Journal of Non-Crystalline Solids 355 (2009) 141–147

Contents lists available at ScienceDirect

Journal of Non-Crystalline Solids

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

Phase separation, crystallization and leaching of microporous glass ceramics

in the CaO–TiO2–P2O5 system
M. Kord, V.K. Marghussian *, B. Eftekhari-yekta, A. Bahrami
Ceramics Div., Dept. of Materials, Iran University of Science and Technology, Narmak, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: Microporous glass ceramics belonging to the CaO–TiO2–P2O5 system were prepared with the assumption
Received 25 April 2008 of a 2:1 mole ratio for b-Ca3(PO4)2:CaTi4(PO4)6, the anticipated crystalline phases in the end product. The
Received in revised form 26 September glasses formulated according to the above composition were melted and cast onto a steel mold and were
2008
crystallized to glass ceramics containing the above phases. Dilatometric/differential thermal analysis
Available online 25 November 2008
(DTA) techniques were utilized to determine the appropriate phase separation–nucleation and crystalli-
zation temperatures. The crystalline products and resulting microstructures in various stages of process
Keywords:
were determined and observed by X-ray diffraction (XRD) and scanning electron microscopy (SEM). By
Crystallization
Glass ceramics
leaching the resulting glass ceramics in HCl, b-Ca3(PO4)2 was dissolved out leaving a porous skeleton
Nucleation of CaTi4(PO4)6. It was found that the volume porosity, specific surface area and mean pore diameter of
Microstructure microporous glass ceramics can be managed through the proper selection of heat treatment conditions.
Nanocrystals In the optimized conditions for fabricating glass ceramics of minimum mean pore size the values of
41 ± 4%, 26 ± 3 m2/g and 14.3 ± 2 nm were obtained for porosity, surface area and pore diameter
respectively.
Ó 2008 Elsevier B.V. All rights reserved.

1. Introduction Although many technical aspects of the process including the

fabrication, details of crystallization, acid leaching and effect of
Porous ceramics have attracted considerable attention in recent composition changes on properties has been investigated and re-
years, and are currently used in various engineering applications as ported in literature [6–11], some aspects of the whole process have
membranes [1], supporting materials for immobilization of en- remained not fully explained and unclear, e.g. despite the marked
zymes [2], catalyst supports [3] and sensors [4]. On the other hand effect of the occurrence of glass-in glass phase separation by
microporous glass ceramic materials, have many advantages over spinodal decomposition mechanism upon the process, it has never
conventional porous ceramics such as a more convenient produc- substantiated experimentally, the scientific basis for the determi-
tion technology, better control over the size and distribution of nation of nucleation and crystallization temperatures and using
pores and the possibility of producing crystalline skeletons of dif- unusually long holding times at these temperatures are unclear,
fering compositions, structures and functionalities. the details of crystallization process has not been explained, the
These materials have also some advantages in comparison with microstructural studies are rare and the effect of temperature
the earlier Vycor porous glasses [5], regarding the chemical and and time on the leaching of glass ceramics has not been studied.
mechanical properties, easier production routes and more versatil- In this paper an effort was made to clarify and describe some as-
ity in the skeletal material properties. pects of the production processes, especially the phase separation,
The porous glass ceramics are usually prepared utilizing crystallization and leaching, in a more comprehensive manner.
phosphate glasses which are mainly located in CaO–TiO2–P2O5 or
related systems. The glasses after undergoing a spinodal type 2. Experimental procedure
phase separation (like Vycor glasses) are heat treated to develop
crystalline phases such as b-Ca3(PO4)2 and CaTi4(PO4)6. The former The studied glass composition, formulated with a b-Ca3(PO4)3:-
phase is usually leached out, leaving behind a porous skeleton of CaTi4(PO4)6 mole ratio of 2:1, is presented in Table 1.The raw mate-
CaTi4(PO4)6 [6–8]. rials used to make the glass were regent grade CaCO3,TiO2 and
P2O5. The glasses were melted in alumina crucibles at 1350 °C in
* Corresponding author. Tel.: +98 21 73912814; fax: +98 21 77240480. an electric furnace for 2 h and were cast onto preheated steel
E-mail address: v_k_marghus@yahoo.com (V.K. Marghussian). moulds. The glass specimens then were annealed at 600 °C for

0022-3093/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jnoncrysol.2008.10.006
142 M. Kord et al. / Journal of Non-Crystalline Solids 355 (2009) 141–147

