ARTICLE IN PRESS

Biomaterials 27 (2006) 24142425

www.elsevier.com/locate/biomaterials

45S5 Bioglasss-derived glassceramic scaffolds for

bone tissue engineering
Qizhi Z. Chena, Ian D. Thompsonb, Aldo R. Boccaccinia,
a
Department of Materials and Centre for Tissue Engineering and Regenerative Medicine, Imperial College London,
Prince Consort Road, London SW7 2BP, UK
b
Oral & Maxillofacial Surgery, GKT Dental Institute, Kings College London, London SE1 9RT, UK
Received 12 August 2005; accepted 9 November 2005
Available online 5 December 2005

Abstract

Three-dimensional (3D), highly porous, mechanically competent, bioactive and biodegradable scaffolds have been fabricated for the
rst time by the replication technique using 45S5 Bioglasss powder. Under an optimum sintering condition (1000 1C/1 h), nearly full
densication of the foam struts occurred and ne crystals of Na2Ca2Si3O9 formed, which conferred the scaffolds the highest possible
compressive and exural strength for this foam structure. Important ndings are that the mechanically strong crystalline phase
Na2Ca2Si3O9 can transform into an amorphous calcium phosphate phase after immersion in simulated body uid for 28 days, and that
the transformation kinetics can be tailored through controlling the crystallinity of the sintered 45S5 Bioglasss. Therefore, the goal of an
ideal scaffold that provides good mechanical support temporarily while maintaining bioactivity, and that can biodegrade at later stages
at a tailorable rate is achievable with the developed Bioglasss-based scaffolds.
r 2005 Elsevier Ltd. All rights reserved.

Keywords: Scaffolds; Bone tissue engineering; Mechanical properties; Bioactivity; Biodegradation; Replication technique

1. Introduction the replication technique [22] (also called the polymer-

Tissue engineering seeks to promote the regeneration
ability of host tissue through a designed scaffold that is
populated with cells and signalling molecules. The specic
criteria for ideal scaffolds used in bone tissue engineering
are summarised as follows [13]: (1) ability to deliver cells,
(2) excellent osteoconductivity, (3) good biodegradability,
(4) appropriate mechanical properties, (5) highly porous
structure: porosity490% [4] and pore sizes 4400500 mm
[5], (6) irregular shape fabrication ability, and (7)
commercialisation potential.
Bioactive glasses meet the rst three criteria: excellent
osteoconductivity and bioactivity [610], ability to deliver
cells [11], and controllable biodegradability [1214]. These
advantages make bioactive glasses promising scaffold
materials for tissue engineering [1517]. Among a variety
of processes for fabrication of porous materials [5,1821],
Corresponding author. Tel.: +44 207 594 6731; fax: +44 207 584 3194.
E-mail address: a.boccaccini@imperial.ac.uk (A.R. Boccaccini).
sponge method) produces porous ceramic structures that
are most similar to those of spongy bone [23,24]. This
technique also satises scaffolds criteria (5)(7) mentioned
above. Thus, all criteria for an ideal tissue engineering
scaffold, except that related to mechanical competence,
could be satised by 45S5 Bioglasss foams fabricated by
the replication method. The replication method has been
applied to produce scaffolds of hydroxyapatite (HA)
[2527]. Surprisingly, this technique, however, has never
been considered before to produce scaffolds from bioactive
glasses. Bioactive glass scaffolds have only been fabricated
by dry-powder processing with porogen additions [2830]
and by solgel and gel-casting techniques [3,31].
The major hurdle in the production of highly porous
Bioglasss-based foam-like scaffolds has been caused by the
following apparently irreconcilable issues of this glass: (a)
it has been reported that crystallisation of 45S5 Bioglasss
turns a bioactive glass into an inert material [32]; (b) full
crystallisation of the glass occurs prior to signicant
densication [33]; (c) extensive densication is required to

