Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: The effect of heat treatment on microstructure and properties of SiO2–Li2O–K2O–Al2O3–ZrO2 glass-ceramics was
Received 18 February 2014 investigated. The glasses were subjected to a multi-stage heat treatment. At the first stage at 650 °C, Li2SiO3 was
Received in revised form 22 May 2014 the main crystalline phase of all samples, which had a dimensional change with prolonging the heating time from
Available online xxxx
3 h to 72 h. After the second stage treatment at 830 °C for 3 h, all the Li2SiO3 crystals decomposed and Li2Si2O5
crystals formed with an opposite dimensional change from that of Li2SiO3. These Li2Si2O5 crystals were small,
Keywords:
Lithium disilicate;
which could contribute to forming high translucency. The third stage at 550 °C would release the internal strain
Heat treatment; and improve samples' bending strength, which could reach up to above 560 MPa. The sample treated for 48 h at
Bending strength; the first stage had the smallest crystals and also obtained a highest real in-line transmission of 29% at 650 nm
Translucency with a thickness of 1.45 mm and a bending strength above 581 ± 25 MPa.
© 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jnoncrysol.2014.05.023
0022-3093/© 2014 Elsevier B.V. All rights reserved.
102 P. Zhang et al. / Journal of Non-Crystalline Solids 402 (2014) 101–105
The glass frits were used for differential scanning calorimetry mea-
surement with DSC-machine (SDT Q600, TA, America). The temperature
Fig. 1. DSC analysis of parent glass (ramp rate of 20 K/min in air).
range was from room temperature to 1150 °C in air and the heating rate
amounted to 10 K/min. A three-stage heating cycle was designed in the
present work. At the first stage, samples named G3, G6, G12, G24, G48 G72, which was relatively higher. The total crystallinities of all samples
and G72 were heat treated at 650 °C for 3 h, 6 h, 12 h, 24 h, 48 h and were similar after the second stage.
72 h, respectively. Subsequently, all of them were heated at 830 °C for
3 h at the second stage, and then all the samples were treated at 3.2. Microstructural features
550 °C for 3 h at the third stage. All treatments were carried out in a
muffle furnace with a heating rate of 5 K/min. Finally, all the treated The microstructural features of glass heated at 650 °C for different
samples were ground on wet silicon carbide papers and then polished holding times are illustrated in Fig. 3. As LM is more easily dissolved in
to get a smooth surface. HF than silica glass matrix, when the glass is etched by HF, left pits
reflect the microstructure of LM crystals. The LM crystals of all samples
2.3. Crystalline phase analysis and performance test were in dendritic shape and had a dimensional change with prolonging
the holding time. Fig. 4 represents the microstructure of samples under
The crystalline phases were determined by X-ray diffraction (XRD) same treatment at 830 °C for 3 h at the second stage. Fig. 4(a) shows
analysis (Empyrean, PANalytical, Netherlands) with Cu Kα radiation rod-like shape LD crystals of G3 in a closely packed and multi
and scanning from 10° to 60° at a scanning speed of 5°/min. The directionally interlocking microstructure. The other samples had the
polished samples were etched by immersing in 5 vol.% HF solution for same microstructure with a gentle variation of the mean size of LD crys-
15 s and coated with a very thin Au layer. A microstructure analysis in tals. Some LD crystals in G72 were relatively coarser and in an irregular
a field emission scanning electron microscope (FE-SEMVEGAII XMU,
Tescan, Czech Republic) then followed. The crystallinities of lithium
disilicate GCs could be estimated by Eq. (1), where Ia is the integrated
intensity of the non-crystalline phase and Ic is the integrated intensity
of the crystalline phase.
∑Ic
Xc ¼ 100%: ð1Þ
∑Ic þ ∑Ia
The SEM results were also used for measuring the crystal length
using Nano-Measuring image analysis software. Three-point bending
strength was tested with a universal testing apparatus (Instron-1195,
Shimadzu, Japan), at a span length of 16 mm and a crosshead speed
of 0.5 mm/min. Translucent values were expressed by real in-line trans-
mission, which was measured in 300–1100 nm range by ultraviolet/
visible spectrum spectrometer (UV-1800; Mapada, Shanghai, China).
3. Results
DSC trace of the investigated glass is shown in Fig. 1. The glass tran-
sition temperature (Tg) is about 480 °C. Two exothermic crystallization
peaks at 645 °C and 827 °C could be associated to the formation of LM
and LD, respectively.
