Raman Spectroscopy and Electrical Transport in 30Li₂O• (67−x) B₂O₃•(x) SiO₂•3Al₂O₃ Glasses

Abstract

We have investigated the influence of the relative proportions of glass formers in a series of lithium alumino-borosilicate glasses with respect to electrical conductivity (σ) and glass transition temperature (T_g) as functions of glass structure, as determined using Raman spectroscopy. The ternary lithium alumino-borate glass exhibits the highest σ and lowest T_g among all the compositions of the glass series, 30Li₂O•3Al₂O₃• (67−x) B₂O₃•xSiO₂. However, as B₂O₃ is replaced by SiO₂, a shallow minimum in σ, as well as a shallow maximum in T_g, are observed near x = 27, where the Raman spectra indicate that isolated diborate/tetraborate/orthoborate groups are being progressively replaced by danburite/reedmergnerite-like borosilicate network units. Overall, as the glasses become silica-rich, σ is minimized, while T_g is maximized. In general, these findings show correlations among T_g (sensitive to network polymerization), σ (proportional to ionic mobility), and the different borate and silicate glass structural units as determined using Raman spectroscopy.

1. Introduction

Recent advancements in glass science have prompted extensive research efforts aimed at comprehending both the fundamental and applied aspects of non-crystalline electrolytes in all-solid-state batteries (ASSBs) [1,2]. Ionic glasses containing sulfide and oxide anions, with alkaline cations, are investigated to be used in ASSBs. The optimization of electrical conductivity in these ionic glasses has been attempted through the incorporation of halides or other salts into the glass structure [3]. However, this strategy often comes at the cost of decreased chemical durability and thermal stability [4].

To address these challenges, researchers have explored [5,6,7,8,9,10,11] the incorporation of multiple glass formers, such as P₂O₅, B₂O_3, and/or SiO₂, into binary glass systems. This has proven to be an effective approach to optimize electrical conductivity while mitigating the adverse effects of added halide salts. Recent studies have been directed toward glasses containing at least two network formers [5]. A phenomenon known as the mixed glass former effect (MGFE) in binary glasses can vary electrical conductivity in non-linear ways by holding alkali content constant while changing the proportions of different glass formers [6].

The lithium boro-phosphate (LBP) system [7] exemplifies the MGFE, where the conductivity of these ternary glasses can be two orders of magnitude greater than that for lithium borate or lithium phosphate binary glasses. Comparable changes in electrical conductivity between ternary and binary glasses demonstrating the MGFE can also be found for other systems, including silver boro-phosphate, lithium boro-tellurite [8], and lithium seleno-borate glasses [9]. This intriguing behavior extends to glasses in the Li₂O•B₂O₃•SiO₂ system, known for their ability to incorporate significant amounts of Li₂O while maintaining stability [10]. Maia et al. [11] found no signs of MGFE in a series of lithium boro-silicate glasses with 40 mol.% of Li₂O. However, Otto [12] reported MGFE in the lithium boro-silicate glasses, with 20 to 30 mol.% Li₂O. These compositional effects in glasses arise from structural modifications, resulting in the formation of different glass former species [13]. To better comprehend these effects, it is crucial to characterize the various network structural units in these ternary glasses. Raman spectroscopy has proven to be an effective technique to detect structural units [14] present in the glasses. This information aids in explaining the compositional dependence of activation energies that impede ion transport in glasses.

The ionic transport in glasses is governed by the Arrhenius equation [15] for ionic conductivity, as expressed in Equation (1).

σ = σ₀ exp (−E_a/kT)

(1)

where:

σ₀ = N₀ α_(Ze)²λ² v/T

(2)

and N₀ represents the total number of mobile ions, T is the absolute temperature, α is the degrees of freedom, Z_e is electric charge of ions, λ is the distance covered by the ion in a single jump, v is the frequency of jump attempts the ion makes, E_a is the activation energy—the energy barrier an ion carrier has to overcome to move from one site to another to contribute to conduction—and k is the Boltzmann constant [16]. According to the Anderson–Stuart model [17], the activation energy for ionic conduction in glass is given by the following:

E_a = E_b + E_s

(3)

where E_b is the electrostatic bond energy and E_s is the network strain energy.

This study centers on a series of lithium borosilicate glasses with 30 mol% lithium oxide, denoted as 30Li₂O•(67−x)B₂O₃•(x)SiO₂•3Al₂O₃ (LABS glass), where x varies from 0 to 67. Aluminum oxide is incorporated to enhance the chemical stability of the glass. By varying the B₂O₃ and SiO₂ ratios, we measure how electrical conductivity (σ), glass transition temperature (T_g), and Raman spectral features change across the glass series. One of the objectives was to investigate if the incorporation of a second network former could lower the net bond energy and/or the strain energy, possibly increasing alkali ion mobility.

Another aspect of this research was to explore the possibility of inducing phase separation in the glass. It is commonly observed that alkali ions tend to segregate in one of the two phases of a glass containing two glass-forming oxides (as in Vycor glass). If such a glass composition decomposes spinodally, then the high concentration of one of two phases may lead to high alkali ion mobility, which is desirable in the solid electrolyte of a battery.

For each glass, van der Pauw’s four-probe method [18] was implemented to measure σ over a temperature range of 50 to 170 °C, which was then used to calculate the glass activation energy. The T_g value was determined for each glass based on differential thermal analysis (DTA). T_g denotes a temperature range during which a super-cooled liquid transforms into a glassy state. Analyzing T_g provides valuable insights into both the structure and electrical conductivity of the glass. Polarized Raman spectra were gathered using a single grating spectrograph-edge filter system, where various spectral features were assigned to pertinent vibrational modes for borate, borosilicate, and silicate glasses and some crystalline phases presented in the literature.

2. Literature Review

2.1. Electrical Conductivity Literature

Research on lithium-conducting glasses has been ongoing for many decades, focusing on both fundamental understanding and practical applications as solid electrolytes in Li-ion batteries. To make such applications work, it is important to know how well these glasses conduct electricity. In earlier studies [19,20,21], electrical conductivity has been measured using two methods: direct current (DC) and alternating current (AC), often with two or four probes. In the DC method, we simply pass current through the sample and measure the voltage and current. For AC methods, like impedance spectroscopy, we use various alternating current frequencies and observe how the sample responds. The electrical resistance of a sample is determined based on impedance spectroscopy [22]. The electrical conductivity is calculated by using the relationship σ = t/RA, where R is the electrical resistance, A is the cross-sectional area, and t is the thickness of the sample [22].

