ARTICLE IN PRESS

Biomaterials 26 (2005) 2255–2264

www.elsevier.com/locate/biomaterials

A new model formulation of the

SiO2–Al2O3–B2O3–MgO–CaO–Na2O–F glass-ceramics
Simeon Agathopoulosa,, Dilshat U. Tulyaganova, Patricia Valériob, José M.F. Ferreiraa
a
Department of Ceramics and Glass Engineering, University of Aveiro, CICECO, 3810-193 Aveiro, Portugal
b
Department of Biophysics and Physiology, Federal University of Minas Gerais, ICB-UFMG, 31270-901 Belo Horizonte, Brazil
Received 16 December 2003; accepted 12 July 2004
Available online 15 September 2004

Abstract

Mono-phase glass-ceramics of akermanite were successfully produced from a Ca-mica and wollastonite via low-temperature
sintering and crystallization. Doping with P2O5 considerably improves sintering behaviour since P2O5 increases the stability of glass
against crystallization at the temperature of sintering onset. The resulting glass-ceramics feature good in vitro acceptance from
osteoblasts, and moderate bioactivity due to the enrichment of the glassy phase with Ca and Si. The good quality of the white colour
at the surface and throughout the bulk, the matching of microhardness with tooth enamel, and the possibility to coat other
biomaterials such as ZrO2, Ti or hydroxyapatite make these materials promising for medical applications.
r 2004 Elsevier Ltd. All rights reserved.

Keywords: Akermanite; Glass; Glass-ceramics; Properties; Biocompatibility; In vitro

1. Introduction material is generally not desirable because the likely

mismatch of thermal expansion coefficients between
Glass-ceramics have been broadly used as dental the different crystalline phases can cause weakening
restorative and osteoconductive materials due to their of material. To our knowledge, there are no similar
biocompatibility and potential bioactivity [1–4]. For studies reporting single-phase glass-ceramics tested or
instance, apatite-wollastonite (A/W) has been used in used in biomedical applications. In this investigation,
orthopaedic prostheses [5–7], while glass-ceramics based akermanite (Ca2MgSi2O7) was the optimum demanded
on potassium mica, such as Dicors, have been applied phase.
in dental restorations due to their good physical, Akermanite and ghelenite are the end members of
chemical and mechanical properties [8,9]. melilite, which belongs to the group of sorosilicates [10],
In this work, we present a new model system, which having a general formula of X2YZ2O7, where X is
comprises components of boron-containing calcium Ca or Na, Y is Al or Mg, and Z is Si or Si and Al. X
mica (CaMg3Si2Al0.5B1.5O10F2) and wollastonite (Ca- is a large 8-coordinated site, while Y and Z are
SiO3). The most important challenge in this work was to tetrahedral ones. Akermanite and ghelenite form solid
determine the optimum ratio between mica and wollas- solutions with a general formula of 2CaO
(1x)
tonite, which would eventually result in a single crystal- MgO
xAl2O3
(2x)SiO2. Melilites usually occur in
line phase. In such systems, several phases of complex igneous and metamorphic rocks and meteorites, but
structure are usually formed. However, a multi-phase also in the slag of blast furnaces [10,11]. These facts
unequivocally evidence their high stability. Akermanite
Corresponding author. Tel.: +351-234-370242; fax: +351-234- exhibits relatively high refractory properties, melting
425300. point at 1454 1C, density of 2.944 g/cm3, and hardness of
E-mail address: [email protected] (S. Agathopoulos). 56 in Moose scale.

0142-9612/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biomaterials.2004.07.030
 ARTICLE IN PRESS
2256 S. Agathopoulos et al. / Biomaterials 26 (2005) 2255–2264

The mechanism, whereby akermanite would be Table 1

produced from mica and wollastonite, is assumed to Chemical compositions of the investigated glasses
be as follows. Micas consist of two sheets made of Glass SiO2 Al2O3 B2O3 CaO MgO P2O5 Na2O CaF2
tetrahedral units, sandwiching one sheet made of
octahedral structural units. The tetrahedral structural MW (wt%) 42.51 2.00 4.10 30.86 9.51 — 4.87 6.14
units of SiO44 exhibit high structural stability, due to (mol%) 40.91 1.14 3.41 31.82 13.64 — 4.54 4.54
MWP (wt%) 41.13 1.94 3.97 29.86 9.20 3.24 4.72 5.94
the four hybridized sp3 orbitals of Si4+, which can give
(mol%) 40.36 1.12 3.46 31.40 13.45 1.35 4.44 4.44
the same configuration of these structural units in both
the glass melt and the crystals of the glass-ceramic. In
the octahedral layer, where there is coexistence of 4s and
2p bonds, the weaker s-bonds are primarily ruptured surface area of 0.719 m2/g for MW and 0.354 m2/g for
over heating, facilitating diffusion of cations to escape MWP, measured by BET technique (Micromeritics,
the octahedral layer [12]. Hence, the Mg2+ ions, Gemini II 2370, USA). Glasses in bulk form were also
accommodated in the octahedral units of Ca-mica, obtained by casting, followed by appropriate annealing
diffuse easily towards the units of wollastonite during to reduce internal stresses.
heating according to the following simplified chemical Powder of glass frit was granulated by mixing in a
equation: 5 vol% polyvinyl alcohol solution (PVA) (wt% propor-
tion frit/PVA=97.5/2.5). Parallelepiped bars
3MgO þ 7ðCaO
SiO2 Þ ! 3ð2CaO
MgO
2SiO2 Þ (4  5  50 mm3) were prepared by uniaxial pressing
þ CaO
SiO2 : (1) (200 MPa). After debinding at 450 1C for 2 h, the bars of
glass-powder compacts were sintered at several tem-
In this work, to synthesize akermanite, the optimal
peratures between 650 and 920 1C for 1 h at a slow
molar ratio between mica and wollastonite was taken as
heating rate (2–31/min) to avoid deformation. Produc-
1:7. Thus, the basic formula, hereafter denoted as MW,
tion of coatings (thickness 100 mm) was also attempted
was (3MgO
0.25Al2O3
0.75B2O3
2SiO2
CaF2)
7(CaO

