Mechanical properties of calcium phosphate scaffolds
fabricated by robocasting
Pedro Miranda,1 Antonia Pajares,2 Eduardo Saiz,3 Antoni P. Tomsia,3 Fernando Guiberteau1
1
Departamento de Electrónica e Ingenierı́a Electromecánica, Universidad de Extremadura, Avda de Elvas s/n. 06071
Badajoz, Spain
2
Departamento de Fı́sica, Universidad de Extremadura, Avda de Elvas s/n. 06071 Badajoz, Spain
3
Materials Sciences Division, Lawrence Berkeley National Laboratory, 1 Cyclotron Rd., Berkeley, California 94720
Received 23 February 2007; revised 14 May 2007; accepted 31 May 2007
Published online 9 August 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.31587
Abstract: The mechanical behavior under compressive
stresses of b-tricalcium phosphate (b-TCP) and hydroxyapatite (HA) scaffolds fabricated by direct-write assembly
(robocasting) technique is analyzed. Concentrated colloidal
inks prepared from b-TCP and HA commercial powders
were used to fabricate porous structures consisting of a 3D tetragonal mesh of interpenetrating ceramic rods. The
compressive strength and elastic modulus of these model
scaffolds were determined by uniaxial testing to compare
the relative performance of the selected materials. The
effect of a 3-week immersion in simulated body fluid (SBF)
on the strength of the scaffolds was also analyzed. The
results are compared with those reported in the literature
for calcium phosphate scaffolds and human bone. The
robocast calcium phosphate scaffolds were found to exhibit excellent mechanical performances in terms of
strength, especially the HA structures after SBF immersion,
indicating a great potential of this type of scaffolds for
use in load-bearing bone tissue engineering applications.
Ó 2007 Wiley Periodicals, Inc. J Biomed Mater Res 85A:
218–227, 2008
INTRODUCTION
structures with customized, complex, 3-D shapes.
Thus, these techniques enable the production of optimal porous structures to attain the desired mechanical behavior and mass transport properties (permeability and diffusion properties) for scaffold applications. Furthermore, if the CAD model is obtained
from medical scan data (e.g., computerized tomography or nuclear magnetic resonance imaging), the
scaffold’s external shape can be made to match
the damaged tissue site. Also, SFF techniques enable
the fabrication of structures with different combinations of materials, to create microstructural and
chemical gradients or patterns with tailored functionality. A feature that distinguishes robocasting
from other SFF techniques is that it uses water-based
inks with a minimal organic content (<1 wt %) and
does not require any sacrificial support material or
mold. Since the use of considerable amounts of binders in other SFF techniques complicates their sintering and densification process, robocasting appears as
a very appealing candidate technology for building
(i.e., printing) ceramic scaffolds.
Consequently, recent work has been directed
towards developing colloidal suspensions with the
appropriate viscoelastic properties to be used as robocasting inks from calcium phosphate–b-tricalcium
The robocasting, or direct-write assembly, technique consists of the robotic deposition through a
specific nozzle of highly concentrated colloidal suspensions (inks) capable of fully supporting their
own weight during assembly.1–3 Thus, lines (rods) of
ink are laid down in a controlled manner to build
up a 3-D structure layer-by-layer, following a computer-aided design (CAD) model. Like other solid
freeform fabrication (SFF) techniques,4–6 it overcomes the limitations of more conventional scaffold
fabrication methods, allowing the fabrication of
Correspondence to: P. Miranda; e-mail: pmiranda@unex.es
Contract grant sponsor: European Community’s Sixth
Framework Program; contract grant number: MOIF-CT2005-7325
Conract grant sponsor: Ministerio de Educación y Ciencia and the Fondo Social Europeo; contract grant number:
MAT2006-08720
Conract grant sponsor: National Institutes of Health
(NIH); contract grant number: 5R01 DE015633
' 2007 Wiley Periodicals, Inc.
Key words: robocasting; hydroxyapatite;
phosphate; scaffolds; strength
b-tricalcium
MECHANICAL PROPERTIES OF ROBOCAST CALCIUM PHOSPHATE SCAFFOLDS
phosphate7 (b-TCP) and hydroxyapatite8,9 (HA)–bioceramic powders. These two calcium phosphate
ceramics are the materials most widely studied for
scaffold applications because their composition is
close to that of the mineral bone phase.10–12 Although the development of these inks has allowed
the fabrication of robocast calcium phosphate scaffolds, there is little information yet available about
their mechanical behavior. Good mechanical performance of porous scaffolds is paramount for their
applications in orthopedics, as they should be able
to withstand some degree of loading during their
use in vivo. Therefore, high strength and reliability
are determining properties, while matching elastic
properties to surrounding tissue is also desirable to
provide the appropriate support and stress level for
bone regeneration. Understanding the fundamental
features controlling these properties in this type of
structures is thus essential.
Accordingly, in this work we analyze the mechanical behavior under compression of b-TCP and HA
scaffolds fabricated by robocasting, with an emphasis on their respective compressive strengths. For
this purpose, prototype scaffolds consisting of a 3-D
tetragonal mesh of interpenetrating ceramic rods
were fabricated using the ink recipes developed in
previous work for each material,7,9 and tested under
uniaxial compression. Tests were performed along
two orthogonal directions and the respective compressive strengths analyzed in detail. The effect of
immersion in simulated body fluid (SBF) on the
strength was also analyzed. Estimates of the elastic
moduli of the scaffolds were obtained from the
load–displacement curves.
