Network Structure and Properties of Lithium Aluminosilicate Glass

Based on lithium aluminosilicate glass, the composition of glass was optimized by replacing SiO2 with B2O3, and the influence of glass composition on structure and performance was studied. With the increase in B2O3 concentrations from 0 to 6.5 mol%, Al2O3 always existed in the form of four-coordinated [AlO4] in the network structure, and B2O3 mainly entered the network in the form of four-coordinated [BO4]. The content of Si-O-Si linkages (Q4(0Al)) was always dominant. The incorporation of boron oxide improved the overall degree of polymerization and connectivity of the lithium aluminosilicate glass network structure. An increase in the degree of network polymerization led to a decrease in the thermal expansion coefficient of the glass and an increase in Vickers hardness and density. The durability of the glass in hydrofluoric acid and NaOH and KOH solutions was enhanced overall.


Introduction
B 2 O 3 has been incorporated into various glass systems to obtain the required chemical and physical properties [1]. Boron-containing silicate glass is a commonly used material in industrial applications and has been widely used in applications in industrial technology fields including pharmaceutical containers, sealing glass, optics, flat-panel displays, electronic packaging and low-expansion glass [2][3][4][5]. The ability of the modifiers and network formers in the glass components to compete for oxygen affects the coordination of boron, that is, the glass component plays an important role in the coordination of boron. Therefore, the structure of boro-aluminosilicate glass is more complicated [6]. In pure boron oxide glass, boron is three-coordinated. With the addition of alkali metal oxides, alkali metal ions compensate for the charge of the [BO4]-unit, and four-coordinated boron begins to appear [7]. Borate or borosilicate glass exhibits nonlinear composition-property relationships, which is a phenomenon known as the "boron anomaly" [8][9][10]. The relationship between the composition, performance and structure of traditional borosilicate glass has been extensively and systematically studied [11][12][13][14][15][16]. For instance, Yun and Bray [17] put forward a model for calculating fraction N4 of tetra-coordinated boron. According to the model, when the ratio of sodium oxide to boron oxide exceeds 0.5, there are enough sodium ions to compensate the negative charge of [BO4] units, and the number of [BO4] units in glass increases; on the contrary, the number of [BO3] units is greater. Jonathan F. Stebbins [18] quantified the number of non-bridging oxygens bonded to boron in sodium borate glass and barium borate glass, which have a major influence on glass properties, and enlightened the quantification of the number of non-bridging oxygens in borate networks
It was likely that a significant proportion of boron volatilized at high temperatures and caused deviations from nominal compositions. With this in mind, an additional 12% B 2 O 3 was added to each sample to compensate for volatilization prior to melting. For example, 1.5 mol% B 2 O 3 for 2# was equivalent to 1.72 wt% oxide, and we added an additional 1.72 × 12% g B 2 O 3 per 100 g of oxide in a well-mixed batch. The practical proportion and volatility of B 2 O 3 measured by chemical analysis are shown in Table 1. The test method was in accordance with the Chinese national standard method. From the quantitative analysis results of ICP and XRF, the volatiles of B 2 O 3 in the samples were not exactly the same, but the actual concentration of B 2 O 3 was close to the nominal value. As shown in Table 1, the deviation of boron content was not more than 5%. Deviations from nominal components were, therefore, ignored in the calculations to simplify the understanding process.