0142-9612/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biomaterials.2005.11.025
 ARTICLE IN PRESS
Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425
strengthen the struts of a foam, which would otherwise be
made of loosely bonded particles and thus be too fragile to
handle. According to these three factors, to maintain the
bioactivity of 45S5 Bioglasss, one should sinter the foam
at a relatively low temperature at which crystallisation does
not take place or does not occur to a great extent.
However, sufcient densication by sintering will not occur
at low temperatures, and therefore a very fragile scaffold
made of loosely packed 45S5 Bioglasss particles is
produced.
The above dilemma might be solved in light of the recent
work of Clupper and Hench [3437], who carried out
quantitative investigations on the effect of crystallinity on
the apatite formation on Bioglasss surfaces in vitro. Their
ndings revealed that the crystal phase Na2Ca2Si3O9
slightly decreased the formation kinetics of an apatite
layer on the Bioglasss sample surface but it did not totally
suppress the formation of such layer [34]. Moreover, it is
recognised that the bioreaction kinetics of a highly porous
network can be very different from that of a dense product
of the same chemical composition due to a high surface
area in the foams. Hence, it might be possible to nd a new
sintering protocol leading to mechanically competent
foams through extensive densication of the struts, while
inducing the formation of a bioactive and biodegradable
crystalline phase. The objectives of this work, therefore,
were to synthesize 45S5 Bioglasss scaffolds using the
replication technique, to achieve mechanically stable 3D
scaffolds through a tailored sintering schedule, and to
assess the bioactivity and biodegradability of the scaffolds.
The nal goal is to create an ideal scaffold for bone tissue
engineering.
2. Materials and experiments
2.1. Materials
The starting material was melt-derived 45S5 Bioglasss powder (particle
size
5 mm). A fully reticulated polyester-based polyurethane foam with
60 ppi (pores per inch) from Recticel UK (Corby) was used as sacricial
template for the replication method. The details of the polyurethane foam
used have been reported by other authors [38]. The foam was supplied in
large samples of 20 mm in thickness and was cut to size 10 mm 
10 mm  20 mm for compression strength tests and 10 mm  10 mm 
60 mm for bending strength tests.
2.2. Scaffold fabrication
The replication method involves preparation of green bodies of ceramic
(or glass) foams by coating a polymer (e.g. polyurethane) foam with a
ceramic (or glass) slurry. The polymer, having the desired pore structure,
simply serves as a sacricial template for the ceramic coating. The polymer
template is immersed in the slurry, which subsequently inltrates the
structure and ceramic (glass) particles adhere to the surfaces of the
polymer. Excess slurry is squeezed out leaving a more or less homogeneous
coating on the foam struts. After drying, the polymer is slowly burned out
in order to minimise damage to the ceramic (glass) coating. Once the
polymer has been removed, the ceramic (or glass) network is sintered to a
desired density. The process replicates the macrostructure of the starting
sacricial polymer foam, and results in a rather distinctive and well-
2415
Ceramic (or glass) powder
Add
Prepare slurry from the powder Binder
Prepare a green body by dipping a
polymer foam in the slurry
Dry, burn out sacrificial polymer
foam, and sinter the green body
Ceramic (or glass) foam
Fig. 1. Flowchart of the polymer-sponge method for fabrication of glass
or ceramic foams.
dened microstructure within the struts. A owchart of the process is
given in Fig. 1.
In our experiments, the slurry for the impregnation of the polyurethane
foam was prepared using the following recipe. Polyvinyl alcohol (PVA)
was dissolved in water, the ratio being 0.01 mol/L. Then 45S5 Bioglasss
powder was added to 100 ml PVA-water solution up to concentration of
40 wt%. Each procedure was carried out under vigorous stirring using a
magnetic stirrer for 1 h.
The polyurethane foams cut to shape were immersed in the above-
prepared slurry and remained in it for 15 min. The foams were manually
retrieved from the suspension as quickly as possible, and the extra slurry
was completely squeezed out. The samples (called green bodies) were then
placed on a smooth surface and dried at ambient temperature for at least
12 h. The coating thickness of a green body could be increased by
repeating the above coating procedure. In this work most green bodies
were prepared by single coating, but few were made by double coating.
The double-coated green bodies will be mentioned where they are used in
this paper.
Post-forming heat treatments for the burnout of the polymer template
and sintering for the 45S5 Bioglasss structure were programmed, as
shown in Fig. 2. The burning condition of the polymer templates was the
same for all samples: 400 1C/1 h. Sintering conditions were designed to be
900 1C/5 h; 950 1C/05 h; and 1000 1C/02 h. The heating and cooling rates
were 2 and 5 1C/min, respectively.
2.3. Characterisation
The density rfoam of the scaffolds was determined from the mass and
dimensions of the sintered bodies. The porosity p was then calculated by
rfoam
p1 1  rrelative , (1)
rsolid
where rsolid 2:7 g=cm3 is the density of solid 45S5 Bioglasss [14].
The microstructure of the foams was characterised in a JEOL 5610LV
scanning electron microscope (SEM), before and after immersion in
simulated body uid (SBF). Samples were gold- or carbon-coated and
observed at an accelerating voltage of 15 kV.
Selected foams were also characterised using X-ray diffraction (XRD)
analysis with the aim to assess the crystallinity after sintering and
formation of HA crystals on strut surfaces after different times of
immersion in SBF. The foams were rst ground into a powder. Then 0.1 g
of the powder was collected for XRD analysis. A Philips PW 1700 Series
automated powder diffractometer was used, employing Cu ka radiation
(at 40 kV and 40 mA) with a secondary crystal monochromator. Data were
 ARTICLE IN PRESS
2416 Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425
900-1000C/0-5hr
Temperature
2C/min
400C/1hr
5C/min
2C/min
R.T.
Time
Fig. 2. Heat treatment program designed for burning-out the polyur-
ethane templates and sintering the 45S5 Bioglasss green bodies.
collected over the range of 2y 51001 using a step size of 0.041 and a
counting time of 25 s per step.
2.4. Mechanical testing
The compression strength of foams was measured using a Zwick/Roell
Z010 mechanical tester at a crosshead speed of 0.5 mm/min. The samples
were rectangular in shape, with dimensions: 10 mm in height and
5 mm  5 mm in cross-section. During compression test, the load was
applied until densication of the porous samples started to occur.
Three-point bending strength tests were carried out using a Hounseld
testing machine. The size of the specimens was
3 mm  4 mm  40 mm.
The load was applied over a 30 mm span and at the mid-point of the
4 mm  40 mm surface. All tests were performed using a cross-head speed
of 0.5 mm/min. The bending strength was calculated according to [39]:
3Pf L
sf , (2)
2bh2
where Pf is the load at fracture, L 30 mm is the sample length over
which the load is applied, bE4 mm is the sample width, and hE3 mm is
the sample height.
2.5. Assessment of bioactivity in simulated body fluid
This part of the study was carried out using the standard in vitro
procedure described by Kokubo et al. [40]. The foams were immersed in
75 ml of acellular SBF in clean conical asks, which had previously been
washed using HCl and deionised water. The conical asks were placed
inside an incubator at controlled temperature of 37 1C. The pH of the
solution was maintained constant at 7.25. The size of all samples for these
tests was 10 mm  10 mm  10 mm. Two samples were extracted from the
SBF solution after given times of 3, 7, 14, and 28 days. The SBF was
replaced twice a week because the cation concentration decreased during
the course of the experiments, as a result of the changes in the chemistry of
the samples. Once removed from the incubation, the samples were rinsed
gently, rstly in pure ethanol and then using deionised water, and left to
dry at ambient temperature in a desiccator.
3. Results
3.1. Porous structure of foams
All foams exhibited porosity of
90%, as determined by
measurement of their mass and dimensions and applying Eq.
(1). The cell size of sintered scaffolds was estimated as follows.
The cell size of the as-received polymer foam was
7401040 mm. The volume shrinkage from a polymer template
to a sintered 45S5 Bioglasss-based scaffold was dened as
V BGfoam =V PU-foam , and it was determined, through measur-
ing the volumes of the starting polymer and sintered 45S5
Bioglasss-based foams, to be 33% on average for the sintering
condition of 1000 1C/1 h. Therefore, the linear shrinkage
V BGfoam =V PUfoam 1=3 would be
70%. Finally, the range
of cell sizes of the foams sintered at 1000 1C for 1 h was
calculated to be 0.70  (7401040) mm 510720 mm.
The macroporous network and the strut microstructure of
typical foams are illustrated in Fig. 3. Highly porous scaffolds
were produced at all sintering conditions. A comparison of
Figs. 3a, c and d shows that the cell struts are considerable
thicker when sintered at 1000 1C for up to 1 h than at
900950 1C for 25 h. It was observed at high magnication
that extensive sintering of 45S5 Bioglasss particles did not
occur at 900 1C even after 5 h sintering (Fig. 3b), but
densication, which occurs by a viscous ow sintering
mechanism in glass, increased signicantly when the foams
were heated up to 950 and 1000 1C (Figs. 3d and 3f). Fine
crystalline grains of
0.5 mm in diameter could be detected by
SEM observation in foams sintered at 1000 1C for 1 h (Fig. 3f).
The combination of extensive densication and the presence of
a crystalline phase in the struts of scaffolds sintered at 1000 1C
for 1 h are expected to lead to improved mechanical properties
of these foams. Hence mechanical tests and assessment of
bioactivity in SBF were carried out on foams sintered at
1000 1C, as described in Sections 3.3 and 3.4.
The hollow nature of a strut and its wall microstructure
are shown in Fig. 4. Similar morphologies have been
reported for a variety of sintered ceramic foams synthesised
by the polymer-sponge method [41]. It can be seen that the
wall of the strut has been nearly fully densied after
sintering at 1000 1C for 1 h.
3.2. Crystallisation
The XRD investigation revealed that crystallisation had
occurred extensively in all samples sintered at 900, 950 and
1000 1C for 5 h (Fig. 5). However, the bonding of particles
was not obvious at the sintering condition of 900 1C/5 h
(Fig. 3a). This observation conrmed the nding of
Clupper and Hench [33] that extensive crystallisation
occurs prior to signicant viscous ow sintering in 45S5
Bioglasss and related bioactive glasses. In Fig. 5, both
angular location and intensity of the peaks match the
standard PDF #22.1455, which indicates that the crystal-
line phase is Na2Ca2Si3O9. The same crystalline phase has
been formed and identied in previous studies on sintered
bioactive glasses [36,37].
The present 45S5 Bioglasss-based foams are in fact made
of a glassceramic, as the crystallinity of the sintered 45S5
Bioglasss material cannot be 100%. From the components
of 45S5 Bioglasss and Na2Ca2Si3O9 (Table 1), one can nd
that the Na2Ca2Si3O9 phase would demand too much CaO
to fully crystallise from Bioglasss. Eventually CaO is
depleted when the crystallinity reaches 80.7 mol% (i.e.
77.4 wt%), which is thus the maximum crystallinity achiev-
able by the 45S5 Bioglasss composition.
 ARTICLE IN PRESS
Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425 2417