The electrical conductivity of a glass depends upon the total concentration of various charge carriers, which is affected by the composition. Bude et al. [19] showed that σ increases from 10–11 to 10⁻⁴ S/cm in a series of yLi₂O•(1−y) B₂O₃ glasses, and from 10⁻¹¹ S/cm to 10⁻⁵ S/cm in a series of yLi₂O•(1−y) SiO₂ glasses, where y = 0.1 to 0.6. Montouillout et al. [20] investigated the xLi₂O•(100−x)B₂O₃ glass system over a wide range of compositions, where x = 0 to 50 mol.%, by characterizing local structure and electrical conductivity using solid-state nuclear magnetic resonance (NMR) and impedance spectroscopy, respectively. They reported that σ increases linearly as x increases up to 32 mol.%, where in the glass structure, NMR indicates that BO₄ populations increase at the expense of BO₃. Kluvanek et al. [21] reported the absence of MGFE while studying (Li₂O)0.4•(B₂O₃) (0.6x)•(Si₂O₄)0.6(1−x) glasses with x = 0 to 0.8 and found an increase in DC conductivity from silicate-rich to borate-rich glasses. Neyret et al. [23] studied the role of alkali in the borosilicate glass structure and showed that ionic conductivity decreases as silica polymerization increases.

2.2. Glass Transition Temperature Literature

Techniques such as differential thermal analysis (DTA) or differential scanning calorimetry (DSC) are commonly used to measure T_g [13]. In general, borate glasses tend to have lower T_g values, in contrast to higher-T_g silicate glasses [24]. The introduction of alkali oxide, such as Li₂O, into the glass composition can induce variations in T_g for borate, borosilicate, and silicate glasses. For instance, Kodama et al. [25] conducted a study on the xLi₂O• (1−x) B₂O₃ glass series and measured changes in T_g from 245 to 490 °C while varying x from 0 to 0.28. Avramov et al. [26] investigated binary xLi₂O•(1−x)B₂O₃ and xLi₂O•(1−x)SiO₂ glasses, where x was varied from 0.01 to 0.6, and reported that T_g increases for borate compositions, reaching a maximum at x = 0.3, while for silicate compositions, T_g decreases linearly with increasing Li contents.

Boekenhauer et al. [27] explored the relationship between glass transition temperature and structure in a series of lithium borosilicate glasses, taking into account the composition-related structural parameters R and K, where R and K as proposed by Yun and Bray [28], Yun et al. [29], and Dell et al. [30] in Na-borosilicate glass, given as follows:

K = [SiO₂] (mol.%)/[B₂O₃](mol.%)

(4)

and R = [M₂O] (mol.%)/[B₂O₃](mol.%)

(5)

where M = alkali atom.

Boekenhauer et al. considered a wide range of Li₂O-to-B₂O₃ ratios, or R, where 2 ≤ R ≤ 10 for several fixed SiO₂-to-B₂O₃ ratios or K, where K = 0.5, 1, 2, or 3. The study discovered two maxima in T_g for each SiO₂-to-B₂O₃ ratio in the borosilicate glass series. The first maximum is associated with the largest tetrahedral BO₄ population, while the second maximum is linked to the glass, separating into lithium borate and lithium silicate domains. Neyret et al. [23] reported that T_g increases with increases in silicate polymerization in lithium, sodium, and potassium borosilicate glasses.

2.3. Raman Spectroscopy Literature

2.3.1. Borate Glasses

The Raman spectra of borate glasses doped with alkali ions have been studied extensively, where various borate structural units have been assigned to specific spectral features [31,32,33,34,35,36]. Krogh-Moe’s work [37] indicates that vitreous B₂O₃ consists of a random network of linked boroxol ([B₃O₆]³⁻) rings and isolated BO₃ triangles. In another study [33], a strongly polarized band near 806 cm⁻¹ is attributed to the symmetric breathing in boroxol rings. In Kamitsos et al. [34], a weak, broad band near 1260 cm⁻¹ is assigned to the B-O stretch in BO₃ triangles within both isolated rings and more polymerized borate networks. However, the introduction of alkali oxide, M₂O (M = Li, Na, K, Rb, Cs), into the borate network generates diverse isolated poly-borate species. This structural transformation has been confirmed in multiple studies [31,32,33,34,35] investigating alkali borate glasses spanning a range of M₂O concentrations. Mozzi and Warren’s findings [38] establish that the addition of M₂O disrupts boroxol rings, altering the coordination of some boron atoms from 3 to 4, which leads to the formation of planar rings containing linked BO₃ and BO₄ configurations (i.e., diborate, triborate, tetraborate, pentaborate, etc.) [32]. This introduction of M₂O gradually decreases the intensity of the 806 cm⁻¹ band, and a new band emerges at 770 cm⁻¹. For M₂O concentrations exceeding 20 mol.%, the 806 cm⁻¹ band disappears, while the 770 cm⁻¹ band shifts to lower frequencies as the M₂O content surpasses 30 mol.% [34]. According to Bril [39], the 770 cm⁻¹ band is assigned to symmetric breathing vibrations of six-membered rings (in an alternating arrangement of three boron atoms and three bridging oxygens (BOs)), each with one BO₄ tetrahedron and two BO₃ triangles (i.e., triborate, tetraborate, pentaborate). At higher alkali contents, the 770 cm⁻¹ band shifts to lower frequencies and is attributed to six-membered rings with two BO₄ tetrahedra and one BO₃ triangle (i.e., diborate, ditriborate, or dipentaborate).

An extensive Raman study of binary borate glasses across a wider range of alkali, M, concentrations [34] yielded comparable assignments for structural groups within the borate networks. For glasses with M₂O compositions below 35 mol.% [40], the 770 cm⁻¹ band corresponds to tetraborate, pentaborate, or triborate groups. In glasses with M₂O concentrations ranging from 15% to 45 mol.%, the 1100 cm⁻¹ band is attributed to diborate units [41,42]. For borate glasses with alkali contents exceeding 40 mol.%, isolated diborate groups are assigned to the 500 cm⁻¹ band [42]. Similarly, in Kamitsos’ study [40] of magnesium sodium borate glasses, the presence of pyroborate (850 cm⁻¹) and orthoborate (945 cm⁻¹) are reported in accordance with the spectra of crystalline pyroborate and orthoborate compounds [31,43].

Dwivedi et al. [44] studied xLi₂O•(1−x) B₂O₃ glasses at 0.1 < x < 0.5, where 760–780 cm⁻¹ features are assigned to breathing vibrations of six-membered rings containing both BO₃ triangles and BO₄ tetrahedra (e.g., ditriborate and dipentaborate units). A band near 855 cm⁻¹ is assigned to pyroborate units, where x ≥ 0.25. A broad band near 500 cm⁻¹ is also assigned to pentaborate, tetraborate, and diborate units, which shifts to near 550 cm⁻¹ as the alkali content increases. In another study of lithium borate glasses [32], the 550 cm⁻¹ band is also assigned to diborate units at 50 mol.% Li₂O.

