Preparation and Properties of MgO-Al₂O₃-SiO₂ Glass–Ceramics with Controllable Crystalline Phases

Abstract

MgO–Al₂O₃–SiO₂ (MAS) glass–ceramics with controllable crystalline phases were successfully prepared using the melting method followed by heat treatment. The effects of the main components of glass on the crystallisation kinetics, nucleation, crystallisation and properties of glass–ceramics were investigated in detail. As the Al₂O₃ and MgO contents increase and SiO₂ content decreases, the crystallisation peak temperature and activation energy of MAS glass decrease, while the crystal growth tends to follow a homogeneous nucleation crystallisation. The MAS glass nucleation temperature and time increase with higher concentrations of Al₂O₃ and MgO and with a lower SiO₂ concentration. Mg₂(Al₄Si₅O₁₈) indialite and MgAl₂O₄ spinel precipitate simultaneously in the MAS glass after crystallisation; the relative proportion of crystalline phases is related to the composition and crystallisation temperature. A higher SiO₂ content allows the formation of a dominant indialite phase, while higher MgO and Al₂O₃ contents promote the formation of a dominant spinel phase. The MAS glass ceramic with a greater indialite phase has good dielectric properties with a dielectric constant of 6.499 and dielectric loss of 0.0064, while that of a higher spinel phase possesses improved mechanical properties, with a Vickers hardness of 715 Hv and a bending strength of 244.9 MPa.

1. Introduction

Glass–ceramics, which are multi-phase composites formed by the controlled crystallisation of glass, have been widely used in industry for a long time because of their excellent properties [1,2,3,4,5,6]. Recently, MgO–Al₂O₃–SiO₂ (MAS) glass–ceramics have been gaining attention due to their low thermal expansion coefficient [7,8,9] and excellent dielectric properties [10,11,12]. MAS glass–ceramics have wide application possibilities, including kitchen heating panels, display panels, back panels and various electronic devices [13,14,15,16,17,18].

Presently, the research on MAS glass–ceramics mainly focuses on the influence of other additives and heat treatment on the microstructure and material properties [19,20,21,22,23,24,25]. Gao et al. discussed the effect of Y₂O₃ on the MAS glass–ceramics and found that the addition of Y₂O₃ can promote the indialite phase formation and reduce the sintering activation energy, thereby controlling the thermal expansion coefficient and mechanical properties [26]. Hu et al. discussed the influence of ZnO and heat treatments on the MAS glass–ceramics and found that the addition of ZnO could promote crystallisation and improve the microstructure. Further, the crystallisation and nucleation time affect the Vickers hardness of the glass–ceramics [27]. He et al. investigated the influence of different nucleating agents (TiO₂, ZrO₂ and TiO₂ + ZrO₂) on the crystallisation and properties of the MAS glass–ceramics. In addition, the nucleating agents, TiO₂ + ZrO_2, can promote the precipitation of cordierite and reinforce the stability of the glass, resulting in the best mechanical performance of the glass–ceramics [21]. Li et al. studied the effect of the SiO₂/Al₂O₃ ratio on the properties of the MAS glass–ceramics and observed that an increase in the SiO₂/Al₂O₃ ratio increased the average thermal expansion coefficient but decreased the density and grain refinement, and the sample of MAS glass–ceramics exhibited the best dielectric properties with a dielectric constant of 8.78 (1 MHz) and a dielectric loss of 3.55 × 10^–3 (1 MHz) when the SiO₂/Al₂O₃ ratio was 1.27 [28]. Wu et al. studied the effect of the MgO/Al₂O₃ ratio on the crystallisation and thermal expansion coefficient of the MAS glass–ceramics; they found that the strongest crystallisation was achieved when the MgO/Al₂O₃ ratio was close to 1:1 [29]. Some studies showed that more crystalline phases often existed in the MAS glass–ceramics, including indialite and spinel phases [30,31]. MAS glass–ceramics only containing controlled-content indialite and spinel phases have not been reported so far. And a few detailed studies have examined the different performance characteristics of glass–ceramics with different crystal phases in the MAS system [32,33,34]. However, the limited scope of work greatly restricts the practical applications of MAS glass–ceramics. To meet the requirements for industrial production development, efficiently controlling the main crystalline phases and optimising the performances of the MAS glass–ceramics have become urgent issues.