as follows. Glass frit, granulated with PVA, was applied
SiO2)
Na2O. The ratio 1:7 was selected because it seems
on pellets of several polished substrates (i.e. zirconia, Ti
that it better approaches the structural and bioactivity
and hydroxyapatite) and crystallized similarly to the
requirements of akermanite glass-ceramics. In particu-
glass compacts at 800 1C for 1 h.
lar, apart the stoichiometry of Eq. (1), i.e. 1:6, the excess
The resulting materials were characterized by the
of wollastonite aimed to react with the other compo-
following techniques: Differential thermal analysis
nents of mica and result in formation of glassy phase.
(DTA, Labsys Setaram TG-DTA/DSC, France, heating
The presence of B and F in the glassy phase was
rate 51/min, in air). Dilatometry (Bahr Thermo Analyse
expected to favour densification of the resulting glass-
DIL 801 L, Germany, heating rate 31/min). X-ray
ceramics, while Ca and Si would potentially enhance
diffraction (XRD, Rigaku Geigerflex D/Mac, C Series,
bioactivity [13].
Cu Ka radiation, Japan). Microstructure observations
A modified composition, denoted as MWP, was
by scanning electron microscopy (SEM, Hitachi S-4100,
derived from MW by doping with P2O5, having a
Japan, 25 kV acceleration voltage) under secondary
formula of (3MgO
0.25Al2O3
0.75B2O3
2SiO2
CaF2)

electron and back scattering modes, equipped with
7(CaO
SiO2)
Na2O
0.3P2O5.
energy dispersive spectroscopy apparatus (EDS, mini-
mum spot diameter 1 mm). The Archimedes method
was used to measure the apparent density of the glass
2. Materials and experimental procedure and glass-ceramic blocks (immersion in ethylenoglycol).
Vickers microhardness was estimated from ten indenta-
2.1. Production of glasses and glass-ceramics and tions for each sample (Shimadzu microhardness tester
characterization techniques type M, Japan, load of 9.8 N). Weibull statistics were
employed to evaluate the mechanical reliability of
Powders of reagent grade of SiO2, Al2O3, CaCO3, materials (i.e. three-point flexural strength of parallele-
H3BO3, 4MgCO3
Mg(OH)2
5H2O, Na2CO3, CaF2 and piped bars, 3  4  40 mm3, Shimadzu Autograph AG
NH4H2PO4 were used. The chemical compositions of 25 TA, 0.5 mm/min displacement). Water absorption
MW and MWP glasses are presented in Table 1. was measured according to the ISO-standard 10545-3,
Homogeneous mixtures of 100 g batches obtained by 1995 (i.e. weight gain of dried bulk samples after
ball milling were preheated at 1000 1C for 1 h for immersion into boiling water for 2 h, cooling for 3 h
decarbonization and then melted in Pt crucibles at and sweeping of their surface with a wet towel). The
1400 1C for 1 h, in air. Glass frit was obtained by linear shrinkage during sintering was also calculated
quenching of the melt in cold water. The frit was dried from the dimensions of the green and the resulting
and then milled. The resulting powders had specific sintered samples.
 ARTICLE IN PRESS
S. Agathopoulos et al. / Biomaterials 26 (2005) 2255–2264 2257

