Short-range and Medium-range Structural Order in CaO–SiO₂–TiO₂–B₂O₃ Glasses

Abstract

The present study aimed to provide a deep insight of short-range and medium-range structural order for CaO–SiO₂–TiO₂–B₂O₃ glasses. From various prospects, four techniques, FTIR, Raman, XPS and NMR, were simultaneously employed and only the direct evidences indicated from the spectra were used for structural analysis. The FTIR and Raman spectra proved that the main silicon-related units were Q⁰(Si), Q¹(Si), Q²(Si) and Q³(Si) and the results of Raman fittings revealed that a B₂O₃ addition resulted in an increase of Q¹(Si) and Q³(Si) at the cost of Q⁰(Si) and Q²(Si), thus inducing an increasing Degree of Polymerization (DOP). An enhanced DOP generally implied a lower abundance of non-bridging oxygen in the networks, which was further demonstrated by the O1s XPS fittings. Additionally, the ¹¹B NMR spectra indicated that the dominant boron-related groups were BO₃ trigonal comprising ring and non-ring BO₃ as well as BO₄ tetrahedral comprising BO₄(1B,3Si) and BO₄(0B,4Si). Furthermore, it was clarified that an increase of B₂O₃ content promoted the ratios of BO₄(1B,3Si) to BO₄(0B,4Si) and BO₄ tetrahedral to BO₃ trigonal, which undoubtedly identified the occurrence of an equilibrium reaction between NBO, BO₄ tetrahedral and BO₃ trigonal in the glass.

1. Introduction

Glass, considering its widespread application nowadays and in history, is a fantastic material due to its unique structures and properties.¹⁾ From a structural viewpoint, it is known that glass phase is considered as a metastable, super-cooled liquid phase.^1,2,3) Nature of the glass state remains a critical knowledge gap in modern science where only limited insights has been acquired, which even makes up an issue to be through lightly.^4,5,6) Nowadays, there are two research fields about glass state to be identified, i.e., the medium-range order structures in a glass and the structure-property relations. To solve these mysteries, an understanding of the glass structures in contiguous length scale is quite significant, i.e., the short-range order (SRO, 2–5 Å), medium-range order (MRO, 5–20 Å) and long-range order (LRO, ≥ 20 Å).⁷⁾ Due to the lack of LRO and the relatively constant SRO structures in a glass, identification of the connectivity of the local structures in the medium range, indeed, become a key step towards uncovering the nature of glass state.

In the family of glasses, borosilicate glass accounts for an important sub-type because of its wide utilizations in the glass-⁸⁾ and metallurgical industry.^9,10) Numerous studies have been thus performed to pursue the network structures of borosilicate glass, which could be categorized into three types based on the three important variables, viz., temperature, pressure and chemical composition. For example, Stebbins et al. found a conversion of BO₃ trigonal to BO₄ tetrahedral with increasing temperature.¹¹⁾ Recently, Edwards et al.¹²⁾ discovered the dynamic process of the BO₃→BO₄ conversion with increasing pressure using a high pressure ¹¹B solid state Nuclear Magnetic Resonance (NMR) spectroscopy. There appears a deformation of the BO₃ planar triangle into trigonal pyramid, which serves as a precursor for the formation of BO₄ tetrahedral. In the present study, the influence of the third variable, chemical composition, on the structures of borosilicate glasses was analyzed and possible structural conversions were identified.

Drawing parallels with silicate glass, there were five structural tetrahedral units of SiO₄ in a silicate glass, which could be denoted as Qⁱ(Si) (i=0, 1, 2, 3, 4), where i represents the number of bridging oxygen.^13,14,15,16) The fractions of these units can be derived based on the deconvolutions of Raman or Magic Angular Spinning Nuclear Magnetic Resonance (MAS-NMR) curves. The present study aimed to apply these methods to borosilicate glasses.

