Crecimiento de Hueso

Paper | Sheng Yue
● Facebook
● LinkedIn
Sheng Yue
● Google
Plus
PhD, MPhil, BSc(Eng), BEng
Menu
Skip to● Welcome
content● Research
❍ Pore network quantification
❍ 3D QA & QC tool
● Honours and Awards
● Publications
● Curriculum Vitae
❍ Education
Category Archives: Paper

Selective Laser Melting: A regular unit cell
approach for the manufacture of porous,
titanium, bone in-growth constructs, suitable
for orthopedic applications
October 12, 2012 by Sheng Yue
Lewis Mullen1,*, Robin C. Stamp1, Wesley K. Brooks1, Eric Jones2, Christopher J. Sutcliffe1
Article first published online: 6 OCT 2008
DOI: 10.1002/jbm.b.31219
Mullen, L., Stamp, R. C., Brooks, W. K., Jones, E. and Sutcliffe, C. J. (2009), Selective Laser
Melting: A regular unit cell approach for the manufacture of porous, titanium, bone in-
growth constructs, suitable for orthopedic applications. J. Biomed. Mater. Res., 89B: 325–334.
http://www.yue.cn.com/category/publication/paper/ (1 de 44) [13/01/2014 02:06:13 p.m.]
Paper | Sheng Yue
doi: 10.1002/jbm.b.31219
Author Information
1. Department of Engineering, The University of Liverpool, Liverpool, United Kingdom
2. Department of Advanced Technology, Stryker Orthopaedics, Cork, Ireland
Email: Lewis Mullen (lewis.mullen@liv.ac.uk)
*Department of Engineering, The University of Liverpool, Liverpool, United Kingdom
Publication History
● Issue published online: 2 APR 2009
● Article first published online: 6 OCT 2008
● Manuscript Accepted: 27 JUN 2008
● Manuscript Revised: 4 JUN 2008
● Manuscript Received: 4 APR 2008
Funded by
● The Engineering and Physical Sciences Research Council (EPSRC)
● Stryker Orthopaedics (New Jersey)
Abstract
In this study, a novel porous titanium structure for the purpose of bone in-growth has been
designed, manufactured and evaluated. The structure was produced by Selective Laser
Melting (SLM); a rapid manufacturing process capable of producing highly intricate,
functionally graded parts. The technique described utilizes an approach based on a defined
regular unit cell to design and produce structures with a large range of both physical and
mechanical properties. These properties can be tailored to suit specific requirements; in
particular, functionally graded structures with bone in-growth surfaces exhibiting
properties comparable to those of human bone have been manufactured. The structures
were manufactured and characterized by unit cell size, strand diameter, porosity, and
compression strength. They exhibited a porosity (10–95%) dependant compression strength
(0.5–350 Mpa) comparable to the typical naturally occurring range. It is also demonstrated
Paper | Sheng Yue
that optimized structures have been produced that possesses ideal qualities for bone in-
growth applications and that these structures can be applied in the production of orthopedic devices.
INTRODUCTION
Total hip and knee replacements are now in widespread use for the treatment of diseased
and damaged human joints. Such implants consist of elements functioning as an articulation
couple together with an element that fixes the implant firmly to the bone. For many years,
the primary method of fixation was with a combination of screws and acrylic bone cement
that provided a rigid bond with the host bone. 1 Such components are inherently stiff and the
nature of the loading leads to under-utilization of the surrounding host bone, with the
inevitable effect of stress shielding, necrosis and eventual failure. 2–4
Press-fit systems have been devised to encourage the bone to take a more active role in the
healing process whereby bone cells attach to the implant surface to effect fixation either by on-
growth or in-growth. 5, 6 Natural progression has led to greater emphasis being placed on
the integration of porous structures in orthopedic devices such that the bone cells are encouraged
to grow into the implant and become integral with it, thereby resulting in a more
physiological solution to total joint replacement.
It has been shown that porous implants can be designed to closely mimic the natural structure
and mechanical properties of bone compared to their solid-cast counterparts. 7-10 This can help
to reduce the mismatch of properties between the implant and host bone leading to a reduction in
the stress shielding effect and its associated bone necrosis.
The use of porous metal foams for the manufacture of orthopedic devices has been shown to
offer significant advantages in comparison to previously used manufacturing methods such as
metal beads and fiber mesh.1, 11-13 One typical structure is an open cellular, tantalum foam
(trade name Hedracell) that is used to produce devices such as acetabular cups. The foam
was designed to mimic the structure of trabecular bone and consequently exhibits an
interconnecting porosity of between 60 and 80%, with dodecahedral pores of size 427 μm ±
303 μm. 10, 14, 15 The foam has been implemented in devices as both a bonded surface coating
and as a shaped bulk structure that have been shown to produce excellent clinical results in the
short term. 16, 17 Research has shown that the optimum pore size for promoting bone in-growth is
in the range of 100–700 μm.14, 18, 19 Pore sizes of less than 100 μm fail to support the growth
of capillaries and also allow the bone cells to bridge pores, whilst pore sizes larger than 700 μm
do promote bone in-growth but at reduced rates and volumes.It has also been shown that porosity
is a key variable, influencing the quantity and rate of bone in-growth. Additionally, the volume
Paper | Sheng Yue
and rate of bone in-growth has been shown to improve as the porosity of the
structure increases.14, 20–23 Mechanical considerations however, limit the range of porosities at
the optimum pore size that can be employed to produce functional structures. Therefore, it
is necessary to consider the range of stresses likely to be encountered by typical structures and set
the overall requirements of mechanical property, porosity and pore size accordingly. Widmer et
al. showed that the maximum compressive stress in the acetabulum occurs during stumbling
and calculated this stress to be 30 Mpa.24
Conventional manufacturing methods 25 are either difficult to employ or are unsuccessful