2 h and furnace cooled. The glass specimens after crystallization 3. Results

heat treatments were leached in 1 M HCl solution in order to dis-
solve the more soluble, b-Ca3(PO4)2 phase and to produce a porous 3.1. Phase separation
skeleton of CaTi4(PO4)6.
Differential thermal analysis (DTA) were performed on the glass As the annealed glass specimens had not shown clearly obser-
specimens (polymer laboratories STA-1640) using samples of vable phase separation, even in very high magnification SEM
<65 lm and 0.5–0.6 mm particle size ranges in order to determine examination, it was decided to carry out a heat treatment opera-
the crystallization temperature and the susceptibility to bulk crys- tion in order to initiate or intensify the separation process in glass
tallization.In DTA runs alumina was used as an inert reference
material and the heating rate was 10 °C min 1.
Glass transition temperature (Tg) and dilatometric softening
point (Td) were also determined by dilatometer (Netzsch 402E)
and used to demarcate the glass-in glass phase separation and
nucleation temperature ranges. Three specimens were used for
each DTA test.
X-ray diffraction (XRD) was used in order to identify the crystal-
lization products in heat treated specimens (Siemens-D500). Cu Ka
radiation was used in XRD examinations in the 2h range of 10–
100°.
The samples after polishing and etching were coated with a thin
film of gold and subjected to SEM examination (scanning electron
microscope, Philips-XL-30). Some fractured surfaces of porous
specimens were also examined.
The specific surface area and the mean pore size of porous spec-
imens were measured utilizing the Nitrogen adsorption and
desorption technique (Micromeritics-2000, ASAP).Three separate
specimens were used for the test.

Table 1
Nominal chemical composition of the glass specimen.

Oxide P2O5 CaO TiO2

Mol% 31.25 43.75 25 Fig. 2. DTA traces for glass specimens of different particle sizes (a) <75 lm, (b) <0.5–
0.6 mm.

Fig. 1. SEM micrographs showing phase separation in glass specimen after heating at 710 °C (a) 6 h, (b) 18 h, (c) 24 h.
 M. Kord et al. / Journal of Non-Crystalline Solids 355 (2009) 141–147 143

Fig. 3. XRD patterns for specimens heat treated at differing temperatures for 4 h
after nucleation at 710 °C for 24 h (a) 745 °C, (b) 755 °C, (c) 765 °C.

specimens. Dilatometric technique was used to determine the

suitable heat treatment range for the phase separation process
which usually occurs in glass transition (Tg)-dilatometric softening
temperature (Td) interval. The above temperatures were deter-
mined as 700 ± 3 and 720 ± 3 °C for Tg and Td, respectively. Fig. 4. XRD patterns for specimens nucleated at 710 °C for differing times (a–d)
SEM micrographs showed phase separation in glass specimens followed by a heat treatment at 765 °C for 48 h (a) 6 h, (b) 12 h, (c) 18 h, (d) 24 h.

after heating at 710 °C for differing times (Fig. 1(a), (b) and (c).
The above temperature was chosen at the middle of Tg and Ts inter-
val. Obviously after a 6 h hold at 710 °C the signs of phase separa-
tion are barely observable, whereas after 18 h a very fine to crystallization in glass specimens. The difference of 50 °C be-
microstructure revealing the occurrence of a phase separation with tween the two first peak temperatures of DTA curves i.e. 790 ± 4
an apparently spinodal decomposition mechanism is observed and 840 ± 4, is an indication of the relative importance of surface
which after a soaking time of 24 h became quite distinct. crystallization in these specimens. The much smaller second DTA
exo-peaks are related to Ca2 P2O7 and TiO2 irrelevant to the present
3.2. Crystallization study.
The preference of bulk crystallization in this work necessitated
Fig. 2 shows DTA traces for specimens possessing fine and the use of bulk specimens and the selection of the corresponding
coarse particle sizes. The DTA exothermic peaks usually correspond exo-peak temperature of 840 °C.

Fig. 5. SEM micrographs of specimens nucleated at 710 °C for (a) 12 h and (b) 24 h followed by a heat treatment at 765 °C for 48 h.
144 M. Kord et al. / Journal of Non-Crystalline Solids 355 (2009) 141–147

