs gradually combine with the disconnected boron triangles which then form boron tetrahedrals in addition to continuous depolymerization of the Si–O network. On the whole, doping CeO₂ has three influences on the structural evolution of the studied glass, namely, depolymerization of the skeletal borosilicate network, enhanced linkage by charge compensation and improved compactness because of the close packing, which causes diverse variation trends in the physical properties in the presence of CeO₂ levels below and above 3–4%, respectively. The revised structural parameter analysis is explained by the observed structural evolution as well as its correlation with the physical properties, and thus can be a useful reference for the regulation of the constituents of CeO₂-doped borosilicate radiation resistant glasses.

Acknowledgements

The authors acknowledge the support of Natural Science Foundation of Chongqing (Grant No. cstc2012jjA50034), the Chinese 973 Fundamental Research (2011CB605806) and National Natural Science Foundation of China (50820145202).

References

1. M. Hubert and A. J. Faber, Phys. Chem. Glasses: Eur. J. Glass Sci. Technol., Part B, 2014, 55, 136.
2. J. N. Cachia, X. Deschanels, C. Den Auwer, O. Pinet, J. Phalippou, C. Hennig and A. Scheinost, J. Nucl. Mater., 2006, 352, 182.
3. T. Konishi, T. Matsumoto, T. Araki, T. Tsuchiya, S. Todoroki and S. Inoue, Appl. Surf. Sci., 2004, 223, 238.
4. S. Peuget, J. M. Delaye and C. Jégou, J. Nucl. Mater., 2014, 444, 76.
5. I. W. Donald, B. L. Metcalfe and R. N. J. Taylor, J. Mater. Sci., 1997, 32, 5851.
6. N. Baydogan and A. B. Tugrul, Res. Chem. Intermed., 2014, 40, 299.
7. R. Kaur, S. Singh and O. P. Pandey, J. Mol. Struct., 2013, 1049, 409.
8. R. Kaur, S. Singh and O. P. Pandey, J. Mol. Struct., 2013, 1049, 386.
9. M. Arora, S. Baccaro, G. Sharma, D. Singh, K. S. Thind and D. P. Singh, Nucl. Instrum. Methods Phys. Res., Sect. B, 2009, 267, 817.
10. G. A. Haynes, NASA technical note, National aeronautics and space administration, Washington, DC, 1970, p. 1.
11. A. Paul, M. Mulholland and M. S. Zaman, J. Mater. Sci., 1976, 11, 2082.
12. Y. Sheng, L. Yang, H. Luan, Z. Liu, Y. Yu, J. Y. Li and N. Dai, J. Nucl. Mater., 2012, 427, 58.
13. B. Speit, E. Radlein, G. H. Frischat, A. J. Marker and J. S. Hayden, Nucl. Instrum. Methods Phys. Res., Sect. B, 1992, 65, 384.
14. Z. Han, Z. Tang, Q. Chen, X. Gong and H. Zhao, J. Chin. Rare Earth Soc., 2010, 28, 275.
15. T. D. Henson and G. K. Torrington, Proc. SPIE, 2001, 4452, 54.
16. V. K. Deshpande and R. N. Taikar, Mater. Sci. Eng., B, 2010, 172, 6.
17. Y. H. Yun and P. J. Bray, J. Non-Cryst. Solids, 1978, 27, 363.
18. Y. H. Yun, S. A. Feller and P. J. Bray, J. Non-Cryst. Solids, 1979, 33, 273.
19. W. J. Dell, P. J. Bray and S. Z. Xiao, J. Non-Cryst. Solids, 1983, 58, 1.
20. S. Wang and J. F. Stebbins, J. Non-Cryst. Solids, 1998, 231, 286.
21. D. Manara, A. Grandjean and D. R. Neuville, J. Non-Cryst. Solids, 2009, 355, 2528.
22. A. Rupesh Kumar, T. G. V. M. Rao, K. Neeraja, M. Rami Reddy and N. Veeraiah, Vib. Spectrosc., 2013, 69, 49.
