The effect of CeO₂ addition on crystallization and thermo-physical properties of lithium zinc silicate (LZS) glasses containing Li₂O-ZnO-SiO₂-Al₂O₃-Na₂O-P₂O₅ was investigated. The changes of CeO₂ contents (2-8 wt.%) had an obvious influence on the transition temperatures (T_g) and crystallization temperatures (T_c) of LZS glass-ceramics, and they increased with CeO₂ content increasing. According to XRD analysis, CeO₂ promoted the formation of cristobalite and β-spodumene crystals, and β-spodumene increased obviously. As the CeO₂ content increasing, the microstructure and microhardness (being 6.88 Gpa at 880 ^oC) of glass-ceramics had great changes. The average thermal expansion coefficient (20 - 450 ^oC) showed first increasing then decreasing, having a wide range. The maximum of thermal expansion coefficient was obtained when the glass-ceramics contained 4 wt.% CeO₂, being 175×10^-7 K^-1 (at 700 ^oC) and 178×10^-7 K^-1 (at 880 ^oC) respectively. Excellent thermo-physical properties indicate the glass has greater potential application, such as being used as sealing glass.

Keywords: Lithum zinc silicate, Crystallization, Thermal expansion coefficient, Microhardness

Currently there has been a considerable amount of interest on crystallization behavior and thermo-physical property of Li₂O-ZnO-SiO₂ glass-ceramics [1-7] due to their beneficial performances, e.g., a wide range of thermal expansion coefficient (α = 50 - 200 × 10^-7 K^-1) by controlling heat treatments and adding different elements, high electrical resistivity and good chemical durability. McMillan and Partridge in 1963 [8] first reported lithium zinc silicate glass-ceramics containing high proportions of zinc oxide. Later, they found glass-ceramics could also be employed to produce superior glass-ceramic-to-metal seals [9]. Donald et al. [10] studied a number of Li₂O-ZnO-SiO₂ materials containing relatively high concentrations of ZnO, including the influence of nucleating species and concentration, and an assessment of the thermal expansion characteristics and mechanical properties. And they summarized recent developments of glass-ceramic-to-metal seals and coatings to gain a better understanding of diffusion and reaction behavior of individual metallic [11]. The glass-ceramics are suitable for matching sealing to lots of metals and alloys with different thermal expansion coefficients such as copper alloys, nickel-based alloys, Fe-Ni-C alloys and stainless steel etc. It is generally agreed that the properties of glass-ceramics are dependent on the crystal contents, phases and their microstructure. Chemical compositions of basic glasses play very important roles in the formation of crystals. The crystallization characteristic of glass-ceramics is usually influenced markedly by adding some specific elements, such as rare earth elements. Recently, Y₂O₃, Nd₂O₃, Fe₂O₃, B₂O₃, CuO, alkali oxides and alkali earth oxides also have been used as flux in some glass-ceramics to study the changes of their crystallization and thermo-physical properties [12-19].
Cerium oxide, as a rare earth oxide flux, has been reported in many glass-ceramics studies. In some studies, it is used as raw material in glasses [20-24]. Mostly CeO₂ is used as additive to regulate the properties of glass-ceramics [25-27]. Sohn et al. [28] found that CeO₂ as a flux markedly decreased viscosity in MgO-Al₂O₃-SiO₂ glasses, and had a little influence on thermal expansion, mechanical properties and chemical durability of glasses. Anmin Hu et al. [29] investigated phase transformations of Li₂O-Al₂O₃-SiO₂ glasses with CeO₂ addition, and showed that the transformations of glass to β-quartz and of β-quartz to β-spodumene were accelerated by addition of CeO₂. Temuujin et al. [30] studied the influence of CeO₂ addition in crystallization behavior and mechanical properties of glass ceramics in the Na₂O-CaO-Al₂O₃-SiO₂ system, and found that CeO₂ added to glass ceramics enhanced hardness because of an increased crystal size. Yongsheng et al. [31] also found CeO₂ could improve the integrity of the glass network structure and enhance fracture toughness in CaO-Al₂O₃-SiO₂ system. In the previous research, we know CeO₂ in different glass systems has various influences on thermo-physical and mechanical properties. Therefore, in order to enhance the properties of lithium zinc silicate (LZS) glass, it is essential to investigate the effect of CeO₂ on LZS glass.
The purpose of this research was to investigate the effect of different content of CeO₂ addition on crystallization and thermo-physical properties of Li₂O-ZnO-SiO₂ glasses. After leaning its effect on the properties, we could adjust them by adding different content of CeO₂ to make the glass-ceramic better seal to various metals.