In order to follow the crystallization process in these glass spec- Fig. 9 depicts SEM microstructures of some heat treated speci-
imens, first the suitable nucleation temperature and time should mens. The crystals that started to grow after 45 min within the
be found. phase separated glass specimen (Fig. 9(c)), were spread rapidly
Since specimens had usually been soaked at a temperature after 60 min (Fig. 9(d)) and began to exhibit different morphologies
around 700 °C for relatively long times for ‘nucleation’ purposes i.e. fibrous and nearly equiaxed after 90 and 120 min (Fig. 9(e,f)).
in previous investigations [6–8], it was supposed that the occur-
rence of an effective nucleation in these specimens required a pre- 3.3. Leaching
cursor phase separation process to occur prior to the actual
nucleation. Hence, first a tentative ‘nucleation’ heat treatment at The leaching of crystallized specimens carried out in 1 M HCl
710 °C for 24 h, the condition of occurrence of distinct phase sepa- solution at differing temperatures up to 7 days were followed by
ration, were considered for the nucleation process. measuring the bulk densities vs. time.
Since the specimens heated at DTA exo-peak temperature of Fig. 10 depicts the results. It can be deduced that the leaching
840 °C subsequent to the nucleation treatment, developed very temperature in the range of ambient temperature 60 °C has no
coarse microstructures it was decided to choose a lower tempera- marked effect upon the leaching process.
ture interval of Tc (Tc 50) for crystallization. Where Tc is DTA exo-
thermic peak onset temperature determined as 785 °C for the
coarse particle size specimen from DTA curve (Fig. 2) In order to
find the suitable crystallization temperature, specimens first were
subjected to the aforementioned ‘nucleation’ treatment, and then
soaked for 4 h at differing crystallization temperatures in the above
interval.
Fig. 3 depicts the XRD patterns of some specimens heat treated
according to the above procedure. Since the microstructures of
glass ceramics controls the pore sizes of final products, in order
to keep the pore diameters in the as low as possible range the low-
est temperature showing distinct crystallization in XRD examina-
tions, i.e. 765 °C, was chosen as the ‘crystallization’ temperature.
In order to assess the actual occurrence of nucleation in the
above conditions, the glass specimens after a nucleation treatment
at 710 °C for differing times (6–24 h), were heated at 765 °C for
48 h and then subjected to XRD examination.
Fig. 4 represents the XRD patterns of the above specimens. It
can be seen that all XRD peaks for the major phases b-Ca3(PO4)2
and CaTi4(PO4)6 and the minor phase Ca(PO3)2 have continu-
ously been intensified by prolongation of the nucleation time
up to 24 h. On the other hand the SEM microstructures of two
specimens nucleated for 24 and 12 h in the above conditions
(Fig. 5) show that the specimen held at the nucleation temper-
ature for 24 h developed a much finer microstructure in com-
parison with the specimen soaked for 12 h, indicating the
formation of much greater population of nuclei in the former
specimen.
Moreover, in order to determine the optimum crystallization
time for the specimens, they were first nucleated for 24 h at
710 °C and subsequently soaked for differing times at 765 °C.
Fig. 6 shows the XRD patterns for the above specimens. Although
it seems that the crystallization products continuously increased
from 8 to 48 h the difference shown between 24 and 48 h is not
great. On the other hand the microstructural examination by
SEM revealed the occurrence of considerable growth and coarsen-
ing of crystals in the period of 24–48 h (Fig. 7).
Therefore in order to keep the pore sizes in the as low as possi-
ble range, it was decided to choose 24 h as the holding time for
crystal growth.
In order to follow the steps of crystallization process, it was
decided to carry out some short time heat treatments on the glass
specimens. In these treatments the specimens after a 24 h nucle-
ation at 710 °C, were heated at 765 °C for 15, 30, 45, 60, 90 and
120 min.
Fig. 8 exhibits the XRD patterns of the heat treated specimens. It
can be seen that after 45 min, considerable crystallization has oc-
curred in the glass specimens that continuously increased up to
120 min. It is interesting to note that the type of crystallization
products, namely b-Ca3(PO4)2 and CaTi4(PO4)6 as the major phases Fig. 6. XRD patterns for specimens nucleated at 710 °C for 24 h and subsequently
and Ca(PO3)2 as the minor phase, were not changed throughout heated at 765 °C for differing times (a) 8 h, (b) 16 h, (c) 24 h, (d) 32 h, (e) 40 h, (f)
these experiments. 48 h.
 M. Kord et al. / Journal of Non-Crystalline Solids 355 (2009) 141–147 145

Fig. 7. SEM micrographs for specimens nucleated at 710 °C for 24 h and subsequently heated at 765 °C for (a) 24 h and (b) 48 h.

Fig. 11 shows the microstructure of a highly porous specimen

after leaching.
Fig. 12 shows the XRD pattern of the porous specimen. It can be
seen that CaTi4(PO4)6 and Ca(PO3)2 are the major and minor phases
remained after the leaching process and there is no evidence for
the presence of b-Ca3(PO4)2 phase.
Table 2 summarizes some physical properties of the specimen.