Fig. 2 presents the XRD patterns of the glass samples subjected to
sequential heat treatments. At 650 °C, LM dominated in all samples
and there was no significant change in crystalline composition until
the heating time reached 72 h, at which a few of silica and LD emerged
in G72. Fig. 2b indicates that LM had completely disappeared and LD
formed, acting as principal crystalline phase in all samples after treated
at 830 °C for 3 h. A few SiO2 as minor phase appeared in all samples.
Fig. 2. XRD patterns of lithium disilicate glass ceramics underwent two-stage heat
Crystallinities of all samples after treated at the first stage are summa- treatment. (a) glasses treated at first stage, at 650 °C for different time, (b) all glasses
rized in Table 1. According to the table, all samples' crystallinities did treated at second stage, at 830 °C for 3 h: Li2SiO3, ICCD card 29-0829; Li2Si2O5, ICCD card
not change obviously with prolonging the heating time except that of 40-0376; SiO2, ICCD card 42-1401, Tridymite; SiO2*: ICCD card 12-0711, Coesite.
P. Zhang et al. / Journal of Non-Crystalline Solids 402 (2014) 101–105 103
Table 1 1100 nm, the RIT of all samples showed a same changing tendency,
Crystallinities of GCs after two stages of treatment; random errors b ±5%. rising dramatically with increase of the wavelength. There was a huge
Crystallinity (%) G3 G6 G12 G24 G48 G72 difference at visible wavelengths among all the samples. The G48
group showed the highest transmission and its RIT at the wavelength
1st stage 34.5 33.27 36.17 38.5 32.85 37.05(LM)
2.73(LD) of 650 nm was 29% but that of G3 only reached 10%.
2.17(SiO2)
2nd stage 65.66 66.53 68.05 65.99 64.67 63.99(LD) 4. Discussion
3.08(SiO2)
The molar ratio of the SiO2/Li2O of glass used in this study was 2.36,
close to that of LD. However, LM crystals nucleated and precipitated
shape, as Fig. 4(b) shows. Fig. 5 shows the dimensional change of both prior to LD, which was different from the crystallization path of stoichio-
LM and LD crystals of different groups. The mean length of LM crystals metric binary component Li2O–SiO2 glass. That is because the viscosity
increased continuously after peaking at 12.39 μm in G48 and decreased of glass is increased with the incorporation of other oxides, such as
to 7.56 μm in G72, while the size of LD showed an opposite trend, ZrO2 and Al2O3 [20-22]. LM is more readily to organize and to nucleate
decreasing slightly from 0.433 μm in G3 to just 0.322 μm in G48, and than LD as the structure of chain silicate units (i.e. LM) is simpler than
then increasing again to 0.396 μm in G72. Namely the larger LM crystals that of layer silicate units (i.e. LD) [23]. When the holding time was
formed at the first stage, the smaller LD crystals formed at the second long enough, there may be a significant change in structural relaxation
stage. of glass, so a few LD crystals started to precipitate after heated for 72 h
during the first stage. Meanwhile, some crystalline SiO2 also precipitated.
3.3. Bending strength and translucency characteristics When treated at 830 °C for 3 h, accompanied by dissolution of LM,
numerous LD crystals formed underlying the state reaction Li2SiO3 +
The bending strength of the GCs is shown in Fig. 6. After the second SiO2 → Li2Si2O5 [14]. In G72, because small amount of SiO2 had formed
stage treatment, the bending strength decreased monotonously. G3 in the first stage, these SiO2 might induce the crystallization of residual
demonstrated the highest strength of 392 ± 27 MPa and G72 showed glass matrix, resulting in more residual silica in G72. The crystallinity of
the lowest strength of 242 ± 31 MPa, which was still higher than G72 at the first stage was higher than that of other groups, which was
those of leucite and mica GCs with a mean flexural strength less than attributed to the precipitation of LD in the G72. After treated at 830 °C
180 MPa [19]. However, after the third stage treatment at 550 °C for for 3 h, all kinds of crystals had crystallized, so the crystallinities of all
3 h, the bending strength of all samples was greatly improved, varying groups were similar.
from 562 ± 32 MPa to 611 ± 27 MPa. The growth of LM crystals is a result of kinetics of crystal growth; the
After the third stage treatment, real in-line transmission (RIT) of all mean size of crystals increases with prolonging the holding time. How-
the GCs is shown in Fig. 7. At wavelength ranging from 300 nm to ever in G72, the mean size of LM crystals decreased which may be
Fig. 3. SEM photographs of etched surfaces of samples after the first stage treatment: (a) G3, (b) G6, (c) G12, (d) G24, (e) G48, (f) G72.