Finally, the results are compared with values
reported in the literature for calcium phosphate scaffolds and bone tissue. Implications for the optimization of the mechanical performance of robocast calcium phosphate scaffolds and their applicability in
load-bearing bone tissue engineering applications
are discussed.
219
consistency was fine-tuned by adjusting its pH with HNO3
or NH4OH as needed. Each addition to the mixture was
followed by shaking for about 1 h, together with a few zirconia grinding balls, in a paint shaker (Red Devil 5400,
Red Devil Equipment, Plymouth, MN). The final powder
contents of the inks were 35 vol % for HA9 and 45 vol %
for b-TCP.7
Robocasting was used to construct scaffolds consisting
of a tetragonal mesh of ceramic rods from the fabricated
inks as illustrated in Figure 1. The printing syringe was
partially filled with the corresponding ink, tapped vigorously under slight vacuum to remove bubbles, and then
placed on the robotic deposition device (3-D Inks, Stillwater, OK), which is controlled by a computer-aided
direct-write program (Robocad 3.0, 3-D Inks). The inks
were deposited through conical nozzles (EFD Inc., East
Providence, RI, USA) of diameter d ¼ 250 lm at a constant
linear printing speed of 20 mm/s. The in-plane line spacing (from center to center), s, in the computer 3-D model
of the structure was set to 400 lm. The layer spacing, h,
was fixed at 225 lm, to have a 25-lm layer overlap that
facilitates printing. The external dimensions of the scaffolds were set at about 10 3 10 3 10 mm so that a total of
44 layers were deposited. Deposition was done in a nonwetting oil bath to prevent nonuniform drying during assembly.
The samples were dried in air at room temperature for
24 h and then at 4008C (18C/min heating rate) for 1 h to
evaporate organics. Finally, the dried samples were sin-
EXPERIMENTAL PROCEDURES
Materials and sample preparation
Robocasting inks were prepared following the recipes
described in detail in previous work, from commercially
available HA9 and b-TCP7 powders. Basically, a concentrated stable suspension of the selected powder in distilled
water was prepared by adding appropriate amounts of
Darvan1 C dispersant (R.T. Vanderbilt, Norwalk, CT).
Then hydroxypropyl methylcellulose (Methocel F4M, Dow
Chemical Company, Midland, MI) was added to the mixture to increase viscosity and the ink was finally gellified
by adding polyethylenimine (PEI) as flocculant. The ink’s
Figure 1. Schematic of robocasting fabrication process.
The ceramic scaffold is built layer by layer from a computer design. A 3-axis robotic arm moves the injection syringe while expressing the ceramic ink through the conical
deposition nozzle, of diameter d, to create a self-supporting 3-D network of ceramic rods immersed in an oil bath.
The layer spacing, h, and in-plane rod spacing, s, are indicated.
Journal of Biomedical Materials Research Part A
220
tered at 13008C (heating rate 38C/min) for 2 h. It was confirmed by X-ray diffraction that after the sintering treatment the HA samples were almost pure HA while the bTCP samples contained small quantities of calcium pyrophosphate as the original powder was calcium-deficient—
the sintering mechanisms and resulting microstructure of
the b-TCP samples were discussed in detail in a previous
work.7 The scaffolds were analyzed using scanning electron microscopy (S-4300SE/N, Hitachi, USA) (Fig. 2). The
average rod diameter was measured to be 220 6 10 lm
and the average rod spacing in the printing plane 300 6
10 lm, with a rod overlap between adjacent layers of
about 50 6 5 lm. Taking into account these dimensions,
one estimates approximately 28% macroscopic porosity for
the scaffold. Total porosity values of around 39% and 40%
were determined by simple weighing for HA and b-TCP,
which implies that the porosity within the rods is approximately 15% and 17% in each case (assuming theoretical
densities of 3.156 and 3.07 g/cm3, respectively). Average
MIRANDA ET AL.
grain size in the HA and b-TCP rods were found to be 3.2
6 0.5 and 7.4 6 0.7 lm, respectively, as determined using
the mean linear intercept method (ASTM Standard E 11288) on a series of section micrographs on thermally etched
samples. It is worth noting the presence of microcracks in
the b-TCP samples [Fig. 2(f)] as a result of the reversible
b–a transformation occurring during sintering.7,13
After sintering some samples were immersed in SBF14 at
378C for 20 days to determine the effect of this in vitro
environment on the mechanical properties of the scaffolds.
Mechanical testing
The compressive strength of the scaffolds was determined by performing uniaxial tests on approximately
cubic blocks of 2-mm side cut from the sintered specimens.
The tests were carried out in air on a universal testing
machine (AG-IS10kN, Shimadzu, Kyoto, Japan) at a con-
Figure 2. SEM micrographs showing the morphology of HA (left) and b-TCP (right) scaffolds after sintering at 13008C
for 2 h: general view (a, b), printing plane view (c, d), and detail of the rod surfaces (e, f). Arrows mark the presence of
microcracks in (f).