2.2. Immersion in SBF After centrifugation at 1000 g for 5 min, each pellet was
re-suspended in 5 ml of RPMI medium supplemented
The dissolution rates and the possible bioactivity of with 10% FBS, 1% antibiotic-antimycotic. The cells
the glasses and the glass-ceramics were investigated by were seeded into 25 ml tissue culture flasks, and led to
immersion of either powders (pulverized from sintered grow in a controlled 5% CO2, 95% humidified
bulk samples) or bulk materials in simulated body fluid incubator at 37 1C. After confluence, the cells were used
(SBF), at 37 1C. SBF solution has an ionic concentration for experiments on passage 2.
of Na+ 142.0, K+ 5.0, Ca2+ 2.5, Mg2+ 1.5, Cl 147.8, To test the stimulation of osteoblasts with a medium
2 2
HCO 3 4.2, HPO4 1.0, SO4 0.5 (in mM), buffered at containing glass powder, osteoblasts were platted
pH=7.25 by tris-hydroxymethyl-amminomethane (Tris, 1  105 cell density, in 24 well plates. After 2 h, the
50 mM) and hydrochloric acid [14]. Sampling took place medium was changed to a medium containing glass
at 1, 3, 6, 10, 24 (1 d), 72 (3 d), 168 (7 d), 336 (14 d), and powder (i.e. MW or MWP; grain size of powder-
744 (31 d) h (d: days). The obtained results were o35 mm). The concentration used was 0.005 g powder/
independent, which means that each sample was 1 ml of culture medium. After 72 h of incubation,
individually treated without interfering with others. osteoblasts’ morphology, viability/proliferation and
An amount of glass powder corresponding to 1 m2 secretion capability (i.e. collagen, alkaline phosphatase)
surface area was immersed in 10 ml of filtered (through were tested.
sterilized filters, cameo 25 AS-MSI, pore size 0.22 mm) The cellular proliferation and viability tests were done
SBF solution and immediately sealed into sterilized as follows: After 72 h of incubation in the presence of
plastic flasks, which were stored afterwards at 37 1C each glass powder, osteoblasts’ viability was evaluated
(70.11). After each experiment, the powder was by MTT assay, based on the reduction of tetrazolium
separated from the liquids by filtering. Similar proces- salt to formazan crystals by dehydrogenase present in
sing was followed for the bulk materials. In this case, the living cells mitochondria. The formazan salt formation
response of material in SBF was tested at both fracture is directly related to the amount of dehydrogenase,
and polished surfaces (mirror finishing, i.e. polishing providing an indirect measurement of cell proliferation
with 1 mm diamond paste, and ultrasonic agitation). [16]. In total, 60 ml of MTT (5 mg/ml) were added to
The analysis of the liquids comprised pH measure- each well. Two hours later, the cell morphology was
ments and determination of the concentration of Ca2+, analysed by inverted optical microscopy and formazan
Mg2+, B3+, P5+, Si4+, Na+, and Al3+ by inductively salts were solubilized with SDS 10% HCl. After
coupled plasma-optical emission spectroscopy (ICP- incubation for 18 h, optical density was measured at
OES, Jobin Yvon, JY 70 plus, France). The dried 595 nm [17].
samples were examined by SEM/EDS. The alkaline phosphatase production was evaluated
by BCIP-NBT assay. This assay is based on a
2.3. In vitro tests with osteoblasts chromagenic reaction initiated by the cleavage of the
phosphate group of BCIP by alkaline phosphatase
For the in vitro tests with osteoblasts the following present in the cells. This reaction produces a proton,
materials were used. Penicillin, streptomycin, fetal which reduces NBT to an insoluble purple precipitate. In
bovine serum, Dulbecco’s phosphate buffered saline, brief, the supernatant of each well was removed and the
trypsin-EDTA, [3(4,5 dimethylthiazol-2yl) 2,5 diphenyl- cell layer was rinsed twice with PBS. Then, 200 ml of
tetrazoliumbromide] MTT, BCIP-NBT kit: Gibco (Bur- BCIP-NBT solution, prepared according to manufac-
lington, Ont., Canada). Crude bacterial collagenase: turer protocol, were added to each well. After 2 h of
Boehringer (Biberach, Germany). RPMI Cell culture incubation, the cells were observed by optical micro-
medium: Sigma (St. Louis, USA), SIRCOL kit: Biocolor scopy and the insoluble purple precipitates were solubi-
(Newtonabbey, N. Ireland) T25 culture flasks and 24 lized with 210 ml of SDS 10% HCl and incubated for
well plates: Nunc products (Naperville, USA). 18 h. The optical density was measured at 595 nm [18].
Osteoblasts were isolated from the calvaria of 1–5 The osteoblast collagen production was analysed by
days old neonatal Wistar rats [15]. The calvaria were SIRCOL assay in the cultures’ supernatant, following
dissected and freed from soft tissue, cut into small pieces the manufacturer instructions. This method is based on
and rinsed in phosphate-buffered saline without calcium the selective binding property of the syrius-red dye to
and magnesium. The calvaria pieces were incubated with the [Gly–X–Y] tri-peptide end sequence of mammalian
1% trypsin-EDTA for 5 min, followed by four sequen- collagen [19]. The collagen present in the supernatant,
tial digestions with 2% collagenase at 37 1C for 45 min precipitated by syrius-red, was solubilized and measured
each. The supernatant of the first collagenase digestion, by an optical density analysis at 595 nm. The collagen
which contains a high proportion of periosteal fibro- concentration was calculated basing on a linear regres-
blasts, were discarded. The other digestions produced a sion from previously known concentrations of type I
suspension of cells with high proportion of osteoblasts. collagen and their optical density measurement.
 ARTICLE IN PRESS
2258 S. Agathopoulos et al. / Biomaterials 26 (2005) 2255–2264

The results are presented as Means7SD (the numbers lower temperatures. Densification is always higher and
of the experiments are mentioned in the legend of the safely occurs until 800 1C, reaching a maximum of
figure). The statistical significance was measured by density (2.89 g/cm3) and shrinkage (13.8%) at 750 1C.
ANOVA and Bonferroni’s post-test. Linear shrinkage values were remarkably constant over
the investigated temperature range. At 650 1C, powdered
frit of MWP achieved higher density than that of bulk
3. Results and discussion glasses, whose density was measured as 2.80 g/cm3 for
both MW and MWP. The dramatic decrease of
3.1. Properties mechanical reliability of MW glass-ceramics sintered at
temperatures X800 1C is characteristically depicted in
The thermal analysis of the glasses, summarized in Fig. 2a, which is in accordance with the values of Table
Fig. 1, provided the necessary information for setting up
the optimum crystallization schedule. For both MW and
MWP glasses, dilatometry measurements showed that Table 2
Density and linear shrinkage of compacts made of glass powders and
the transition temperature (Tg) was at 625 1C and the
sintered at different temperatures
softening point at 670 1C. Tg was detected by DTA at
slightly higher temperature (640 1C). The DTA Sample Sintering temperature (1C)
showed a single exothermic peak for crystallization with
650 700 750 800 825 850 900 920
maximum at 780 1C for MW and 797 1C for MWP. The
decomposition of PVA was also detected at 400 1C Density (g/cm ) 3

(endothermic peak). Accordingly, after debinding MW 2.51 2.82 2.74 2.77 2.46 2.34 2.13 2.04
(450 1C, 2 h), sintering of bars of glass-powder compacts MWP 2.83 2.88 2.89 2.86 2.78 2.67 2.50 2.09
Linear shrinkage (%)
was carried out starting at 650 1C, i.e. slightly lower than
MW 4.4 12.7 12.2 12.6 10.8 10.0
the dilatometric softening point. Then, sintering of these MWP 13.6 13.6 13.8 13.6 13.5 13.3
samples was conducted at seven different temperatures,
i.e. 700, 750, 800, 825, 850, 900 and 920 1C.
The influence of sintering temperature on density and
linear shrinkage as well as the mechanical properties of
the resulting glass-ceramics are presented Table 2 and 2
Fig. 2, respectively. Several results indicate that the 700ºC
modified MWP composition features superior sintering 750ºC
ln{ln[1/(1-Pi)]}

behaviour than MW. In particular, sintering starts at 800ºC

0

y=6.1094x-28.158
R2=0.9892
1.0 -2
670 y=16.105x-74.039
625 y=15.347x-69.953
R2=0.9892
∆l/lo (%)