Fig. 3. Pore structure and strut microstructure of 45S5 Bioglasss-derived foams sintered at (a)(b) 900 1C for 5 h; (c)(d) 950 1C 2 h; and (e)(f) 1000 1C
for 1 h.

at 1000 1C for 1 h. A typical compressive stressstrain curve

Fig. 4. The hollow centre of a single strut in a Bioglasss derived foam
sintered at 1000 1C for 1 h.
3.3. Mechanical properties
Compressive and bending strength tests were carried out
on foams prepared by single or double coating and sintered
is shown in Fig. 6, which is jagged and has three distinct
regimes. The foams tend to crack rst in thin struts at
stress-concentrating sites, causing the apparent stress to
drop temporarily. But the foam, as a whole, still had the
ability to bear higher loads, causing the stress to rise again.
The repetition of this procedure gave a jagged stressstrain
curve.
In stage I (Fig. 6), the stressstrain curve has a positive
slope until a maximum stress is reached. This maximum
stress causes the thick struts of the foam to fracture and as
a result the stressstrain curve has a negative slope in stage
II. In Stage III, densication of the fractured foams occurs
as stress increases, which is the typical behaviour of foams
under compression [42].
The raw data of compressive strength are plotted against
the foam porosities in Fig. 7. The compressive tests were
frequently accompanied by shearing, which was mainly
caused by the end effects imposed on the specimen during
the test. It has been reported that if the faces of the foam
sample are slightly misaligned with the loading platen,
large stress concentrations can occur causing local buck-
ling, which in turn leads to shearing and thus results in an
 ARTICLE IN PRESS
2418 Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425

1000
900
800
700
600
500
400

Apatite
300
200
Intensity (a.u.)

100
1000C/1hr
800
700
600
500
400
Apatite

300
200
100
900C/5hrs
200
100
As-Received
0
0 20 40 60 80 100
2
()

Fig. 5. XRD spectra of 45S5 Bioglasss powder unsintered and sintered at 900 1C for 5 h and 1000 1C for 1 h. All spectra were obtained using 0.1 g powder.
The major peaks of the phase Na2Ca2Si3O9 [35] are marked by (X).

strengths are collectively higher than compressive strengths

at similar porosities. For instance, when porosity is
90%
Table 1 the highest compressive and bending strengths are in the
Components of 45S5 Bioglasss and crystalline phase Na2Ca2Si3O9 range of 0.30.4 and 0.40.5 MPa, respectively. This result
(mol.%)
is in agreement with the general ndings in ceramics and it
45S5 Bioglasss Na2Ca2Si3O9 is related to the statistic nature of strength value of highly
porous brittle materials [44].
SiO2 46.134 50
Na2O 24.35 16.667
CaO 26.912 33.333
P2O5 2.6038 0 3.4. Bioactivity assessment in SBF