in producing such porous devices with the tight constraints of porosity, optimum pore size,
and mechanical strength that are required. Rapid Prototyping (RP) methods have been proposed as
a way to overcome this issue,7, 26 and the process of Selective Laser Melting (SLM) is a
technology that shows great promise.SLM is ordinarily used to produce solid
metal components, 28 but is also suited to the manufacture of intricate porous geometries.
The majority of RP processes use a layered manufacturing technique, whereby discrete
layers constituting the structure are stacked and manufactured sequentially. 29
The SLM process has been designed to operate with any processable powder with the
correct rheology to flow through the metering mechanism, and these powders (either elemental
or alloy) are fused together by a scanning laser beam, 30 which fully melts the exposed powder.
This full melting is possible because of the high absorptance of the powder bed to the laser
light.31, 32 It has been shown by Santos et al. that by careful selection of the
manufacturing methodology, and the use of a much higher laser energy density, this process
can produce both fully dense solid, and porous geometries from a variety of bio-
compatible materials.30 In particular titanium and its alloys.
Open cellular foam structures produced by SLM can achieve controlled pore interconnectivity
and porosities by using small diameter wire-frame elements to create porous structures. Dimano
et al. demonstrated that these structures were able to support and encourage rapid and
sustained osteo-integration.33The structures approximate well to the interconnected network that
is created by the architecture of cancellous bone 34 however there is limited data available using
this method.
The objective of this article is to describe how a unit cell approach can be implemented to design
and manufacture novel bone in-growth structures by SLM, and to evaluate and more
importantly control the physical and mechanical properties of the structures produced.

Paper | Sheng Yue
MATERIALS AND METHODS

Samples for single strand testing (236–456 μm diameter by 5 mm high), compression and
porosity testing (cylinders 15Ø × 30 mm2 at various unit cell sizes) and SEM analysis (10 cells ×
10 cells × 2 cells) were produced from the most appropriate commercially pure titanium
(Sumitomo, Japan) using an MCP Realizer 2, 250 SLM system (MCP Tooling Technologies, UK).
The system uses an ytterbium fiber laser (200 W power CW, λ = 1.06 μm). Figure 1 shows
the machine and a schematic of the optical system used to control the movement of the nominal
50 μm diameter focused laser spot on the build area to a positional accuracy of ±5 μm. The
system operates in an over pressure argon environment with processing chamber oxygen levels
below 0.2%. The atmosphere within the chamber is circulated and filtered to remove process
bi-products (Ti nano powder formed from condensed Ti vapor) from the recycled gas.
Figure1.Schematic of the SLM process as used in the MCP Realizer system. [Color figure can
be viewed in the online issue, which is available at www.interscience.wiley.com.]
Parts were built in a layer-wise fashion on a substrate plate connected to an elevator that
moves vertically downwards allowing the controlled deposition of powder layers at 50-μm
intervals. Upon completion of the build the substrate plate was removed from the build chamber
and all un-fused powder was recycled. Test pieces were then cut from the substrate plates using
wire erosion. All individual parts were ultrasonically cleaned (20 min), dried (120°C for 24 h),
Paper | Sheng Yue
and stored prior to testing. Individual identification of all parts was built into the structures using
the SLM process (Magics, Materialise, Belgium). Processing conditions and the unit
cell configurations used in the study are shown in Table I.
TableI.The Variables that were Altered During the Study