R2=0.9903
MW 0.5 -4
4.0 4.2 4.4 4.6 4.8
MWP
(a) ln (T)
0.0
797 2
700ºC
780
750ºC
3
ln{ln[1/(1-Pi)]}

Exo
Heat flow (mV)

0 800ºC
Endo

1 y=7.3746x-33.124
2
MWP -2
R =0.9885

y=8.0658x- 38.338
y=26.679x-125.94
-1 2
R =0.9884 2
R =0.9779
-4
MW 396 641
3.9 4.1 4.3 4.5 4.7 4.9
-3 (b) ln (T)
0 200 400 600 800 1000
Temperature (ºC) Fig. 2. Weibull statistics of flexural strength (3 point bending) of glass-
ceramics obtained from (a) MW and (b) MWP glasses at different
Fig. 1. Thermal analysis (dilatometry top curves and DTA bottom crystallization temperatures (T: bending strength, P: fracture prob-
curves) of MW and MWP glasses. ability).
 ARTICLE IN PRESS
S. Agathopoulos et al. / Biomaterials 26 (2005) 2255–2264 2259

2. On the other hand, MWP glass-ceramics sintered akermanite embedded in a glassy phase. The MWP
at 800 1C anticipate the highest mechanical reliability features similar microstructure, but glassy phase is more
(Fig. 2b). From these curves, the maximum average evident. In the case of MW, the dramatic decrease of
flexural strength is calculated at 95 MPa for MW density, shrinkage and mechanical properties recorded
sintered at 750 1C and 110 MPa for MWP sintered at at temperatures4800 1C (Table 2, Fig. 2a), would be
800 1C. attributed to secondary porosity, probably developed
Consequently, the sintering temperature ranges be- after completion of the crystallization process [21].
tween 700 and 800 1C for the MW composition, having a Doping with P2O5 evidently suppressed such detrimental
span of about 1001. The modified MWP composition effects on the properties of the MWP glass-ceramics.
features a broader sintering span of about 1501, since Consequently, P2O5 improved the sintering ability of
remarkable consistency in the values of the properties the glass compacts because it shifted crystallization
were recorded for the materials sintered between 650 towards higher temperatures and broadened the tem-
and 800 1C. For glass-ceramics sintered at 800 1C (1 h) perature range of densification. Owing to the fact that
both materials have homogeneous white colour coarsening usually reduces sintering ability [22,23], the
throughout the entire bulk and the surface. Vickers effect of P2O5 is more sounded since the MWP powder
microhardness was 4.01 (70.02) GPa for MW and 4.43 comprised more coarsened particles than MW. Evi-
(70.66) GPa for MWP. Water absorption was 0.26% dently, P2O5 should considerably influence the diffusion
for MW and 0.16% for MWP. The linear thermal process of the components. Most probably, phosphor-
expansion coefficient (CTE) calculated from the expan- ous oxide should increase the stability of the glass
sion curves (not shown) between 70 and 500 1C was against crystallization at the temperature of sintering
9.4  106 K1 for MW and 9.9  106 K1 for MWP. onset. Earlier studies have reported that small addition
Fig. 3 shows that at 650 1C both materials are still of several oxides, such as P2O5, to certain glasses can
amorphous. Crystallization occurs at X700 1C, resulting impart excellent sintering ability [24,25].
in formation of a single phase, akerminite. A single- Hence, for both the investigated compositions,
phase formation was correctly anticipated from the sintering should have started at temperatures close to
single exothermic peak of DTA (Fig. 1). It should be
mentioned that in this type of systems, heating rate often
affects the temperature of crystallization onset (i.e.
lower heating rate favours evolution of crystalline
phases at lower temperatures) [20].
Comparison of the intensity of the XRD peaks
(Fig. 3) indicates that MWP generally featured lower
crystallinity than MW. This finding agrees fairly well
with the microstructure of glass-ceramics sintered at
800 1C, for 1 h (Fig. 4). The dense and highly crystallized
structure of MW glass-ceramics comprises a network
of interlocking elongated and prismatic crystals of

800 ºC MW
700 ºC

650 ºC

800 ºC
MWP
700 ºC
650 ºC

10 20 30 40 50 60
2θ Fig. 4. Characteristic microstructures of (a) MW and (b) MWP glass-
ceramics sintered at 800 1C for 1 h observed at fracture surfaces of bulk
Fig. 3. Akermanite exclusively forms at X700 1C (JCPDS card of samples after immersion in SBF for 31 days. The insets show clearly
akermanite-synthetic No. 87-0052). Higher crystallinity is observed in that the surface of both materials was homogeneously covered with
the MW than in the MWP composition. submicron particles enriched in Ca-P (verified by EDS).
 ARTICLE IN PRESS
2260 S. Agathopoulos et al. / Biomaterials 26 (2005) 2255–2264