Glass Testing and Characterization
Solid-state 29 Si NMR spectra were obtained using a Bruker AVANCE III HD 400 MHz solid-state NMR spectrometer, which had an H/X dual resonance solid probe, a 4 mm ZrO 2 rotor and a rotor speed of 14 kHz; the detection resonance frequency of the 29 Si NMR spectrum is 79.49 MHz, and the magnetic field strength is 9.4 T, while scan times are generally 512 times an hour. 27 Al NMR spectra were obtained using a Bruker AVANCE III HD 400 MHz solid-state NMR spectrometer, which also had an H/X dual-resonance solid-state probe, a 4 mm ZrO 2 rotor, and a rotating speed of 14 kHz. The detection resonance frequency of 27 Al is 104.26 MHz, and the number of scans is generally 4096 times an hour. 11 B solid-state NMR spectra were acquired using a Bruker AVANCE NEO 400 WB solid-state superconducting NMR spectrometer (Fällanden, Switzerland), with samples loaded in 4 mm ZrO 2 sample tubes with a Bruker 4 mm standard probe, using the singlepulse technique. The resonance frequency of the boron spectrum is 128.39 MHz. The deconvolution of the NMR spectra was performed with DM-fit software [40]. When only the central part of the spectrum was considered, the non-linear least squares method was used to iteratively determine the positions of different components and calculate the corresponding areas. The accuracy of the chemical shift value of the NMR spectrum was estimated to be 0.2 ppm, and the accuracy of the area quantification was estimated to be 3%. Raman excitation was applied using a solid-state 532 nm wavelength laser and an output power set to 40 milliwatts. The acquisition time for a given sample was usually between 20 and 60 s, and 6 scans were accumulated. In order to determine whether the glass samples were uniform, at least 4 spectra were collected for each sample. Subtract the background using a third-order polynomial function anchored in the unbanded region:~200 cm −1 , 850 cm −1 , 1200 cm −1 , and 1600 cm −1 . Raman spectra were acquired at room temperature without correction for the temperature and frequency dependence of scattering intensity. At room temperature, a third-order polynomial function anchored in the non-band region was used to subtract the spectrum from the spectrum background from 200 to 1800 cm −1 .
Vickers hardness was measured with a microhardness tester. The experimental loading time was 10 s, and the loading pressure was 1.96 N; each sample was tested 5-10 times to reduce experimental errors and improve the accuracy and reliability of the experiment. Archimedes' buoyancy method was used to measure the density of the samples, and the molar volume was calculated according to Formulas (1) and (2) combined with the density. The dynamic method was used to measure the thermal expansion coefficient of the sample. The instrument used was an American Orton DIL2010STD thermal expansion instrument. The shaped and annealed glass samples were cut into strips with a length × width × height of 25.4 × 8 × 5 mm, the heating rate was 5 • C/min, and the heating temperature range was 25~500 • C.The shaped and annealed glass samples were cut into strip samples with a size of 25.4 × 8 × 5 mm, at a heating rate of 5 • C/min and a heating temperature range of 25~500 • C. The sample was cut into a size of 30 mm × 30 mm × 1 mm and optically polished for chemical stability testing. The glass slides were immersed in a 10 vol% hydrofluoric acid solution at room temperature for 30 min, and some glass plates were immersed in a solution of equal volumes of 1 mol/L NaOH and 1 mol/L KOH at 90 • C for 4 h. The chemical stability of glass is mainly expressed by the weight loss ratio (WLR) [22].
where ρ g is the density of the glass sample; ρ w is the density of water at room temperature, and the unit of density is g/cm 3 ; M is the mass of the glass in air (g); and m is the equilibrium reading of glass in water (g).
where ρ is the glass density; x i is the mole fraction of oxide component I; and M i is the molecular weight.
where m 1 is the mass of ultrasonic cleaning before etching; m 2 is the mass of ultrasonic cleaning after etching; and s is the surface area of glass sample.