Konijnendijk and Stevels [32] studied the Raman spectra of xM₂O• (1−x) B₂O₃ glasses, where M = Na and K, and 0.05 ≤ x ≤ 0.35. They observed that the introduction of M₂O up to 20 mol.% disrupts boroxol ring formation, resulting in the creation of tetraborate groups. As the M₂O content increases, tetraborate groups (assigned to the 770 cm⁻¹ band) transform into diborate groups, corresponding to the 755 cm⁻¹ band. However, recent research on lithium diborate (Li₂O•2B₂O₃) glasses [34] has questioned the 755 cm⁻¹ diborate assignment and has instead assigned internal diborate displacements to a band near 1100 cm⁻¹. Additionally, various other structural units with non-bridging oxygens (NBOs), such as pyroborate (840 cm⁻¹), orthoborate (940 cm⁻¹), ring-type metaborate (630 cm⁻¹), and chain-type metaborate (730 cm⁻¹), are created, where the alkali content is greater than 30 mol.% [32,33,34]. In a study of Cs₂O•B₂O₃ glasses, Kamitost et al. [35] similarly assign 725, 675, and 625 cm⁻¹ bands to chain-type metaborate groups based on comparisons with the spectra of crystalline Li₂O•B₂O₃, which contains these structural units.

The Raman scattering observed in the high-frequency region (1300–1500 cm⁻¹) corresponds to the B-O stretching vibrations. These vibrations specifically involve NBOs and are integral components of a connected borate network. The band centered around 1400 cm⁻¹ is assigned to BO₃ units bonding to BO₄ units, while the band at around 1480 cm⁻¹ is assigned to BO₃ units bonding to other BO₃ triangles [30,31,32,33,45], which has also been used for borosilicate glasses [45,46].

2.3.2. Silicate Glasses

Silica glasses consist of a fully polymerized three-dimensional framework of SiO₄ tetrahedra, and the addition of alkali oxide to the glass results in the depolymerization of the network and the creation of NBOs [13]. Larger NBO populations are linked with increasing alkali contents. Alkali silicate glass consists of various species of silicate tetrahedra with different numbers of BOs and NBOs. The Raman spectra of alkali silicate glasses can be divided into two frequency ranges depending on the type of vibrational modes associated with the spectral features [47,48,49,50,51]. The highly polarized broad envelope below 650 cm⁻¹ is due to longer-range displacements within the silicate tetrahedral network, such as Si-O-Si symmetric bend modes [47,48]. In contrast, the envelope between 850 and 1140 cm⁻¹ is comprised of four to five component bands assigned to localized Q-species Si-O stretch modes within SiO₄ tetrahedra linked to zero to four neighboring silicate tetrahedra (Q⁰ to Q⁴ units), respectively [51]. Components between 1040 and 1140 cm⁻¹ have been assigned differently. McKeown et al. [50] reported that a component at 1100 cm⁻¹ is due Q³ species. On the other hand, when studying xLi₂O•yNa₂O• (100−x−y) SiO₂ glasses where (x = 23, 12; y = 11, 22), Seuthe et al. [52] suggest that Q³ modes are assigned to two bands near 1040 cm⁻¹, and Q⁴ units are assigned to a component near 1140 cm⁻¹. By fitting Gaussian components to this Q-species envelope, Zotov and Keppler [53,54] determined that the component peaks near 950, 1020, 1080, and 1140 cm⁻¹ are due to the Q², Q^3′, Q^3″, and Q⁴ Si-O stretch modes, respectively. A Q^3′ tetrahedron is connected to several combinations of Q³ and Q⁴ tetrahedra, where at least one of the three nearby tetrahedra is Q^3′. On the other hand, Q^3″ units are connected to three Q⁴ tetrahedra [51].

2.3.3. Borosilicate Glasses

Regarding borosilicate glasses, the Raman spectra have vibrational components originating exclusively from the borate or silicate structural units, as well as from structural units containing linked borate and silicate polyhedra. Raman features within the 450–800 cm⁻¹ frequency range stem from longer-range bond-bending vibrations occurring in various BO environments, including Si-O-Si [47], B-O-B [55], and B-O-Si [47,55]. The band near 550 cm⁻¹ is attributed to the Si-O-Si bending vibrations [47,48]. Manara et al. [45] observed a prominent band near 586 cm⁻¹ in the spectra of Na-borosilicate glasses and assigned this feature to a breathing mode within reedmergnerite rings comprised of three SiO₄ tetrahedra and one BO₄ tetrahedron [55,56]. The mineral, reedmergnerite, has its most intense Raman peak at 586 cm⁻¹ [57]. Similarly, another borosilicate structure to consider is the danburite ring, which has two SiO₄ tetrahedra and two BO₄ tetrahedra [56]. The mineral, danburite, has its most intense Raman peak at 614 cm⁻¹ [58,59]. As a result, Manara et al. [45] report a band near 630 cm⁻¹ for their Na-borosilicate glasses that may correspond to the breathing mode of danburite rings.

At higher frequencies, two broad Raman envelopes are observed. In the 850–1250 cm⁻¹ frequency range of the silicate Q-species, it is likely that some contributions come from Si-O and B-O stretching within mixed Q-species modes, including SiO₄ and BO₄ tetrahedra. In the 1300–1500 cm⁻¹ range, the band centered around 1380 cm⁻¹ is assigned to BO₃ units bonded to BO₄ units, while the band at around 1475 cm⁻¹ is assigned to BO₃ units linked to BO₃ units, as seen in alkali borate systems. Indeed, in the alkali borosilicate glass system, the glass network is comprised of silicate or borosilicate Q-species units, possibly linked to BO₃ triangles in varying proportions determined by silica, borate, and alkali contents [45,55,60].

2.4. Spinodal Decomposition in Glasses

Alkali silicate and borosilicate glasses have historically been known to exhibit phase separation [61]. During the cooling process, liquid silicate mixtures undergo phase separation into two distinct phases. This separation occurs as a kinetic process like nucleation or spinodal decomposition. Nucleation predominates when the volume fraction of particles is small, whereas spinodal decomposition occurs when the volume fraction of the separating phases is nearly identical, resulting in an interconnected structure [13].