Journal of Biomedical Materials Research Part A
MECHANICAL PROPERTIES OF ROBOCAST CALCIUM PHOSPHATE SCAFFOLDS
stant crosshead speed of 0.6 mm/min. Tests were performed both in the direction perpendicular to the printing
plane (direction 3 in Fig. 1) and along one of the two
equivalent rod directions (directions 1 or 2 in Fig. 1). The
load–displacement curve was registered during the tests.
The compressive strength of the structure was calculated
as the maximum applied load divided by the measured
square section of the sample. A minimum of eight samples
were tested in each testing condition in order to get statistically reliable values. The elastic moduli of the scaffolds
were estimated from the slopes of the load–displacement
curves, taking into account the compliance of the testing
machine (20 kN/mm).
Intrinsic mechanical properties of the individual calcium
phosphate rods were also evaluated. Instrumented indentation (Nanotest, Micro Materials, Wrexham, UK) was
used to determine the elastic modulus and hardness of the
rods. The indentation tests were performed using a diamond Berkovich indenter on polished sections of the scaffolds (to 1 lm finish) perpendicular to the rod axis. Single
indentations of about 5.5-lm deep were placed at the center of b-TCP and HA rods as shown in Figure 3. The
indent size, of about 40-lm side, is large enough compared
to grain size to provide meaningful information about the
mechanical properties of the rods (and not of individual
grains) but small enough to avoid the influence of the free
surface of the rods. The Poisson’s ratio was assumed to be
the same for the two materials and equal to 0.28 (Ref. 15)
to calculate the Young’s modulus, E, from the reduced
modulus, E ¼ E=ð1 m2 Þ, obtained in the indentation
tests. The inert fracture strength of the rods was determined from 3-point bending tests performed in air at a
constant crosshead speed of 30 mm/min, in individual
rods printed and sintered for this purpose. Since the
fracture load for rods of 220-lm diameter was close to
the sensitivity limit of the load cell, thicker rods of 360lm diameter were used instead. The final density and
microstructure of these rods were undistinguishable
from those of the scaffolds. The load–displacement
curves from these tests were also used to estimate the
Young’s modulus of the rods, confirming the results
obtained from the instrumented indentation.
RESULTS AND ANALYSIS
Figure 4 shows some typical load–displacement
curves obtained in uniaxial tests performed on the
HA and b-TCP scaffolds in the printing direction
(direction 3 in Fig. 1)—curves corresponding to
direction 1 or 2 (rods directions) are similar. The significant drops in load observed in these plots are
associated with the development of longitudinal
cracks in the structure, as already reported in a previous work.9 The considerable mechanical resistance
retained by the structure after several cracking
events, as evidenced by the gradual decline in the
load, is also analyzed in that earlier study.9 It is evident from these curves that the compressive strength
of the HA scaffolds is clearly superior to that of the
221
Figure 3. SEM micrographs of Berkovich indents on a
polished cross-section of the rods: (a) 2-N load imprint on
HA and (b) 1-N load imprint on b-TCP. Note that the
imprints (40 lm size) are far from the free edges of the
rods.
b-TCP structures. As is appreciable in the figure, the
HA scaffolds are also stiffer than the b-TCP structures (E ¼ 7 6 2 GPa vs. E ¼ 2 6 1 GPa). These
results are easily explained by considering the intrinsic mechanical properties of HA and b-TCP rods—
determined as described in the previous section—
which are summarized in Table I. As is evident from
these results, the HA rods are significantly stiffer,
harder, and stronger than are the b-TCP rods, thus
qualitatively explaining the differences observed in
Figure 4.
Figure 5 shows the Weibull plot corresponding to
the compressive strength data. This plot shows the
failure probability as a function of applied stress for
HA (circles) and b-TCP (squares) scaffolds tested
along direction 1 or 2 (filled symbols) and direction
3 (open symbols). The straight lines are the best fits
using the Weibull probability function16–18
P ¼ 1 exp½ðr=r0 Þm ð1Þ
with P being the failure probability, and where the
Weibull modulus, m, and central value, r0, are adjustable parameters. In particular, it was found that
the average compressive strength obtained in tests
Journal of Biomedical Materials Research Part A
222
MIRANDA ET AL.
Figure 4. Load–displacement curves obtained during uniaxial compression tests performed on HA and b-TCP scaffolds along the direction orthogonal to the printing plane
(direction 3 in Fig. 1). Drops in load mark longitudinal
crack pop-in events.
performed in direction 3 was slightly lower than in
the other directions. Also, as clearly shown in Figure
5, the data scatter is significantly larger in this direction. For instance, the Weibull modulus—which is
inversely related to data scatter—for HA scaffolds
changes from m ¼ 3.2 6 0.2 in direction 3 to m ¼ 9.3
6 0.6 in direction 1 or 2. This difference in Weibull
modulus of the data could be associated with a
greater localization of the tensile stresses when load
is applied along direction 3.9 Higher spatial concentration of the tensile stresses implies a lower probability of finding flaws within the tensile region, and
this translates into a greater dispersion of the
strength data.19 Also evident in this figure are the
differences in strength between the two materials.