From the prospect of structure-property relations, measurement of the structures of borosilicate glasses was expected to throw much light on the evolution of the macroscopic properties. Studies on viscous flow behaviors of the borosilicate melts^17,18,19,20) generally show a decreasing viscosity with increasing B₂O₃ content in the melts. In addition, in a previous study on crystallization behaviors,²¹⁾ we found that a B₂O₃ addition of 5% brought about a drastic change of primary crystalline phase in a CaO–SiO₂–MgO–Al₂O₃–TiO₂–B₂O₃ melt, viz., from CaTiO₃ to TiO₂. All these forgoing variations of macroscopic properties could be connected to the microscopic structure of the glasses; the present study was therefore motivated. A glass of CaO–SiO₂–TiO₂–B₂O₃ was designed herein and different techniques were employed to complementarily confirm the structural analysis, which made up a main methodology followed by this study. As a consequence, it was feasible that only the direct evidence indicated from the spectra was used for the structural analysis. Another innovation was the quantification of the local structures and their connectivity in the glasses, which could be achieved by deconvolutions of the spectral curves obtained.

2. Experimental and Methods

2.1. Sample Preparation

In this study, glass samples of the quaternary CaO–SiO₂–TiO₂–B₂O₃, with a fixed basicity of 1.2 (mass ratio of CaO to SiO₂), were prepared, using the conventional melting-quenching method. Reagent purity grade materials of CaO (99.5%), SiO₂ (99.5%), TiO₂ (99.8%) and H₃BO₃ (99.5%) (Produced by Alfa Aesar company) were thoroughly mixed and placed in a Pt crucible (Φ40×45×H40 mm). The mixture was melted at 1823 K (1550°C) in a tube furnace in air and soaked for 2 hours to in order to confirm the homogeneity. After pre-melting, the liquid melts were rapidly poured from the crucible into cold water to be quenched so that it formed a glassy state. Subsequently, the samples were dried in air at 378 K (105°C), crushed and ground to less than 300 meshes for spectral measurements. The compositions of the samples prepared are presented in Table 1. In order to confirm the amorphous nature, the samples were characterized by X-ray diffraction (XRD) technique (D/Max 2500, Rigaku, Japan).

Table 1. Chemical composition of pre-melted samples (mass%).

Sample	CaO	SiO2	B2O3	TiO2	CaO/SiO2
CSBT-1	41%	34%	0	25%	1.20
CSBT-2	38%	32%	5%	25%	1.20
CSBT-3	35%	30%	10%	25%	1.20
CSBT-4	33%	27%	15%	25%	1.20

2.2. Spectroscopic Measurements

The structures of the glass samples were complementally characterized using FTIR, Raman, MAS-NMR and XPS spectra. The FTIR spectra were recorded in the wavenumber range of 4000–400 cm⁻¹ using a Tensor 27 spectrometer (Bruker, USA), with a spectral resolution of 2 cm⁻¹. About 2.0 mg samples was mixed thoroughly with 200 mg pure KBr, ground well in an agate motor and pressed into a disc of 13.0 mm diameter for the infrared measurement. The Raman spectra in the range of 200–2000 cm⁻¹ were collected at room temperature using a JY-HR800 spectrometer (Jobin Yvon Company, France) with a light source of a 1 mW semiconductor. The excitation wavelength was 532 nm. In order to further clarify the structural roles of boron in the glass samples, solid state ¹¹B MAS-NMR measurements were conducted using a 400M FT-NMR spectrometer (Avance III 400M, Bruker, USA) with a MAS probe of a 4 mm ZrO₂ rotor and two pairs of Dupont Vespel caps. In order to identify the functionalities of O atom in the glasses, X-ray photoelectron spectroscopy (XPS) were employed using an Imaging Photoelectron Spectrometer (AXIS-Ultra, Kratos Analytical), with monochromatic Al Kα radiation (225 W, 15 mA, 15 kV) and low energy electron flooding for charge compensation. In addition, a C 1s hydrocarbon peak at 284.8 eV was used to calibrate the bond energies.