23. P. Trocellier, A. Haddi, S. Poissonnet, P. Bonnaillie and Y. Serruys, Nucl. Instrum. Methods Phys. Res., Sect. B, 2006, 249, 145.
24. G. Yang, G. Mobus and R. J. Hand, Micron, 2006, 37, 433.
25. C. Gautam, A. K. Yadav and A. K. Singh, ISRN Ceram., 2012, 2012, 1.
26. A. K. Yadav and P. Singh, RSC Adv., 2015, 5, 67583.
27. S. L. Lin and C. S. Hwang, J. Non-Cryst. Solids, 1996, 202, 61.
28. M. R. Adams, R. Engelmann, P. D. Grannis, J. Horstkotte, L. Godfrey, S. L. Linn, M. D. Marx, J. Timms, P. M. Tuts and J. Willins, Nucl. Instrum. Methods Phys. Res., Sect. A, 1985, 238, 333.
29. P. Kaur, G. P. Singh, S. Kaur and D. P. Singh, J. Mol. Struct., 2012, 1020, 83.
30. G. P. Singh, P. Kaur, S. Kaur and D. P. Singh, Phys. B, 2012, 407, 4168.
31. Y. M. Lai, X. F. Liang, S. Y. Yang, J. X. Wang, L. H. Cao and B. Dai, J. Mol. Struct., 2011, 992, 84.
32. J. Du and L. Kokou, J. Am. Ceram. Soc., 2011, 94, 2393.
33. G. El-Damrawi and K. El-Egili, Phys. B, 2001, 299, 180.
34. E. Mansour, J. Non-Cryst. Solids, 2011, 357, 1364.
35. Y. Y. Ivanova, E. J. Spassova, E. P. Kashchieva, M. A. Bursukova and Y. B. Dimitriev, J. Non-Cryst. Solids, 1995, 192–193, 674.
36. Q. Zhao, M. Guerette and L. Huang, J. Non-Cryst. Solids, 2012, 358, 652.
37. X. Li, L. Jiang, X. Zhang and Y. Yan, J. Non-Cryst. Solids, 2014, 385, 1.
38. W. C. Oliver and G. M. Pharr, J. Mater. Res., 2004, 19, 3.
39. G. M. Pharr, Mater. Sci. Eng., A, 1998, 253, 151.
40. A. M. Efimov, V. G. Pogareva and A. V. Shashkin, J. Non-Cryst. Solids, 2003, 332, 93.
41. R. A. Brooker, S. C. Kohn, J. R. Holloway and P. F. McMillan, Chem. Geol., 2001, 174, 241.
42. A. K. Sandhu, S. Singh and O. P. Pandey, Mater. Chem. Phys., 2009, 115, 783.
43. E. Mansour, J. Non-Cryst. Solids, 2012, 358, 454.
44. C. R. Gautam, D. Kumar, O. M. Parkash and P. Singh, Bull. Mater. Sci., 2013, 36, 461.
45. C. R. Gautam, D. Kumar and O. M. Parkash, Bull. Mater. Sci., 2010, 33, 145.
46. M. Massot, C. Julien and M. Balkanski, Infrared Phys., 1989, 29, 775.
47. M. S. Gaafar and S. Y. Marzouk, Phys. B, 2007, 388, 294.
48. G. H. Kim and I. Sohn, J. Non-Cryst. Solids, 2012, 358, 1530.
49. Y. Cheng, H. Xiao, S. Chen and B. Tang, Phys. B, 2009, 404, 1230.
50. E. I. Kamitsos, M. A. Karakassides and G. D. Chryssikos, J. Phys. Chem., 1987, 91, 5807.
51. S. Song, Z. Wen, Y. Liu, Q. Zhang, X. Wu, J. Zhang and J. Han, Ceram. Int., 2009, 35, 3037.
52. J. Bonfils, S. Peuget, G. Panczer, D. Ligny, S. Henry, P. Y. Noël, A. Chenet and B. Champagnon, J. Non-Cryst. Solids, 2010, 356, 388.
53. D. Manara, A. Grandjean and D. R. Neuville, Am. Mineral., 2009, 94, 777.
54. V. E. Eremyashev, A. A. Osipov and L. M. Osipova, Glass Ceram., 2011, 68, 205.
55. O. N. Koroleva, L. A. Shabunina and V. N. Bykov, Glass Ceram., 2011, 67, 340.
56. A. G. Kalampounias, Bull. Mater. Sci., 2011, 34, 299.
57. T. Yano, S. Shibata and T. Maehara, J. Am. Ceram. Soc., 2006, 89, 89.
58. Y. Ou, S. Baccaro, Y. Zhang, Y. Yang and G. Chen, J. Am. Ceram. Soc., 2010, 93, 338–341.
59. A. F. Maged and F. Abdel Rehim, Appl. Radiat. Isot., 1991, 42, 763.

---