Assessment of bioactivity was carried out on foams

underestimation of both Youngs modulus and strength
[43]. In Fig. 7, the apparent strength values of the sheared
foams (marked by D) were much lower than those of purely
compressed samples (marked by solid triangles m). It is
reasonable to consider that the strength values obtained
from pure compression tests represent the compressive
strength of the foams.
Fig. 8 illustrates a typical forcedisplacement curve in a
three-point bending strength test. Like in the compressive
strength test, thin struts cracked rst at stress-concentrat-
ing sites, giving a typical jagged curve. When a maximum
stress (bending strength) was reached, the sample fractured
into two pieces, causing the stress to drop to zero abruptly.
The raw data of three-point bending strength of as-
sintered and coated foams are given in Fig. 9. Bending
sintered at 1000 1C for 0.5 and 1 h. Similar XRD results
were obtained for both groups of foams. Fig. 10 shows the
XRD spectra of the foams sintered at 1000 1C for 1 h and
then immersed in SBF for 328 days, together with the
XRD patterns of 45S5 Bioglasss in as-received and as-
sintered conditions.
A signicant phenomenon, in addition to the growing
peaks of HA-like phase detected in the spectra of soaked
samples, was that the crystallinity of the sintered foams
decreased with increasing immersion time in SBF. Even-
tually the sharp diffraction peaks of the Na2Ca2Si3O9
phase disappeared from the XRD spectrum after soaking
in SBF for 28 days, leaving a typical broad halo (produced
by an amorphous phase) overlapped by the sharp diffrac-
tion peaks of the HA phase. This indicates that at least
under the detection limits of XRD, the sintered 45S5
 ARTICLE IN PRESS
Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425 2419

0.5

I II III

Compressive Stress (MPa)

0.4

0.3

0.2

0.1

0.0
0 20 40 60 80 100
Compressive strain (%)

Fig. 6. A typical compressive stress-strain curve of the 45S5 Bioglasss-based foams sintered at 1000 1C for 1 h. The porosity of the foam was 91.0%.

Theoretical strength (ti / t=0) 1.4

Bending strength (MPa)

Theoretical strength (ti / t=0.5)
Experimental strength with shearing involved
1.2
Experimental strength without shearing 1
Reported strength of HA-based foams in literature [25-27]
0.9 0.8
0.8 0.6
Compressive strength (MPa)

0.7 0.4
0.6 0.2
0.5 0
0.76 0.78 0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94
0.4
Porosity
0.3
0.2 Fig. 9. Bending strength values of Bioglasss-based scaffolds. The foams
with porosity lower than 89% were prepared by double coating.
0.1
0
0.84 0.86 0.88 0.9 0.92 0.94 0.96
Bioglasss-derived material was mainly composed of an
Porosity
amorphous phase and crystalline apatite after soaking in
Fig. 7. Theoretical and experimental compressive strength values of SBF for 28 days.
Bioglasss-based scaffolds in the present work, and those of hydroxyapa- The microstructural evolution in both groups of foams
tite-based foams reported in literature [2527]. The foams with porosity
(sintered at 1000 1C for 0.5 and 1 h) is summarized in
lower than 89% were prepared by double coating. Theoretical values were
obtained from Eq. (3). Table 2. Figs. 11(ad) illustrate typical surface morphol-
ogies of samples sintered at 1000 1C for 0.5 h followed by
5
immersion in SBF for different time periods.

4 4. Discussion

3 In this section, the results of the investigation are

Force (N)

discussed in relation to mechanical properties, microstruc-

2 ture, and bioactivity.

1 4.1. Comparison of 45S5 Bioglasss-based foams with

spongy bone
0
0 0.2
The foams produced by using the polymer-sponge
Displacement (mm) method are very similar to spongy bone (also called
Fig. 8. A typical forcedisplacement curve of 45S5 Bioglasss-based foam cancellous bone) in terms of their pore structure. It is thus
in bending test. of importance to nd out whether or not the mechanical
 ARTICLE IN PRESS
2420 Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425

200
0 As-received
1000
800

600

400

0 As-sintered
600
400
Intensity (a.u.)

200
0 3 days
600
400

200
0 7 days
400
200
0 14 days
400
200
28 days
0
0 10 20 30 40 50 60 70 80 90 100
2
()

Fig. 10. XRD spectra of 45S5 Bioglasss-based foams sintered at 1000 1C for 1 h, and immersed in SBF for 3, 7, 14, and 28 days. All spectra were obtained
using 0.1 g powder. The major peaks of Na2Ca2Si3O9 phase and hydroxyapatite are marked by (X) and (K), respectively.

Table 2
Summary of characteristics of 45S5 Bioglasss-derived foams after immersion in SBF

Immersion time in SBF 1000 1C/30 min 1000 1C/1 h

3 days Sparsely distributed apatite precipitates Very few apatite precipitates

1 week Strut surface was unevenly covered by aggregated Sparsely distributed apatite precipitates
apatite spheres
2 weeks Apatite spheres were fused together. Strut surface was fully covered by a large amount of
apatite spheres, size being
1 mm
4 weeks The whole foam is made of amorphous calcium Apatite spheres grew, size being
2.5mm
phosphate and crystalline hydroxyapatite
 The apatite could be a mixture of amorphous and crystalline calcium phosphates.

strength of the foams is comparable to that of cancellous
bone.
There have been many reports on the mechanical
properties of cancellous bone, which have been reviewed
in Ref. [23]. It is generally accepted that the mechanical
properties of struts in cancellous bone are close to those of
cortical bone. Typical values are: 12 GPa for Youngs
modulus, 136 MPa for compressive strength, and 105 MPa
for tensile strength [24]. The compressive strength of
spongy bone (not the strut) is in the range of 0.24 MPa,
when the relative density is
0.1 [24]. Hence, the measured
compressive strength (0.30.4 MPa) of the present foams
falls in this range, but lies closer to the lower bound. Our
experience indicates that the strength of 0.30.4 MPa is
sufcient for the foam to be handled with, such as
manipulating during SBF tests and cutting of the samples
for mechanical tests.
In addition, it has been reported that the compressive
strength of a HA scaffold signicantly increases (e.g. from