3. Results and Discussion

3.1. FTIR Spectra for the Glass Samples

In order to understand the influence of B₂O₃ on the network structures, these glass samples were first characterized using FTIR spectra. The results are presented in Fig. 1. Several remarkable variations appeared on the spectra. The effective spectral range of 400–1600 cm⁻¹ could be divided into four regions, i.e., the 400–600 cm⁻¹ and 600–800 cm⁻¹ regions with low intensity, and the 800–1200 cm⁻¹ and 1200–1600 cm⁻¹ regions with high intensity. These spectral bands could be parallelly labeled with the help of the Raman spectra because of the complex structural groups in the networks.

Fig. 1.

FTIR spectra for the glasses.

The 400–600 cm⁻¹ band, assigned to the O–Si–O bending vibrations,^22,23) got less pronounced from sample CSBT-1 to CSBT-4 because of the lower concentration of SiO₄ tetrahedron with increasing B₂O₃ content. On the contrary, the 600–800 cm⁻¹ band, associated with B–O–B bending vibrations,^23,24) became more profound with increasing B₂O₃ content, which indicated increasing boron atoms were interconnected into the networks as B₂O₃ substituted CaO and SiO₂. Especially, a band located at ~710 cm⁻¹, generally attributed to the oxygen bridges between two BO₃ trigonal,^25,26) gradually grew with increasing B₂O₃ content, indicating one main type of boron-related structural group in the glasses could be BO₃ trigonal.

The 800–1200 cm⁻¹ band was a remarkable region in the FTIR spectra of silicate glasses, which was generally associated with the stretching vibrations of SiO₄ tetrahedral.^22,27) Interestingly, a shoulder centered near 845 cm⁻¹, assigned to Q⁰(Si), slightly became less apparent with increasing B₂O₃ content, which could suggest an increasing degree of polymerization (DOP) of the networks. In the spectral region of 1200–1600 cm⁻¹, two bands, centered at 1215 cm⁻¹ and 1374 cm⁻¹, gradually appeared only with the existence of B₂O₃ in the glasses. These two bands could be assigned to the BO₄ tetrahedral and BO₃ trigonal, respectively,^25,26) which largely accounted for the dominant types of boron-related structural groups present in the borosilicate glasses.

3.2. Raman Spectra for the Glass Samples

To further explore and verify the structural analysis, Raman measurements were carried out. The spectra obtained are presented in Fig. 2. Similar to the FTIR curves, these Raman curves showed a continuous variation with varying B₂O₃ contents. The whole Raman spectral range between 100 cm⁻¹ and 1600 cm⁻¹ could be divided into four regions, viz., the 600–1200 cm⁻¹ band with low intensity, the 1200–1600 cm⁻¹ band with high intensity and the 100–250 cm⁻¹ and 250–600 cm⁻¹ regions with medium intensity.

Fig. 2.

Raman spectra for the glasses.

The band in the range of 250–600 cm⁻¹ could be attributed to the O–Si–O bending vibrations and those due to Ca–O complex,^15,16) the shape of which slightly varied. In addition, it can be noted, in range of 1200–1600 cm⁻¹, a small band centered at 1370 cm⁻¹,^25,26) originating from BO₃ trigonal, gradually appeared with B₂O₃ additions from sample CSBT-1 to CSBT-4; the variation trend of this band also indicated a dominant existence of BO₃ trigonal in the networks, which was in agreement with the FTIR spectra. While compared to the FTIR spectra, the BO₄ tetrahedral was not observed in the Raman spectra, which should be further confirmed using other techniques in the following sections.

