Journal of Non-Crystalline Solids 402 (2014) 101–105

Contents lists available at ScienceDirect

Journal of Non-Crystalline Solids

journal homepage: www.elsevier.com/ locate/ jnoncrysol

Effect of heat treatment on the microstructure and properties of lithium

disilicate glass-ceramics
Pei Zhang ⁎, Xinghua Li, Jianfeng Yang, Shaochun Xu
State Key Laboratory for Mechanical Behavior of Materials, Xi'an Jiaotong University, 710049 Xi'an, China

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.

1. Introduction Li2O–Al2O3–SiO2–B2O3–K2O–P2O5 system, and found that LD content

Glass-ceramics (GCs) are polycrystalline materials obtained by
crystallization of suitable glasses through an appropriate heat treatment
[1–3]. Many properties of GCs are determined by the precipitated
crystals and their microstructures [4,5]. Lithium disilicate (Li2Si2O5,
hereafter referred as LD) GC is one of the most thoroughly studied GCs
for its good crystallization ability, translucency and mechanical proper-
ties. This glass has a wide range of practical applications [6–9]. The
excellent properties depend on the composition of the parent glass
and thermal treatment. In the case of the composition, several oxides
have been introduced for developing LD GC featuring properties. Hugo
R. et al. found that the addition of Al2O3 and K2O could improve the
chemical durability, densification and mechanical strength, which can
reach up to 201 MPa [10]. TiO2 and P2O5 were used as efficient nucleating
agents, contributing to a fine-grained interlocking microstructure in
GCs which could lead to a high mechanical strength [11,12].
In the case of heat treatment, the crystallization in the multicompo-
nent glasses took place in two heat-treat stages [13]: first lithium
metasilicate (Li2SiO3, hereafter referred to as LM) crystallized in the
glass then transformed into LD, this transformation was believed to
underlie the solid-state reaction of Li2SiO3 + SiO2 → Li2Si2O5 [14]. The
heat treatment at both stages would have a significant influence on
final crystal precipitation, crystal species and content. Borom et al. [15]
studied the effect of different heating routes on the microstructure of
and crystal size remained fairly constant with increasing growth time.
Zheng et al. [16] studied one-stage and two-stage treatments on
Li2O–SiO2–ZnO–K2O–P2O5 glass and found that two-stage treatment
was favorable for the growth of stable LD crystals. Some researchers
also focused on the extended heat-treatment of the glasses, McCollister
et al. [17] heat treated the SiO2–Li2O–Al2O3–K2O–P2O5–B2O3 glass for a
time period sufficient to cause growth of cristobalite and LM producing
a glass-ceramic having a specific thermal expansion coefficient. Burgner
et al. [18] heated the 34.5Li2O–65.6SiO2 (mol%) glass for over 100 h and
found that LM phases did not persist at long times in lithium disilicate
glass. But few reports systematically studied the effect of different
holding times at the LM precipitation stage on the microstructure,
mechanical and optical properties of final LD GCs. In this study, experi-
mental lithium disilicate GCs in the SiO2–Li2O–K2O–Al2O3–ZrO2–P2O5
system were prepared by melt-casting and three-stage heat treatment
was carried out. Different holding times, especially extending to 72 h
at the first stage, were performed to evaluate the developed microstruc-
ture, mechanical and optical properties. Furthermore, the interrelation
of microstructure, mechanical and the optical properties of the GC was
also discussed in detail.
2. Experimental
2.1. Glass preparation

The investigated glass composition was designed as follows: 65.0

⁎ Corresponding author at: State Key Laboratory for Mechanical Behavior of Materials,
Xi'an Jiaotong University, No. 28 Xianning West Road, 710049 Xi'an, Shaanxi, China.
Tel.: +86 13468675140; fax: +86 029 82667942.
E-mail address: zappersp2@gmail.com (P. Zhang).
SiO2–27.5 Li2O–2.2 Al2O3–2.3 K2O–2.0 ZrO2–1.0 P2O5 (mol%). Standard
laboratory reagent-grade Li2CO3, SiO2, Al2O3, K2CO3, ZrO2 and
NH4H2PO4 powders (99.99 mass%, CNCM, China) were chosen as raw

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

materials. Homogeneous mixtures of batches were obtained by ball

milling and melted in high-purity alumina crucibles at 1450 °C for 2 h,
in air. The melts were poured into preheated steel molds and annealed
at 500 °C for 2 h to relieve stress. Annealed glass samples were cut into
30 mm × 3 mm × 4 mm for bending test and 1.45 mm × 15 mm ×
15 mm for translucency evaluation. Small amounts of the glasses were
poured into water to obtain glass frits.