10 to
30 MPa [45]) due to tissue ingrowth in vivo. It has
also been speculated that it might not be necessary to
fabricate a scaffold with a mechanical strength equal to
bone because cultured cells on the scaffold and new tissue
formation in vitro will create a biocomposite and will
 ARTICLE IN PRESS
Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425
Fig. 11. Hydroxyapatite formed on the surfaces of foam struts after immersion in simulated body uid (SBF) for (a) 3 days, (b) 7 days, (c) 14 days, and (d)
28 days. The foams were sintered at 1000 1C for 30 min.
increase the time-dependent strength of the scaffold
signicantly [16]. An ideal scaffold, however, should have
at least a proper strength and fracture toughness to allow it
to be manipulated adequately for tissue engineering
applications. The present 45S5 Bioglasss-based scaffolds
possess such an appropriate mechanical competence.
There is no reliable fracture toughness data available for
cancellous bone. But it is predicted that the current foams
will be more brittle than spongy bone. A further study
involving the incorporation of poly(D,L-lactic acid) into the
45S5 Bioglasss-based foams is on-going, aiming at
improving the fracture toughness of these scaffolds.
4.2. Comparison of the present foams with previous
investigations
There are few reports available on porous 45S5
Bioglasss and related bioactive glassceramics with low
porosity (2142%) [2830], but no work has been
published on highly porous (p490%) 45S5 Bioglasss-
based foams, to the best knowledge of the authors.
Therefore, the comparison carried out here is between the
present scaffolds and highly porous (p470%) foams made
from other bioactive ceramics and glasses, including HA,
b-tricalcium phosphate (b-TCP), and 70S30C solgel
derived glass foams.
Table 3 summarises characteristics of highly porous
bioactive ceramic and glass foams developed for bone
engineering, including method of fabrication, pore structure,
and compressive strength data. In general, the compressive
strength varies signicantly with foam porosity. For example,
the compressive strength of HA foams, which were
2421
synthesised by the polymer-sponge method, decreased from
0.29 to 0.03 MPa when the porosity increased from 69 to 86%
[27]. Some compressive strength data of porous HA-based
foams reported in literature [25,26] have been collected and
are shown in Fig. 7 (marked by ). It is obvious that the
present 45S5 Bioglasss-based foams are in general stronger
than the HA-based foams of similar porosities.
It is unwise to directly compare foams exhibiting partially
open pore structure with completely open pore scaffolds
fabricated by the polymer-sponge method. The high mechan-
ical strength of the former [46,47] is obviously achieved at the
cost of a less interconnected pore structure. The windows on
the wall of pores in these foams are mainly in the range of
30120 mm, which is considerably smaller than the required
size (
400 mm) for osteoblast penetration [5].
It is apparent that gel-casting combined with the
polymer-sponge technique produces stronger HA foams
[48,49] than the simple polymer-sponge method [26].
However, a comparison of the present 45S5 Bioglasss-
based foams (0.42 MPa at porosity 89%) with HA foams
produced by Ramay and Zhang [48] (0.55 MPa at porosity
77%) indicates that the polymer-sponge method developed
here can produce as strong foams as the gel-casting/
polymer-sponge combined technique, and that the well-
sintered and crystallised 45S5 Bioglasss-based scaffolds
can be as strong as HA foams.
4.3. Comparison of experimental and theoretical strength
data
The modelling of the mechanical behaviour of highly porous
materials has been presented by Gibson and Ashby [24].
 ARTICLE IN PRESS
2422 Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425

Table 3
Overview of structural characteristics and mechanical properties of highly porous bioactive ceramic or glass foams for bone tissue engineering

Technique Material Porosity (%) Pore size (mm) Closed (C) or Compressive Ref.
open (O) strength (MPa)

Polymer-sponge 45S5 Bioglasss 8992 510720 O 0.270.42 Present work

Glass-reinforced HA 8597.5 420560 O 0.010.175 [25]
HA 86 420560 O 0.21 [26]
6986 4901130 O 0.030.29 [27]
Gel-casting/ HA 76.780.2 201000 Partly O/C 4.47.4 [46]
foamed by
vigorous stirring
HA Cell: 100500 Partly O/C 1.65.8 [47]
Window: 30120
Polymer-sponge HA 7077 200400 O 0.555 [48]
b-TCP+HA 73 200400 O 9.8 [49]
Solgel/foamed by Bioactive glasses (e.g. 7095 Cell: up to 600 Partly O/C 0.52.5 [31]
vigorous stirring 70S30C)
Windows: 80120

The theoretical compressive collapse stress stheo can be
expressed as a function of the relative density rfoam =rsolid
of a cellular structure and the size of the central hollow
struts by Eq. (3):