Composition and preparation of the glass
The chemical compositions (wt.%) of the glasses were given in Table 1. We used the ingredients SiO₂, ZnO, Li₂O and Na₂O. P₂O₅ was used as nucleating agent. The starting materials were SiO₂ (≥99.0%), ZnO (≥99.0%), Li₂CO₃ (≥97.0%), Na₂CO₃ (≥99.8%), NH₄H₂PO₄ (≥99.0%) and CeO₂ (≥99.0%). The weighted errors of all raw materials were controlled in the range of ±0.01 g.
Glasses were prepared by fusing regent-grade chemicals in a platinum crucible. The glass samples were melted and maintained at 1,400 ^oC in an electric furnace for 3 h. The molten glasses were then poured into a preheated (480 ^oC) graphite mould and immediately transferred to an annealing furnace set at 480 ^oC, being held for 2 h for annealing before cooling down to room temperature. The transparent and bubble-free glasses were obtained. Based on the DSC curves, we made the nucleation temperature and crystallization temperature to obtain glass-ceramics.
 
Analytical methods
Differential scanning calorimeter (DSC, Netzsch 404PC, Germany) was used to determine the glass transition temperature (T_g) and crystallization temperature (T_c) of the glass samples. The DSC measurements were performed using ~15 mg of powdered samples (45-53 μm), which were placed in an alumina crucible. The temperature range of the samples was between 20 and 1,000 ^oC at heating rate of 10 K/min. The measurement error was ±2 ^oC.
X-ray diffraction (XRD, D/max, 2500 model, Rigaku, Japan) was used to investigate the crystalline phases of the samples. The diffractmeter was with Cu Kα radiation in the 2θ range from 10^o to 80^o at 0.02 steps, which operated at 40 kV and 50 mA at a scanning rate of 4 ^o/min. The crystalline phases were identified by matching the peak positions of the intense peaks with PCPDF standard cards.
The microstructure of glass-ceramics samples were examined by the scanning electron microscope (SEM, FEI Quanta-200, America). The samples were polished to under 1μm and eroded by HF (4 wt.%) for 60 s, then transferred them to an ultrasonic cleaning equipment for 1 h to eliminate impurities that did not dissolve in the acid before being sputtered with a gold coating. Crystal morphology could be observed clearly.
Vickers hardness (H_v) tests were tested on polished glass-ceramics samples, using a Matsuzawa microhardness tester (HVS-1000) with a pyramid shaped diamond indenter, applying loads 1.96 N for 30 s. At least 10 different positions measurements were taken for each sample to take the average values. Indentation diagonals were measured to calculate hardness values in GPa.
The thermal expansion coefficient (TEC) measurements were carried out in a thermo-mechanical analyzer (Netzsch DIL 402EP, Germany) in the temperature range 20-450 ^oC at heating rate of 2 K/min, using a silica probe in an inert atmosphere using argon (50 ml/min). The samples were machined into 3 mm × 3 mm × 20 mm cuboid test bars. The average thermal expansion coefficient values were calculated in the temperature range of 20-450 ^oC.