the glass transition point. A liquid phase was formed glasses’ surface during immersion in SBF. The concen-
and wetted the surface of the powder grains. Heating at tration of Si reached a plateau, descending afterwards.
temperatures close to the softening point causes low- There are no considerable changes of the dissolution of
ering of viscosity of the liquid and densification Mg2+ and B3+ over immersion time, while there is no
advances via viscous coalescence. In several glass- clear evidence about the possible Na+ exchange
ceramic systems, crystallization process has been attrib- between SBF and the investigated materials. Finally,
uted to have an inhibition effect on densification the concentration of Al3+ was always very low, being at
occurring via viscous flow [26–28]. the limit of the sensitivity of the ICP apparatus.
Comparison of the curves of Fig. 5 with earlier
reports indicates that the evolution of ionic concentra-
3.2. Dissolution and mineralization capability in SBF tions over immersion time strongly resembles the
published reports of oxides containing CaO–SiO2
Fig. 5 outlines the change of ionic concentrations in [29–34], rather than the calcium phosphates, such as
the liquid over immersion time of the powders of both apatites, TCP, etc. [35,36]. The reactivity of the
glasses in SBF at 37 1C. Evidently, alkaline reaction was investigated glasses should have followed the mechan-
observed in all cases. Dissolution spontaneously took ism proposed by Kokubo: (i) Ca2+ ions are exchanged
place, even after 1 h. MW dissolves slightly faster than with H+ resulting in a pH increase, (ii) a silica gel layer
MWP. This finding further supports the aforementioned forms and provides favourable nucleation sites for
speculation whereby phosphorous oxide can increase the apatite, and (iii) apatite layer forms and thickens on
stability of the glassy phase, not only with regard to the the silica gel layer.
delay of crystallization at the temperature of sintering In the light of this mechanism, the investigated glasses
onset but apparently also towards SBF. showed evidences of mild biomineralization capability.
However, the plots feature similar characteristics for Fig. 6a shows that submicron precipitates, enriched in
both glasses. In particular, the concentration of Ca2+ Ca and P (found by EDS), were uniformly formed on
was continuously increased after 1 month of immersion. the surface of MWP glass after 7 days of immersion, but
Phosphorous concentration was rapidly decreased, not earlier (i.e. 3 days or shorter). Their tiny size
indicating the active role of P on the transformation of hampered their complete characterization by EDS (e.g.
Ca/P ratio), XRD and FTIR (attempts with all these
techniques gave ambiguous results and thus omitted).
9.8
30 2.0 Extensive cracking is also obvious at the surface of the
25 glass. In literature, there are studies reporting similar
P,Si, Al, Na/100 (mM)

9.4
1.5 intensive corrosion of very well-known bioactive glasses
Ca, Mg, B (mM)

20
9.0 [37]. Chemical analysis by EDS at steep angles showed
pH

15 1.0 that the glass surface was rich in P and Ca, comparing to
8.6
the bulk (fracture surface) of (as cast) glass (Fig. 6b).
10
8.2
0.5 This finding agrees fairly well with the conclusion with
7.8
5 regard to the active participation of P ions in surface
transformation process of glass during immersion in
0 0.0
7.4
1 10 100 1000 SBF. The increase of Al should be likely apparent, due
(a) Immersion time (h) to the leaching of the other elements during the first
stage (i) of the aforementioned mineralization mechan-
Ca Mg B pH P Si Na Al ism, since SBF does not contain Al ions.
30 2.0 Nevertheless, prolonged immersion in SBF only
9.8
widened surface corrosion but did not result in
25
P,Si, Al, Na/100 (mM)

9.4 formation of apatite layer. Similar submicron Ca-P-rich

Ca, Mg, B (mM)

1.5
20 precipitates (verified by EDS) were formed all over the
9.0
surface of both MW and MWP glass-ceramics after 1
pH

15 1.0
8.6 month of immersion (Fig. 4). In this case, the remaining
10 glassy phase should have favoured the formation the
8.2
0.5 Ca-P precipitates while the crystalline structure main-
5
7.8 tained the structural integrity of materials and sup-
7.4
0 0.0 pressed corrosion [38].
1 10 100 1000 However, the surface of the glasses appears to enable
(b) Immersion time (h) suitable modifications resulting in a layer, which could
Fig. 5. Evolution of ionic concentrations in the liquid over immersion act as nucleation centre of Ca-P precipitates, according
time of: (a) MW and (b) MWP glass powders in SBF at 37 1C. to the aforementioned mineralization mechanism. For
 ARTICLE IN PRESS
S. Agathopoulos et al. / Biomaterials 26 (2005) 2255–2264 2261