Moreover, there was a more remarkable band in the range of 600–1200 cm⁻¹, which could be attributed to the stretching vibrations of SiO₄ tetrahedra and titanium-related structures.^{15,16,28,29,30,31)} In particular, according to the previous studies, this region could be divided into two parts, i.e., the band of ~600–800 cm⁻¹ originating from titanium-related structural units^28,29,30,31) and the band of ~800–1150 cm⁻¹ originating from SiO₄ tetrahedral.^15,16,27) In order to determine the number of these structure units, in an effort to quantification, deconvolution of the Raman curves was conducted in the range of 600–1150 cm⁻¹, strictly following the rules proposed by Mysen et al.³²⁾ First, it is assumed that the Raman curves can be divided into several Gaussian functions stemming from different local structures. Second, the peaks, shoulders or bands are inserted into the spectral curves only in the regions strictly proved by the studies. As for the silicon-related structures, the Raman spectra have been numerously investigated and generally, the peaks at ~870, ~960, ~990, and ~1050 cm⁻¹ could be assigned to the stretching modes of Q⁰(Si), Q¹(Si), Q²(Si), and Q³(Si), respectively.^13,14,15,16) As for the titanium-related structures, many researches demonstrated that two bands at around 710 cm⁻¹ and 790 cm⁻¹ could be assigned to the O–Ti–O deformation and the TiO₄⁴⁻ monomers, respectively,^15,28,29,30) which was also clearly observed in the present Raman spectra. In summary, six Gaussian functions originating from silicon-related and titanium-related local structures were adopted to deconvolute the Raman curves in the range of 600–1150 cm⁻¹.

The deconvolution results of different Raman curves are presented in Fig. 3. It can be observed these six Gaussian functions fitted well with the Raman envelopes, which, actually, was a confirmation of the applicability of the present methodology. As one of the main objective of this study was to quantify the glass structures, after Raman fittings, the mole fractions (X_i) (i=0, 1, 2, 3) of various SiO₄ tetrahedral (Q⁰(Si), Q¹(Si), Q²(Si) and Q³(Si)) could be further derived according to the peak areas (A_i), calibrated by the Raman scattering coefficient (S_i), by means of Eq. (1):^14,33,34)   

$$X_i\text{=(}{A}_i/S_i\text{)/}\left(\sum _{i=0}^3{A}_i/S_i\right)$$(1)

Where S₀, S₁, S₂ and S₃ equaled 1, 0.514, 0.242 and 0.09, respectively.

Fig. 3.

Deconvolution of Raman spectra: (a) CSBT-1, (b) CSBT-2, (c) CSBT-3 and (d) CSBT-4.

The results are presented in Fig. 4. As can be observed, the mole fraction of Q¹(Si) and Q³(Si) increased significantly while those of Q⁰(Si) and Q²(Si) decreased with increasing B₂O₃ content. This indicated that the SiO₄ tetrahedrons with more bridging oxygen (BO) prominently increased; in other words, with B₂O₃ addition, there appeared an equilibrium reaction between the NBO and the BO that the NBO was consumed and more BO was formed, which would be further analyzed by the O1s XPS spectra. Moreover, the variation trend of various Qⁱ(Si) indicated an increasing DOP of the glass structures, which was in agreement with the results of FTIR spectra. These results, indeed, proved that B₂O₃ acted as a typical network former contributing to polymerization of the silicate glasses. In the case of pure B₂O₃ forming a glass phase, the only structural units, are BO₃ trigonal, as shown in previous study.³⁾ Thus it can be deduced that the BO₃ trigonal existing caused by the B₂O₃ addition, captured the NBO of Q⁰(Si) and Q²(Si) and got transformed into BO₄ tetrahedral, resulting in the formation of more Q¹(Si) and Q³(Si). Based on the foregoing analysis, the reactions occurring in the glasses with B₂O₃ addition could be described by the following equations:   

$${\text{BO}}_{\text{3}}{\text{+Q}}^0\left(\text{Si}\right)\to {\text{BO}}_{\text{4}}{\text{+Q}}^1\left(\text{Si}\right)$$(2)

  

$${\text{BO}}_{\text{3}}{\text{+Q}}^2\left(\text{Si}\right)\to {\text{BO}}_{\text{4}}{\text{+Q}}^3\left(\text{Si}\right)$$(3)

Fig. 4.

Mole fractions of various Qⁱ(Si) units in the networks by Raman spectra fittings.