In this work, we prepared MAS glass–ceramics containing controlled indialite and spinel phases and investigated the influence of their compositions on the nucleation, crystallisation and other properties of the MAS glass–ceramics. MgO, Al₂O₃ and SiO₂ compositions and heat treatment of the MAS can be manipulated to control the formation of the main crystalline phases, Mg₂(Al₄Si₅O₁₈) indialite and MgAl₂O₄ spinel, and their relative proportion in the MAS glass–ceramics determines their mechanical and dielectric properties. The as-prepared MAS glass–ceramics with controllable crystalline phases may be used as display panels, the back panels of various electronic devices.

2. Materials and Methods

2.1. Glass Preparation

Compositions of the batches of MgO–Al₂O₃–SiO₂ (MAS) glass–ceramics used in this experiment are summarised in

Table 1. Compositions of MgO–Al₂O₃–SiO₂ (MAS) glass–ceramics/wt%.

Compounds	SiO2	MgO	Al2O3	Li2O	ZnO	ZrO2	P2O5	Na2O	Sb2O3
MA1	39	14	35	0.5	0.5	4.5	3.5	2	1
MA2	37	15	36	0.5	0.5	4.5	3.5	2	1
MA3	35	16	37	0.5	0.5	4.5	3.5	2	1
MA4	37	17	34	0.5	0.5	4.5	3.5	2	1
MA5	37	13	38	0.5	0.5	4.5	3.5	2	1

. SiO₂, MgO and Al₂O₃ were the main glass components, and Li₂O, Na₂O and ZnO were the fluxing agents. The fluxing agents reduced the melting temperature of the MAS glass and promoted the uniformity of the melted glass. ZrO₂ and P₂O₅ were used as nucleation agents to effectively reduce crystallisation activation energy of glass and improve the nucleation and crystal growth of the glass [35]. Sb₂O₃ was used as a clarifier to remove bubbles in the melted glass [36]. All raw materials were purchased from Shanghai Aladdin Bio-Chem Technology Co., Ltd. (Shanghai, China) without further purification.

The melting process of the MAS glass was as follows. All raw materials were accurately weighed, mixed for 24 h by ball milling and sieved through a 100-mesh screen. The mixed raw materials were then melted at 1600 °C for 6 h and quickly moulded to obtain glass in a pre-heated die. Further, the moulded glass was annealed at 500 °C for 6 h to eliminate internal stress. The annealed glass samples were prepared for heat treatment.

2.2. Nucleation and Crystallisation

The MAS glass samples were pre-nucleated at 760–860 °C (every 20 °C) for 0.5–2 h with a heating rate of 5 °C·min⁻¹. The pre-nucleated samples were rapidly cooled, grounded and sieved through a 200-mesh screen to obtain glass powder. A differential scanning calorimeter (DSC; NETZSCH STA 449F5, Selb, Germany) with a heating rate of 10 °C·min⁻¹ was then used to perform the differential thermal analysis of the glass powders. The instrument had a temperature resolution of 0.1 °C, DSC resolution of 0.01 μV and weight resolution of 0.0001 mg. The mass of the sample was in the range of 25–30 mg. The gas used was 99.9% N₂ and the pressure was 10 MPa. Suitable nucleating temperature and time for MAS glass was obtained according to the peak crystallisation temperatures (T_p, the temperature corresponding to the main exothermic peak of the DSC curve) at different pre-nucleating temperatures and time.