stheo rfoam 3=2 1 ti =t2
0:2 q , (3)
sfs rsolid
1  ti =t2
where ti =t is the ratio of the void and strut sizes on a cross-
section of a strut (see Fig. 4) and sfs is the modulus of
rupture of the strut. Theoretical calculations show that the
modulus of rupture of a brittle material is typically about
1.1 times larger than the tensile strength [24]. In our
calculation, the tensile strength sts 42 MPa of bulk 45S5
bioglasss (annealed) [14] was used for the strength of the
partially crystallised material. The ratio ti =t was estimated
to be 0.5 according to Fig. 4.
Using Eq. (3), the compressive strength of the present
foams (with ti =t 0:5) and the lower bound of theoretical
strength (when ti =t 0) were calculated. The results are
illustrated in Fig. 7. The experimental strengths determined
by compressive strength tests are generally above the lower
bound, and most of them are in good agreement with the
theoretical strengths when ti =t 0:5. This indicates that
the cell walls have been sintered to be fully dense at 1000 1C
for 1 h.
Two points should be mentioned. (i) The tensile strength
of sintered HA is 40 MPa, which is very close to that of
dense 45S5 Bioglasss (42 MPa). Hence theoretically, the
mechanical strength of a 45S5 Bioglasss-based scaffold
should be similar to, if not higher than, that of a HA
scaffold with a similar porous structure. (ii) As ti =t
increases, the foam becomes stronger. In other words, the
hollow tubular structure is benecial to the mechanical
performance of the foam, which is a direct result of the
derivation of Eq. 3 [24].
4.4. Possible mechanisms for the transition from
Na2Ca2Si3O9 to an amorphous phase
Since the sintered 45S5 Bioglasss material is in fact a
glassceramic, one might argue that the bioactivity of the
sintered material could be attributed to the residual glass
phase. We suggest that the bioactivity remains also with the
crystalline phase Na2Ca2Si3O9, based on two reasons: (1)
the bioactivity of pure Na2Ca2Si3O9 phase has been
reported [35], and (2) the transition from Na2Ca2Si3O9 to
an amorphous phase provides an explanation for the
nding that the presence of Na2Ca2Si3O9 decreased the
kinetics of apatite formation but did not inhibit the growth
of an apatite layer on the form surfaces, which has been
reported in the literature [34].
The mechanisms behind the transformation of Na2Ca2-
Si3O9 to an amorphous phase might be based in the well-
known bone-bonding mechanisms of bioactive glasses,
which were originally proposed by Hench and colleagues
[14]. In the sequence of interfacial reactions on the surface
of Bioglasss in contact with body uids, the bioactive glass
rst dissolves to form a silica-gel layer; then an amorphous
calcium phosphate is formed from the hydrated silica-gel;
and nally apatite crystallites nucleate and grow from the
amorphous calcium phosphate. We suggest that the general
idea of the reaction sequence should be applicable to
Na2Ca2Si3O9 crystallites as well, which however dissolves
at a slower rate than the glass phase. Hence, the
amorphous phase detected by XRD after immersion in
SBF for 28 days (Fig. 10) could be the amorphous calcium
phosphate, according to Hench et al.s theory [14]. This
suggestion has been proved by energy dispersive X-ray
(EDX) analysis, as shown elsewhere [50].
Although the kinetics of the transformation has yet to be
fully understood, it is believed that the high surface area
(including hollow centre of the struts) in the porous
 ARTICLE IN PRESS
Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425
network is of relevance in maintaining bioactivity and
biodegradability of the sintered 45S5 Bioglasss-based
foams. The high surface energy should make possible that
the transformation of Na2Ca2Si3O9 to the amorphous
phase of calcium phosphate occurs at a reasonably fast rate
at the body temperature. This assumption is supported by
the fact that the bioactive reactions only occur at the
surface of a bulk solid glass. It is based on this fact that
bioactivity is dened to be the interfacial ability to bond to
bone [14]. Nevertheless, the transformation mechanism
from a crystal to an amorphous phase, as found in this
material system in contact with SBF, remains a subject for
future dedicated research.
4.5. Significance of the transformation of Na2Ca2Si3O9 to
the amorphous phase
An ideal scaffold for bone engineering serves as a
temporary frame to foster new bone growth. It is expected
to provide a temporary mechanical support and later to
degrade at a rate matching the regeneration rate of new
bone tissue. Unfortunately, this is not the manner
conventional bioceramics behave in biological conditions.
Crystalline HA, for instance, can provide reasonably
strong support; but it degrades very slowly in contact
with body uids, degradation time being of the order of
years. Amorphous HA degrades much faster than crystal-
line HA; but it is too fragile to build highly porous
scaffolds. Bioactive glasses encounter a similar hurdle:
they have excellent bioactivity and tailorable biodegrad-
ability; at the same time they possess poor mechanical
reliability.
The above problem could be solved by designing
scaffolds following the discovery of this work, which has
shown that the mechanically strong crystalline phase
Na2Ca2Si3O9 (which is formed in 45S5 Bioglasss upon
sintering) can transform into an amorphous calcium
phosphate (the good resorbability of which has been well
documented [51,52]) in a simulated body uid environ-
ment. Based on this nding, it is possible to sinter 45S5
Bioglasss green foams at optimised conditions such that
both signicant densication and Na2Ca2Si3O9 crystal-
lisation take place. The extensive densication and ne
crystalline grains of Na2Ca2Si3O9 confer the scaffold a
temporary good mechanical performance. The transforma-