Unit Cell Size (μm) Laser Power (W) Laser Exposure (μs)
600 62 200
800 96 400
1000 132 600
1200 167 800
1400 1000
A series of tests were undertaken prior to manufacturing to determine the powder rheology.
These tests included flowability, powder particle size distribution, and morphology. The
flowability tests were carried out using the Hall flow meter test, which measures the time taken for
a known quantity of powder to flow through a standardised funnel assembly according to
ASTM B213-03.
Two laser diffraction systems were used to measure the particle size distributions (Beckman
Coulter LS, dry dispersal, and Malvern Mastersizer 2000, wet dispersal, UK). Both systems
employed the Fraunhofer laser diffraction method fitted with a Mie mathematical model.
The powder particle morphology was assessed using SEM (Hitachi S-246N, Japan, resolution 10
nm). This technique enabled the high magnification viewing of the particles with a large depth of
field permitting detailed qualitative and quantitative assessment of particle morphology. SEM
was also used to confirm the pore size of the structures post manufacture.
The porosities of the structures were determined gravimetrically and the pieces were
subsequently subjected to compression testing. Equation 1 was used to determine the porosities
of the structures relative to fully dense CpTi (ρTi= 4507 kg/m3):
● (1)

Paper | Sheng Yue
Compression tests were carried out on an Instron 4505 test machine at room temperature
according to the Ashby method for compression of porous structures.35 The dimensions of
each sample were such that they contained at least seven cells in all directions to avoid edge
effects, and had a height-to-thickness ratio exceeding 1.5.
A statistically significant sample set of five samples for each test condition were assessed and
all individual data points were included in the results.
THE UNIT CELL APPROACH

The Unit Cell Approach describes the process of creating three-dimensional cellular structures from
a single repeating unit cell. Typically the approach can be summarized as:
1. A candidate CAD geometry is selected and a bounding box enclosing the entire CAD geometry
is calculated.
2. The bounding box is completely populated with cuboids of a defined unit cell size.
3. The cuboids are individually populated with the specified unit cell geometry.
4. The repeating cell geometry is trimmed to the candidate CAD geometry.
5. The final geometry is sliced at appropriate layer thickness and output to the appropriate
machine format.
It is possible to produce many cell geometries using this technique. This article is concerned with
the simple octahedral shown in Figure 2.
Figure2.Single octahedral unit cell (left) and examples of unit cells multiplied to populate shapes
Paper | Sheng Yue
and CAD geometries.
SLM has the ability to manufacture the wire elements of the structure whilst leaving the remainder
of the cell completely open. The major and interconnecting pore size diameters for an octahedral
unit cell are given by Eqs.2 and 3, whilst the representation of these pores are shown in Figure 3:
● (2)
● (3)
Figure 3. Schematic defining major and inter connected pore size in an octahedral
wireframe structure.
Paper | Sheng Yue
A software suite (Manipulator , University of Liverpool, UK) has been developed to design
porous geometries based on this approach. The small pore sizes and high porosities required for
bone in-growth structures necessitate that the diameter of the wireframe strands be made as small
as possible. Therefore, to minimize strand diameter the structures are built from single laser
point exposures (i.e. not by raster scanning, which is the scanning of multiple points to form lines,
as is the norm in typical laser based RP techniques). This outcome results in the strand
diameter being equal to the diameter of the laser melt pool produced by each exposure.
RESULTS
Powder Choice
Eight powders from four different manufacturers were rheologically analyzed as shown
in Figures 4 and 5. Figure 4 displays the results from the Hall flow tests and indicates that
the powders, although similar in specification, in reality show a wide range of flowability. Some of
the powders were unsuitable because they would not flow through the test rig, whilst others were
so flowable as to be un-meterable by the powder delivery system. Close analysis of the
powder fraction illustrated in Figure 5 indicates that a wide powder distribution with an
average particle size of less than 45 μm is ideal, hence the choice of Sumitomo <45 μm
powder. Figure 6 is a micrograph of the Sumitomo <45 μm powder showing the typical range
of particle sizes expected.

Paper | Sheng Yue
Figure 4. Powder characterization graph showing the flow rates for the tested powders.
Figure 5. Powder characterization graph showing the particle size distribution.

Paper | Sheng Yue
Figure 6. SEM image of Sumitomo <45 μm gas atomized titanium powder.
Unit Cell Properties

The three factors that determine the properties of a unit cell structure are the strand diameter,
the unit cell size and the porosity. As these factors can all be independently modified it
allows structures to be created with the same base cell geometry but with selectable and
controllable physical and mechanical properties. The following sections describe experimental
results characterizing the effect of these factors on the structures.
Strand Diameter
As previously stated, in order to produce high porosity material of relatively small unit cell size
the strand diameter of the constructs must be minimized.
These diameters are governed by the laser energy, the laser spot size, and the absorpivity and
heat transfer coefficients of the powder.