The occurrence of phase separation is guided by the Gibbs free energy of the potential phases. In regions of the phase diagram where the second derivative of the Gibbs free energy with respect to composition is negative, there is no barrier to phase growth [62]. According to Cahn [62], this transformation proceeds through a continuous alteration in the composition of the growing phases while keeping their extent unchanged. The composition shift occurs within a regular three-dimensional array and continues until the compositions of the two phases reach equilibrium values, forming an interconnected structure. The properties of glasses change the phase separation. Notably, the electrical conductivity of a phase-separated glass increases, mainly if the phase with higher conductivity forms a continuous network.

3. Experimental

3.1. Glass Synthesis

A glass series with the nominal composition 30Li₂O•(67−x) •B₂O₃•(x)SiO₂•3Al₂O₃, with x = 0, 7, 17, 27, 37, 47, 57, 67, was prepared using conventional melting and quenching methods. Reagent-grade chemicals including Li₂CO₃ (Alfa Aesar, Thermo Fisher Scientific, Ward Hill, MA, USA), SiO₂, B₂O₃ (Sigma Aldrich, St. Louis, MO, USA), and Al₂O₃ (Alfa Aesar) were mixed in various stoichiometric ratios. Each mixture was melted in an alumina crucible placed in a Deltech Inc., Denver, CO, USA, (DT29-BL56-E2404) furnace. The glass was formed by quenching the corresponding viscous melt between two copper plates. The series end-member glasses are named 67B and 67Si to signify lithium alumino-borate and lithium alumino-silicate glass, respectively. The other glasses are labeled as 67−xBxSi, with the x-values outlined above (Figure 1).

X-ray diffraction (XRD), X-ray fluorescence spectroscopy (XRF), and differential thermal analysis (DTA) were all performed using glass powder produced after crushing and grinding a few pieces of each quenched glass. For Raman and electrical conductivity measurements, we used 1 mm uniform thickness 2000-grit polished glass fragments. XRD measurements from 10° to 80° 2θ were performed using a Rigaku Americas, The Woodlands, TX, USA, SmartLab-SE θ-θ diffractometer that verified the glass samples were amorphous. XRF analyses were run on each glass to determine the chemical composition using a PANalytical Wavelength dispersive Axios max advanced system with SuperQ4 analysis software. Lithium and boron characteristic XRF lines are too low in energy to be routinely measured by the system. As a result, Li₂O and B₂O₃ target concentrations from each glass recipe are listed in

Table 1. LABS glass series chemical compositions according to XRF. XRF values have a ±10% uncertainty, with the associated glass former ratio, K; oxide modifier-to-boron oxide ratio, R; critical composition, R* [30]; and T_g.

Glass ID	Li2O (Mole %) (Nominal)	Al2O3(Mole %)		B2O3 (Mole %) (Nominal)	SiO2(Mole %)		B2O3 (%Mole)	SiO2 (%Mole)	$\mathbf{K}=\frac{\left[{\mathit{S}}_{\mathit{i}}{\mathit{O}}_2\right]}{\left[{\mathit{B}}_2{\mathit{O}}_3\right]}$	$\mathbf{R}=\frac{\left[\mathit{L}{\mathit{i}}_2\mathit{O}\right]}{\left[{\mathit{B}}_2{\mathit{O}}_3\right]}$	R* = 0.5 + 0.0625 K	Tg (°C)
		(Nominal)	(XRF)		(Nominal)	(XRF)
67B	30	3	2.60	67	0	0.00	100	0	0	0.40	0.5	442
60B7Si	30	3	3.37	60	7	6.15	89.5	10.5	0.12	0.50	0.51	446
50B17Si	30	3	3.60	50	17	15.79	74.6	25.4	0.34	0.60	0.52	450
40B27Si	30	3	4.40	40	27	24.43	59.7	40.3	0.68	0.75	0.54	461
30B37Si	30	3	3.52	30	37	36.00	44.8	55.2	1.23	1.00	0.58	458
20B47Si	30	3	3.51	20	47	45.21	29.9	70.1	2.35	1.50	0.65	456
10B57Si	30	3	3.60	10	57	58.04	14.9	85.1	5.70	3.00	0.86	463
67Si	30	3	3.65	0	67	66.40	0	100	---	---	---	475

.

DTA curves were obtained using a Perlin Elmer DAT7 system at a heating rate of 10 °C/min from 200 °C to 1000 °C (Figure 2). DTA system calibration was performed using quartz, alumina, and gold standards. The glass transition temperature was determined from the onset of the increase in the DTA signal (mW) of the DTA scan (Table 1).

3.2. Electrical Conductivity Measurements

DC resistivity (ρ) was measured for each sample under ohmic conditions using an MMR Technologies H50 van der Pauw four-probe apparatus [63] that was scanned from 50 to 170 °C to a standard uncertainty of 0.5 °C. Silver electrodes were attached to the top edges of each sample using silver paste. Each sample was then mounted on the top of a Watlow ceramic heater with thermal grease. The current was varied from 10⁻⁶ to 10⁻¹¹ A, and the apparatus temperature was scanned from 170 to 50 °C. All electrical measurements were performed in a vacuum below 8 to 10 millitorr. The reciprocal of the resistivity obtained from the measurement provides the glass conductivity values. The relative uncertainty is 5.5% of the measured conductivity value (Figure 3).

The activation energy for the electrical conduction is obtained from the slope of the Arrhenius plots (Figure 4) [17] to a standard uncertainty of 0.02 eV (Figure 5).

3.3. Raman Spectroscopy

Raman spectra were gathered on polished glass pieces using a WITec alpha-300 RA+ micro-Raman system with a 532 nm solid-state DPSS laser. An 1800 gr/mm grating was used to disperse the Raman signal onto a 1024 × 128 element Peltier-cooled CCD camera (Andor Technology, Model DV401A-BVF-352, Belfast, UK). A 50× Zeiss objective was used for data collection, producing a ~1 µm diameter laser spot on the sample. A laser power of 36 mW was measured at the sample position. An analyzer polarizer was inserted into the Raman-scattered light path, while the incident laser light polarization could be rotated to be parallel or perpendicular with respect to the scattered light polarization direction to collect parallel and cross-polarized spectra, respectively. The spectra were the frequency calibrated to the Si 520 cm⁻¹ mode. The parallel polarized Raman spectrum for each glass was reduced and rescaled [64] for the purposes of comparison and analysis (Figure 6).

