The Li₂Si₂O₅ whiskers were synthesized in a one-step hydrothermal process. The molar mass proportion of LiOH·H₂O to SiO₂·H₂O was maintained at 1:1 according to the stoichiometric composition. After mixing the constituents, batches were dissolved in deionized water for 4 h. The resultant solutions were transferred and sealed in a Teflon-lined stainless-steel autoclave under autogenous pressure and heated to 150°C for 6 h. The solutions were then cooled naturally to room temperature, sieved, washed several times in turn with distilled water and ethanol, and finally dried at 80°C for 24 h; the resultant white precipitates were recovered. The detailed process is the same as our previous studies (Liu et al., 2022).

Reagent-grade powders of Li₂CO₃, SiO₂, NH₄H₂PO₄, Al₂O₃, K₂CO₃, and La₂O₃ were used as raw materials. The composition of the base glass is given in Table 1. After mixing the constituents, batches were placed in a Pt crucible and melted in an electric furnace at 1,450°C for 30 min in air. Then, the glass melts were quenched in deionized water to obtain frits for milling. The dried glass frits were ball milled with high-purity zirconia balls in an ethanol environment. The blended powders were washed and then dried to obtain glass powders (Liu et al., 2022).

To investigate their crystallization effect on glass-ceramics, Li₂Si₂O₅ whiskers (0, 1, 3, and 5 wt%) were then added to the glass powders. These glasses were represented as LDW0, LDW1, LDW3, and LDW5, respectively. The glasses were wet-mixed with zirconia balls in 99.7% anhydrous alcohol for 2 h. After drying, the mixtures were placed in a hardened-steel die and uniaxially pressed under 20 MPa. Then, the samples were sintered in a vacuum furnace at 900°C for 1.5 h. Finally, after cooling to ambient temperature, the surface layers of the samples were removed for subsequent characterization.

The crystalline phases of the samples were characterized by X-ray diffraction analysis (XRD-7000S, Japan) using Cu-Kα radiation with a scanning velocity of 5°/min, a step width of 0.02°, a scanning range of 10–60°, an acceleration voltage of 40 kV, and a current of 30 mA. In addition, the relative crystallinities of the samples were calculated by Jade 6.0 software using the number, relative intensity, and location of diffraction peaks according to the XRD pattern. The relative crystallinity of the glass-ceramics is estimated according to the following equation: X c = ∑ ​ I c ∑ ​ I c + K I a × 100 % (1) where Xc is the crystallinity, Ic is the integrated intensity of the crystal diffraction peaks, Ia is the integrated intensity of the amorphous fraction, and K is a constant related to the measurement conditions and glass compositions.

The Li₂Si₂O₅ glass-ceramic samples were polished and etched using a 10 vol% HF solution for 15 s, and their microstructures were observed by scanning electron microscopy (SEM, JSM6701F, Japan). The three-point flexural strength of the specimens (24 mm × 4 mm × 1.5 mm after chamfering, smoothing, and polishing) was determined by a universal mechanical machine (Instron 3366, United States) with a span (center-to-center distance between support rollers) of 16 mm and a crosshead speed of 0.5 mm/min according to ISO 6872. Ten test bars were prepared to obtain average values. The three-point flexure is calculated using the following equation: σ = 3 P l 2 w b 2 (2) where σ is the flexural strength in units of MPa, P is the breaking load in N, and l, w, and b represent the sample test span, width, and thickness in mm respectively.

The fracture toughness of the specimens (24 mm × 4 mm × 3 mm) was measured following the single-edge V-notched beam (SEVNB) method. A U-shaped groove was notched at the surface of the 3 × 24 mm side with a diamond cutting wheel cooled with water, and a V-shaped notch with a tip radius of less than 15 μm was machined on the bottom of the groove. The notched side of the bar was placed under tension in the three-point bending apparatus with a span of 16 mm and a crosshead speed of 0.5 mm/min. Fracture toughness tests were carried out for each group of five samples per group of glass-ceramic specimens to verify accuracy and dispersion.

The SEM morphology and XRD pattern of the Li₂Si₂O₅ whiskers synthesized by the hydrothermal method are shown in Figures 1A,B, respectively. It was clearly observed that the Li₂Si₂O₅ whiskers were rod-like crystals with an average length of 1.37 ± 0.23 μm, width of 0.13 ± 0.02 μm, and aspect ratio of 5–12 (Figure 1A). XRD analysis showed no impurity peaks, as seen in Figure 1B, indicating that the obtained samples had high purity, and the diffraction peaks were consistent with those of ICDD PDF #33-0816 (Liu et al., 2022). Therefore, the synthesis of the Li₂Si₂O₅ whiskers by the hydrothermal method laid the foundation for the subsequent Li₂Si₂O₅ whisker-reinforced glass-ceramic. In this work, we report a simple hydrothermal approach for the synthesis of Li₂Si₂O₅ whiskers with regular, new morphology. At present, the hydrothermal method has broad application prospects in the synthesis of nanophase materials under low-temperature conditions. The hydrothermal method is environmentally friendly, as its reaction is carried out under closed-system conditions, saving energy (Alemi et al., 2014a; Alemi et al., 2014b).