tion of Na2Ca2Si3O9 to the amorphous calcium phosphate,
which is expected to occur upon exposure to a body uid
environment, ensures the bioactivity and degradability of
the scaffold.
The transformation of a crystalline phase to a degrad-
able amorphous phase is not an exclusive phenomenon of
45S5 Bioglasss material. HA and related calcium phos-
phates also show a similar transition in an in vivo
environment [53]. The difference with Bioglasss is that
the transition in HA is too slow to match clinical
expectation. It has been shown that only a thin layer of
amorphous phase on the surface of crystalline HA particles
2423
(
0.5 mm) is formed after implantation for 3 months, and
that HA particles do not degrade considerably even after
implantation for 6 months [53]. Tissue engineering
applications demand that the degradation kinetics of a
scaffold should match the regeneration kinetics of new
bone in vitro and/or in vivo. In general, the degradation
time should be less than 6 months, depending on the
anatomic site for regeneration, the mechanical loads
present at the site, and the desired rate of osseointegra-
tion [1].
Hence, the signicance of the Na2Ca2Si3O9 to amor-
phous phase transition in our 45S5 Bioglasss-based foams
lies in its kinetics which seems to be sufciently fast for
application of the material in bone engineering. More
importantly, the kinetics of the transformation and of the
scaffold degradation can be controlled by factors such as
initial crystallinity, porosity in struts, and grain size of
Na2Ca2Si3O9, all of which can be tailored by the sintering
conditions. Therefore, the goal of an ideal scaffold that
provides good mechanical support temporarily while
maintaining bioactivity, and that can biodegrade later at
a tailorable rate can be achieved with the developed 45S5
Bioglasss-derived scaffolds.
5. Conclusions
This work has successfully synthesized highly porous
(porosity:
90%, cell diameter: 510720 mm), mechanically
competent, bioactive and biodegradable 45S5 Bioglasss-
derived glassceramic scaffolds for bone engineering, using
the replication technique. When sintered under an optimal
condition (1000 1C/1 h), the nearly full densication and
the ne crystals of Na2Ca2Si3O9 confer the scaffolds
competent mechanical strength. A signicant nding is
that the mechanically strong crystalline phase can trans-
form into a bioactive and biodegradable amorphous
calcium phosphate upon immersion in SBF. Therefore,
the goal of an ideal scaffold that provides sufcient
mechanical support temporarily, and that can biodegrade
later at a tailorable rate is achievable with the present
Bioglasss-derived glassceramic scaffolds.
Acknowledgements
Helpful discussions on the sintering experiments with
Mr. Jonny Blaker (Imperial College London) are gratefully
acknowledged. Recticel UK at Corby is gratefully
acknowledged for providing the polyurethane foam used
in this research. Helpful discussions with Prof. Larry
Hench are greatly appreciated.
References
[1] Temenoff JS, Lu L, Mikos AG. Bone tissue engineering using
synthetic biodegradable polymer scaffolds. In: Davies JE, editor.
Bone engineering. Toronto: EM Squared; 2000. p. 45562.
 ARTICLE IN PRESS
2424 Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425
[2] Bruder SP, Caplan AI. Bone regeneration through cellular engineer-
ing. In: Lanza RP, Langer R, Vacanti J, editors. Principles of tissue
engineering. 2nd ed. California: Academic Press; 2000. p. 68396.
[3] Jones JR, Boccaccini AR. Cellular ceramics in biomedical applica-
tions: tissue engineering. In: Schefer M, Colombo P, editors.
Cellular ceramics: structure, manufacturing, processing and applica-
tions. Weinheim: Wiley-VCH Verlag GmbH & Co. KgaA; 2005.
p. 55073.
[4] Freyman TM, Yannas IV, Gibson LJ. Cellular materials as porous
scaffolds for tissue engineering. Prog Mater Sci 2001;46:27282.
[5] Hutmacher DW. Scaffolds in tissue engineering bone and cartilage.
Biomaterials 2000;21:252943.
[6] Wilson J, Pigot GH, Schoen FJ, Hench LL. Toxicology and
biocompatibility of bioglass. J Biomed Mater Res 1981;15:80511.
[7] Oonishi H, Kutrshitani S, Yasukawa E, Iwaki H, Hench LL Wilson
J, Tsuji E, Sugihara T. Particulate bioglass compared with hydro-
xyapatite as a bone graft substitute. Clin Orthop Relat Res
1997;334:31625.
[8] Hench LL, Splinter RJ, Allen WC. Bonding mechanisms at the
interface of ceramic prosthetic materials. J Biomed Mater Res Symp
1971;2(part 1):11741.
[9] Hench LL, Paschall HA. Direct chemical bond of bioactive
glassceramic materials to bone and muscle. J Biomed Mater Res
Symp 1973;4:2542.
[10] Hench LL, Paschall HA. Histochemical response at a biomaterials
interface. J Biomed Mater Res Symp 1974;5(part 1):4964.
[11] Gatti AM, Valdre G, Andersson OH. Analysis of the in vivo
reactions of a bioactive glass in soft and hard tissue. Biomaterials
1994;15:20812.
[12] Clark AE, Hench LL. Calcium phosphate formation on solgel
derived bioactive glasses. J Biomed Mater Res 1994;28:6938.
[13] Hench LL. Solgel materials for bioceramic applications. Curr Opin
Solid State Mater Sci 1997;2:60410.
[14] Hench LL, Wilson J. Surface-active biomaterials. Science 1984;226:
6306.
[15] Laurencin CT, Lu HH, Khan Y. Processing of polymer scaffolds:
polymerceramic composite foams. In: Atala A, Lanza RP, editors.
Methods of tissue engineering. California: Academic Press; 2002.
p. 70514.
[16] Jones JR, Hench LL. Regeneration of trabecular bone using porous
ceramics. Curr Opin Solid State Mater Sci 2003;7:3017.
[17] Boccaccini AR. Bioresorbable and bioactive composite materials
based on polylactide foams lled with and coated by Bioglasss
particles for tissue engineering applications. J Mater Sci: Mater Med
2003;14:350443.
[18] Leong KF, Cheah CM, Chua CK. Solid free-form fabrication of