The lower compressive strength values measured for
b-TCP are attributable to the development of microcracks during sintering in this material,7 as evidenced in Figure 2(f) (arrows).
The effect of immersion in SBF14 for 20 days on
the compressive strength of the scaffolds is shown in
the Weibull plots of Figure 6. The data corresponds
to tests performed in direction 3 but the effect is
TABLE I
Intrinsic Mechanical Properties of HA and b-TCP Rods
HA
TCP
a
E* (GPa)a
mb
E (GPa)
H (GPa)
rF (MPa)
83 6 4
38 6 8
0.28
0.28
82 6 4
36 6 7
2.9 6 0.3
1.5 6 0.8
68 6 12
27 6 9
Reduced modulus, E ¼ Eð1 m2 Þ.
From Ref. 15.
b
Journal of Biomedical Materials Research Part A
Figure 5. Weibull compressive strength plot (i.e., failure
probability vs. applied stress) for HA (circles) and b-TCP
(squares) scaffolds tested uniaxially along direction 3
(open symbols) and direction 1 or 2 (filled symbols). The
straight lines are linear fits using a Weibull probability
function [Eq. (1)].
analogous for direction 1 or 2. As can be clearly
seen, the SBF had no effect on the mechanical behavior of the b-TCP scaffolds but markedly increased
the compressive strength of HA, without affecting
significantly the Weibull modulus of the data. This
effect is associated with the in vitro formation of
bone-like apatite (or octacalcium phosphate, OCP20)
on the surface of the HA rods, as shown in Figure
7(a) [cf. Fig 2(e)]. No apatite formation is observed
in the b-TCP scaffolds [Fig. 7(b), cf. Fig. 2(f)], and
therefore, no analogous effect in the strength is
observed for this material. This inability of b-TCP to
induce apatite growth in vitro has recently been
reported in the literature.20
Figure 8 shows a comparative summary of the experimental results obtained in the uniaxial tests for the
average compressive strength of the scaffolds, with the
error bars representing one standard deviation of
the data. Again, the superior resistance properties of
HA relative to b-TCP are clear, especially after the
approximately twofold increase in HA strength
obtained by immersion in SBF. In fact, the average
compressive strength of the HA scaffolds was more
than three times greater than that of the b-TCP structures before immersion in SBF, but after the immersion
that difference increased to more than sevenfold.
Finally, in Figure 9 our strength and modulus data
are compared with results from the literature for HA
(circles) and TCP (squares) scaffolds fabricated by
conventional and SFF processing techniques11,21–44—
different literature sources are not distinguished.
The results are plotted as a function of material den-
MECHANICAL PROPERTIES OF ROBOCAST CALCIUM PHOSPHATE SCAFFOLDS
223
Figure 6. Weibull compressive strength plot (i.e., failure
probability vs. applied stress) for HA (circles) and b-TCP
(squares) scaffolds tested uniaxially along direction 3,
before (open symbols) and after (filled symbols) immersion
for 20 days in SBF. The straight lines are linear fits using a
Weibull probability function [Eq. (1)].
which increased the value to levels similar to those
corresponding to cortical bone of the same dry density. To the best of our knowledge, this striking
effect has not before been reported in the literature.
In principle, there is no reason to believe that the
same immersion procedure cannot be applied to
increase the strength of scaffolds fabricated by other
techniques, though evidently the amount of
improvement may depend on the preexisting surface
conditions of the scaffold.
Regarding the elastic moduli, Figure 9(b) shows
that the values estimated from the uniaxial compression curves for our scaffolds lie well below the levels
reported in the literature for scaffolds fabricated by
more conventional techniques, which usually lie
above the values corresponding to bone tissue of the
same apparent dry density. However, the only value
found in the literature for an HA scaffold fabricated
from an SFF technique38 (with coincidentally the
same total porosity) is even lower than our value.
Nevertheless, estimates of the elastic modulus of the
scaffolds from the load–displacement curves in uniaxial tests can be unreliable, even after correcting the
sity (bottom axis), calculated from reported porosity
values (top axis) assuming a theoretical density of
3.1 g/cm3 for both HA and TCP. Also included for
comparison are estimated values for bone tissue following an empirical model due to Keller.45 This
model, derived from an extensive data collection corresponding to both femoral and spinal tissue, relates
the properties of bone to its apparent dry density
through a simple power-law. Those estimates, including the data dispersion, are represented by the
shaded bands in Figure 9.
The compressive strength plot [Fig. 9(a)] clearly
shows that our as-fabricated robocast scaffolds have
a level of mechanical integrity at least comparable to
those of scaffolds fabricated by other techniques
from the same materials. In particular, the robocast
HA scaffold presents greater strength than the single
available literature value with the same total porosity—which corresponds to a scaffold fabricated by
another SFF technique.38 Higher levels of compressive strength have only been claimed in a recent
paper on scaffolds fabricated by freeze casting (the
two open circles inside and above the shaded band
within the cortical bone region).21 However, those
high values were obtained for a lamellar structure
with interlamella spacing of *15 lm and only when
tested along a direction parallel to the lamina. Also,
the control of the pore structure of the scaffolds
achieved in freeze casting, though better than in
most conventional techniques, do not reach the levels attainable in robocasting.