3.3. O1s XPS Spectra for the Samples

In order to verify the occurrence of the equilibrium reaction between the NBO and BO in borosilicate glasses and to further clarify this mechanism, the O1s XPS spectra were obtained. The results are presented in Fig. 5. As can be seen, the O1s XPS spectra could be divided into two parts, a peak corresponding to higher bond energy, attributed to BO and a peak with lower bond energy, attributed to NBO.^23,35) A variation trend was that the center of this band gradually moved to the left side with higher binding energy with increasing B₂O₃ content, which indicated an increasing BO content and thus a higher DOP in the networks. In addition to the variation trend of the mole fractions of various Qⁱ(Si), it was further indicative of an equilibrium reaction between the NBO and BO in the glasses,³⁵⁾ as described by means of the following equations:   

$${\text{Q}}^0\left(\text{Si}\right)-\text{NBO}\to {\text{Q}}^1\left(\text{Si}\right)$$(4)

  

$${\text{Q}}^2\left(\text{Si}\right)-\text{NBO}\to {\text{Q}}^3\left(\text{Si}\right)$$(5)

Fig. 5.

O1s XPS spectra for the samples.

To further quantify the relative abundance of NBO and BO in the glasses, two Gaussian functions, associated with NBO and BO, respectively, could be applied to deconvolute the O1s XPS spectra. This method has, actually, been demonstrated as an effective way to analyze the state of oxygen in a glass.²³⁾ As shown in Fig. 6, two Gaussian functions could exactly fit these spectra. Based on the areas of the two peaks associated with BO and NBO, the content of NBO and BO could be deduced, as displayed in Fig. 7. It can be clearly seen that the BO content gradually increased with the B₂O₃ content increasing from 0 to 15% with a small increment, which confirmed the occurrence of the reactions (4 and 5) in the glasses. On the other hand, an increase of the BO content was indicative of an increase of DOP of the glass structures, which, again, was in consistent with the results of FTIR spectra and Raman deconvolutions.

Fig. 6.

Deconvolution of O1s XPS spectra: (a) CSBT-1, (b) CSBT-2, (c) CSBT-3 and (d) CSBT-4.

Fig. 7.

Content of BO and NBO in the networks by O1s XPS spectra fittings.

3.4. ¹¹B MAS-NMR Spectra for the Samples

After quantifying the structural units of SiO₄ tetrahedral and oxygen in the glasses, the structural information obtained in the present work can further be extended to the derivation of the boron-related structural units such as the relative content of BO₄ tetrahedral and BO₃ trigonal, since the B₂O₃ content acted as a main variable in the present glasses. Therefore ¹¹B MAS-NMR spectra measurements were further performed, as presented in Fig. 8. Based on the NMR spectra, several characteristics could be clarified. First, the NMR spectra could be divided into two main regions, i.e., a region associated with BO₃ trigonal at higher chemical shifts and a region associated with BO₄ tetrahedral at lower chemical shifts, respectively.^{8,36,37,38,39)} Second, generally, the higher chemical shift region was composed of two bands, assigned to two types of BO₃ trigonal, namely BO₃ trigonal rings and BO₃ trigonal that do not form rings.^8,38,39)

Fig. 8.

¹¹B MAS-NMR spectra for the glasses.

Third, the lower chemical shift region, originating from the vibration modes of BO₄ tetrahedral, was mainly made up of a peak at around 2.15 ppm, which could be assigned to the BO₄ tetrahedral connected to three silicon atoms and one boron atom (BO₄ (1B, 3Si)).^39,40) An apparent trend was that, with increasing B₂O₃ content, this peak became more pronounced, indicating that more BO₄ (1B, 3Si) was formed in the glasses. In addition, another weak shoulder was noticed appearing at around −3.85 ppm, which originated from the BO₄ tetrahedral connected to four silicon atoms (BO₄ (0B, 4Si)).^8,38,39) In order to quantify the relative content of BO₄ tetrahedral and BO₃ trigonal in the glasses, it is crucial to deconvolute the NMR spectra. However, a reasonable fitting of an NMR curve was not easy because the higher chemical shift region, associated with BO₃ trigonal, could not be simply simulated using several Gaussian function.^8,38,39) But the lower chemical shift region associated with BO₄ tetrahedral, as observed, could be fitted using two Gaussian functions related to BO₄ (1B, 3Si) and BO₄ (0B, 4Si). Therefore a feasible approach could be that the lower chemical shift region was first tentatively deconvoluted using two Gaussian functions and then the differences of initial NMR values and these two Gaussian functions were taken as the band of BO₃ trigonal.