27 Al NMR Analysis
The 27 Al NMR spectra of boron-doped lithium aluminum silicate glass are shown in Figure 1a. The 27 Al MAS NMR peaks of the seven glass samples were all asymmetric composite peaks with a relatively broad distribution due to the quadrupole-broadening and localized-bonding environment related to compositional complexity, and the chemical shift in the resonance peak was~42 ppm; the peak shapes were also roughly similar, indicating that the morphologies of the aluminum in the glass samples were relatively similar. With the increase in boron oxide content, the chemical shift in the peak shifted to high field, which may have been due to the fact that the oxygen in the Al-O bond carries low formal charge, and the amount of non-bridging oxygen in the Al-O network may be decreasing or the second element adjacent to it may change (B III or Si is preferably to B IV and Al IV ) [41]. In addition, the spectral peak widened asymmetrically toward the low-chemical-shift direction, which may have been related to the appearance of the high coordination of aluminum ions and the change in quadrupole-coupling parameters.
In order to describe the coordination state of aluminum ions in the network structure and their respective contents, the reference and DM-fit software were used to fit and analyze the spectral peaks of the 27 Al nuclear magnetic resonance of lithium aluminum silicate glass with the GIM model. GIM model fitting parameters included the following: the FWHM CS value was set to 14; the EM au value was set to −200; the d value was set to 5. The size of the computed spectra showed good rendering for 2048, but it was better with 8192, which was the preferred value.
Step CSA, i.e., the step for sampling the isotropic CSA distribution in amorphous models, is typically 50, depending on the apodization factor.
Step nuQ, i.e., the step for sampling the Gaussian distribution of quadrupolar interaction in amorphous or int2Q models, is typically 21, which can give decent results depending upon the final apodization factor. Sw mult, i.e., the sweep width multiplier, should be set to 4 in principle.
The chemical shift in the Al IV peak in the network structure is about 45.8 ppm [42]. The peak shifts of Al V and Al VI are around 38-28 ppm and (-5)-15 ppm [43]. The GIM model fitting and calculation results of 27 Al NMR spectra are shown in Figure 1b and Table  1. In the alumina network of boron-doped LAS glass, aluminum was always four-coordinated. The content of Al V was only second to that of Al IV . It can be seen from Table 2 that with the increase in boron content, the contents of Al IV , Al V and Al VI in the aluminumoxygen network changed little, and the spectral peak of Al IV gradually moved toward the direction of low chemical shift.   The chemical shift in the Al IV peak in the network structure is about 45.8 ppm [42]. The peak shifts of Al V and Al VI are around 38-28 ppm and (-5)-15 ppm [43]. The GIM model fitting and calculation results of 27 Al NMR spectra are shown in Figure 1b and Table 1. In the alumina network of boron-doped LAS glass, aluminum was always four-coordinated. The content of Al V was only second to that of Al IV . It can be seen from Table 2 that with the increase in boron content, the contents of Al IV , Al V and Al VI in the aluminum-oxygen network changed little, and the spectral peak of Al IV gradually moved toward the direction of low chemical shift.  Figure 2a shows the NMR spectra of glass 11 B. When a relatively weak magnetic field of 9.39 T was used, the NMR spectra contained two parts of the signal that could be resolved, which was consistent with the three-coordinated and four-coordinated boron atoms in the glass structure. It appeared that the symmetrical and relatively narrow Gaussian line shape in the~0 ppm region was attributed to the [BO 4 ] tetrahedron (i.e., B IV ), while the complex-shaped, low-intensity asymmetric broad signal with chemical shifts of NMR peaks close to~10 ppm corresponded to [BO 3 4 ] associated with the two species changed little, so the properties of the entities did not appear to change significantly during the substitution. With the increase in boron oxide content, the absorption intensity of the boron-oxygen tetrahedral NMR spectral peak located at~0 ppm increased gradually, that is, the content of B IV increased.  Figure 2a shows the NMR spectra of glass 11 B. When a relatively weak magnetic field of 9.39 T was used, the NMR spectra contained two parts of the signal that could be resolved, which was consistent with the three-coordinated and four-coordinated boron atoms in the glass structure. It appeared that the symmetrical and relatively narrow Gaussian line shape in the ~0 ppm region was attributed to the [BO4] tetrahedron (i.e., B IV ), while the complex-shaped, low-intensity asymmetric broad signal with chemical shifts of NMR peaks close to ~10 ppm corresponded to [BO3] units (B III ). The chemical shifts of [BO3] and [BO4] associated with the two species changed little, so the properties of the entities did not appear to change significantly during the substitution. With the increase in boron oxide content, the absorption intensity of the boron-oxygen tetrahedral NMR spectral peak located at ~0 ppm increased gradually, that is, the content of B IV increased.  The 11 B NMR peaks were deconvolved with a Gaussian model to determine the components (B III and B IV ). The area percentages of B IV and B III were calculated, along with their respective contents (N 4 and N 3 ; N 4 + N 3 = 1). The connectivity of the silicate and borate networks could also be reflected by 11 B NMR, and the fitting results are shown in Figure 2c. The connection of four-coordinated boron atoms with silicon atoms included: B IV -2Si (one four-coordinated boron atom linked to two silicon atoms via bridging oxygen), 2.5 ppm; and B IV -3Si (one four-coordinated boron atom connected with three silicon atoms via bridging oxygen),~0.5 ppm [43].

] units (B III ). The chemical shifts of [BO 3 ] and [BO
From the fitting results in Figure 2b,c and Table 3, it can be seen that the network structure of glass was dominated by B IV . The bonding states of free oxygen and boron atoms provided by alkaline earth metals or alkali metal oxides were related to their molar ratio, that is, Ψ = (O-Al 2 O 3 )/B 2 O 3 (where O is the molar number of free oxygen); when Ψ ≥ 1, boron mainly existed in tetra-coordination. In the boron-doped glass system, when all Ψ > 1, the network structure was dominated by B IV . With the increase in the boron oxide content, N 4 increased, and the area percentage of the B IV -3Si spectral peak at~0.5 ppm was always more than the B IV -2Si spectral-peak area percentage at~2.5 ppm.