The reduced parallel polarized spectra were then fitted with Gaussian peak functions using the program IGOR [65]. To interpret these spectra, we employed a fitting procedure involving 13 Gaussian components. Fitting of the 10 highest-frequency Gaussian components to the experimental spectra allowed us to provide a vibrational assignment to each component, as outlined in the literature [31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55]. We ignore spectral trends below 400 cm⁻¹ due to the dominance of Rayleigh-scattered intensity that contain little, if any, structural information. Initially, we focused on the two end members of the glass series. By determining the Gaussian peak parameters to the spectral features for these two glasses, we determined trends with respect to the composition for the whole glass series. Subsequently, Gaussian component vibrational assignments were checked to ensure consistency with those from the literature for borate [31,32,33,34,35,36,37,38,39,40,41,42,43,44], silicate [47,48,49,50,51,52], and borosilicate [45,55,66,67] glasses (Figure 7a–d).

4. Results

4.1. Glass Transition Temperature (T_g)

T_g can be regarded as being proportional to bond strength and polymerization within a glass structure [13]. In general, for the glass series studied here, T_g increases with silica content, indicating an overall increase in bond strength (Table 1, and Figure 5), which would be anticipated as more weakly bonded and isolated borate units are replaced by more polymerized silicate tetrahedra. However, T_g has a localized maximum at 40B27Si glass and then rises significantly to the silicate glass end member. This non-linear behavior indicates that the structure is undergoing a secondary transition near this composition, which may indicate more subtle structural changes compared those occurring for the whole glass series.

4.2. Electrical Conductivity (σ)

Electrical conductivity σ relationships with temperature can be shown in the log σ versus 1000/T plots (Figure 4), which are based on the Arrhenius equation. The plotted data confirm that temperature plays an important role in facilitating the mobility of Li⁺ charge carriers, such that increasing temperatures boost electrical conductivity in these glasses.

We generally observe a decrease in σ from boron-rich (67B) to silica-rich (67Si) glass (Figure 3). 67B glass has the highest σ (1.41 × 10⁻⁷ S/cm at 50 °C) with the lowest E_a (0.50 ± 0.02 eV), while 67Si exhibits the lowest σ (4.27 × 10⁻⁹ S/cm at 50 °C) with the highest E_a (0.68 ± 0.02 eV) in the 50 to 170 °C temperature range. We also observe a shallow minimum in σ for 40B27Si glass throughout this temperature range, indicating that some secondary structural adjustments are taking place compared with the overall structural changes across the glass series, where B is replaced by Si. This minimum generally becomes more pronounced at lower temperatures. As expected, there is a consistent increase in activation energy (calculated from σ) from 67B to 67Si glass, with a shallow maximum for 40B27Si glass (Figure 5).

4.3. Raman Spectroscopy

Reduced parallel- and cross-polarized Raman spectra for the glass series provide a basis for our glass structure determinations and corresponding structural changes with respect to glass chemistry. The parallel polarized spectra have more information and show continuous changes (Figure 6) that have been quantified by Gaussian fitting, where the Gaussian peak position and area results are presented in

Table 2. Borate-rich glasses fitted Gaussian component values from the program IGOR [65] with corresponding vibrational assignments from the literature. Areas values are in arbitrary units.

Vibrational Assignment	67B			60B7Si			50B17Si
	Position (cm−1)	Peak Width	Area	Position (cm−1)	Peak Width	Area	Position (cm−1)	Peak Width	Area
Diborate (506 cm−1) [29,39]	527	133	575	527	135	746	543	154	994
Ring-type metaborate (600–650 cm−1) [29,30,31,32]	612	84	148	607	83	142	599	90	0
Chain-type metaborate (700–735 cm−1) [29,30,31,32]	703	113	646	701	126	890	690	123	790
Tetraborate (740–775 cm−1) [29,38]	772	59	649	772	63	612	769	72	561
Pyroborate (820 cm−1) [29,30,31,32,40]	857	89	347	854	91	364	861	105	373
Orthoborate (875–1000 cm−1) [29,39,62]	971	127	745	973	133	1009	981	134	1127
Diborate (1000–1110 cm−1) [34,52,57]	1114	106	254	1109	110	326	1098	104	357
BO3 symmetric stretch (1200 cm−1) [31,35,43]	1236	127	168	1233	125	195	1233	125	179
BO3-BO4 (1300–1450 cm−1) [28,29,30,31]	1381	122	479	1384	125	517	1393	125	358
BO3-BO3 (1450–1600 cm−1) [28,29,30,31]	1483	117	733	1480	112	783	1475	109	597

,

Table 3. Borosilicate glasses fitted Gaussian component values with corresponding vibrational assignments from the literature. Conventions from Table 2 are followed.

Vibrational Assignment	40B27Si			30B37Si
	Position (cm−1)	Peak Width	Area	Position (cm−1)	Peak Width	Area
Vibration of bridge bonds B-O-B, B-O-Si, Si-O-Si (500–600 cm−1) [44,45,52]	543	155	1072	539	144	980
Danburite and reedmergnerite rings [42,52,53]	681 *	118 *	742 *	664	121	809
Tetraborate (740–775 cm−1) [29,38]	765	79	505	764	102	505
Orthosilicate–pyroborate (850 cm−1) [40,67]	858	120	380	888	87	241
Q2 (950 cm−1) [46,47,48]	952	93	548	942	59	174
Q3′ (1020 cm−1) [46,47,48]	1014	80	546	1020	121	1369
Q3′‘ (1080 cm−1) [46,47,48,51]	1079	90	448	1085	54	138
Q4 (1140 cm−1) [46,47,48]	1148	88	117	1138	68	148
Symmetric stretching of BO3 units (1200 cm−1) [31,35,43]	1234	121	143	1227	92	50
BO3-BO4 (1300–1450 cm−1) [28,29,30,31]	1391	117	273	1415	77	172
BO3-BO3 (1450–1600 cm−1) [28,29,30,31]	1473	107	529	1478	78	255

* The Gaussian component at 681 cm⁻¹ for 40B27Si glass can be assigned to the chain-type metaborate units.

and

Table 4. Silica-rich glasses’ fitted Gaussian component values with corresponding vibrational assignments from the literature. Conventions from Table 2 are followed.