Crystalline phases and microstructure
Fig. 1 shows the DSC curves obtained from as-cast LZS glasses. There were obviously endothermic valleys on the differential thermal curves. The start points of the endothermic valleys (T_g, marked temperatures in Fig. 1) were from 485 to 510 ^oC with CeO₂ content increasing. It indicated the glass transition temperature (T_g) increased. Exothermic peaks were also obvious on the curves, and the peaks had changed. The temperatures of first peaks increased but the peak intensity (height and area) weakened at the same time. Crystallization peaks correspond to the formation of different crystalline phases. The curve shape of C₀ is similar with that of C₂, it showed that low content (under 2 wt.%) of CeO₂ would not have obvious influence on crystalline phases. As the CeO₂ content increasing, when it was above 4 wt.%, it had a bigger impact on the formation of crystals. The curves of C₄ and C₆ showed new peaks at 804 ^oC and 809 ^oC, and C₆ and C₈ both appeared another peaks at 867 ^oC and 874 ^oC. Table 2 given a summary of the results of differential scanning calorimetry: glass transition temperatures (T_g) and the peak temperatures of crystallization (T_p). The main crystals of LZS are lithium zinc silicate and cistobalite, and the formation temperatures of them are about 650 ^oC and 750 ^oC [34]. However, β-spodumene crystal formation temperature is about 850 ^oC. According to the DSC curves, CeO₂ might promote the formation of new phase β-spodumene. On account of the analysis of DSC curves above, the nucleation temperature of the samples was set at 520 ^oC for 2 h, and the crystallization temperatures were set at 700, 830 and 880 ^oC for 2 h, respectively, to obtain glass-ceramics. We could know the crystal growth process from X-ray diffraction patterns.
Fig. 2 and 3 show the X-ray diffraction patterns of LZS glass-ceramics at different crystallization temperatures (700 ^oC and 880 ^oC). Fig. 2 shows the XRD patterns of the samples containing different amounts of CeO₂ crystallized at 700 ^oC. Crystallization consequence of the CeO₂-free sample is quite simple, as C₀ showing. The major phases were Li₂ZnSiO₄ (PDF# 24-0677) and Li₃Zn_0.5SiO₄ (PDF# 24-0667), which was the same to LZS system. The curves of C₀, C₂ and C₄ in Fig. 2 are similar. It means that low content of CeO₂ (under 4 wt.%) in LZS system would not have obvious effect on the main crystals in the glass-ceramics at low crystallization temperature (700 ^oC). However, the amounts of main crystals increased with CeO₂ content increasing. When CeO₂ content was over 6 wt.%, β-spodumene (PDF# 35-0797) appeared, which could be observed from C₆ and C₈ curves in Fig. 2. According to the XRD analysis result in Fig. 2, the Li₂ZnSiO₄ and Li₃Zn_0.5SiO₄ crystals were promoted by CeO₂, and the amount of cristobalite (PDF# 39-1425) also increased. High content of CeO₂ (above 6 wt.%) could accelerate crystals growth, especially for β-spodumene.
Fig. 3 shows the XRD patterns for the samples crystallized at 880 ^oC. Li₂ZnSiO₄ and Li₃Zn_0.5SiO₄ also were identified as major crystallization phases along with cristobalite phase increasing when the amount of CeO₂ was under 2 wt.%, according to C₀ and C₂ curves in Fig. 3. C₀ curve in Fig. 3 had no obvious difference with C₀ in Fig. 2, but at 880 ^oC, C₂ appeared β-spodumene, and cristobalite increased. It indicated that temperature also had a distinct effect on the formation of β-spodumene and cristobalite. At high crystallization temperature (880 ^oC), CeO₂ could better promote crystals growth. The CeO₂-free samples had no β-spodumene phase, regardless of temperature. The amounts of Li₂ZnSiO₄, Li₃Zn_0.5SiO₄ and cristobalite also increased as CeO₂ increasing at 880 ^oC. The samples containing high amount of CeO₂ (6 wt.%) at low crystallization temperature (700 ^oC) had the same crystalline phases with the samples containing low amount of CeO₂ (2 wt.%) at high crystallization temperature (880 ^oC). When the content of CeO₂ reached to 8 wt.%, the amount of β-spodumene was highest at 880 ^oC. By contrasting Fig. 2 and Fig. 3, it indicated CeO₂ could accelerate crystals growth in LZS system, and it is more obvious at higher temperature. Table 3 shows more details of crystalline phase changes of the samples containing different content of CeO₂.
The microstructures of the samples are showed in Fig. 4 and 5. The chosen scanning electron microscope pictures were representative, which were C₀, C₂, C₆, C₈ heat-treated at 700 ^oC and 880 ^oC for 2 h, respectively. The morphology in Fig. 4(a) and (b) was similar. The interconnected network microstructure formed by tiny dendritic crystals, and the crystals in picture (b) was much bolder. According to the XRD analysis, the main crystals in Fig. 4(a) and (b) were Li₂ZnSiO₄ and Li₃Zn_0.5SiO₄ at 700 ^oC. The microstructures of glass-ceramics had changed significantly when CeO₂ content were 6 and 8 wt.%, showing in Fig. 4(c) and (d). A lot of β-spodumene crystals formed. The morphologies become smaller network-shaped microstructure, and crystals density increased. The development and disorderly distribution of massive thin small dendritic crystals resulted in the formation of honeycomb-shaped network structure. As the amount of CeO₂ increasing, the dendritic crystals became coarser, as shown in Fig. 4(d).
When the heat-treated temperature was 880 ^oC, the effect of CeO₂ on the crystals was more obvious. The microstructures of CeO₂-free sample appeared different parts (Fig. 5(a)). The spherical crystals could be observed, which was cristobalite crystal. Contrasting with Fig. 4(d), Fig. 5(b) had similar crystalline structure with it. It showed that CeO₂ would have a much stronger impact on promoting crystals formation at a high temperature (880 ^oC). The Fig. 5(c) has a looser net-work structure than that of picture (b). The reason was that with the increasing of CeO₂, the effect of partial substitution of Si⁴⁺ by Al³⁺ ions became stronger [32] and caused crystallization and development of β-spodumene. The grain of cristobalite grew bigger, distributing in tiny dendritic β-spodumene crystals, which could be seen in Fig. 5(d).
 