29 Si NMR Analysis
The peaks of the 29 Si NMR spectra ranged from −90 to~−130 ppm (Figure 3a), centered on a broad resonance of −110 ppm. With the increase in boron oxide content, the spectral peaks tended to shift to the low field of~−150 ppm, which was related to the existence of aluminum or boron atoms in the silicon-oxygen tetrahedron and the nonbridging oxygen number in the silicon-oxygen tetrahedron. The chemical shift in the spectral peaks of glass sample No. 7# was the most obvious. The presence of boron or aluminum resulted in a shift in the chemical shifts, and the substitution of [SiO 4 ] with Al IV or B IV increased the chemical shift values by approximately 5 ppm [44]. Similarly, the presence of non-bridging oxygen atoms within the silicon coordination polyhedron caused the peaks to shift toward the high fields due to the high formal charge carried by oxygen [41]. From the fitting calculation results of the 29 Si NMR spectra (Figure 3b), it can be seen that the main species in the glass network structure was Q4 (0Al) , followed by Q4 (1Al) ; the area percentage of the two exceeded 60%, and Q4 (3Al) and Q4 (4Al) contents were low. Among them, the Q4 (0Al) content of glass samples Nos. 1#~6# increased with the increase in boron oxide content, while the content of Q4 (0Al) of glass sample No. 7# decreased compared with that of No. 6#. However, the area percentage of Q4 (0Al) eventually dominated.   The deconvolution of the 29 Si NMR spectra could quantify the quantity of Q n m (Al) in the network, where 0 ≤ m ≤ n ≤ 4, with n being the coordination number of silicon and m the aluminum atom connected with the silicon atom through the bridging-oxygen quantity. Q n m(Al) was used to describe the connectivity of the silicon-oxygen network and aluminum-oxygen network. The connection between silicon and four-coordinated aluminum included: Q 4 (0Al) connected with four silicon atoms (−102~−115 ppm); Q 4 (4Al) connected with four aluminum atoms (−82~−90 ppm); Q 4 (3Al) connected with three aluminum atoms (−85~−94 ppm); Q 4(2Al) connected with two aluminum atoms (−92~−98 ppm); Q 4 (1Al) connected with an aluminum element (−96~−107 ppm) [45]. The Gaussian model fitting results of the 29 Si NMR spectra of glass are shown in Figure 3b and Table 4. From the fitting calculation results of the 29 Si NMR spectra (Figure 3b), it can be seen that the main species in the glass network structure was Q 4 (0Al) , followed by Q 4 (1Al) ; the area percentage of the two exceeded 60%, and Q 4 (3Al) and Q 4 (4Al) contents were low. Among them, the Q 4 (0Al) content of glass samples Nos. 1#~6# increased with the increase in boron oxide content, while the content of Q 4 (0Al) of glass sample No. 7# decreased compared with that of No. 6#. However, the area percentage of Q 4 (0Al) eventually dominated. Figure 4 shows the Raman spectra of lithium aluminosilicate glass. The relationships between the vibrations of the main functional groups in the spectra and the corresponding wave numbers are shown in Table 5 The wave-number Raman spectra of the glass sample in the range of 200-1800 cm −1 are shown in Figure 4; there were characteristic peak intensity responses at 400-470 cm −1 , 700-900 cm −1 and 900-1200cm −1 . The vibrational peaks of 1#~7# were concentrated at 480 cm −1 , which indicated that more 5-and 4-membered rings formed in 2#, 3# and 4#. The peak at 470 cm −1 was attributed to the mixed tensile bending vibration mode of Si-O-Si, and with the increase in B 2 O 3 (mol%), the peak intensities of 1#~6 # gradually increased, while the peak intensity of 7# decreased. Combined with 29 Si NMR, it can be seen that the content of Q 4 (0Al) decreased, and the connectivity of the silicate sub-network of glass changed. The vibrational peak at~760cm −1 was the stretching vibration of the Si-O-Al linkages between [AlO 4 ] and [SiO 4 ] formed after Al 3+ in the composition entered the network. The peak observed at~770 cm −1 was attributed to the tetrahedral coordination of boron in glass due to the higher vibrational intensity of glass with high boron and low Na 2 O contents (e.g., 6#), while the opposite was true for glass with high Na 2 O and low B 2 O 3 contents, which indicated that the modifier (such as sodium ions) functioned as a charge compensator, changing the neutral [BO 3 ] unit into [BO 4 ] -, thereby improving the aggregated state of the glass structure. It may have been formed by the breathing vibration of [BO 3 ] of the three-membered ring at~803cm −1 . Combined with 11 B NMR, it can be seen that its effect was relatively weak. In the range of 900-1200cm −1 , we observed the Si-O stretching vibration in the silicon-oxygen tetrahedron, while the band at 920cm −1 could be attributed to the [BO 4 ]containing structural unit, while the stretching vibration mode of the mixed Si-O-B linkages was also expected to function in this range [46]. In borosilicate glass, it is generally believed that for low-modified oxide additions (R < 0.5), triangular [BO 3 ] 0 converts to a [BO 4 ]tetrahedron, and many NMR results are still interpreted to support the existence of [BO 4 ]only being connected with the four silicate tetrahedra and not with the borate network. The boron content of the lithium-aluminosilicate glass system studied by the subject was low, and the band intensity related to borate was not as strong as the Si-O stretching vibration peak in the silicon-oxygen tetrahedron in the range of 900~1200 cm −1 . With the increase in B 2 O 3 (mol%), the peak intensity at 770 cm −1 gradually increased, which indicated that the content of [BO 4 ] in glass increased, which was consistent with the calculation results of 11    The wave-number Raman spectra of the glass sample in the range of 200-1800 cm −1 are shown in Figure 4; there were characteristic peak intensity responses at 400-470 cm −1 , 700-900 cm −1 and 900-1200cm −1 . The vibrational peaks of 1#~7# were concentrated at ~480 cm −1 , which indicated that more 5-and 4-membered rings formed in 2#, 3# and 4#. The peak at 470 cm −1 was attributed to the mixed tensile bending vibration mode of Si-O-Si, and with the increase in B2O3 (mol%), the peak intensities of 1#~6 # gradually increased, while the peak intensity of 7# decreased. Combined with 29 Si NMR, it can be seen that the content of Q4 (0Al) decreased, and the connectivity of the silicate sub-network of glass changed. The vibrational peak at ~760cm −1 was the stretching vibration of the Si-O-Al linkages between [AlO4] and [SiO4] formed after Al 3+ in the composition entered the network. The peak observed at ~770 cm −1 was attributed to the tetrahedral coordination of boron in glass due to the higher vibrational intensity of glass with high boron and low Na2O contents (e.g., 6#), while the opposite was true for glass with high Na2O and low B2O3 contents, which indicated that the modifier (such as sodium ions) functioned as a charge compensator, changing the neutral [BO3] unit into [BO4] -, thereby improving the aggregated state of the glass structure. It may have been formed by the breathing vibration of [BO3] of the three-membered ring at ~803cm −1 . Combined with 11 B NMR, it can be seen that its effect was relatively weak. In the range of 900-1200cm −1 , we observed the Si-O stretching vibration in the silicon-oxygen tetrahedron, while the band at 920cm −1 could be attributed to the [BO4] -containing structural unit, while the stretching vibration mode of the mixed Si-O-B linkages was also expected to function in this range [46]. In borosilicate glass, it is generally believed that for low-modified oxide additions (R < 0.5), triangular [BO3] 0 con-