Vibrational Assignment	20B47Si			10B57Si			67Si
	Position (cm−1)	Peak Width	Area	Position (cm−1)	Peak Width	Area	Position (cm−1)	Peak Width	Area
Breathing vibration in four-membered rings (485 cm−1) [45,50]	531	138	1119	509	127	855	490	98	781
Breathing vibration in three-membered rings (600 cm−1) [52,63,65]	641 **	138 **	772 **	601	131	1279	585	131	1790
Stretching plus bending of Si-O-Si bond (654 cm−1) [44,45]	690	121	436	673	117	735	685	130	635
Si-O-Si bending modes (800 cm−1) [47,66]	777	89	314	774	89	333	790	83	280
Orthosilicate (850 cm−1) [67]	891	87	239	881	87	156	873	41	47
Q2(950 cm−1) [46,47,48,55]	941	59	253	950	76	774	952	79	811
Q3′(1020 cm−1) [46,47,48,68]	1018	105	1442	1016	72	841	1014	56	513
Q3′‘(1080 cm−1) [46,47,48]	1084	70	484	1076	87	1715	1077	83	2234
Q4(1140 cm−1) [46,47,48,55]	1145	68	176	1159	68	150	1156	88	407
Symmetric stretching of BO3 units (1200 cm−1) [31,35,43]	1227	92	68	1229	92	64	---	---	---
BO3-BO4(1300–1450 cm−1) [28,29,30,31]	1416	62	121	1449	87	201	---	---	---
BO3-BO3(1450–1600 cm−1) [28,29,30,31]	1476	74	202	1512	45	30	---	---	---

** The Gaussian component at 641 cm⁻¹ for 20B47Si glass can be assigned to danburite and reedmergnerite rings.

. The general Raman spectral features and trends can be compared by dividing the spectra into borate-dominated, borosilicate, and silicate-dominated glass groups. According to vibrational assignments presented in the literature, we can group these spectral trends into 450–800 cm⁻¹, 850–1250 cm⁻¹, and 1300–1500 cm⁻¹ frequency ranges.

4.3.1. Borate-Dominated 67B, 60B7Si, and 50B17Si Glasses

The low-frequency range can be described by four Gaussian bands, centered near 527, 612, 703, and 772 cm⁻¹ (Table 2). We assign the band near 527 cm⁻¹ for 67B glass to diborate units (Figure 7a and Figure 8a,b) [29,41]. The minor peak near 610 cm⁻¹ is assigned to ring-type metaborate units, while the band near 703 cm⁻¹ is assigned to chain-type metaborate (polymerized BO₃) modes (Figure 7a) [39,40]. The prominent 772 cm⁻¹ feature has been assigned to tetraborate, pentaborate, or triborate groups (Figure 7b) in the literature [31,38,40]. Since tetraborate groups contain pentaborate and triborate configurations, we simplify our assignment of this 772 cm⁻¹ band to symmetric vibrations within the linked tetraborate rings (Figure 7b).

In the 850–1200 cm⁻¹ frequency range, the small band at 857 cm⁻¹ for the 67B glass spectrum is assigned to pyroborate groups (Figure 7a) [35]. Based on Kamitsos’s study [42], we assign the band near 970 cm⁻¹ to orthoborate units (Figure 7a) [32,43,67]. We assign more localized (possibly B-O (non-bridging) stretch) displacements in diborate units for the band near 1114 cm⁻¹ (Table 2) based on findings in the literature for alkali borate glasses [55,66,68]. The band near 1228 cm⁻¹ for 67B glass is assigned to symmetric B-O stretching of planar BO₃ units [32,46,55].

The 1300–1500 cm⁻¹ envelope in the spectrum for 67B glass is fit with two Gaussian components. The 1380 cm⁻¹ band is assigned to vibrations of BO₃ units linked to BO₄ tetrahedra, while the 1480 cm⁻¹ band is assigned to BO₃ linked to other BO₃ triangles [31].

The 60B7Si spectrum is nearly identical to that of 67B glass (Figure 8b). Diborate group vibrations are assigned to the band at 527 cm⁻¹, which increases in area and shifts to 543 cm⁻¹ for 50B17Si glass (Table 2). Increasing the SiO₂ concentration in 50B17Si likely causes the further rupturing of the more complex triborate, tetraborate, and pentaborate groups (Figure 7b) while promoting the formation of the simpler diborate species. The band near 600 cm⁻¹ assigned to ring-type metaborate units loses area and disappears for 50B17Si glass. The area under the 700–735 cm⁻¹ shoulder for the 67B glass (assigned to chain-type metaborate units) shifts to 690 cm⁻¹ for 50B17Si glass. Similarly, the major band near 772 cm⁻¹ steadily decreases in area for 60B7Si and 50B17Si. There is no substantial alteration observed in the pyroborate band around 850 cm⁻¹ and the orthoborate band near 950 cm⁻¹ in the 67B, 60B7Si, and 50B17Si glass compositions. Diborate units persist in the 60B7Si and 50B17Si glasses, as shown by the increasing area for the Gaussian component near 1110 cm⁻¹. The areas for the 1380 cm⁻¹ and 1480 cm⁻¹ BO₃-related bands decrease from 67B to 50B17Si glass, showing fewer BO₃-BO₃ and BO₃-BO₄ linkages.

4.3.2. 40B27Si and 30B37Si Borosilicate Glasses

The introduction of 27 mol.% silica to 40B27Si glass results in noticeable changes in the Raman spectra compared with the more borate-rich glasses (Figure 6). This silica increase appears to disrupt the six-membered borate rings, thereby facilitating the formation of borosilicate rings within these glasses. The Gaussian component, centered near 540 cm⁻¹, can be attributed to B-O-B [55] and B-O-Si [48,55] bending modes (Table 3). A similar band is observed for the 30B37Si glass (Figure 9a,b), where the area is smaller and can be assigned to B-O-B [55], B-O-Si [48,55], and Si-O-Si [47,48] motions. A Gaussian component at 681 cm⁻¹ for 40B27Si is assigned to the chain-type metaborate units.

Additionally, borate tetrahedra in mixed danburite-type four-membered rings (Figure 7d) may be forming [55]. For the SiO₂–Na₂O–B₂O₃ glass system, Manara et al. [45] attributed the band at 630 cm⁻¹ to danburite ring breathing modes; however, in 30B37Si glass, Li replaces Na with respect to the glasses studied by Manara et al., which has a smaller ionic radius and larger charge density that can shift this danburite ring feature to higher frequencies. Consequently, we assign the Gaussian component band at 664 cm⁻¹ to danburite ring motions within 30B37Si glass (Figure 9a,b and Table 3). Like in other alkali borate glasses [34,36], the Gaussian component band at 772 cm⁻¹ is assigned to tetraborate displacements and decreases in area as the glasses become more silica-rich, until this component disappears in the Gaussian fitting procedure for 20B47Si and the more silica-rich glasses.

In the silicate Q-species’ 850–1250 cm⁻¹ frequency range, it is possible that some contributions come from the Si-O and B-O stretch mixed Q-species modes, which can include silicate and borate tetrahedra (Figure 7d). The area under the orthosilicate–pyroborate Gaussian component near 850 cm⁻¹ remains relatively constant (Table 3). The areas decrease for the 950 and 1080 cm⁻¹ bands, which can be assigned to the Q² and Q^3″ species Si-O stretch modes (Figure 7c), respectively [46,47,48,55,68]. The Q^3′ and Q⁴ species components (Figure 7c) at 1020 and 1140 cm⁻¹, respectively, have increasing areas from 40B27Si to 30B37Si glass, indicating a more polymerized borosilicate network.