2.2. DSC analysis and heat treatment

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

3.1. Phase formation

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)

shape, as Fig. 4(b) shows. Fig. 5 shows the dimensional change of both
LM and LD crystals of different groups. The mean length of LM crystals
increased continuously after peaking at 12.39 μm in G48 and decreased
to 7.56 μm in G72, while the size of LD showed an opposite trend,
decreasing slightly from 0.433 μm in G3 to just 0.322 μm in G48, and
then increasing again to 0.396 μm in G72. Namely the larger LM crystals
formed at the first stage, the smaller LD crystals formed at the second
stage.
3.3. Bending strength and translucency characteristics
The bending strength of the GCs is shown in Fig. 6. After the second
stage treatment, the bending strength decreased monotonously. G3
demonstrated the highest strength of 392 ± 27 MPa and G72 showed
the lowest strength of 242 ± 31 MPa, which was still higher than
those of leucite and mica GCs with a mean flexural strength less than
180 MPa [19]. However, after the third stage treatment at 550 °C for
3 h, the bending strength of all samples was greatly improved, varying
from 562 ± 32 MPa to 611 ± 27 MPa.
After the third stage treatment, real in-line transmission (RIT) of all
the GCs is shown in Fig. 7. At wavelength ranging from 300 nm to
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
prior to LD, which was different from the crystallization path of stoichio-
metric binary component Li2O–SiO2 glass. That is because the viscosity
of glass is increased with the incorporation of other oxides, such as
ZrO2 and Al2O3 [20-22]. LM is more readily to organize and to nucleate
than LD as the structure of chain silicate units (i.e. LM) is simpler than
that of layer silicate units (i.e. LD) [23]. When the holding time was
long enough, there may be a significant change in structural relaxation
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.
When treated at 830 °C for 3 h, accompanied by dissolution of LM,
numerous LD crystals formed underlying the state reaction Li2SiO3 +
SiO2 → Li2Si2O5 [14]. In G72, because small amount of SiO2 had formed
in the first stage, these SiO2 might induce the crystallization of residual
glass matrix, resulting in more residual silica in G72. The crystallinity of
G72 at the first stage was higher than that of other groups, which was
attributed to the precipitation of LD in the G72. After treated at 830 °C
for 3 h, all kinds of crystals had crystallized, so the crystallinities of all
groups were similar.
The growth of LM crystals is a result of kinetics of crystal growth; the
mean size of crystals increases with prolonging the holding time. How-
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
second stage, the LD nucleated frequently at the sites where LM
dendrites grew [24]. Larger LM may have higher efficiency and supply
more nucleation sites for LD, which may refine the LD crystals, so larger
LM crystals may lead to smaller LD crystals. After the second stage, all
LM crystals decomposed and transformed to LD. In G72, a bit of LD
crystals had formed at the first stage, which made the finally formed
LD of G72 exhibiting a coarse microstructure. Its mean size was larger
than that of G48.
In the LD GCs, intergranular fracture may be the dominant fracture
mode. The crack development always takes place at the weaker inter-
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
mechanical strength. On the other hand, there is discrepancy of both
density and thermal expansion coefficient between crystalline phase
and glass matrix, which may cause severe internal strain and multiple
microcracks during crystallization [26]. During the process of grain
growth, the microcrack in samples may increase to weaken the bending
strength of samples. After the third stage, annealing at 550 °C, the inter-
nal stress was released and microcracks healed, so the bending strength
increased effectively.
The translucency of GCs is largely dependent on light scattering [27].