104 P. Zhang et al. / Journal of Non-Crystalline Solids 402 (2014) 101–105
Fig. 4. SEM photographs of etched surfaces of samples after the second stage treatment: (a) G3, (b) G72.
attributed to the fact that some of the crystals transformed to LD. In the high of all groups because the grain size of LD was smaller than the
second stage, the LD nucleated frequently at the sites where LM wavelength. The grain size of G48 was smaller than other groups after
dendrites grew [24]. Larger LM may have higher efficiency and supply the second stage treatment, so it showed a higher RIT at all the test
more nucleation sites for LD, which may refine the LD crystals, so larger wavelengths. Thirdly, with prolonging the holding time, the ions of
LM crystals may lead to smaller LD crystals. After the second stage, all other oxides diffused and the composition of glass matrix became
LM crystals decomposed and transformed to LD. In G72, a bit of LD more homogeneous, which may further reduce the scattering. All the
crystals had formed at the first stage, which made the finally formed factors mentioned above showed the good explanation for the fine
LD of G72 exhibiting a coarse microstructure. Its mean size was larger translucency performance in G48. In G72, the coarser microstructure
than that of G48. and non-uniformity in grain size caused more scattering. Moreover,
In the LD GCs, intergranular fracture may be the dominant fracture the extra SiO2 crystals acting as the secondary phase inclusions caused
mode. The crack development always takes place at the weaker inter- extra scattering, thus the sample became less translucent than the G48.
faces between crystals and propagates through the residual glass matrix
[25]. The closely packed and multi directionally interlocking structure of
the tested GC will deflect the crack propagation, leading to a high 5. Conclusion
mechanical strength. On the other hand, there is discrepancy of both
density and thermal expansion coefficient between crystalline phase The crystalline process in multicomponent Li2Si2O5 glass was always
and glass matrix, which may cause severe internal strain and multiple divided into two stages: at the first stage, the Li2SiO3 precipitated at
microcracks during crystallization [26]. During the process of grain 650 °C, then at the second stage, this unstable intermediate phase
growth, the microcrack in samples may increase to weaken the bending decomposed completely and transformed to Li2Si2O5 at 830 °C. The
strength of samples. After the third stage, annealing at 550 °C, the inter- effect of annealing time at the first stage on the properties of finally
nal stress was released and microcracks healed, so the bending strength formed Li2Si2O5 glass-ceramic was analyzed in this study. Prolonging
increased effectively. the annealing time did not change the dominant crystalline phase of
The translucency of GCs is largely dependent on light scattering [27]. Li2SiO3 but enlarged the mean size of these crystals. The larger the
The factors of light scattering include crystalline content, crystal size mean size of Li2SiO3 crystals, the smaller the mean size of Li2Si2O5, and
and refractive index. [28]. In our present work, the crystalline contents the translucency of final glass-ceramic would increase when the size
of all samples are similar, so refractive index and the crystal size may of Li2Si2O5 reduced. Prolonging the annealing time also introduced
be the main factors. Firstly, translucent GCs need the refractive index internal stress, which would be released after the third stage treatment.
of the crystal phase close to that of glass matrix. The LD crystals have a The sample treated for 48 h at the first stage finally showed a perfect
refractive index of 1.55, close to that of the glass matrix of 1.50 [29]. Sec- bending strength of 581 ± 25 MPa and a highest real in-line transmis-
ondly, the particles which are similar in size to the light wavelength sion of 29% at 650 nm. Additionally, with excessive extension of holding
have the greatest scattering effect [27]. The grain sizes of all groups time (72 h), small amount of Li2Si2O5 and SiO2 crystals may precipitate
were around 500 nm and close to the visible wavelength, causing a at the first stage, which would slightly reduce the bending strength and
strong light scattering in the GC, thus the RIT declined significantly at the translucency of finally formed glass-ceramics.
the visible wavelength range. At infrared wavelengths, RIT values were
Fig. 5. Mean crystal length of samples after different stages of heat treatment. Fig. 6. Bending strength of samples after the second and third stage treatments.
P. Zhang et al. / Journal of Non-Crystalline Solids 402 (2014) 101–105 105