The 1300–1500 cm⁻¹ envelope is comprised of two Gaussian components that are centered near 1400 cm⁻¹ and 1480 cm⁻¹. As the glasses become more silica-rich, both components decrease in area, indicating fewer BO₃ triangles and a reduction in B-O-B [36], B-O-Si [38], and Si-O-Si [28,31] motions.

4.3.3. Silicate-Dominated 20B47Si, 10B57Si, and 67Si Glasses

In the lower-frequency region from 450 to 800 cm⁻¹, most mode assignments are associated with Si-O-(Si, B) displacements. We observe two major Gaussian components near 530 and 585 cm⁻¹ (Table 4) that are associated with Si-O-Si symmetric bending motions [48,53]. The 530 cm⁻¹ feature loses area and decreases frequency for the two more silica-rich glasses, suggesting that BO₄ tetrahedra and BO₃ triangles are being replaced by the higher-mass SiO₄ tetrahedra in the network. For the more silica-rich 20B47Si glass, the Gaussian component at 641 cm⁻¹ may be assigned to displacements within more silica-rich reedmergnerite rings (Figure 7d) [55]. A band near 601 cm⁻¹ for 10B57Si glass is associated with Si-O-Si, Si-O-B, and B-O-B bending vibrations [53,67,68]. Moreover, a weak feature near 690, 673, and 685 cm⁻¹ in the spectrum for the 20B47Si, 10B57Si, and 67Si glasses (Figure 6 and Figure 10a,b), respectively, is assigned to the stretching–bending motions in Si-O-Si bonds found for a calculated 654 cm⁻¹ mode for Na-silicate glasses [48]. A weak Gaussian component within the 770 to 790 cm⁻¹ range for the 20B47Si and 10B57Si glasses may be due to remnant contributions from tetraborate units, while for the 67Si glass, this component becomes more prominent (Figure 6 and Figure 10b) and may be equivalent to (Si, Al)O₄ tetrahedral ‘cage’ vibrations in Na-aluminosilicate glasses [50] or Si-O-Si modes in (Na, Ca) aluminosilicate glasses [69].

The prominent Si-O stretch Q-species 850–1250 cm⁻¹ envelope was fit with five Gaussian components for the 20B47Si, 10B57Si, and 67Si glasses. The band near 850 cm⁻¹ was fit with the 890 to 870 cm⁻¹ Gaussian component (Table 4) and is assigned Si-NBO stretch motions in orthosilicate or Q⁰ structural units [70]. The Gaussian component near 940 to 950 cm⁻¹ is assigned to Q² species modes [51,71]. Gaussian bands near 1015 cm⁻¹ and 1080 cm⁻¹ are assigned to Q^3′ and Q^3′‘ species, respectively [49,50,51,72], and these bands generally shift to slightly lower frequencies from 20B47Si to 67Si glass, possibly due to Si replacing B in the network (Table 4). The Gaussian component near 1150 cm⁻¹ is assigned to Q⁴ species [71]. The Gaussian band areas at these frequencies generally increase as the boron content decreases. In the borate-dominated glasses, a small Gaussian component near 1228 cm⁻¹, assigned to B-O symmetric stretch within BO₃ triangles [38,40], is observed for 20B47Si and 10B57Si glasses and is absent in the 67Si glass spectrum (Table 4, Figure 6 and Figure 10b).

Borosilicate glass Raman features from 1300 cm⁻¹ to 1600 cm⁻¹ are due to motions within BO₃ triangles. As B content decreases, both the lower-frequency BO₂O-BO₄ stretch Gaussian component near 1430 cm⁻¹ and the BO₂O-BO₃ stretching component near 1490 cm⁻¹ [31,55,60] decrease in area and are absent for the 67Si glass (Figure 11).

5. Discussion

The Raman spectra of the glass series reveal distinct structural changes linked to the composition. The 67B glass clearly has isolated larger borate structural unit populations that include B and O six-membered rings in tetraborate and diborate units, as well as isolated smaller borate structural units that include orthoborate and pyroborate. However, as silica replaces borate, significant alterations are observed for bands associated with the lower-frequency borate species, including metaborate, tetraborate, and pyroborate, as well as the higher-frequency borate species including BO₃-BO₃ and BO₃-BO₄. These changes indicate that borate units become less polymerized and fewer in number as the silica content increases (Figure 11). Additionally, in the 850–1200 cm⁻¹ frequency range assigned to more localized atomic displacements, the borosilicate glass spectra show likely superpositions of ortho- and pyro-borate modes with silicate tetrahedra Q-species modes within the borosilicate networks.

Substantial changes in the Raman spectra are observed between 40B27Si and 30B37Si glasses, particularly from 500 cm⁻¹ and 1200 cm⁻¹. These changes likely indicate the formation of Si-O-B and Si-O-Si linkages in 4-membered borosilicate rings, such as those in the danburite and reedmergnerite crystal structures [56,57].

Yun and Bray [28] used the R and K ratios to describe mechanisms of B/Si mixing and NBO creation in simplified ternary Na-borosilicate glasses with K < 8 based on experimental 11B, 17O, 23Na, and 29Si NMR analyses, as well as Raman spectroscopy. According to them, when R < 0.5, M₂O-B₂O₃-SiO₂ glass has structural units like M₂O-B₂O₃ glass, and when R > 0.5, borate groups start to mix with silicate groups. Moreover, considering the role of K in the formation of borosilicate units, Dell et al. [30] proposed the critical composition R*, where:

R* = 0.5 + 0.0625K

(6)

when R < R*, Na⁺ interacts as a charge compensator for borate tetrahedra in the domain R < 0.5. For R > R*, as Na-borate glass is diluted by silica, NBOs start to form on the silicate tetrahedra [58]. For the glass series studied here, K ranges from 0 (67B) to 5.7 (60B7Si), where Li⁺ interacts as a charge compensator for BO^4-, when R < 0.5 (Figure 1, and Table 1). Moreover, NBOs start to form on silicate tetrahedra in 40B27Si glass, where K = 0.68 and R = 0.75, which results in R* = 0.54, or R > R*. It is important to note that for R > 0.5, mixed borosilicate units are formed, as we see evidence of danburite-like and possibly reedmergnerite-like 4-membered borosilicate rings in the Raman spectrum of the 30B37Si glass (R = 1) (Table 1).