Summary of the NMR and Raman Observations
The coordination state of aluminum ions in the glass network structure is related to the ratio of the alumina content in glass to the modifier oxide content, that is, Al 2 O 3 /M 2 O(MO) (where M is the modifier cation). When Al 2 O 3 :M 2 O(MO) < 1, a sufficient number of oxygen ions in the system convert aluminum ions into a tetrahedral coordination, and Al 3+ acts as a network former [47]. In the boron-doped glass system, Al 2 O 3 :M 2 O(MO) < 1, and the number of oxygen ions was sufficient. Because Al 3+ has a larger nuclear charge number than B 3+ , its ionic field is stronger, and the content of boron is smaller than that of aluminum, so it is more favorable for Al 3+ to preferentially combine with free oxygen. Therefore, the incorporation of boron did not substantially affect the coordination of aluminum ions. The increase in the coordination of aluminum ions stemmed from the competition for short and strong oxygen bonds between modifier cations and aluminum atoms. Modifiers with smaller radii and higher charges have a stronger ability to compete for oxygen, which promotes the high coordination of aluminum ions [48].
When modifier oxide was added to glass to promote the formation of B IV , the oxygen ion (O 2− ) in the oxide destroyed the network connection, converted bridging oxygen (BO) to NBO, and changed the coordination of boron. The conversion relationship between B IV and B III is shown below.
B IV B III +NBO (4) Modifier field strength, boron concentration, aluminum and total modifier content have important effects on the coordination of boron in a network structure [49]. At high temperatures, the reaction skews to the right. The effect of temperature on N 4 depends on the NBO content, which, in turn, depends on the composition. To test the interpretation of the 29 Si NMR spectrum in terms of Q n (mAl) composition, the main assumptions regarding the location of non-bridging oxygen, which is assumed to only be present in the silicon environment, and the alignment of aluminum-oxygenaluminum bonds according to Lowenstein's rule are first emphasized, with missing or negligible ratio. Furthermore, the properties of the Q n (mAl) structural environment take into account the chemical shift ranges given by Engelhardt et al. [50] and are derived from extensive studies of crystalline mineral morphology [51].
The introduction of boron oxide resulted in a shift in the signal center of gravity towards the lower-chemical-shift direction and an increase in signal asymmetry with the emergence of separation towards the lower-chemical-shift direction. Therefore, this evolution of the 29 Si NMR spectra may have reflected an increase in silicate network aggregation, especially due to the formation of Q 4 (0Al) species, or a decrease in Si-O-Al linkages, which was to be expected here. With the addition of boron, the connectivity of Al in the tetra-coordinated silicon environment evolved, and the increase in Q n (0Al) entities explained the shift in the spectrum to high fields. The content of Q 4 (1Al) in the network structure was only second to that of Q 4 (0Al) , and the sum of the two exceeded 60%. The content of Q n m(Al) (2 ≤ m ≤ 4) showed a decreasing trend as a whole with the increase in boron oxide content. The area percentage of Al V was only second to that of Al IV , relatively speaking; the number of Al IV entering the network framework decreased, which was beneficial to the formation of Q 4 (0Al) [43]. The decrease in the content of Q n m(Al) (2 ≤ m ≤ 4) meant that the connectivity between the pure silicon-oxygen network and the aluminum-oxygen network was weakened. Raman spectroscopy showed increased interconnectivity of the borate and silicate sub-networks in the glass network, which was due to the fact that silicates consist of  3 ] structure, which has a certain degree of negative impact on the integrity of the surrounding Si-O bonds. As a result, the bond fracture of the silica skeleton occurs, resulting in the destruction of the integrity of the silicon-oxygen-silicon group. From the uniformity of the structure distribution, when the layered conformation is inserted into the three-dimensional conformation and there is no obvious phase separation, the jamming phenomenon very easily occurs inside the network. This is also the reason why B 2 O 3 can be introduced into some glass systems as a flux. Combined with the fitting results of NMR spectra, it could be inferred that the vibration peaks of [SiO 4 ] with different bridging-oxygen numbers in the range of 800-1100 cm −1 were Q n , and Q 1 mainly included Q 4 (1Al) ; Q 2 mainly included Q 4 (2Al) and B IV -2Si; Q 3 mainly included Q 4 (3Al) and B IV -3Si; and Q4 mainly included Q 4 (4Al) .