Especially remarkable was the effect of the immersion in SBF on the strength of the HA scaffolds,
Figure 7. SEM micrographs showing the morphology of
the surfaces of the scaffold rods after immersion for 20
days in SBF: (a) HA surface showing the bone-like apatite
crystals that grew during the in vitro immersion, and (b)
unchanged b-TCP surface after the immersion. Microcracks that developed in b-TCP during sintering are apparent in (b).
Journal of Biomedical Materials Research Part A
224
MIRANDA ET AL.
The strength differences between the two materials
increased markedly after in vitro immersion in SBF for
20 days. The HA scaffolds doubled their compressive
strength—reaching values similar to those of cortical
bone with the same dry density—due to bone-like apatite growth on the surface of the HA rods, while the
b-TCP scaffolds remained unaltered. Hence, controlled immersion in SBF would seem to be a simple
and inexpensive means of improving the mechanical
performance of complex 3-D structures fabricated by
robocasting from apatite-forming materials. In all
Figure 8. Bar chart summarizing the average compressive
strength data obtained for each type of scaffold. Error bars
are one standard deviation.
curve to take the machine compliance into account.
This is especially relevant for our robocast scaffolds
where the irregular contact surface can complicate
the coupling between the sample and the compression plates. Because of the intrinsic low strength of
the calcium phosphate materials, cracking may occur
at the contacts9 or at isolated points of the structure
even at low load; thus preventing the system from
showing a linear region in the load–displacement
curve. That is the case in the curves of Figure 4,
where the ‘‘linear’’ region is more like a sigmoidal
curve of very slowly varying slope. More accurate
measurement methods (e.g., using extensometers) or
computational techniques (e.g., finite element simulations) would be needed to reliably estimate the
actual elastic response of this type of structure.
CONCLUSIONS AND IMPLICATIONS
The compressive strengths of b-TCP and HA scaffolds fabricated by a direct-write assembly (robocasting) technique have been analyzed and compared in this work. Prototype porous structures consisting of a 3-D tetragonal mesh of interpenetrating
ceramic rods fabricated from concentrated colloidal
inks prepared from b-TCP and HA commercial powders were used. The results showed that the HA
scaffolds had much greater (more than twofold)
compressive strength than b-TCP in all testing configurations. Such significant differences are attributable to microcrack formation during sintering in the
b-TCP materials.7 The microcracks reduce the intrinsic strength of the rods, which in turn reduces the
compressive strength of the scaffolds.
Journal of Biomedical Materials Research Part A
Figure 9. Comparative plot of (a) strength and (b) modulus data from this work and literature reports,11,21–44 for
both HA (circles) and TCP (squares) scaffolds-different literature sources are not distinguished. The results are plotted as a function of material density (bottom axis) and porosity (top axis) for both HA and TCP. The shaded bands
represent bone properties (including data dispersion) as a
function of apparent bone dry density (bottom axis), estimated using a power-law empirical model due to Keller.45
MECHANICAL PROPERTIES OF ROBOCAST CALCIUM PHOSPHATE SCAFFOLDS
probability this procedure could also be successfully
applied to scaffolds fabricated by other SFF and more
conventional techniques. However, in some of the
conventional techniques the apatite growth may
reduce the pore interconnectivity, which is a critical
parameter to promote bone ingrowth. Also, SBF
immersion provides an additional means (together
with that of controlling the sintering process) of modifying the surface properties of the scaffolds, which are
known to have a major influence on cell affinity for
the structure. In particular, it has been shown in vitro
that the formation of a bone-like layer increases the
number of cells adhered to the HA surface.46
Considering both its superior strength and its ability to induce apatite formation in vitro, HA seems a
much better material than b-TCP for the fabrication
of load-bearing scaffolds for bone tissue engineering
applications. However, the inability of b-TCP to
induce apatite/OCP growth in vitro is not necessarily
an indication of low bioactivity in vivo. Indeed, bTCP is known to exhibit great biodegradability and
osteoconductivity, significantly superior to those of
synthetic HA.47,48 For this reason, b-TCP is a better
choice for tissue engineering applications and has
been successfully used in maxillary reconstruction or
as a filler in polymer composites.49–51 However, our
results suggest that b-TCP has still a long way to go
to become a suitable material for load-bearing bone
tissue engineering applications. In particular, the
intrinsic strength of b-TCP would have to be
improved, for example by avoiding the formation of
microcracks during sintering. Since those microcracks form as a consequence of a reversible b–a
transformation,7,13 they could be avoided by selecting appropriate sintering conditions. Also, starting
powder quality (size, morphology, and composition)
may be one of the key parameters in fabrication and
performance. Further research is therefore needed to
determine the optimal processing route to improve
the mechanical performance of b-TCP.
Also, our results suggest that both the strength and
the reliability of this type of tetragonal mesh structure
are superior when tested along the rod directions. If
this is confirmed for different scaffold designs, it
would be something to take into account when orienting these scaffolds in real applications. Since reliability
is reduced (i.e., the dispersion of the strength values is
increased) by stress concentrations, a rule of thumb in
scaffold design would be to build the scaffold so that
tensile stresses are as evenly distributed as possible
over the entire structure. Finite element modeling
would certainly be an invaluable tool for this task.