The factors of light scattering include crystalline content, crystal size
and refractive index. [28]. In our present work, the crystalline contents
of all samples are similar, so refractive index and the crystal size may
be the main factors. Firstly, translucent GCs need the refractive index
of the crystal phase close to that of glass matrix. The LD crystals have a
refractive index of 1.55, close to that of the glass matrix of 1.50 [29]. Sec-
ondly, the particles which are similar in size to the light wavelength
have the greatest scattering effect [27]. The grain sizes of all groups
were around 500 nm and close to the visible wavelength, causing a
strong light scattering in the GC, thus the RIT declined significantly at
the visible wavelength range. At infrared wavelengths, RIT values were
Fig. 5. Mean crystal length of samples after different stages of heat treatment.
high of all groups because the grain size of LD was smaller than the
wavelength. The grain size of G48 was smaller than other groups after
the second stage treatment, so it showed a higher RIT at all the test
wavelengths. Thirdly, with prolonging the holding time, the ions of
other oxides diffused and the composition of glass matrix became
more homogeneous, which may further reduce the scattering. All the
factors mentioned above showed the good explanation for the fine
translucency performance in G48. In G72, the coarser microstructure
and non-uniformity in grain size caused more scattering. Moreover,
the extra SiO2 crystals acting as the secondary phase inclusions caused
extra scattering, thus the sample became less translucent than the G48.
5. Conclusion
The crystalline process in multicomponent Li2Si2O5 glass was always
divided into two stages: at the first stage, the Li2SiO3 precipitated at
650 °C, then at the second stage, this unstable intermediate phase
decomposed completely and transformed to Li2Si2O5 at 830 °C. The
effect of annealing time at the first stage on the properties of finally
formed Li2Si2O5 glass-ceramic was analyzed in this study. Prolonging
the annealing time did not change the dominant crystalline phase of
Li2SiO3 but enlarged the mean size of these crystals. The larger the
mean size of Li2SiO3 crystals, the smaller the mean size of Li2Si2O5, and
the translucency of final glass-ceramic would increase when the size
of Li2Si2O5 reduced. Prolonging the annealing time also introduced
internal stress, which would be released after the third stage treatment.
The sample treated for 48 h at the first stage finally showed a perfect
bending strength of 581 ± 25 MPa and a highest real in-line transmis-
sion of 29% at 650 nm. Additionally, with excessive extension of holding
time (72 h), small amount of Li2Si2O5 and SiO2 crystals may precipitate
at the first stage, which would slightly reduce the bending strength and
the translucency of finally formed glass-ceramics.
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
Fig. 7. Real in-line transmission spectra of all samples.
Acknowledgments
This work was supported by the National Natural Science Foundation
of China (Grant Nos. 51272205, 51072157, A3 Foresight Program-
51161140399), by the High-Tech R&D Program of China (863, No.
2011AA030102), by the Doctoral Fund of Ministry of Education of
China (Grant No. 20100201110036), and by the Science and Technology
Bureau of Yulin (Grant No. 2011YLZX04).
References
[1] P.W. McMillan, The crystallisation of glasses, J. Non-Cryst. Solids 52 (1982)
67–76.
[2] D.G. Burnett, R.W. Douglas, Nucleation and crystallisation in soda–baria–silica
system, Phys. Chem. Glasses 12 (1971) 117–124.
[3] P.F. James, Glass-ceramics: new compositions and uses, J. Non-Cryst. Solids 181
(1995) 1–15.
[4] P. Riello, P. Canton, N. Comelato, S. Polizi, M. Verita, G. Fagherazzi, Nucleation and
crystallization behaviour of glass-ceramic material in the Li2O–Al2O3–SiO2 system
of interest for their transparency properties, J. Non-Cryst. Solids 288 (2001)
127–139.
[5] W.D. Scott, J.A. Pask, Nucleation and growth of sodium disilicate crystals in sodium
disilicate glass, J. Am. Ceram. Soc. 44 (1961) 181–187.