Thermal Expansion Analysis
The coefficient of thermal expansion reflects the ability of the glass structure to resist deformation when heated and is one of the ways to characterize the properties of glass. The thermal expansion curve of Li 2 O-Al 2 O 3 -SiO 2 series glass with different B 2 O 3 contents is shown in Figure 5. The thermal expansion coefficients of the seven groups of samples were 6.90 × 10 −6 /K, 6.65 × 10 −6 /K, 6.50 × 10 −6 /K, 6.45 × 10 −6 /K, 6.34 × 10 −6 /K, 6.00 × 10 −6 /K and 5.93 × 10 −6 /K. B2O3 was arranged in the glass structure in the form of B III and only relied on weak van der Waals forces to connect between the layered structures; the binding force that the particle vibration needed to overcome became smaller, resulting in an increase in the thermal expansion coefficient. During the appearance of highly coordinated boron (B IV ) in glass, the bonding force of the structural connection was greatly strengthened within a certain range (mainly compared with the van der Waals force), and the layered structure of the glass was repaired and accumulated into a spatial grid, which directly led to the reduction in the thermal vibration amplitude of the particles. The smaller the distance between the particles was, the smaller the thermal expansion coefficient of the glass was; the influence of structural unit B IV on the glass network structure was similar to that of the [AlO4] tetrahedron, and the broken network could also be reconnected. Therefore, B IV had a positive effect on the thermal expansion of the glass. It can be seen from Figure 5 that after adding B2O3, the thermal expansion coefficient showed obvious regularity. The thermal expansion coefficients of glass samples 1#~7# decreased with the increase in boron oxide content, which was related to the network structure.
Combined with the GIM fitting calculation results of 27 Al NMR, it can be seen that when the contents of B2O3 (mol%) were 2.5 and 6.5, the contents of Al IV decreased when B2O3 (mol%) had values of 1.5 and 5.5, while the high-coordinated molecules entering the B 2 O 3 was arranged in the glass structure in the form of B III and only relied on weak van der Waals forces to connect between the layered structures; the binding force that the particle vibration needed to overcome became smaller, resulting in an increase in the thermal expansion coefficient. During the appearance of highly coordinated boron (B IV ) in glass, the bonding force of the structural connection was greatly strengthened within a certain range (mainly compared with the van der Waals force), and the layered structure of the glass was repaired and accumulated into a spatial grid, which directly led to the reduction in the thermal vibration amplitude of the particles. The smaller the distance between the particles was, the smaller the thermal expansion coefficient of the glass was; the influence of structural unit B IV on the glass network structure was similar to that of the [AlO 4 ] tetrahedron, and the broken network could also be reconnected. Therefore, B IV had a positive effect on the thermal expansion of the glass. It can be seen from Figure 5 that after adding B 2 O 3 , the thermal expansion coefficient showed obvious regularity. The thermal expansion coefficients of glass samples 1#~7# decreased with the increase in boron oxide content, which was related to the network structure.
Combined with the GIM fitting calculation results of 27 Al NMR, it can be seen that when the contents of B 2 O 3 (mol%) were 2.5 and 6.5, the contents of Al IV decreased when B 2 O 3 (mol%) had values of 1.5 and 5.5, while the high-coordinated molecules entering the network were reduced. The amount of aluminum was relatively high, and the density of the network structure was relatively low. Therefore, when the content of B 2 O 3 (mol%) increased from 2.5 to 3.5 and from 5.5 to 6.5, the decreasing trend of the thermal expansion coefficient was relatively gentle.

Microhardness
Microhardness is the most commonly used index to characterize the mechanical properties of materials. It is affected by the degree of aggregation of the network structure. Figure 6 shows the Vickers hardness of Li 2 O-Al 2 O 3 -SiO 2 series glass with different B 2 O 3 contents. Vickers hardness was between 605 and 720 MPa. When the content of B2O3 (mol%) increased from 0 to 5.5, boron and aluminum mainly existed in four-coordinated forms, acting as network formers. N4 and Q4 (0Al) increased with the increase in boron content, and the network structure shrank, while the density and integrity of the structure were enhanced, and the ability to resist external force damage was enhanced.
When B2O3(mol%) = 6.5, the values of N4, Q4 (0Al) and Al IV were all lower than when B2O3(mol%) = 5.5. The content of SiO2 in the glass exceeded 60 mol%, and the network modifiers containing N3, Al V and Al VI were not only connected with the silicon skeleton but also entered the glass cavity to improve the network density and the content of N 4 , while Q4 (0Al) and Al IV always dominated. Therefore, the increased content of network modifier did not change the overall change trend of Vickers hardness but only slowed down its growth trend.