It is worth noting that our robocast scaffolds had
greater strength than have the most reported ones,
but that is at least partially attributable to the fact
that they have lower porosity than most scaffolds
reported in the literature. This low volumetric poros-
225
ity may induce one to think that our scaffolds would
have less capacity to induce bone ingrowth. However, rather than the volume of porosity, it is the
interconnections between the pores which are mainly
involved in allowing bone ingrowth (i.e., cell migration and nutrient transport processes).11,52 Indeed,
pore interconnectivity has been shown to positively
influence bone deposition rate and depth of infiltration both in vitro53 and in vivo.54 Therefore, the optimal amount of total porosity for bone tissue regeneration is really still an open question. Most published
estimates have been derived from data corresponding to scaffolds fabricated conventionally,55 where
pore interconnection size is much smaller than pore
size, and therefore in which a high volumetric porosity is needed to ensure enough pore interconnections
of the necessary size. Robocasting and other SFF
methods allow the construction of macropore structures where pores and pore interconnections are
similar in size, and therefore lower total volumetric
porosities are needed to achieve the same functionality. Also, these regular interconnected pores provide
spacing for the vasculature required to nourish new
bone and remove waste products.56–59 Indeed,
Rekow et al.60 have shown that bone tissue grows
easily inside the pore structure of HA scaffolds fabricated by robocasting, colonizing even the microporosity within the rods.
In short, robocasting is a scaffold fabrication technique that allows the design of the macropore structure and the control of the microporosity within the
rods by adjusting the sintering conditions.7 The latter,
together with in vitro treatment by immersion in SBF,
provide means to control the cell affinity for the scaffold surfaces. These capabilities will eventually allow
the total porosity of the scaffolds to be reduced to
maintain the suitable mechanical strength without
jeopardizing their osteoconductive properties. Of
course, the pore architecture required for optimal
osteoconductivity and mechanical performance (i.e.,
strength and modulus matching the tissue properties)
has yet to be determined, and work in this direction is
under way in our laboratories. The ability to tailor porosity is obviously essential for the fabrication of loadbearing bone tissue engineering scaffolds, and in this
sense robocasting is a very promising technique. By
printing materials from a computer design, robotic
deposition offers the unique opportunity to explore
systematically the factors controlling the behavior of
porous scaffolds and to fabricate structures with optimal mechanical and biological behavior.
References
1. Cesarano J III, Segalman JR, Calvert P. Robocasting provides
moldless fabrication from slurry deposition. Ceram Ind
1998;148:94–102.
Journal of Biomedical Materials Research Part A
226
2. Cesarano J, Calvert P. Freeforming objects with low-binder
slurry. U.S. Patent 6,027,326, 2000.
3. Smay JE, Cesarano J, Lewis JA. Colloidal inks for directed assembly of 3-D periodic structures. Langmuir 2002;18:5429–
5437.
4. Sachlos E, Czernuszka JT. Making tissue engineering scaffolds work. Review on the application of solid freeform fabrication technology to the production of tissue engineering
scaffolds. Eur Cells Mater 2003;5:29–40.
5. Hollister S. Porous scaffold design for tissue engineering. Nat
Mater 2005;4:518–524.
6. Tay BY, Evans JRG, Edirisinghe MJ. Solid freeform fabrication of ceramics. Int Mater Rev 2003;48:341–370.
7. Miranda P, Saiz E, Gryn K, Tomsia AP. Sintering and robocasting of beta-tricalcium phosphate scaffolds for orthopaedic
applications. Acta Biomater 2006;2:457–466.
8. Michna S, Wu W, Lewis JA. Concentrated hydroxyapatite
inks for direct-write assembly of 3-D periodic scaffolds. Biomaterials 2005;26:5632–5639.
9. Miranda P, Pajares A, Saiz E, Tomsia AP, Guiberteau F. Fracture modes under uniaxial compression in hydroxyapatite
scaffolds fabricated by robocasting. J Biomed Mater Res A.
Available online DOI: 10.1002/jbmr.a.31272.
10. Ducheyne P, Radin S, King L. The effect of calcium-phosphate ceramic composition and structure on in vitro behavior. 1. Dissolution. J Biomed Mater Res 1993;27:25–34.
11. Sous M, Bareille R, Rouais F, Clement D, Amedee J, Dupuy
B, Baquey C. Cellular biocompatibility and resistance to compression of macroporous beta-tricalcium phosphate ceramics.
Biomaterials 1998;19:2147–2153.
12. Klein CPAT, Driessen AA, Degroot K, Vandenhooff A. Biodegradation behavior of various calcium-phosphate materials in
bone tissue. J Biomed Mater Res 1983;17:769–784.
13. Ryu HS, Youn HJ, Hong KS, Chang BS, Lee CK, Chung SS.
An improvement in sintering property of beta-tricalcium
phosphate by addition of calcium pyrophosphate. Biomaterials 2002;23:909–914.
14. Ohtsuki C, Kokubo T, Yamamuro T. Mechanism of apatite
formation on CaO–SiO2–P2O5 glasses in a simulated bodyfluid. J Non-Cryst Solids 1992;143:84–92.