comparing to other elements. This conclusion is in

accordance with the low Al3+ concentration of Fig. 5.
We have also obtained similar evidences in our earlier
study in the system fluorapatite-anorthite, where Al was
not detected in the SBF solution after 108 days of
immersion of glass powders in SBF [43]. Therefore, the
strong adherence of Al in the glass structure should
cause intrinsic structural constraints with regards to the
freedom of the glass structure for undergoing the
necessary transformations anticipated in stage (ii) at
the maximum extent. In the literature, Hench and
(a) Andersson have conjectured that alumina inhibits some
of the stages of apatite formation retarding therefore the
P mineralization of osteoids into bone [44]. Kokubo
O Al Ca experimentally determined the Al2O3-concentration
threshold between bioactivity and bioinertness regimes
Si
C in Al2O3-containing wollastonite glasses, which is in the
Na Cl Surface range of 1.5–1.7 mol%, while glasses with 5% Al2O3
Mg Ca
K were entirely inert [45]. The investigated glasses had,
however, lower amount of Al2O3 content (Table 1).
Bulk Therefore, the apparent enrichment of the surface of the
glass in Al after the stage (i) of leaching should have
0 1 2 3 4 5 locally exceeded this concentration threshold.
(b) Energy (keV)
3.3. In vitro tests with osteoblasts
Fig. 6. (a) Formation of Ca-P-rich submicron precipitates and
corrosion evidences at the surface of MWP- glass bulk sample after
7 days of immersion in SBF at 37 1C. (b) Comparison of the elemental Under light microscopy, osteoblasts showed no
analysis of the surface shown in (a) and the bulk (fracture surface) of evidence of morphological changes after 24, 48 or 72 h
the virgin (i.e. as cast) glass, obtained by EDS analysis done at very of incubation in the presence of MW and MWP glasses,
speed angles. when compared to control. The viability/proliferation of
osteoblasts was not altered in the presence of both glass
powders, when compared to control (Fig. 7a). The
instance, there was an incubation time of 7 days for the production of alkaline phosphatase was slightly higher
formation of the first calcium-phosphate precipitates in the presence of the glasses (Fig. 7b).
[29,39]. Kokubo has mentioned 10 days of incubation The osteoblasts collagen secretion increased about
time for the 20Na2O–80SiO2 glass [2]. Moreover, the 20% in the presence of MW and MWP, when compared
fact that the pH value approaches the acidity constant of to control (Fig. 7c). These results are in accordance with
silicic acid, pKSi(OH)2=9.6 (Fig. 5a) [40], probably earlier findings [46], likely attributed to the silicon
supports the existence of stage (ii), whereby silica is content of the glasses. It has been reported that release
hydrolysed to silicic acid. Silicates, which also greatly of silicilic acid enhances production of collagen type I
dissolve from the glasses (Fig. 5) may be actively [47] as well as that bioactive glasses containing
involved in this process [4,32,40,41]. wollastonite components stimulates in vitro biominer-
In earlier studies, dissolution capability and bioactiv- alization [48,49]. Accordingly, the results of Fig. 7c can
ity performance have been directly related to network support fairly well the discussion of Section 3.2, with
connectivity of glasses [3,4,42]. In brief, glasses compris- regards to the evidences of the occurrence of the stages
ing networks with dense cross-linking feature more rigid of biomineralization mechanism in the investigated
structures, which undergo less dissolution, while flexible glasses.
structures with lower network connectivity can promote
dissolution and then mineralization. This work proposes 3.4. Potential applications, coatings
that not only the structure of the initial glass but also the
resulting structure of the surface of the glass, derived The results showed that both MW and MWP glass-
after the leaching of stage (i) of the mineralization ceramics exhibited few properties that may be suitable
mechanism should be enough flexible to favour apatite for biomedical applications. These glass-ceramics have
formation. homogenous colour and high whiteness, good matching
Under this perspective, the experimental results of of microhardness with enamel [50], formation of a single
Fig. 6b clearly show that Al has low leaching capability, (and not multi) phase material during crystallization,
 ARTICLE IN PRESS
2262 S. Agathopoulos et al. / Biomaterials 26 (2005) 2255–2264

0.3 shows typical microstructures of the interfaces devel-

oped between MWP glass-ceramic and the three
substrates after 1 h heating at 800 1C. In the case of
zirconia, the good matching of the CTEs between TZP
Optical density

0.2
and MWP apparently resulted in a continuous interface,
free of cracks or gaps (Fig. 8a). Therefore, good affinity
between the contacting phases might be also assumed
0.1
[52]. In the case of coatings on Ti, a reaction zone of
TiO2 (by EDS) was clearly visible in back-scattering
mode (Fig. 8b). Strong chemical reaction can be
0.0 suggested since the interface is not planar but
(a) MW MWP Control the reaction zone was seemingly diffused towards both
0. 5

0. 4
Optical density

0. 3

0. 2

0. 1

0. 0
(b) MW MWP Control
75
Collagen secretion(mg/ml)

*
50

25

0
(c) MW MWP Control

Fig. 7. (a) Viability/proliferation. (b) Alkaline phosphatase produc-

tion. (c) Collagen secretion. 1  105 osteoblasts were plated in the
presence of MW and MWP powder. After 72 h of incubation, viability
and proliferation were evaluated by MTT assay (a), alkaline
phosphatase production by NBT-BCIP assay (b), and the collagen
present in the supernatant was measured by SIRCOL method (c). In
the latter case (c), the supernatant of all samples, from each condition,
were pooled and measured. In (a) and (b) results represent Mean7SD
of duplicates from five different experiments (Po0:05) and (c)
represent Mean7SD of three different measurements (Po0:05).

and good matching of CTE (9–10  106 K1) with

other materials used in biomedicine, such as ZrO2 and
Ti [51], and therefore may be appropriate for dental Fig. 8. Microstructure of interfaces developed between MWP glass-
ceramic and (a) zirconia (TZP), (b) Ti, and (c) hydroxyapatite (HA)
applications.
after crystallization at 800 1C, 1 h. The grooves shown in (a) and (b)
These advantages motivated a preliminary testing of were purposefully introduced by polishing with a 6 mm diamond paste
producing coatings using MWP glass-ceramic on to give a better perspective of the interface and allow the comparison
zirconia (TZP), titanium and hydroxyapatite. Fig. 8 between the hardness of the contacting phases.
 ARTICLE IN PRESS
S. Agathopoulos et al. / Biomaterials 26 (2005) 2255–2264 2263