The foregoing fitting process was performed in this study and the results are presented in Fig. 9. The three bands observed, ascribed to BO₄(1B, 3Si), BO₄ (0B, 4Si) and BO₃ trigonal, respectively, can exactly and smoothly simulate the ¹¹B MAS-NMR spectra, providing a possible solution to the structure clarification using the present methodology. Based on areas of the bands from the NMR fittings, the relative content of different boron-related structures could be evaluated. The calculation results are presented in Fig. 10. First, the ratio of BO₄ (1B, 3Si) to BO₄ (0B, 4Si) could be determined by calculating the ratio of the areas of these two Gaussian functions. As can be seen, an increase of B₂O₃ content gave rise to a higher ratio of BO₄ (1B, 3Si) to BO₄ (0B, 4Si). Actually, an important effect of BO₄ group avoidance reported in literatures^8,39) was that the BO₄ tetrahedral showed a preferential trend for interconnection with near silicon atoms leading to less B–O–B bonds in the borosilicate glasses. This was confirmed by the results obtained in the present study. However, with increasing B₂O₃ content, the content of SiO₂ decreased simultaneously, which in turn, would lead to fewer lattices for BO₄ insertion. Thus more BO₄ (1B, 3Si) was formed, increasing the number of B–O–B bonds. This phenomenon, indeed, indicated another equilibrium reaction between BO₄ (1B, 3Si) and BO₄ (0B, 4Si), as described by the following equation:   

$${\text{2BO}}_4\text{(0B,4Si)}\to {\text{BO}}_4{\text{(1B,3Si)+SiO}}_4$$(6)

Fig. 9.

Deconvolution of ¹¹B MAS-NMR spectra: (a) CSBT-2, (b) CSBT-3 and (c) CSBT-4.

Fig. 10.

Content of BO₃ trigonal and BO₄ tetrahedral in the networks by ¹¹B MAS-NMR spectra fittings.

Moreover, another important phenomenon observed was that the ratio of BO₄ tetrahedral to BO₃ trigonal increased with increasing B₂O₃ content, which finally demonstrated the equilibrium reaction between these two species. On the other hand, more BO₄ tetrahedral was present in the structures as compared with BO₃ trigonal related to the newly added B₂O₃ in the glasses. Assuming that the added B₂O₃ originally existed as pure BO₃ trigonal, these BO₃ trigonal could get linked to the NBO and gave rise to BO₄ tetrahedral. Considering the varying content of BO-NBO in the glasses, the equilibrium reaction could be described as follows:^11,12)   

$${\text{BO}}_{\text{3}}\text{+NBO}\iff {\text{BO}}_{\text{4}}$$(7)

This equation was, in fact, another form of Eqs. (2) and (3). Based on Eq. (7), another parameter, equilibrium constant of this reaction, K, could be introduced, defined as:^11,40)   

$$\text{K=}\frac{{\text{X}}_{{\text{BO}}_{\text{4}}}}{{\text{X}}_{{\text{BO}}_3}{\text{X}}_{\text{NBO}}}$$(8)

where the X_BO4, X_BO3 and X_NBO represent the content of BO₄ tetrahedra, BO₃ trigonal and NBO respectively. In this work, with increasing B₂O₃ content, initially the content of BO₃ trigonal show an increase; then the reaction proceeded to the right side and the content of NBO decreased as a consequence. This, once again, was in conformity with the results of Raman and O1s XPS fitting that an increasing B₂O₃ content in the glasses resulted in an increasing BO content in the networks, overall.