4. Discussion

On the basis of the results described above it can be inferred

that the most critical step in the preparation of these glass ceram-
ics is the phase separation process. The occurrence of phase sepa-
ration by an apparently spinodal decomposition mechanism
(Fig. 1(c)) in the glass specimen provides the required conditions
for the formation of interconnected crystalline collections of main
phases,Ca3(PO4)2 and CaTi4(PO4)6 in the respective phase separated
regions of glasses enriched in CaO and TiO2 respectively. Without
these interconnected phases the whole process of leaching and fab-
rication of bodies possessing high values of fine nano-size open
porosity would be impossible. On the other hand the comparison
of the microstructure of a glass ceramic specimen which prior to
the crystallization was subjected to a heat treatment at the phase
separation temperature of 710 °C for 24 h with a specimen heated
at the same temperature for 12 h (Fig. 5), indicates the existence of
a much finer microstructure and hence the formation of much
greater number of nuclei at the former specimen. This also resulted
in much higher bulk crystallization values as proved by XRD pat-
terns shown in Fig. 4. Therefore it can be inferred that an actual
nucleation process was progressing along with the glass-in glass
phase separation at 710 °C in these glasses. On the other hand
the temperature interval between Tg and Td in which the occur-
rence of a suitable phase separation becomes possible is quite
restricted. Lower temperatures would halt the already slow phase
separation process and the acceleration of the phase separation
process by raising the temperature may be detrimental to whole
process, since this may disrupt the interconnected spinodal micro-
structure and convert it to a microstructure of isolated droplets
distributed within a matrix. Hence it can be concluded that the rel-
atively slow phase separation process (Fig. 1) is mainly responsible
for the slow nucleation rates in these glass ceramics. The crystalli-
zation (growth) temperature is another important factor that re-
Fig. 8. XRD patterns for specimens nucleated at 710 °C for 24 h and subsequently
quires careful consideration. Although DTA exo-peak temperature
heated at 765 °C for differing times (a) 15 min, (b) 30 min, (c) 45 min, (d) 60 min, (e) of 840 °C (Fig. 2), can be taken as the suitable crystallization
90 min, (f) 120 min. temperature, the relatively coarse microstructures developed in
146 M. Kord et al. / Journal of Non-Crystalline Solids 355 (2009) 141–147

Fig. 9. SEM micrographs for specimens nucleated at 710 °C for 24 h and subsequently heated at 765 °C for (c) 45 min, (d) 60 min, (e) 90 min, and (f) 120 min.

specimens heated at this temperature necessitated the use of low- although the condition is more flexible in the case of growth. In
er temperatures, around the exothermic peak onset temperature this case a relatively limited temperature range, located between
(Fig. 2) e.g. 765 °C as described above. The relatively slow growth To, several degrees lower than DTA crystallization temperature of
rate encountered at the chosen temperature in this experiment Tc = 840 °C and Tf = 765 °C can be imagined within which the vari-
necessitated the use of long crystallization times, (at least 24 h). ation of pore size by the variation of crystallization temperature
Increasing the temperature could lead to considerable coarsening becomes practically possible and useful.
of the crystals and larger pore in the final product. Therefore it The comparison of Fig. 9(f) with Fig. 5(b), showing the crystal-
seems that in the case of growth process, similar to the nucleation lization of specimens after 2 and 24 h heat treatment respectively,
process, a relatively restricted condition exists in respect to the reveals that the equiaxed particles eventually adopted plate like
selection of suitable time/temperature for the heat treatment, shapes, whereas the fibrous particles preserved their fibrous or
rod-shape morphology.
Also the comparison of the microstructure of porous specimens
(Fig. 11) with microstructure of crystallized specimens before
leaching (Fig. 5(b) or Fig. 7(a)) reveals that the rod-shape particles
are those which remained in the porous specimen after the leach-

Fig. 10. Bulk density vs. time curves for leaching glass ceramic specimens at (a) Fig. 11. SEM micrograph showing the porous specimen after leaching (fractured
ambient temperature, (b) 40 °C, and (c) 60 °C. surface).
 M. Kord et al. / Journal of Non-Crystalline Solids 355 (2009) 141–147 147

ature, as a inherent difficulty of this fabrication method in control-

ling the characteristics of the resulting porous material through
variation of heat treatment conditions it seems that only the com-
positional changes leading to higher mobility at relatively low tem-
peratures for the ions taking part in phase separation process, with
no adverse effect on the spinodal mechanism, could bring about
the useful high separation and nucleation rates. The same problem
is encountered in growth process. The relatively slow growth rate
occurring at the chosen temperature in this experiment necessi-
tated the use of long crystallization times, (at least 24 h). Increas-
ing the temperature usually leads to considerable coarsening of
the crystals and larger pores in the final product. This problem
probably may also be solved by selecting a suitable composition,
as discussed above.