the ceramic and the metal phase. Although continuous [5] Kokubo T, Shigematsu M, Nagashima Y, Tashiro M, Nakamura
and strong interface without gaps was observed, the T, Yamamuro T, Higashi S. Apatite- and wollastonite-containing
glass-ceramics for prosthetic application. Bull Inst Chemi Res
possible brittleness of the reaction zone in conjunction
1982;60:260–8.
with the possible mismatch of the thermal and the elastic [6] Kokubo T. A/W glass-ceramic: processing and properties. In:
properties between the three contacting phases might Hench LL, Wilson J editors. An introduction to bioceramics.
cause failure of the Ti/MWP joint, especially under Advanced series in ceramics, vol. 1. Singapore: World Scientific;
fatigue conditions. The MWP coating was ideally joined 1993. p. 75–88.
with hydroxyapatite. Evidently, the two phases diffused [7] Yamamuro T. A/W glass-ceramic: clinical applications. In: Hench
LL, Wilson J editors. An introduction to bioceramics. Advanced
one to another and no interface can be observed series in ceramics, vol. 1. Singapore: World Scientific; 1993.
(Fig. 8c). Therefore, production of porous materials, p. 89–103.
which will function as scaffolds for tissue ingrowth, [8] Beall GN. Glass-ceramics: recent development and applications.
by combining MWP glass-ceramics and other bio- Ceram Trans 1993;30:241–66.
active ceramics into a composite material structure is [9] Wood DJ, Bubb NL, Clifford A, Hill RG, Knowles JC.
An investigation into the crystallization of Dicor glass-ceramic.
possible [53]. J Mater Sci Lett 1999;18:1001–2.
[10] Yang H, Hazeu RM, Downs RT, Finger LW. Structural change
associated with the incommensurate transitions in akermanite,
Ca2MgSi2O7, at high pressure. Phys Chem Min 1997;24:510–9.
4. Conclusions [11] Fredericci C, Zanototto ED, Ziemath EC. Crystallization
mechanism and properties of blast furnace slag glass. J Non-
Glass-ceramics of akermanite from a Ca-mica and Cryst Solids 2000;273:64–75.
wollastonite were successfully designed and produced. [12] Avgustinik AI. Ceramics. Leningrad: Stroiizdat; 1975 (in Rus-
Accordingly, the liquid phase formed during sintering of sian).
[13] Hench LL, Wilson J. Introduction. In: Hench LL, Wilson J
glass-compacts facilitated densification due to the editors. An introduction to bioceramics. Advanced series in
presence of B and F and favoured moderate bioactivity ceramics, vol. 1. Singapore: World Scientific; 1993. p. 1–24.
due to the presence of Ca and Si. Doping with P2O5 [14] Kokubo T, Kushitani H, Sakka S, Kitsugi T, Yamamuro T.
improved the sintering behaviour of the compacts Solutions able to reproduce in vivo surface structure changes in
because P2O5 increases the stability of glass against bioactive glass-ceramics A-W. J Biomed Mater Res 1990;24:
721–34.
crystallization at the temperature of sintering onset and [15] Silver IA, Deas J, Erecinska M. Interactions of bioactive glasses
broadens the range of sintering temperatures. Crystal- with osteoblasts in vitro. Biomaterials 2001;22:175–85.
lization occurred at X700 1C and resulted in only one [16] Amaral M, Costa MA, Lopes MA, Silva RF, Santos JD. Si3N4-
crystalline phase, akermanite, which maintained materi- bioglass composites stimulate the MG63 osteoblast-like cells.
al’s structural integrity towards body fluids. Biomaterials 2002;23:4897–906.
[17] Meleti Z, Shapiro IM, Adams CS. Inorganic phosphate induces
With regards to the use in biomedical applications, apoptosis of osteoblastic-like cells in culture. Bone 2000;27:
the most important properties of the resulting glass- 359–66.
ceramics are the good in vitro acceptance from [18] Klein B, Bem-Bassat H, Solomon V. Structurally different
osteoblasts, the good quality of the white colour at the bisphosphonates exert oposite effects in Alp. J Cell Biochem
surface and throughout the bulk, the matching of 1998;68:186–94.
[19] Tullberg-Reinert H, Jundt G. In situ measurement of collagen
microhardness with tooth enamel, and the possibility synthesis by human bone cells with Sirius-Red based colorimetric
to coat other biomaterials, such as Zirconia, Ti or microassay. Histochem Cell Biol 1999;112:271–6.
hydroxyapatite. [20] Karamanov A, Pelino M, Hreglich A. Sintered glass-ceramics
from municipal solid waste incinerator fly ashes, part I: the
influence of the heating rate on the sinter crystallization. J Eur
Ceram Soc 2003;23:817–32.
Acknowledgements [21] Lira C, Oliveira APN, Alarcon OE. Sintering and crystallization
of CaO–Al2O3–SiO2 glass powder compacts. Glass Technol
This work was supported by the Portuguese FCT and 2001;42:91–6.
[22] Siligardi C, D’Arrigo MC, Leonelli C. Sintering behaviour of
CICECO.
glass-ceramic frits. Am Ceram Soc Bull 2000;79:88–93.
[23] Tulyaganov DU, Ribeiro MJ, Labrincha JA. Development of
glass-geramics by sintering and crystallization of fine powders
References of calcium–magnesium–aluminisilicate glass. Ceram Int 2002;28:
515–20.
[1] Hench LL. Bioceramics. J Am Ceram Soc 1998;81:1705–28. [24] Miller DM. Sintered cordierite glass-ceramic bodies. USA Patent
[2] Kokubo T, Kim HM, Kawashita M, Nakamura T. Novel 3.926.648. 29 May 1976.
ceramics for biomedical applications. J Aust Ceram Soc [25] Tulyaganov DU, Ismatov AA. Development and application of
2000;36:37–46. anorthite-diopside containing glass-ceramics. Ceram Trans
[3] Hill R. An alternative view of the degradation of bioglass. J Mater 1993;29:221–4.
Sci Lett 1996;15:1122–5. [26] Clark TJ, Reed JS. Kinetic processes involved in the sintering
[4] Karlsson KH. Bioactivity of glass and its relation to glass and crystallization of glass-powders. J Am Ceram Soc 1986;69:
structure. Glass Phys Chem 1998;24:280–4. 837–46.
 ARTICLE IN PRESS
2264 S. Agathopoulos et al. / Biomaterials 26 (2005) 2255–2264