Physical Properties
Density is the most basic performance parameter of an object. The density of glass is sensitive to changes in structure, and structural changes are basically reflected in changes in density. The molar volume of glass depends on the packing density of atoms in the glass network structure and the density and integrity of the glass network, which reflect the porosity of the glass structure. Figure 7 shows the experimentally obtained density and molar volume results. Vickers hardness was between 605 and 720 MPa. When the content of B 2 O 3 (mol%) increased from 0 to 5.5, boron and aluminum mainly existed in four-coordinated forms, acting as network formers. N 4 and Q 4 (0Al) increased with the increase in boron content, and the network structure shrank, while the density and integrity of the structure were enhanced, and the ability to resist external force damage was enhanced.
When B 2 O 3 (mol%) = 6.5, the values of N 4 , Q 4 (0Al) and Al IV were all lower than when B 2 O 3 (mol%) = 5.5. The content of SiO 2 in the glass exceeded 60 mol%, and the network modifiers containing N 3 , Al V and Al VI were not only connected with the silicon skeleton but also entered the glass cavity to improve the network density and the content of N 4 , while Q 4 (0Al) and Al IV always dominated. Therefore, the increased content of network modifier did not change the overall change trend of Vickers hardness but only slowed down its growth trend.

Physical Properties
Density is the most basic performance parameter of an object. The density of glass is sensitive to changes in structure, and structural changes are basically reflected in changes in density. The molar volume of glass depends on the packing density of atoms in the glass network structure and the density and integrity of the glass network, which reflect the porosity of the glass structure. Figure 7 shows the experimentally obtained density and molar volume results.

Physical Properties
Density is the most basic performance parameter of an object. The density of glass is sensitive to changes in structure, and structural changes are basically reflected in changes in density. The molar volume of glass depends on the packing density of atoms in the glass network structure and the density and integrity of the glass network, which reflect the porosity of the glass structure. Figure 7 shows the experimentally obtained density and molar volume results.  It can be seen from Figure 7 that the density of the glass increased with the increase in the boron oxide content, and the molar volume gradually decreased with the increase in the boron oxide content. When the content of B 2 O 3 (mol%) increased from 0 to 5.5, combined with the fitting calculation results of 27 Al, 11 B and 29 Si NMR spectra, it can be seen that boron and aluminum atoms mainly existed in four-coordinated forms, acting as network formers; N 4 and Q 4 (0Al) increased, and the network structure shrank, while the structure density and integrity were strengthened.
When B 2 O 3 (mol%) = 6.5, compared with B 2 O 3 (mol%) = 6.5, the values of N 4 , Q 4 (0Al) and Al IV in the network structure decreased. However, the density change trend did not change, which was related to the way in which [BO 3 ] was connected with the silica skeleton. The SiO 2 content of the entire glass system exceeded 60 mol%. [BO 3 ] and [BO 4 ] in the high silicon glass system were mainly connected with the silica skeleton, and some [BO 3 ] may have entered the glass cavity, such as the Al V or Al VI of the network modifier, causing the network structure to shrink. Therefore, the increasing trend of density became flat.

Chemical Stability
The chemical stability of glass is affected by the composition, structure and other factors of glass. The erosion effect of acid on glass is to change, destroy or leach the network modifiers, such as alkaline earth metals or alkali metal oxides, in the glass structure; alkalis not only act on the network modifiers but also dissolve the silica skeleton of glass. Figure 8 shows the weight loss ratio of lithium aluminum silicate glass with different boron oxide content in strong acid and strong alkali solutions. It can be seen from Figure 7 that the density of the glass increased with the increase in the boron oxide content, and the molar volume gradually decreased with the increase in the boron oxide content. When the content of B2O3 (mol%) increased from 0 to 5.5, combined with the fitting calculation results of 27 Al, 11 B and 29 Si NMR spectra, it can be seen that boron and aluminum atoms mainly existed in four-coordinated forms, acting as network formers; N4 and Q4 (0Al) increased, and the network structure shrank, while the structure density and integrity were strengthened.
When B2O3 (mol%)=6.5, compared with B2O3 (mol%)=6.5, the values of N4, Q4 (0Al) and Al IV in the network structure decreased. However, the density change trend did not change, which was related to the way in which [BO3] was connected with the silica skeleton. The SiO2 content of the entire glass system exceeded 60 mol%.
[BO3] and [BO4] in the high silicon glass system were mainly connected with the silica skeleton, and some [BO3] may have entered the glass cavity, such as the Al V or Al VI of the network modifier, causing the network structure to shrink. Therefore, the increasing trend of density became flat.