As aforementioned, the clarification of the structural units of the glasses itself was extremely interesting and this could also enable an exploration of the structure-property relations. First, for the viscosity change,^17,18,19,20) the dominant existence of the two-dimension BO₃ trigonal in the three-dimension networks could weaken the whole rigidity of the structures. This effect could thus cause a decrease of the melt viscosities, despite the increasing DOP of the glasses. Second, the added B₂O₃, as a typical acid oxide, could act as a network former in the glasses, resulting in an increase of DOP. Specially, another main boron-related local structure in the glasses was BO₄ tetrahedral, which required that the Ca²⁺ cations acted as change balancing cations. This, in turn, restricted the connectivity of Ca²⁺ and TiO₄⁴⁻ and consequently enhanced the precipitation of TiO₂, as demonstrated by our previous study.²¹⁾ Thus, the present study not only leaded to a better understanding of the structure of borosilicate glasses, but also provided a possible insight of the structure-property relationships especially in the medium-range order.

4. Conclusions

This present work attempted an understanding of the short-range and medium-range structural order in CaO–SiO₂–TiO₂–B₂O₃ glass. In order to confirm the structural analysis, FTIR, Raman, O1s XPS and ¹¹B MAS-NMR spectra were complementarily employed. It was discovered that the fraction of Q¹(Si) and Q³(Si) increased while fraction of Q⁰(Si) and Q²(Si) showed a decrease with increasing B₂O₃ content. The DOP of the glasses was consequently enhanced. This agreed well with the results of O1s XPS fittings that the content of BO increased while that of the NBO decreased with more B₂O₃ additions. Additionally, not only the FTIR but also the ¹¹B MAS-NMR results demonstrated the existence of BO₃ trigonal and BO₄ tetrahedral in the networks. BO₃ trigonal could generally be divided into ring and non-ring BO₃ and the BO₄ tetrahedral was made up of BO₄(1B,3Si) and BO₄(0B,4Si). Based on ¹¹B MAS-NMR deconvolutions, it was found that an increase of B₂O₃ content gave rise to the ratio of BO₄ tetrahedral to BO₃ trigonal. Most importantly, an equilibrium reaction between NBO, BO₄ tetrahedral and BO₃ trigonal was identified in the glass, considering the variations of the relative amounts of BO–NBO and BO₃–BO₄.

Acknowledgements

The authors gratefully acknowledge financial support by the Common Development Fund of Beijing, the National Natural Science Foundation of China (51172001, 51074009 and 51172003), the National High Technology Research and Development Program of China (863 Program, 2012AA06A114).