[6] M. Guazzato, M. Albakry, S.P. Ringer, M.V. Swain, Strength, fracture toughness and
microstructure of a selection of all-ceramic materials, part I, Dent. Mater. 20
(2004) 441–448.
105
[7] H. Fischer, R. Marx, Fracture toughness of dental ceramics: comparison of bending
and indentation method, Dent. Mater. 18 (2002) 12–19.
[8] L.C. Sobrinho, M.J. Cattell, J.C. Knowles, Fracture strength of all-ceramic crowns, J.
Mater. Sci. Mater. Med. 9 (1998) 555–559.
[9] W. Höland, M. Schweiger, M. Frank, A comparison of the microstructure and prop-
erties of the IPS Empress2 and the IPS Empress glass-ceramics, J. Biomed. Mater.
Res. A 53 (2000) 297–303.
[10] H.R. Fernandes, D.U. Tulyaganov, I.K. Goel, J.M.F. Ferreira, Crystallization process and
some properties of Li2O–SiO2 glass-ceramics doped with Al2O3 and K2O, J. Am.
Ceram. Soc. 91 (2008) 3698–3703.
[11] G.A. Khater, M.H. Idris, Role of TiO2 and ZrO2 on the crystallizing phases and micro-
structure in Li, Ba aluminosilicate glass, Ceram. Int. 33 (2007) 233–238.
[12] S.C.V. Clausbruch, M. Schweiger, W. Höland, V. Rheinberger, The effect of P2O5 on
the crystallization and microstructure of glass-ceramics in the SiO2–Li2O–K2O–
ZnO–P2O5 system, J. Non-Cryst. Solids 263 (2000) 388–394.
[13] W. Höland, E. Apel, C.V. Hoen, V. Rheinberger, Studies of crystal phase formations in
high-strength lithium disilicate glass-ceramics, J. Non-Cryst. Solids 352 (2006)
4041–4050.
[14] P. Goharian, A. Nemati, M. Shabanian, A. Afshar, P. Goharian, A. Nemati, M.
Shabanian, Properties, crystallization mechanism and microstructure of lithium
disilicate glass-ceramic, J. Non-Cryst. Solids 356 (2010) 208–214.
[15] M.P. Borom, A.M. Turkalo, R.H. Doremus, Strength and microstructure in lithium
disilicate glass-ceramics, J. Am. Ceram. Soc. 58 (1975) 385–391.
[16] X. Zheng, G. Wen, L. Song, X.X. Huang, Effects of P2O5 and heat treatment on crystal-
lization and microstructure in lithium disilicate glass ceramics, Acta Mater. 56
(2008) 549–558.
[17] H.L. McCollister, S.T. Reed, Glass ceramic seals to inconel, U.S. Patent 4414282 A.
1983;11–8.
[18] L.L. Burgner, P. Lucas, M.C. Weinberg, P.C. Soares Jr., E.D. Zanotto, On the persistence
of metastable crystal phases in lithium disilicate glass, J. Non-Cryst. Solids 274
(2000) 188–194.
[19] A.S. Rizkalla, D.W. Jones, Mechanical properties of commercial high strength ceramic
core materials, Dent. Mater. 20 (2004) 207–212.
[20] Y. Iqbal, W.E. Lee, D. Holland, P.F. James, Metastable phase formation in the early
stage crystallisation of lithium disilicate glass, J. Non-Cryst. Solids 224 (1998) 1–16.
[21] C.C. Lin, P. Shen, H.M. Chang, Y.J. Yang, Composition dependent structure and
elasticity of lithium silicate glasses: effect of ZrO2 additive and the combination of
alkali silicate glasses, J. Eur. Ceram. Soc. 26 (2006) 3613–3620.
[22] H.R. Fernandes, D.U. Tulyaganov, The role of P2O5, TiO2 and ZrO2 as nucleating
agents on microstructure and crystallization behaviour of lithium disilicate-based
glass, J. Mater. Sci. 48 (2013) 765–773.
[23] P.F. James, Y. Iqbal, U.S. Jais, Crystallisation of silicate and phosphate glasses, J.
Non-Cryst. Solids 219 (1997) 17–29.
[24] H.A. Elbatal, Z.S. Mandouh, H.A. Zayed, S.Y. Marzouk, G.M. Elkomy, A. Hosny, Hatem A.
Elbatal, Zeinab S. Mandouh, Hamdia A. Zayed, Thermal, structure and morphological
properties of lithium disilicate glasses doped with copper oxide and their glass-
ceramic derivatives, J. Non-Cryst. Solids 358 (2012) 1806–1813.
[25] E. Apel, J. Deubener, A. Bernard, Phenomena and mechanisms of crack propagation
in glass-ceramics, J. Mech. Behav. Biomed. Mater. 1 (2008) 313–325.
[26] M.H. Lewis, J. Metcalf-Johansen, P.S. Bell, Crystallization mechanisms in glass-
ceramics, J. Am. Ceram. Soc. 62 (1979) 278–288.
[27] G.C. Wei, Transparent ceramics for lighting, J. Eur. Ceram. Soc. 29 (2009)
237–244.
[28] G.H. Beall, D.A. Duke, Transparent glass-ceramics, J. Mater. Sci. 4 (1969) 340–352.
[29] M.J. Heffernan, S.A. Aquilino, A.M. Diaz-Arnold, Relative translucency of six all-
ceramic systems, part I: core materials, J. Prosthet. Dent. 88 (2002) 4–9.