XRD patterns of the Li₂Si₂O₅ glass-ceramics are shown in Figure 2A. The main precipitated crystalline phase of all samples was Li₂Si₂O₅ (ICDD PDF#40-0376). With an increase in the amount of Li₂Si₂O₅ whiskers, the intensity of the diffraction peaks of the Li₂Si₂O₅ crystals shifted slightly, owing to changes in crystallinity and crystal size. Under an identical preparation process, the intensity of the diffraction peaks of the Li₂Si₂O₅ crystals increased slightly from LDW0 to LDW5. As seen in Figure 2B, the crystallinity of the samples remained relatively constant (from 75.98 ± 5.72% to 78.09 ± 8.89%) upon varying the whisker content from 0 to 1 wt%, while it increased significantly to 86.86 ± 7.18% in LDW5.

Enormous efforts have been made to optimize the Li₂Si₂O₅ crystals embedded in glass-ceramics in multicomponent systems with nucleating agents or glass modifiers. Among these additives, P₂O₅ is known to be the most effective for increasing the nucleation rate since it promotes the bulk nucleation of Li₂Si₂O₅ by forming a steep compositional gradient in the vicinity of the amorphous Li₃PO₄ phase, where it acts as the heterogeneous nucleation site for both Li₂SiO₃ and Li₂Si₂O₅ phases (Clausbruch et al., 2000; Huang et al., 2013). As for the Al₂O₃ component, it can reduce phase-separation trends and increase thermal stability, and a small amount is beneficial for improving the crystallization controllability of the Li₂O–SiO₂ system (Thieme and Rüssel, 2014). As for K₂O, it can promote the rupturing of bridging oxygen bonds between silicon and oxygen tetrahedrons, improving the O/Si ratio in the system, which is conducive to the precipitation of Li₂SiO₃ crystals (Fernandes et al., 2012). In addition, La₂O₃ can be added to reduce the viscosity of the glass system.

In this study, phase separation was promoted during the crystallization heat treatment to form Li-rich and Si-rich phases, after the addition of P₂O₅ to the Li₂Si₂O₅ glass system. Li₂O interacts with P₂O₅ in the Li-rich regions to form Li₃PO₄ crystal nuclei, which can occur at non-uniform nucleation sites of Li₂SiO₃ and Li₂Si₂O₅ crystals according to the following reaction scheme, which was conducive to reducing the nucleation energy (Wen et al., 2007; Lodesani et al., 2020). P 2 O 5 ( glass ) + 3 Li 2 O ( glass ) = 2 Li 3 PO 4 ( crystal ) (3) Li 2 O ( glass ) + SiO 2 ( glass ) = Li 2 SiO 3 ( crystal ) (4) Li 2 SiO 3 ( crystal ) + SiO 2 ( glass ) = Li 2 Si 2 O 5 ( crystal ) (5)

Upon increasing the amount of Li₂Si₂O₅ whiskers, the whisker content in the Li-rich regions also increased. The added Li₂Si₂O₅ whiskers could act as nuclei sites, so the simultaneous effect of surface nucleation from the glass powders and that induced by the whiskers increased the degree of crystallization (Zhao et al., 2021). Consequently, the nucleation rate rose, and the crystallinity increased.

Although pure SiO₂ glass crystallizes above 1,000°C, a small amount of quartz appeared at lower temperatures in this study. This was because Li₃PO₄ and the added Li₂Si₂O₅ whiskers may have provided some heterogeneous nucleation sites to induce the crystallization of quartz (Zhao et al., 2019).

The SEM morphologies of the Li₂Si₂O₅ glass-ceramics are shown in Figure 3. The Li₂Si₂O₅ crystals exhibited a closely packed and rod-like morphology, forming multi-directionally interlocking microstructures. However, the sizes of the crystals were quite different. The crystal sizes of the LDW0 samples were uniform, and no large crystals were found. In the LDW1, LDW3, and LDW5 samples, a bimodal crystal size distribution in which both large, elongated Li₂Si₂O₅ crystals and fine crystals existed was observed. This was because the surface nucleation of some whiskers began to occur, where the glass phase nucleates to form Li₂Si₂O₅ crystals on the surfaces of these whiskers. These crystals then grow from the interfaces between whiskers and the glass phase in the center of the adjacent glass region. The large rod-like Li₂Si₂O₅ crystals were believed to result from the epitaxial growth of the Li₂Si₂O₅ whiskers, while the small crystals were directly crystallized from the Li₂Si₂O₅ glass powders. The specific crystal size and porosity data are summarized in Figure 4. These results showed that the average length and width of the Li₂Si₂O₅ crystals increased from 1.85 ± 0.53 to 3.23 ± 1.14 μm and 0.39 ± 0.07 to 0.66 ± 0.2 μm, respectively, moving from LDW0 to LDW5. Further, the porosity increased after first decreasing. The lowest porosity of 1.15 ± 0.11% was obtained in the LDW1 sample.