The fluctuation in E_a corresponds with changes in T_g (Figure 5). According to the Anderson–Stuart model [73] (Equation (3)), the E_a for glasses is dependent on the E_b between an anion site and a neighboring alkali cation, along with the E_s needed to deform the glass network. E_b is associated with bond energy for Li⁺, which should stay the same or similar throughout the glass series as the baseline for the E_a at 0.5 eV (Figure 5). Larger T_g values are linked to more strongly bonded networks, requiring increased E_s values to deform the network. This, in turn, leads to elevated E_a and impeded ion mobility. Based on the borate-rich to silicate-rich glasses studied here, increases in T_g infer increases in the glass network connectivity, which hampers Li⁺ mobility, resulting in decreased σ and increased E_a values (Table 1 and Figure 5). The reversal in E_a and T_g indicates the presence of danburite rings in the 30B37Si and 20B47Si glasses.

An interesting trend was observed in the electrical conductivity of the glass series. Generally, for borate-rich to silica-rich compositions, there is an overall decrease in electrical conductivity. However, there is a localized minimum for this trend at 40 mol.% B₂O₃, where conductivity increases from 40B27Si to 20B47Si (Figure 4). The binary 67B glass is the most conductive glass in the series, suggesting no mixed glass former effects (MGFE) across the glass series. Tatsumisago et al. [74] observed MGFE in rapidly quenched borosilicate glasses with larger alkali-oxide concentrations (Li₂O > 60 mol.%) than those for the LABS glasses studied here.

Mai et al. [11] explored a broader spectrum of K and R ratios, spanning from 0.22 to 18 and 0.67 to 6.17, respectively, resulting in critical R* values ranging from 0.51 to 1.63 in their study of the glass series 0.40Li₂O•0.60xB₂O₃•1.2(1−x) SiO₂, where x = 0 to 1, prepared by annealing the melt. Our study covers K and R ratios ranging from 0.12 to 5.70 and 0.45 to 3.00, respectively, producing R* values within the 0.51 to 0.86 range. However, the observed trend of decreasing electrical conductivity from boron-rich to silica-rich glasses, as reported in their study, aligns with our findings. Their reported E_a values of 0.65 eV for the most-conductive borate-rich glass and 0.80 eV for the least-conductive silica-rich glass are slightly higher than the values in our study, which are 0.50 eV for the most-conductive 67B glass and 0.68 eV for the least-conductive 67Si glass (Figure 5). Our glasses obtained from quenching the melt contain small amounts of Al₂O₃ (Table 1), which likely enhance ionic conduction, potentially resulting in lower E_a values for the 67B and 67Si glasses. Furthermore, the Li₂O/SiO₂ ratio in the silica-rich glass studied by Mai et al. (0.33) is lower than that for 67Si (0.45), which might contribute to the higher E_a values observed in their silica-rich glass.

Furthermore, Montouillout et al. [20], in their study on the correlation between the ionic conductivity of lithium borate glass, (xLi₂O•(1−x) B₂O₃, where 0 ≤ x ≤ 0.50), reported that from x ≥30, the formation of NBO sites accompanied other structural changes, which aligns with the presence of pyroborate and orthoborate species in our 67B glass (Figure 8a,b). The presence of more BO₄ sites with ionic bonding to Li⁺, along with some NBOs, results in the highest conductivity of the 67B glass among the series [75]. As the silica content increases in the glass composition, the borate polyhedra intermix with silicate tetrahedra, resulting in more polymerized borosilicate units such as danburite-like rings, which contribute to the slight increase in conductivity, especially for the 30B37Si and 20B47Si glasses. As the boron content continues to decrease and mixed borosilicate populations continue to diminish, the conductivity decreases slightly as the glass becomes more silica-rich (Figure 4 and Figure 12).

6. Conclusions

In summary, this study revealed the relationships between glass composition, structural changes determined from Raman spectroscopy, σ, and T_g. In 67B glass, isolated diborate, tetraborate, and orthoborate groups are prominent. However, substitution of boron with silicon leads to dissociation of linked borate polyhedra, resulting in the formation of borosilicate units as well as larger populations of isolated and linked silicate tetrahedra (Figure 12). 30B37Si and 20B47Si glass structures can be characterized as containing borosilicate rings, similar to those found in danburite and reedmergnerite, along with tetraborate, BO₃-BO₄, and BO₃-BO₃ units. As the boron content continues to decrease, increasing numbers of silicate tetrahedra become more polymerized. The effects of such structural alterations are indicated by the relationships of σ and T_g with respect to the glass composition. For 67B to 40B27Si glass, σ steadily decreases, while T_g steadily increases, until the trends reverse slightly for 30B37Si and 20B47Si glasses. This reversal in σ can be attributed to the presence of danburite-like rings, which contain BO₄^- tetrahedra that act as local charge compensators for Li⁺. At the same time, the reversal in T_g may be caused by the increasing numbers of more weakly bonded borosilicate units compared to the polymerized silicate network units. However, the BO breathing mode for Si₃O₉ planar 3-membered rings is assigned at 608 cm⁻¹ for silica glass [76]. The trends in σ and T_g supports the danburite assignments with localized charge imbalances in the borosilicate network. Subsequent reductions in B₂O₃ content leads to fewer BO₄^- local charge compensators and more polymerized SiO₄ tetrahedra that increasingly hinder Li⁺ mobility, resulting in decreasing σ values. In this case, charge neutrality results when a SiO₄ tetrahedron glass former links with four other SiO₄ tetrahedra to create localized SiO₄⁰ in a silicate-dominated glass. Additionally, a BO₃⁰ triangle would be formed if it is linked to three SiO₄ tetrahedra. As more strongly bonded SiO₄ tetrahedra replace more weakly bonded BO₄ and BO₃ units, stiffening of the borosilicate and silicate networks likely causes a further increase in T_g.

The glasses studied here were prepared via conventional melting and quenching. The Raman spectra for the glass series do not show any evidence of phase separation. However, spinodal decomposition can occur in these glasses with a different heat treatment routines. We plan to explore how phase separation of two different amorphous phases can improve electrical conductivity of the glasses in subsequent studies of lithium borosilicate glass. The understanding of the mixed glass former effect and phase separation induced by spinodal decomposition in lithium borosilicate glasses holds potentially important implications for the development of glass compositions for optimized electrical conductivity. Such optimizations can improve the performance of glasses as solid-state electrolytes in Li⁺ batteries. These observations and resulting insights offer a foundation for further research and innovation in the field of glass science and materials engineering, ultimately driving the development of novel materials to meet the evolving demands of technology and industry.