After the nucleation at a suitable nucleating temperature and time, the nucleated glass samples underwent crystallisation at T_p and 1100 °C using a heating rate of 5 °C·min⁻¹. The samples were cooled rapidly, grounded and sieved through a 200-mesh screen. The sample phases were identified by performingX-ray diffraction (XRD) analysis using a Panalytical X’Pert³ X-ray diffractometer using nickel-filtered Cu Kα radiation in the range of 2θ = 10–80° with a scanning speed of 5°·min⁻¹. Scanning electron microscopy (SEM, SU-8010, Tokyo, Japan) was then used to observe the crystal morphology of the crystallised glass samples. The surface of the samples was etched with hydrofluoric acid (HF, 3 wt%) for 60 s for further observation.

2.3. Performance Characterisation

After nucleation and crystallisation, the MAS glass–ceramics were prepared for performance characterisation. The conventional water displacement method was used to study the sample density. The thermal expansion coefficient was recorded using a DIL-402C dilatometer (NETZSCH, Waldkraiburg, Germany) at a heating rate of 5 °C·min⁻¹. The mechanical properties (bending strength and elastic modulus) were measured by three-point bending tests conducted using a CMT5202 electric material performance measurement equipment (200 kN) with a span of 30 mm and a cross-head rate of 1 mm·min⁻¹. The Vickers hardness was obtained using a Vickers hardness tester (Tukon2500, Wilson, Norwood, MA, USA) and the formula: $\mathrm{H}\mathrm{V}=F/A$, wherein HV is the Vickers hardness, F is the applied load of 0.5 kg and A is the area of the contact surface of the indenter [37]. The fracture toughness was calculated using an electronic universal testing machine (CMT6103) and according to the formula: $K_{IC}=\frac{P_{f}S}{B{W}^{1.5}}\mathrm{f}\left(\frac{z}{W}\right)$, wherein K_IC is the fracture toughness, P_f is the maximum fracture load, S is the support span, B is the specimen thickness, Z is the size of the pre-crack, and W is the specimen width [38]. Dielectric properties (dielectric constant and dielectric loss) were measured using a network analyser (E8363C PNA, Santa Rosa, CA, USA).

3. Results

3.1. DSC of MAS Glass and Crystallisation Kinetics

The DSC curves of the annealed MAS glass samples are shown in Figure 1, and the transition temperature (T_g) and crystallisation peak temperature (T_p) are listed in

Table 2. Glass transition temperature (T_g) and crystallisation peak temperature (T_p) of MAS annealed glass samples.

	MA1	MA2	MA3	MA4	MA5
Tg/°C	747	754	753	742	764
Tp/°C	982	956	950	951	949

. For the MA1, MA2, MA3, MA4 and MA5 glass samples, the T_g values were 747 °C, 754 °C, 753 °C, 742 °C and 764 °C, respectively, and the T_p values were 982 °C, 956 °C, 950 °C, 951 °C and 949 °C, respectively. In this MAS system, the combined amounts of SiO₂, MgO and Al₂O₃ were kept constant, with only their relative contents being changed. For MA1, MA2 and MA3 samples, the relative SiO₂ content was reduced, and the corresponding relative contents of MgO and Al₂O₃ increased. It was observed that the T_g of the MAS glass sample increased slightly, while the T_p decreased. For the MA2, MA4 and MA5 samples, the SiO₂ content was kept constant and the relative concentrations of MgO and Al₂O₃ varied. When the relative content of MgO increased from 13 wt% to 17 wt% and the corresponding relative content of Al₂O₃ decreased from 38 wt% to 34 wt%, the T_g of the MAS glass sample decreased from 764 °C to 742 °C, while the T_p increased slightly from 949 °C to 951 °C. Among the MA1, MA3 and MA5 samples, the MA1 sample had a higher SiO₂ content, hence showing a higher T_p and a lower T_g value. In comparison, the MA3 and MA5 samples had higher MgO and Al₂O₃ contents, respectively; therefore, their corresponding T_g values were higher and their T_p values were lower than those of the MA1. These results indicated that a higher SiO₂ content restricted MAS glass crystallisation, while increased MgO and Al₂O₃ contents not only reinforced the network structure of the glass but also improved its crystallisation. In a glass network, an increase in SiO₂ content will increase the viscosity of the glass melt and hinder substance diffusion, thereby increasing T_p. Meanwhile, an increase in the MgO content, as the glass network modifier, can destroy the Si–O bond and break the oxygen bridges, thereby worsening the glass structure and reducing the viscosity of the glass melt to promote glass crystallisation [39].