15. Grenoble DE, Dunn KL, Katz JL, Gilmore RS, Murty KL.
Elastic properties of hard tissues and apatites. J Biomed
Mater Res 1972;6:221–233.
16. Weibull W. A statistical distribution function of wide applicability. J Appl Mech Trans ASME 1951;18:293–297.
17. Sanchez-Gonzalez E, Miranda P, Diaz-Parralejo A, Pajares A,
Guiberteau F. Effect of sol–gel thin coatings on the fracture
strength of glass. J Mater Res 2004;19:896–901.
18. Sanchez-Gonzalez E, Miranda P, Diaz-Parralejo A, Pajares A,
Guiberteau F. Influence of zirconia sol–gel coatings on the
fracture strength of brittle materials. J Mater Res 2005;20:
1544–1550.
19. Miranda P, Pajares A, Guiberteau F, Cumbrera FL, Lawn BR.
Role of flaw statistics in contact fracture of brittle coatings.
Acta Mater 2001;49:3719–3726.
20. Xin RL, Leng Y, Chen JY, Zhang QY. A comparative study of
calcium phosphate formation on bioceramics in vitro and in
vivo. Biomaterials 2005;26:6477–6486.
21. Deville S, Saiz E, Tomsia AP. Freeze casting of hydroxyapatite scaffolds for bone tissue engineering. Biomaterials
2006;27:5480–5489.
22. Almirall A, Larrecq G, Delgado JA, Martinez S, Planell JA,
Ginebra MP. Fabrication of low temperature macroporous
hydroxyapatite scaffolds by foaming and hydrolysis of an
alpha-TCP paste. Biomaterials 2004;25:3671–3680.
23. Charriere E, Lemaitre J, Zysset P. Hydroxyapatite cement
scaffolds with controlled macroporosity: Fabrication protocol
and mechanical properties. Biomaterials 2003;24:809–817.
Journal of Biomedical Materials Research Part A
MIRANDA ET AL.
24. del Real RP, Wolke JGC, Vallet-Regi M, Jansen JA. A new
method to produce macropores in calcium phosphate
cements. Biomaterials 2002;23:3673–3680.
25. Kawata M, Uchida H, Itatani K, Okada I, Koda S, Aizawa M.
Development of porous ceramics with well-controlled porosities and pore sizes from apatite fibers and their evaluations.
J Mater Sci Mater Med 2004;15:817–823.
26. Milosevski M, Bossert J, Milosevski D, Gruevska N. Preparation and properties of dense and porous calcium phosphate.
Ceram Int 1999;25:693–696.
27. Ramay HRR, Zhang M. Biphasic calcium phosphate nanocomposite porous scaffolds for load-bearing bone tissue engineering. Biomaterials 2004;25:5171–5180.
28. Ramay HR, Zhang MQ. Preparation of porous hydroxyapatite scaffolds by combination of the gel-casting and polymer
sponge methods. Biomaterials 2003;24:3293–3302.
29. Tian JT, Tian JM. Preparation of porous hydroxyapatite. J
Mater Sci 2001;36:3061–3066.
30. Ioku K, Yanagisawa K, Yamasaki N, Kurosawa H, Shibuya K,
Yokozeki H. Preparation and characterization of porous apatite ceramics coated with beta-tricalcium phosphate. Biomed
Mater Eng 1993;3:137–145.
31. Landi E, Celotti G, Logroscino G, Tampieri A. Carbonated
hydroxyapatite as bone substitute. J Eur Ceram Soc 2003;23:
2931–2937.
32. Liu DM. Influence of porosity and pore size on the compressive strength of porous hydroxyapatite ceramic. Ceram Int
1997;23:135–139.
33. Ota Y, Kasuga T, Abe Y. Preparation and compressive
strength behavior of porous ceramics with beta-Ca(PO3)2
fiber skeletons. J Am Ceram Soc 1997;80:225–231.
34. Bignon A. Optimization of the Porous Structure of Calcium
Phosphate Implants for Bone Substitutes and In Situ Release
of Active Principles. Doctoral Thesis. National Institute of
Applied Science, Lyon, 2002.
35. Barralet JE, Grover L, Gaunt T, Wright AJ, Gibson IR. Preparation of macroporous calcium phosphate cement tissue engineering scaffold. Biomaterials 2002;23:3063–3072.
36. Pilliar RM, Filiaggi MJ, Wells JD, Grynpas MD, Kandel RA.
Porous calcium polyphosphate scaffolds for bone substitute
applications—In vitro characterization. Biomaterials 2001;22:
963–972.
37. Tancred DC, McCormack BAO, Carr AJ. A synthetic bone
implant macroscopically identical to cancellous bone. Biomaterials 1998;19:2303–2311.
38. Chu TMG, Orton DG, Hollister SJ, Feinberg SE, Halloran JW.
Mechanical and in vivo performance of hydroxyapatite
implants with controlled architectures. Biomaterials 2002;23:
1283–1293.
39. Dong J, Kojima H, Uemura T, Kikuchi M, Tateishi T, Tanaka
J. In vivo evaluation of a novel porous hydroxyapatite to sustain osteogenesis of transplanted bone marrow-derived osteoblastic cells. J Biomed Mater Res 2001;57:208–216.