References

• 1)   J.  Couzin: Science, 309 (2005), 83.
• 2)   P. I. K.  Onorato and  D. R.  Uhlmann: J. Non-Cryst. Solids, 22 (1976), 367.
• 3)   Y. D.  Yiannopoulos,  G. D.  Chryssikos and  E. I.  Kamitsos: Phys. Chem. Glasses-Eur. J. Glass Sci. Technol. Part B, 42 (2001), 164.
• 4)   R. E.  Youngman: Science, 345 (2014), 998.
• 5)   Z. W.  Wu,  M. Z.  Li,  W. H.  Wang and  K. X.  Liu: Nat. Commun., 6 (2015), 6035.
• 6)   P. W.  Anderson: Science, 267 (1995), 1609.
• 7)   S. R.  Elliott: Nature, 354 (1991), 445.
• 8)   X.  Wu,  R. E.  Youngman and  R.  Dieckmann: J. Non-Cryst. Solids, 378 (2013), 168.
• 9)   G. H.  Kim and  I.  Sohn: Metall. Mater. Trans. B, 45 (2014), 86.
• 10)   Y. Q.  Sun and  Z. T.  Zhang: Metall. Mater. Trans. B, 46 (2015), 1549.
• 11)   J. F.  Stebbins and  S. E.  Ellsworth: J. Am. Ceram. Soc., 79 (1996), 2247.
• 12)   T.  Edwards,  T.  Endo,  J. H.  Walton and  S.  Sen: Science, 345 (2014), 1027.
• 13)   P.  McMillan: Am. Mineral., 69 (1984), 622.
• 14)   J. D.  Frantz and  B. O.  Mysen: Chem. Geol., 121 (1995), 155.
• 15)   B. O.  Mysen and  D. R.  Neuville: Geochim. Cosmochim. Acta, 59 (1995), 325.
• 16)   D. R.  Neuville,  D. D.  Ligny and  G. S.  Henderson: Rev. Mineral. Geochem., 78 (2014), 509.
• 17)   Y. Q.  Sun,  J. L.  Liao,  K.  Zheng,  X. D.  Wang and  Z. T.  Zhang: JOM, 66 (2014), 2168.
• 18)   A. I.  Priven: Glass Phys. Chem., 27 (2001), 360.
• 19)   D. B.  Dingwell,  R.  Knoche,  S. L.  Webb and  M.  Pichavant: Am. Mineral., 77 (1992), 457.
• 20)   S.  Ren,  J.  Zhang,  L.  Wu,  W.  Liu,  Y.  Bai,  X.  Xing,  B.  Su and  D.  Kong: ISIJ Int., 52 (2012), 984.
• 21)   Y. Q.  Sun,  Z. M.  Li,  L. L.  Liu,  X. D.  Wang and  Z. T.  Zhang: ISIJ Int., 55 (2015), 158.
• 22)   J. H.  Park,  D. J.  Min and  H. S.  Song: ISIJ Int., 42 (2002), 344.
• 23)   X.  Zhu,  C.  Mai and  M.  Li: J. Non-Cryst. Solids, 388 (2014), 55.
• 24)   D.  Seo,  S.  Park and  J.  Lee: Curr. Appl. Phys., 9 (2009), 72.
• 25)   E. I.  Kamitsos,  M. A.  Karakassides and  G. D.  Chryssikos: J. Phys. Chem., 91 (1987), 1073.
• 26)   I. J.  Hidi,  G.  Melinte,  R.  Stefan,  M.  Bindea and  L.  Baia: J. Raman Spectrosc., 44 (2013), 1187.
• 27)   S. A.  MacDonald,  C. R.  Schardt,  D. J.  Masiello and  J. H.  Simmons: J. Non-Cryst. Solids, 275 (2000), 72.
• 28)   Z.  Wang,  Q.  Shu and  K.  Chou: ISIJ Int., 51 (2011), 1021.
• 29)   K.  Zheng,  J. L.  Liao,  X. D.  Wang and  Z. T.  Zhang: J. Non-Cryst. Solids., 376 (2013), 209.
• 30)   K.  Zheng,  Z. T.  Zhang,  L. L.  Liu and  X. D.  Wang: Metall. Mater. Trans. B, 45 (2014), 1389.
• 31)   Y. Q.  Sun,  Z. T.  Zhang,  L. L.  Liu and  X. D.  Wang: J. Non-Cryst. Solids., 420 (2015), 26.
• 32)   B. O.  Mysen,  D.  Virgo and  C. M.  Scarfe: Am. Mineral., 65 (1980), 690.
• 33)   B. O.  Mysen and  J. D.  Frantz: Am. Mineral., 78 (1993), 699.
• 34)   Y. Q.  Wu,  G. C.  Jiang,  J. L.  You,  H. Y.  Hou and  H.  Chen: Acta Phys. Sin., 54 (2005), 961.
• 35)   J. B.  Kim and  I.  Sohn: J. Non-Cryst. Solids, 379 (2013), 235.
• 36)   J. F.  Stebbins and  Z.  Xu: Nature, 390 (1997), 60.
• 37)   R. E.  Youngman,  S. T.  Haubrich,  J. W.  Zwanziger,  M. T.  Janicke and  B. F.  Chmelka: Science, 269 (1995), 1416.
• 38)   L. S.  Du and  J. F.  Stebbins: J. Phys. Chem. B, 107 (2003), 10063.
• 39)   L. S.  Du and  J. F.  Stebbins: J. Non-Cryst. Solids, 315 (2003), 239.
• 40)   R. J.  Araujo: J. Non-Cryst. Solids, 58 (1983), 201.