The essence of the crystallization mechanism could be understood through the different samples with and without the Li₂Si₂O₅ whiskers, as compared in Figure 5. In the LDW0 specimen, the parent glass powders, with high surface energy, directly precipitated small, rod-like crystals. In the case of LDWx (x ＞ 0), in addition to the glass powders, Li₂Si₂O₅ whiskers were added to induce crystallization and encourage epitaxial growth to form large, rod-like crystals, thus forming the coexistence of multi-scale crystals.

The improvement in the mechanical properties of the samples was also a consequence of obtaining an appropriate microstructure, crystalline phase composition, and lower porosity. The flexural strength and fracture toughness of the four specimen groups are listed in Figure 6. Compared with the whisker-free glass-ceramics (LDW0), flexural strength was found to increase upon adding Li₂Si₂O₅ whiskers. With increasing whisker amount, the flexural strength increased and then decreased after its peak value, increasing from 356.8 ± 8.4 MPa for LDW1 to 389.5 ± 11.77 MPa for LDW3, then slightly decreasing for LDW5. The fracture toughness of the glass-ceramics strongly improved with the addition of the Li₂Si₂O₅ whiskers, reaching a maximum (3.46 ± 0.1 MPa·m^1/2) for the LDW5 specimens.

Concerning the LDW0 sample, the flexural strength and fracture toughness were lower than those of LDW1, LDW3, and LDW5. Although the LDW0 sample had a relatively low porosity, its crystallinity was not high, and the sizes of its crystal distribution were concentrated, which cannot effectively hinder crack propagation and transfer interfacial stress. The flexural strength of the LDW3 sample was the highest. In part, this was because the crystallinity was more suitable, and the porosity was within an acceptable range. Further, the microstructure was more compact, and the average aspect ratio of the sample was suitable.

Micro residual stress existed in the Li₂Si₂O₅ glass-ceramics owing to the thermal expansion coefficient mismatch between the glass matrix and crystal (Pinto et al., 2007). At room temperature, there existed radial residual compressive stress in the crystal and tangential residual tensile stress in the glass matrix (Serbena and Zanotto, 2012; Serbena et al., 2015). With the increase of crystal size, the residual stress level increased (Li et al., 2016). In the three-point bending process, the residual tensile stress in the glass matrix overlapped with the macroscopic external tensile stress on the tension side of the specimen, which was beneficial to crack propagation in the glass matrix and reducing the bending fracture load. In this study, although LDW5 showed the highest crystallinity (86.86%), its flexural strength was not optimal. There were several reasons to explain these phenomena. Firstly, the coarse crystals were also controlled by the micro residual stress effect, the existence of which led to the decrease in strength. Secondly, increasing the amount of whiskers led to an increased porosity. Thus, the optimal flexural strength appeared in LDW3.

As for fracture toughness, the values of the glass-ceramics increased with the addition of Li₂Si₂O₅ whiskers, with the highest value (3.46 ± 0.10 MPa·m^1/2) exhibited by the LDW5 samples. The increase in the fracture toughness of the LDWx (x ＞ 0) samples may be related to the interlocking microstructure of coexisting elongated Li₂Si₂O₅ crystals with smaller ones, which was related to crack bridging and crack deflection. A microstructure consisting of bimodal crystal size distributions could cause crack blunting or branching, which was advantageous for enhancing the mechanical properties. The presence of rod-like whiskers was equivalent to a bridge between two crack surfaces and provided a force that brought the crack surfaces close to each other, which canceled out the effect of applied stress to a certain extent and significantly reduced the strength of the effective stress field at the crack tip. As the crack expanded further, the increase in the distance between the crack surfaces was bound to be inhibited and constrained by the bridge action of the crystals, which increased the crack propagation resistance of the material. The crack-propagated in the glass matrix and through some small crystals, while it was deflected by large crystals. As seen in Figure 3, the LDW5 sample had a large amount of abnormally enlarged Li₂Si₂O₅ crystals (L_max of 6.81 μm, and W_max of 1.83 μm), which resulted in a larger angle of crack deflection. This deflection meant that the cracks had to surmount this additional surface area as they progressed through the material. As a result, more energy was absorbed, which contributed to the purpose of increasing the toughness. Consequently, LDWx (x ＞ 0) in the present study had a better fracture toughness than LDW0.

Table 2 compares the mechanical properties of the Li₂Si₂O₅ whisker-reinforced glass-ceramics investigated in this work with those of other reported studies. The three-point flexural strength was higher than conventional sintered Li₂Si₂O₅ glass-ceramics and even comparable to commercial IPS e.max Press and IPS e.max CAD (Ivoclar Vivadent, Liechtenstein), with flexural strengths reaching 400 and 360 MPa according to the manufacturer’s specifications, respectively. The fracture toughness of LDW5 was superior to those of most reported studies, including IPS e.max Press (2.75 MPa∙m^1/2) and IPS e.max CAD (2.25 MPa∙m^1/2). Thus, we improved the mechanical properties of Li₂Si₂O₅ glass-ceramics by adding Li₂Si₂O₅ whiskers. The results showed that Li₂Si₂O₅ whisker reinforcement has great potential in improving the mechanical properties of Li₂Si₂O₅ glass-ceramics.