Fig. 12. XRD pattern of the glass ceramic specimen after leaching. 5. Conclusions

It was shown that glass-in glass phase separation with a spinod-

Table 2
Physical properties of porous glass ceramics.
al decomposition mechanism is the precursor step for the occur-
rence of an effective nucleation, in the glass ceramics studied
Bulk density (g/cm3) Volume porosity (%) Specific surface Average pore herein. In this way phase separation plays a very decisive role in
area (m2/g) diameter (nm)
controlling the crystallization process and affecting the micro-
1.66 ± 0.05 41 ± 4 26 ± 3 14.3 ± 2 structure of the glass ceramics, hence determining the size and
vol.% of the final porosity in the end product. The optimum phase
separation–nucleation temperature and time determined as 710 °C
ing operation i.e. CaTi4(PO4)6 and the plate like particles are b- and 24 h, in fact should be considered as an almost invariable con-
Ca3(PO4)2 crystals which were leached out along with some resid- dition for the fabrication of these type of porous materials (if com-
ual glass phase. Fig. 11 also reveals that the seemingly rod-shape position was kept constant) with acceptable properties and nano-
particles of CaTi4 (PO4)6 are in fact comprised of many tiny crystals. size porosity. The chosen crystallization (growth) condition of
This can be explained by the crystallization mechanism exhibited 765 °C and 24 h was also proved to be quite effective in producing
in Fig. 5(a), which is the manifestation of a spherulitic growth. In bodies of acceptable values of porosity (41 ± 4%) and quite small
this mechanism several crystals begin to grow from an identical (perhaps the minimum possible) mean pore diameter of
central nucleus in the form of radiating fibers, which branch in var- 14.3 ± 2 nm. The obtained glass ceramics after leaching in 1 M
ious directions and soon covered with numerous tiny crystallites HCl solution at the ambient temperature, 40 °C and 60 °C for up
[12]. to 7 days did not show any considerable differences in the leaching
In close observation of Fig. 11 it seems that pores seen in this rate or final bulk density and volume porosity values.
figure, which are mainly in the 70–200 nm diameter range, are
much larger than the measured mean pore diameter of 14.3 nm. References
This can be explained by the existence of numerous fine pores
[1] T. Yazawa, H. Tanaka, in: K. Ishizaki (Ed.), Porous Materials, American Ceramic
on the external and internal pore walls, which are observable with Society, Westerville, OH, 1993, p. 213.
higher magnifications and is consistent to the findings of other [2] R.A. Messing, J. Non-Cryst. Solids 26 (1977) 482.
investigators [6]. [3] I. Aso, M. Nakao, N. Yamazoe, T. Seeiyama, J. Catal. 57 (1979) 287.
[4] M. Uo, M. Numata, M. Suzuki, I. Karube, A. Mkishima, J. Ceram. Soc. Jpn. 100
Although it was not possible to differentiate between the crys-
(1992) 430.
tallization steps of different phases, from Figs. 9 and 11 it can be [5] H.P. Hood, M.E. Nordberg, US patent 2, 106, 774 (1934).
inferred that both major phases begin to crystallize simultaneously [6] H. Hosono, Z. Zhang, Y. Abe, J. Am. Ceram. Soc. 72 (1989) 1587.
from different phase separated regions, rich in CaO or TiO2, and re- [7] H. Hosono, Y. Sakai, M. Fasano, Y. Abe, J. Am. Ceram. Soc. 73 (1990) 2536.
[8] H. Hosono, Y. Abe, J. Am. Ceram. Soc. 75 (1992) 2862.
mained confined within their respective boundaries to the end of [9] H. Hosono, Y. Abe, Non-Cryst. 190 (1995) 185.
crystallization process. [10] K. Yamamoto, T. Kasuga, Y. Abe, J. Am. Ceram. Soc. 80 (1997) 822.
[11] T. Kasuga, M. Nogami, Y. Abe, J. Am. Ceram. Soc. 82 (1999) 765.
[12] V.K. Marghussin, A. Arjomandnia, Phys. Chem. Glasses 40 (1999) 311.
4.1. Suggestion for further work

On the basis of the above discussion regarding the restricted

conditions for selection of nucleation and growth time and temper-