Chemical Stability
The chemical stability of glass is affected by the composition, structure and other factors of glass. The erosion effect of acid on glass is to change, destroy or leach the network modifiers, such as alkaline earth metals or alkali metal oxides, in the glass structure; alkalis not only act on the network modifiers but also dissolve the silica skeleton of glass. Figure 8 shows the weight loss ratio of lithium aluminum silicate glass with different boron oxide content in strong acid and strong alkali solutions. With the increase in boron oxide content, the weight loss ratio of glass in hydrofluoric acid and alkali solutions showed a trend that first decreased and then increased. The alkaline earth metal ions in the glass formed low-solubility silicates with the silicate ions With the increase in boron oxide content, the weight loss ratio of glass in hydrofluoric acid and alkali solutions showed a trend that first decreased and then increased. The alkaline earth metal ions in the glass formed low-solubility silicates with the silicate ions generated after erosion and adhered to the glass surface to form a protective film, which can slow down further erosion, so the overall quality of glass lost in the alkaline solution was maintained at lower levels. The degree of damage of H + and OHto the damaged network structure was related to the number of non-bridging oxygens in the network structure. According to the NMR spectra and the Raman test results, with the increase in boron content, the aluminum ion was dominated by tetra-coordination, while the values of N 4 and Q 4 (0Al) increased, and the number of bridging oxygens increased. The ternary network skeleton structures of [SiO 4 ], [BO 4 ] and [AlO 4 ] tetrahedra were more compact, and the resistance of glass in hydrofluoric acid and alkali solutions increased. The reaction difficulty was increased, and the chemical stability was improved.
The acid resistance is not only affected by the structural integrity of a glass network but is also related to the cation species. Compared with Al 3+ , B 3+ can enhance the acid resistance of glass within a certain range [22]. The corrosion of glass by HF mainly relies on the electrophilic corrosion of non-bridging oxygen by high concentrations of H + , which undergoes a substitution reaction with R 2+ [52]. As more B 3+ was introduced into bonding sites in the form of -Si-O-B IV , it was difficult for the electrophilic corrosion mechanism of HF in the form of ion exchange to occur in the glass, so the acid resistance of glass became better. Since the glass itself reacted with HF (Al 2 O 3 , SiO 2 , etc.), acid weight loss was generally at a high level.
When B 2 O 3 (mol%) = 6.5, the weight loss rate of glass in hydrofluoric acid solution and alkali solution increased, and when B 2 O 3 (mol%) = 5.5, the chemical stability of glass appeared to have an extreme value. When B 2 O 3 (mol%) = 6.5, the contents of [AlO 6 ] and [AlO 5 ] in the glass were the highest. [AlO 6 ] and [AlO 5 ] generally cause the prominent Si-O bond to break around it. The unstable polar covalent bond requires more R 2+ for charge neutralization and increases the sensitivity to acid or alkaline olutions due to its relatively relaxed structure that promotes the acceleration of ion transport.

Conclusions
In this paper, the effects of different contents of B 2 O 3 (mol%) on the structure, thermal properties, chemical stability and mechanical properties of Li 2 O−Al 2 O 3 −SiO 2 series glass were studied.
According to the fitting calculation results of the NMR spectra of 27 Al, 11 B and 29 Si, it could be seen that the aluminum atoms in the network structure were dominated by four-coordinated Al IV and that the content of Al IV did not change with B 2 O 3 (mol%); boron was mainly tetra-coordinated [BO 4 ], and the N 4 value increased with the increase in boron oxide content. The content of Q 4 (0Al) in the silicon spectrum dominated and increased with the increase in B 2 O 3 (mol%); at B 2 O 3 (mol%) = 6.5, Q 4 (0Al) decreased. The overall density of the glass network structure constructed by the borate network, the silicon-oxygen network and the aluminum-oxygen network increased with the increase in B 2 O 3 (mol%). Combined with the fitting results of the nuclear magnetic resonance spectra, it could be inferred that the types of [SiO 4 ] vibration peaks in the range of 800-1100 cm −1 in the Raman spectra of different bridging-oxygen numbers were Q n , where Q 1 mainly included Q 4 (1Al) ; Q 2 mainly included Q 4 (2Al) and B IV -2Si; Q 3 mainly included Q 4 (3Al) and B IV -3Si; and Q4 mainly included Q 4 (4Al) . The incorporation of boron oxide improved the degree of polymerization and connectivity of the glass network structure as a whole, and its corresponding mechanical properties and physical properties were also optimized. When B 2 O 3 (mol%) = 5.5, the thermal expansion coefficient of glass was 6.00 × 10 −6 /K; the flexural strength and hardness were 180.665 Mpa and 715 Mpa, respectively; the density was 2.41g/cm 3 ; and the acid resistance was improved by about 43%. On the whole, when B 2 O 3 (mol%) = 5.5, the performance of glass was the best.