The influence of main glass compositions on the crystallisation kinetics of the MAS glass samples was investigated using the Kissinger equation and Augis–Bennett equation [40,41,42,43,44], which are expressed as follows:

$$\ln \left(\frac{T_{\mathrm{p}}^2}{a}\right)=\frac{E}{R{T}_{\mathrm{p}}}+ln\frac{E}{R}-ln\nu .$$

(1)

$$n=\frac{2.5}{\mathsf{\Delta }T}\times \frac{{RT}_{\mathrm{p}}^2}{E}.$$

(2)

where E is the activation energy (kJ·mol⁻¹); R is the gas constant; ν is the frequency factor and $a$ is the DSC heating rate (°C·min⁻¹). Further, $n$ represents the crystallisation index, and ΔT is the temperature difference at half-height of the DSC exothermic peak. The activation energy (E) reflects the crystallising ability of the sample. A lower E indicates a higher crystallisation tendency of the glass. The crystallisation index (n) is related to the manner of crystallisation of the glass. It can be inferred that n = 1, 2, 3 and 4 can approximately represent surface crystallisation, two-dimensional growth crystallisation, volumetric crystallisation and homogeneous nucleation crystallisation, respectively [21,42].

Figure 2a, Figure 3a and Figure 4a show the DSC curves of the MA1, MA3 and MA5 samples, respectively, at different heating rates. The T_p values from the DSC curves at different heating rates are also listed in

Table 3. T_p of MAS glass samples at different heating rates.

Heat Rate/°C·min−1	5	10	15	20	25
MA1/K	1236	1255	1268	1279	1286
MA3/K	1201	1218	1227	1237	1243
MA5/K	1222	1239	1251	1260	1266

. These results show that T_p increased with an increase in the heating rate. The rates of nucleation and crystal growth also affect T_p [42]. A high heating rate for a glass caused less time to absorb heat, thereby shifting T_p to a higher value. The relationship between ln(T_p²/a) and 1/T_p was plotted, as shown in Figure 2b, Figure 3b and Figure 4b, to calculate the effective activation energy and crystal index.

The values of the activation energy and crystal index are presented in

Table 4. Crystallisation kinetic parameters of MAS glass samples.

Crystallisation Parameter	MA1	MA3	MA5
E/kJ·mol−1	399.90	457.05	265.92
n (a = 10 °C·min−1)	3.4	2.2	3.7

. Compared with MA1 sample (399.90 kJ·mol⁻¹), MA3 sample has a higher E (457.05 kJ·mol⁻¹) and MA5 sample has a lower E (265.92 kJ·mol⁻¹). This indicates that MA3 sample has a lower crystallisation tendency, while MA5 sample has a better crystallisation tendency compared to MA1. Further, the Al₂O₃ content of the MA5 is higher than that of MA3, indicating that a higher Al₂O₃ content improves glass crystallisation. The n of the MA1 is 3.4, which is close to 3, suggesting volumetric crystallisation. MA3 has a n value of 2.2, which is closer to 2, suggesting a two-dimensional growth crystallisation. The n of the MA5 sample is 3.7 and suggests homogeneous nucleation crystallisation, as it is closer to 4. In combination with the concentration of SiO₂, MgO and Al₂O₃ of glass, the appropriate contents of SiO₂, MgO and Al₂O₃ can not only improve glass crystallisation but also affect the manner of crystallisation. Increased MgO and Al₂O₃ contents can lead to the two-dimensional crystallisation growth of glass. In conclusion, the results confirm that the main glass components play an important role in determining the nature of MAS glass crystallisation.