[27] Boccaccini AR, Schawohl J, Kern H, Schunck B, Rincon JM, [41] Ylanen H, Karlsson KH, Itala A, Aro HT. Effect of immersion in
Romero M. Sintered glass-ceramics from municipal incinerator fly SBF on porous bioactive bodies made by sintering bioactive glass
ash. Glass Technnol 2000;41:99–105. microspheres. J Non-Cryst Solids 2000;275:107–15.
[28] Tulyaganov DU, Agathopoulos S, Kharton VV, Marques FMB. [42] Agathopoulos S, Ferro MC, Xu JY, Oliveira JM, Marques
Glass-ceramics in the former Soviet Union: a review on industry- PAAP, Correia RN, Fernandes MHV. Structural interpretation
oriented developments. Ind Ceram 2003;23:101–15. of the in vitro reactivity of SiO2–MgO–Na2O glasses. Key Eng
[29] Ohtsuki C, Kokubo T, Yamamuro T. Mechanism of apatite Mater 2003;240–242:217–20.
formation on CaO–SiO2–P2O5 glasses in a simulated body fluid. [43] Agathopoulos S, Tulyaganov DU, Marques PAAP, Ferro MC,
J Non-Cryst Solids 1992;143:84–92. Fernandes MHV, Correia RN. The fluorapatite-anorthite system
[30] Kim HM, Miyaji F, Kokubo T, Ohtsuki C, Nakamura T. in biomedicine. Biomaterials 2003;24:1317–31.
Bioactivity of Na2O–CaO–SiO2 glasses. J Am Ceram Soc [44] Hench LL, Andersson O. Bioactive glass coatings. In: Hench LL,
1995;78:2405–11. Wilson J editors. An introduction to bioceramics. Advanced
[31] Chen Q, Miyata N, Kokubo T, Nakamura T. Effect of heat series in ceramics, vol. 1. Singapore: World Scientific; 1993.
treatment on bioactivity and mechanical properties of PDMS- p. 239–59.
modified CaO–SiO2–TiO2 hybrids via sol–gel process. J Mater Sci [45] Ohtsuki C, Kokubo T, Yamamuro T. Compositional dependence
Mater Med 2001;12:515–22. of bioactivity of glasses in the system CaO–SiO2–Al2O3: its in
[32] Siriphannon P, Kameshima Y, Yasumori A, Okada K, Hayashi S. vitro evaluation. J Mater Sci Mater Med 1992;3:119–25.
Influence of preparation conditions on the microstructure and [46] Tworzydo M, Laczka M, Turyna B. Evaluation of biocompat-
bioactivity of a-CaSiO3 ceramics: formation of hydroxyapatite in ibility of apatite-wollastonite ceramics in fibroblast cultures. Folia
simulated body fluid. J Biomed Mater Res 2000;52:30–9. Histochem Cytobiol 2000;38:133–41.
[33] Liu X, Ding C, Wang Z. Apatite formed on the surface of plasma- [47] Reffitt DM, Ogston N, Jugdaohsingh R, Cheung HFJ, Evans
spayed wollastonite coating immersed in simulated body fluid. BAJ, Hampson GN. Orthosilisilic acid stimulates collagen type I
Biomaterials 2001;22:2007–12. synthesis and osteoblastic differentiation in human osteoblast like
[34] Liu X, Ding C. Reactivity of plasma-spayed wollastonite coating cells in vitro. Bone 2003;32:127–35.
in simulated body fluid. J Biomed Mater Res 2002;59:259–64. [48] Matsuoka H, Akyama H, Kokubo T, Nakamura T. In vitro
[35] Radin SR, Ducheyne P. The effect of calcium phosphate ceramic analysis of bone formation by highly bioactive apatite and
composition and structure on in vitro behavior II. Precipitation. wollastonite. J Biomed Mater Res 1999;47:176–88.
J Biomed Mater Res 1993;27:35–45. [49] Sautier JM, Kokubo T, Nefussi JR. Bioactive glass ceramic
[36] Radin SR, Ducheyne P. Effect of bioactive ceramic composition containing crystalline apatite and wollastonite initiates biominer-
and structure on in vitro behavior III. Porous versus dense alization in bone cell culture. Calcif Tissue Int 1994;55:458–66.
ceramics. J Biomed Mater Res 1994;28:1303–9. [50] Agathopoulos S, Nikolopoulos P, Salomoni A, Tucci A,
[37] Mahmood TA, Davies JE. Incorporation of amino acids within Stamenkovic I. Preparation and properties of binary oxide
the surface reactive layers of bioactive glass in vitro: an XPS bioceramics. J Mater Sci Mater Med 1996;7:629–36.
study. J Mater Sci Mater Med 2000;11:19–23. [51] Agathopoulos S, Pina S, Correia RN. A review of recent
[38] Salinas AJ, Roman J, Vallet-Regi M, Oliveira JM, Correia RN, investigations on zirconia joining for biomedical applications.
Fernandes MH. In vitro bioactivity of glass and glass-ceramics of Ceram Trans 2003;138:135–47.
the 3CaO
P2O5–CaO
SiO2–CaO
MgO
2SiO2 system. Bioma- [52] Tomsia AP, Saiz E, Dalgleish BJ, Cannon RM. Wetting and
terials 2000;21:251–7. strength issues in ceramic/metal bonding. In: Proceedings of the
[39] van Kemenade MJJM, de Bruyn PL. A kinetic study of 4th Japan International SAMPE Symposium Challenging to
precipitation from supersaturated calcium phosphate solutions. Future Advanced Materials Aiming for Intelligence and Harmo-
J Colloid Interface Sci 1987;118:564–85. nization, September 25–28, 1995. p. 347–56.
[40] Karlsson KH, Backman R, Hupa M. An equilibrium study of [53] Lorenzo LMR, Regi M, Ferreira JMF. Fabrication of porous
phosphate precipitation on bioactive glass. Key Eng Mater hydroxilapatite bodies by a new direct consolidation method:
2002;218–220:103–8. starch consolidation. J Biomed Mater Res 2002;60:232–40.