Hardness (H_v) and Thermal expansion coefficient (α)
Vickers hardness of the glass-ceramics was an important property. The samples crystallized at 730 ^oC and 880 ^oC were tested, and the results were showed in Table 4. The microhardness of LZS glass-ceramics increased with CeO₂ increasing, as showing in Fig. 6. It was already pointed out that β-spodumene and cristobalite crystals increased in the glass-ceramics, which both have higher bond strength than Li₃Zn_0.5SiO₄ and Li₂ZnSiO₄ crystals. That was the reason of increase in hardness [33]. The hardness curves crystalized at 700 ^oC in Fig. 6 had biggest changes when the amount of CeO₂ was from 2 to 4 wt.%. However, at 880 ^oC, the changes happened when the amount of CeO₂ was from 0 to 2 wt.%. They were both reached to the maximums, 6.68 GPa and 6.88 GPa respectively, when CeO₂ content was 8 wt.%. It suggested high temperature could enhance the effect of CeO₂ in LZS glass-ceramics, which is similar with the XRD results. The increase of microhardness was in favour of advancing the sealing application of LZS glass-ceramics.
The LZS glasses are mainly used to seal to metals, and the thermo-physical property of the glasses is vital. The average thermal expansion coefficient at 20-450 ^oC was investigated. The results were showed in Table 4. Fig. 6 showed the change of thermal expansion coefficients at different crystallization temperatures versus CeO₂ content. The thermal expansion coefficients was first increasing then decreasing, as showing in Fig. 6, and reached to the maximums when the amount of CeO₂ was 4 wt.%. According to the XRD results in Fig. 2, when the amount of CeO₂ is free, the main crystals are Li₂ZnSiO₄ and Li₃Zn_0.5SiO₄, which have low thermal expansion coefficients (α ≈ 110×10^-7 K^-1). However, the amount of cristobalite (α ≈ 270×10^-7 K^-1) increased with CeO₂ increasing. The thermal expansion coefficient of LZS glass-ceramics increased, and reached to the maximum. With the amount of CeO₂ increasing (above 4 wt.%), it promoted the formation of β-spodu-mene, which has a lower thermal expansion coefficient (α ≈ 19×10^-7 K^-1). The thermal expansion coefficient of LZS glass-ceramics decreased with β-spodumene increasing. When the crystallization temperature reached to 880 ^oC, the effect of CeO₂ promoting β-spodumene and cristobalite formation was more obvious, and β-spodumene crystals increased more apparently with CeO₂ content increasing. The glass-ceramics had lower thermal expansion coefficient when the amount of CeO₂ was above 6 wt.% at 880 ^oC. A small amount of CeO₂ (under 4 wt.%) could increase the average thermal expansion coefficients of LZS glass-ceramics, but it would markedly decrease when high CeO₂ content promoted the formation of β-spodumene. It could be concluded that LZS glass-ceramics could have a wide range of thermal expansion coefficients with the changes of temperature and CeO₂ content. The change of thermal expansion coefficient would make it better seal to more metals. 

The amount of CeO₂ could affect the crystallization and thermo-physical properties of lithium zinc silicate glass-ceramics. The glass transition temperature (T_g) increased, but CeO₂ promoted the formation of β-spodumene and cristobalite crystals, which made the structure more rigid and caused the change of thermo-physical property, and the effects were more obvious at high crystallization temperature. The microhardness increased with CeO₂ content increasing; however, the average thermal expansion coefficients (20-450 ^oC) of LZS glass-ceramics showed first increasing then decreasing, and it obtained a wider range and reached to maximums when the amount of CeO₂ was 4 wt.%. The analysis indicated that CeO₂ played an important role in adjusted the crystallization and physical properties. And a higher microhardness and a wider average thermal expansion coefficient indicated the LZS glass-ceramics could better seal to metals.

The work was financially supported by Science and Technology Research Foundation of Jiangxi Education Department (No. GJJ170537/GJJ170498) and Startup Foundation for Docotors of Jiangxi University of Science and Technology (No. jxxjbs17019).