3.2. Nucleation of MAS Glass

The nucleation stage is very important during the heat treatment of glass–ceramics. DSC analysis is regarded as an effective method to study the glass nucleation stages. Because many nano-scale nuclei are formed during pre-nucleation [35,42,43], the glass will absorb considerable thermal energy during continuous heat treatment, indicating that the T_p will shift to lower temperatures in the DSC curve. In this work, we utilised this mechanism to study the nucleation of the MA1, MA3 and MA5 glass samples. The range of the nucleating temperature should be between T_g and T_p [35,42]. According to the DSC results (Figure 1), the pre-nucleating temperatures of the MA1, MA3 and MA5 samples are in the ranges of 760–800 °C, 780–860 °C and 780–860 °C (temperature was increased in steps of 20 °C), respectively. The pre-nucleating times of the MA1, MA3 and MA5 samples are in the range of 0.5–2 h (time was increased in steps of 0.5 h) at suitable pre-nucleating temperatures.

Figure 5a,b show the DSC curves of the MA1 after pre-nucleation at 760–800 °C (every 20 °C) for 2 h and at 760 °C for 0.5–2 h, respectively. The relationship of T_p with the pre-nucleating temperature and time are shown in Figure 5c,d, respectively. Pre-nucleation clearly influences the T_p, as the latter increases with an increase in the pre-nucleating temperature. T_p first decreases and reaches its minimum value at 1 h before it increases in the pre-nucleating time, followed by another increase. For MA3 (Figure 6a), T_p decreases slowly, reaches its minimum at 820 °C and then increases again with an increase in the pre-nucleating temperature; after, T_p increases slightly, followed by another decrease (Figure 6b). For MA 5 (Figure 7a), with an increase in the pre-nucleating temperature, T_p declines first and reaches its minimum at 820 °C before an increase. Meanwhile, with higher pre-nucleating times, T_p increases first before decreasing again (Figure 7b). Hence, the suitable nucleation temperatures for the MA1, MA3 and MA5 are 760 °C, 820 °C and 820 °C, respectively, and the suitable nucleation times are 1, 0.5 and 2 h, respectively. A comparison of the samples in terms of T_p and nucleation temperatures shows that higher MgO or Al₂O₃ content will reduce the T_p of the MAS glass and considerably increase the nucleation temperature.

3.3. Crystallisation of MAS Glass

To better explore the influence of the main glass compositions on the crystallisation of MAS glass samples, MA1, MA3 and MA5 were crystallised for 2 h at T_p and at a higher crystallisation temperature (1100 °C), after suitable nucleation and according to the DSC curves of annealed glass (Figure 1). Precipitation of the main phase of the Mg₂(Al₄Si₅O₁₈) indialite (PDF#82-1884) after crystallisation at T_p for the three MAS glass samples is presented in Figure 8a. After crystallisation at T_p, the diffraction peak intensities of the Mg₂(Al₄Si₅O₁₈) indialite in the MA1 and MA5 samples are stronger than that of the MA3 sample. Further, the MgAl₂O₄ spinel (PDF#86-2258) phase is clearly observed of three samples, coexisting with the Mg₂(Al₄Si₅O₁₈) indialite phase when the crystallisation temperature is raised to 1100 °C, as shown in Figure 8b. Meanwhile, a strong diffraction peak of the MgAl₂O₄ spinel is observed, which exceeded the diffraction peak intensity of the Mg₂(Al₄Si₅O₁₈) indialite in the MA3 sample after crystallisation at 1100 °C. These results indicate that the main glass components and crystallisation temperature can control the formation of the main crystalline phase. MA1 sample has a higher SiO₂ and lower MgO and Al₂O₃ contents. MA5 sample has higher SiO₂ and Al₂O₃ contents and a lower MgO content, which is beneficial for the generation of the Mg₂(Al₄Si₅O₁₈) indialite. MA3 sample has a lower SiO₂ content and higher MgO and Al₂O₃ contents, which favours the precipitation of the MgAl₂O₄ spinel. Thus, controlled crystalline phases of the MAS glass–ceramics can be achieved by adjusting the main glass compositions and heat treatment.