40. Gao TJ, Tuominen TK, Lindholm TS, Kommonen B, Lindholm TC. Morphological and biomechanical difference in
healing in segmental tibial defects implanted with Biocoral1
or tricalcium phosphate cylinders. Biomaterials 1997;18:219–
223.
41. Landi E, Tampieri A, Celotti G, Langenati R, Sandri M, Sprio
S. Nucleation of biomimetic apatite in synthetic body fluids:
Dense and porous scaffold development. Biomaterials 2005;
26:2835–2845.
42. Liu DM. Preparation and characterisation of porous hydroxyapatite bioceramic via a slip-casting route. Ceram Int 1998;
24:441–446.
43. Arita IH, Wilkinson DS, Mondragon MA, Castano VM. Chemistry and sintering behavior of thin hydroxyapatite ceramics
with controlled porosity. Biomaterials 1995;16:403–408.
MECHANICAL PROPERTIES OF ROBOCAST CALCIUM PHOSPHATE SCAFFOLDS
44. Miao X, Tan LP, Tan LS, Huang X. Porous calcium phosphate
ceramics modified with PLGA-bioactive glass. Mater Sci Eng
C. 2007;27:274–279.
45. Keller TS. Predicting the compressive mechanical-behavior of
bone. J Biomech 1994;27:1159–1168.
46. Kizuki T, Ohgaki M, Katsura M, Nakamura S, Hashimoto K,
Toda Y, Udagawa S, Yamashita K. Effect of bone-like layer
growth from culture medium on adherence of osteoblast-like
cells. Biomaterials 2003;24:941–947.
47. Kondo N, Ogose A, Tokunaga K, Umezu H, Arai K, Kudo N,
Hoshino M, Inoue H, Irie H, Kuroda K and others. Osteoinduction with highly purified beta-tricalcium phosphate in
dog dorsal muscles and the proliferation of osteoclasts
before heterotopic bone formation. Biomaterials 2006;27:
4419–4427.
48. Ouhayoun JP, Shabana AHM, Issahakian S, Patat JL, Guillemin G, Sawaf MH, Forest N. Histological-evaluation of natural coral skeleton as a grafting material in miniature swine
mandible. J Mater Sci Mater Med 1992;3:222–228.
49. Horch HH, Sader R, Pautke C, Neff A, Deppe H, Kolk A.
Synthetic, pure-phase betatricalcium phosphate granules
(Cerasorb1) regeneration in the ceramic for bone reconstructive surgery of the jaws. Int J Oral Maxillofac Surg 2006;
35:708–713.
50. Jung UW, Moon HI, Kim C, Lee YK, Kim CK, Choi SH. Evaluation of different grafting materials in three-wall intra-bony
defects around dental implants in beagle dogs. Curr Appl
Phys 2005;5:507–511.
51. Yoneda M, Terai H, Imai Y, Okada T, Nozaki K, Inoue H,
Miyamoto S, Takaoka K. Repair of an intercalated long bone
defect with a synthetic biodegradable bone-inducing implant.
Biomaterials 2005;26:5145–5152.
227
52. Daculsi G, Passuti N. Effect of the macroporosity for osseous
substitution of calcium-phosphate ceramics. Biomaterials
1990;11:86–87.
53. Bignon A, Chouteau J, Chevalier J, Fantozzi G, Carret JP,
Chavassieux P, Boivin G, Melin M, Hartmann D. Effect of
micro- and macroporosity of bone substitutes on their mechanical properties and cellular response. J Mater Sci Mater
Med 2003;14:1089–1097.
54. Hing KA, Best SM, Tanner KE, Bonfield W, Revell PA Mediation of bone ingrowth in porous hydroxyapatite bone graft
substitutes. J Biomed Mater Res Part A 2004;68A:187–200.
55. Karageorgiou V, Kaplan D. Porosity of 3D biornaterial scaffolds and osteogenesis. Biomaterials 2005;26:5474–5491.
56. Rose FR, Cyster LA, Grant DM, Scotchford CA, Howdle SM,
Shakesheff KM. In vitro assessment of cell penetration into
porous hydroxyapatite scaffolds with a central aligned channel. Biomaterials 2004;25:5507–5514.
57. Jin QM, Takita H, Kohgo T, Atsumi K, Itoh H, Kuboki Y.
Effects of geometry of hydroxyapatite as a cell substratum in
BMP-induced ectopic bone formation. J Biomed Mater Res
2000;51:491–499.
58. Woodard JR, Hilldore AJ, Lan SK, Park CJ, Morgan AW, Eurell JAC, Clark SG, Wheeler MB, Jamison RD, Johnson AJW.
The mechanical properties and osteoconductivity of hydroxyapatite bone scaffolds with multi-scale porosity. Biomaterials
2007;28:45–54.
59. Hing KA, Annaz B, Saeed S, Revell PA, Buckland T. Microporosity enhances bioactivity of synthetic bone graft substitutes. J Mater Sci Mater Med 2005;16:467–475.
60. Rekow D, Van Thompson P, Ricci JL. Influence of scaffold
meso-scale features on bone tissue response. J Mater Sci
2006;41:5113–5121.
Journal of Biomedical Materials Research Part A