three-dimensional scaffolds for engineering replacement tissue and
organs. Biomaterials 2003;24:3632378.
[19] Widmer MS, Miko AG. Fabrication of biodegradable polymer
scaffolds for tissue engineering. In: Patrick CW, Mikos AG, Mcintir
LV, editors. Frontier in tissue engineering. New York: Elsevier
Science; 1998. p. 10720.
[20] Atala A, Lanza RP, editors. Methods of tissue engineering.
California: Academic Press; 2002.
[21] Maquet V, Boccaccini AR, Pravata L, Notingher I, Jerome R. Porous
poly(a-hydroxyacid)/Bioglasss composite scaffolds for bone tissue
engineering. I: Preparation and in vitro characterisation. Biomaterials
2004;25:418594.
[22] Schwartzalder K, Somers AV. Method of making a porous shape of
sintered refractory ceramic articles. United States Patent no. 3090094,
1963.
[23] Cowin SC, editor. Bone mechanics. Boca Raton, FL: CTC Press;
1989. p. 14 pp. 10-110-23.
[24] Gibson LJ, Ashby MF. Cellular solids: structure and properties. 2nd
ed. Oxford: Pergamon; 1999. p. 429452.
[25] Callcut S, Knowles JC. Correlation between structure and compres-
sive strength in a reticulate glass-reinforced hydroxyapatite foam.
J Mater Sci Mater Med 2002;13:4859.
[26] Kim HW, Knowles JC, Kim HE. Hydroxyapatite porous scaffold
engineered with biological polymer hybrid coating for antibiotic
vancomycin release. J Mater Sci Mater Med 2005;16:18995.
[27] Miao X, Lim G, Loh KH, Boccaccini AR. Preparation and
characterisation of calcium phosphate bone cement. Mater Proc
Prop Perf (MP3) 2004;3:31924.
[28] Livingston T, Ducheyne P, Garino J. In vivo evaluation of a bioactive
scaffold for bone tissue engineering. J Biomed Mater Res 2002;62:
113.
[29] Kaufmann EABE, Ducheyne P, Shapiro IM. Evaluation of
osteoblast response to porous bioactive glass (45S5) substrates by
RT-PCR analysis. Tissue Eng 2004;6:1928.
[30] Yuan H, de Bruijn JD, Zhang X, van Blitterswijk CA, de Groot K.
Bone induction by porous glass ceramic made from Bioglasss (45S5).
J Biomed Mater Res: Appl Biomater 2001;58:2706.
[31] Jones JR, Hench LL. Factors affecting the structure and properties of
bioactive foam scaffolds for tissue engineering. J Biomed Mater Res
B: Appl Biomater 2004;68B:3644.
[32] Li P, Zhang F, Kokubo T. The effect of residual glassy phase in a
bioactive glassceramic on the formation of its surface apatite layer in
vitro. J Mater Sci Med 1992;3:4526.
[33] Clupper DC, Hench LL. Crystallization kinetics of tape cast bioactive
glass 45S5. J Non-Crys Solid 2003;318:438.
[34] Peitl O, LaTorre GP, Hench LL. Effect of crystallization on apatite-
layer formation of bioactive glass 45S5. J Biomed Mater Res 1996;
30:50914.
[35] Peitl O, Zanotto ED, Hench LL. Highly bioactive P2O5Na2O
CaOSiO2 glassceramics. J Non-Crys Solid 2001;292:11526.
[36] Clupper DC, Mecholsky Jr. JJ, LaTorre GP, Greenspan DC.
Sintering temperature effects on the in vitro bioactive response of
tape cast and sintered bioactive glassceramic in Tris buffer.
J Biomed Mater Res 2001;57:53240.
[37] Clupper DC, Mecholsky Jr. JJ, LaTorre GP, Greenspan DC.
Bioactivity of tape cast and sintered bioactive glassceramic in
simulated body uid. Biomaterials 2002;23:2599606.
[38] Haugen H, Will J, Kohler A, Hopner U, Aigner J, Wintermantel E.
Ceramic TiO2-foams: characterisation of a potential scaffold. J Eur
Ceram Soc 2004;24:6618.
[39] Dowling NE. Mechanical behaviour of materials: engineering
materials for deformation, fracture, and fatigue. 2nd ed. Englewood
Cliffs, NJ: Prentice-Hall Inc.; 1998. p. 603648 [Chapter 13].
[40] Kokubo T, Hata K, Nakamura T, Yamamura T. Apatite formation
on ceramics, metals, and polymers induced by a CaOSiO2-based
glass in simulated body uid. In: Boneld W, Hastings GW,
Tanner KE, editors. Bioceramics, vol. 4. London: Guildford,
Butterworth-Heinemainn; 1991. p. 11320.
[41] Montanaro L, Jorand Y, Fantozzi G, Negro A. Ceramic foams by
powder processing. J Eur Ceram Soc 1998;18:133950.
[42] Gibson LJ, Ashby MF. Cellular solids: structure and properties. 2nd
ed. Oxford: Pergamon; 1999. pp. 117118 and 209212.
[43] Turner CH, Burr DB. Experimental techniques for bone mechanics.
In: Cowin SC, editor. Bone Mechanics. Boca Raton, FL: CTC Press;
1989. p. 14 pp. 7-97-11.
[44] Davidge RW. Mechanical behaviour of ceramics. Cambridge: Cam-
bridge University Press; 1979. p. 139.
[45] Tamai N, Myoui A, Tomita T, Nakase T, Tanaka J, Ochi T. Novel
hydroxyapatite ceramics with an interconnective porous structure
inhibit superior osteoconduction in vivo. J Biomed Mater Res 2002;
59:1107.
[46] Sepulveda P, Binner JGP, Rogero SO, Higa OZ, Bressiani JC.
Production of porous hydroxyapatite by the gel-casting of foams and
cytoxic evaluation. J Biomed Mater Res 2000;50:2734.
[47] Sepulveda P, Bressiani AH, Bressiani JC, Meseguer L, Konig Jr. B. In
vivo evaluation of hydroxyapatite foams. J Biomed Mater Res 2002;
62:58792.
[48] Ramay HRR, Zhang M. Preparation of porous hydroxyapatite
scaffolds by combination of the gel-casting and polymer sponge
methods. Biomaterials 2003;24:3293302.
 ARTICLE IN PRESS
Q.Z. Chen et al. / Biomaterials 27 (2006) 24142425
[49] Ramay HRR, Zhang M. Biphasic calcium phosphate nanocomposite
scaffolds for load bearing bone tissue engineering. Biomaterials
2004;25:517180.
[50] Chen QZ, Boccaccini AR. Poly(D,L-lactic acid) coated 45S5
Bioglasss-based scaffolds: processing and characterisation. J Biomed
Mater Res A, accepted for publication.
[51] Passuti N, Daculsi G. Calcium-phosphate ceramics in orthopedic-
surgery. Presse Med 1989;18:2831.
2425
[52] Knaack D, Goad MEP, Aiolova M, Rey C, Toghi A, Chakravarthy
P, Lee DD. Resorbable calcium phosphate bone substitute. J Biomed
Mater Res 1998;43:399409.
[53] Chen QZ, Wong CT, Lu WW, Cheung KMC, Leong JCY, Luk
KDK. Strengthening mechanisms of bone bonding to crystalline
hydroxyapatite in vivo. Biomaterials 2004;25:424354.