The SEM images in Figure 9a–c show that after crystallisation at T_p, the crystal shape of the three samples (MA1, MA3 and MA5) is spherical with nano-scale crystal size (10–50 nm). After crystallisation at 1100 °C, the crystal shape of the samples is still spherical with a crystal size of about 30–100 nm, as seen in Figure 9d,e. In comparison, some holes with a size of approximately 20–100 nm appear on the surface of the MA5 sample (Figure 9f), which is the result of grain corrosion. The EDX mapping images of MA1, MA3 and MA5 samples (after crystallisation for 2 h at T_p) indicate a uniform distribution of elemental Mg, Al, Si and O (Figures S1–S3). The EDX test results are consistent with the XRD test results.

3.4. Performance of MAS Glass–Ceramics

Table 5. Physical, mechanical, thermal and electrical properties of the MAS glass–ceramics. (All numbers in this table are the mean values of corresponding parameters).

Properties	MA1	MA3	MA5
Density/g·cm−3	2.52	2.51	2.53
Fracture toughness/MPa m1/2	2.07	2.80	3.00
Vickers hardness/Hv	706	715	709
Elastic modulus/GPa	105.92	107.20	106.40
Thermal expansion coefficient/×10−6 °C−1	4.09	5.46	4.42
Dielectric constant (1 MHz)	6.499	6.618	6.510
Dielectric loss (1 MHz)	0.0064	0.0068	0.0066
Bending strength/MPa	217.4	244.9	225.3

and Figure 10 present the physical, mechanical, thermal and electrical properties of the MA1, MA3 and MA5 samples after nucleation and crystallisation. The three samples have different physical, mechanical, thermal and electrical properties. Among the three samples, MA1 has better thermal and dielectric properties with a lower thermal expansion coefficient (Figure 10a), dielectric constant and dielectric loss (Figure 10d); MA3 possesses superior mechanical properties with a higher bending strength (Figure 10c), Vickers hardness and elastic modulus (Figure 10b,c); and MA5 has higher fracture toughness (Figure 10b).

Based on the analysis of the results above, we concluded that the main crystalline phases of the MA1 and MA5 samples were dominated by Mg₂(Al₄Si₅O₁₈) indialite, while the main phase of MA3 was influenced by the MgAl₂O₄ spinel. These microstructural and crystal phase characteristics resulted in differences among the samples in terms of their performance. These results further confirm that the main glass composition and heat treatment can control the microstructure and properties of the MAS glass–ceramics.

4. Conclusions

MAS glass–ceramics with controlled crystalline phases were successfully prepared, and the effects of the main glass composition on the crystallisation, microstructure and performance of the MAS glass–ceramics were investigated using DSC, XRD, SEM and other techniques. The main glass compositions determined the crystallisation kinetics, nucleation, crystallisation and performance of the MAS glass–ceramics. In terms of the main glass composition, high SiO₂ contents restrict the crystallisation of glass, while high Al₂O₃ contents can improve crystallisation. The Mg₂(Al₄Si₅O₁₈) indialite and MgAl₂O₄ spinel phases precipitate in the MAS glass after crystallisation at high temperatures; furthermore, high MgO and Al₂O₃ contents can promote the formation of the dominant MgAl₂O₄ spinel. The MAS glass–ceramics rich in SiO₂ have better thermal and dielectric properties, whereas those rich in MgO and Al₂O₃ possess superior mechanical properties; moreover, ceramics with high Al₂O₃ contents have higher fracture toughness. The resultant MAS glass–ceramics with controllable crystalline phases and adjustable properties are promising materials applicable to various fields, such as optoelectronics, aerospace and communications.