Paper | Sheng Yue
As outlined in Table I, five unit cell sizes were manufactured with the powers and exposures
to produce strand diameters ranging from 150 to 420 μm and corresponding single strand wires.
The results displayed in Figure 7 show that when strand diameter is plotted as a function of
beam energy, there is a logarithmic relationship. However, there is a clear difference between
the strand diameters achieved when processing vertical single strands compared with the
strand diameters achieved when constructing unit cells. Figure 8 shows SEM images of
structures built with the same unit cell size but with different laser energies.
Figure 7. Relationship between the laser beam energy and the strand diameters of wire
frame octahedral structures and thin strands.

Paper | Sheng Yue
Figure 8. SEM images of a 1200 μm octahedral, unit cell structure built at different laser
energies (structure on right = 0.0496 J, structure on left = 0.1336 J).
Unit Cell Size

Although the pore size is an important parameter for the control of the biological response to
porous in-growth structures, the unit cell approach described here is such that pore size is
actually controlled by the unit cell geometry. For example, pore size can be readily modified by
the inclusion of a strut between the diagonals of the octahedral structure thereby reducing the
size and perhaps conferring some additional strength element with little loss of
porosity. Consequently pore size was calculated as a function of the unit cell size at a defined
strand diameter using Eqs. (2) and (3) and was therefore not experimentally investigated any further.
Figure 9 displays the strand diameter as a function of unit cell size over the full range appropriate
for this investigation. It indicates that at high beam energy, smaller unit cells exhibit increased
strand diameter in comparison to larger unit cells at the same beam energy. This “blooming”
will clearly have an effect on the porosity of the constructs.

Paper | Sheng Yue
Figure 9. Showing how the strand diameter can vary with the unit cell size at constant beam energy.
Porosity
The secondary defining parameter for the selection of materials for bone in-growth applications
is porosity. The relationship of porosity to unit cell size is shown in Figure 10. The effect of
the increase in strand diameter described previously is apparent. At small cell sizes this
blooming effect on the strand diameter, as inferred from Figure 9, causes a decrease in the porosity
of the samples. This is an important result as it gives a lower limit for the cell size at a
particular porosity and will be fully researched at a later date.

Paper | Sheng Yue
Figure 10. Porosity characteristics at a range of beam energies of SLM manufactured CpTi
porous constructs over the full range of cell sizes. [Color figure can be viewed in the online
issue, which is available at www.interscience.wiley.com.]
Compressive Strength of Porous Structures

The third qualifying parameter defining the porous structures is compressive
strength. Figure 11 shows the effect that varying the porosity has on the compressive strength of
the structures. It is clear that the unit cell size selected has an effect on the compressive strength
of the structure at a particular porosity. For a given porosity, smaller unit cell sizes have
lower compressive strength. This again is an important result as it indicates that from a
strength point of view there is little to be gained by building constructs with unnecessarily small
cells. It is also apparent from Figure 11 that the only way to achieve very high strength, low
porosity materials with this unit cell and this optical arrangement is by using small cell sizes as
there is a maximum strand diameter that can be produced for each configuration.

Paper | Sheng Yue
Figure 11. The variation of compression strength with porosity for various unit cell sized
octahedral structures.
DISCUSSION
To keep close control on the process it was necessary to work with a wide powder particle
size distribution. Referring to Figure 5 the <45 μm Sumitomo powder was found to be
optimum. During the SLM operation un-fused material was recovered from the builds and
build chamber and then sieved. This ensured that the overall particle size distribution of
the reclaimed powder remained sensibly the same as powder particle size distributions have
been shown to have a significant impact on flowability, and hence are used as a measure
of consistency from run to run. 36 The sieving operation removes agglomerates of size greater
than 70 μm. All recovered powder was returned to the powder lot for subsequent reuse. Samples
were taken from the powder lot before each build and were retained and analyzed periodically
for both particle size distribution and rheology.
A small effect on size distribution was noted with a reduction in the number of finer particles
being observed. This reduction increased the flowability of the powder slightly but did not affect
the part quality, strength or microstructure.

Paper | Sheng Yue
The unit cell geometry selected for this study was based on an octahedron because of its suitability
for manufacture by the SLM process. More complex structures such as the dodecahedron
(the structure selected by Bobyn 10) do not process as well because they contain many horizontal
and low angle strands, which are difficult to build because they are unsupported along their
length. The importance of actual pore shape remains uncertain as citations in the literature refer
to porous structures based on many disparate shapes that still promote bone in-growth.1, 37
For example, regular wire meshes with a specific pore size and shape maybe compared in terms
of their strength, mechanical properties and bone in-growth performance to random
structured, foamed metals with irregular pore shapes and a distribution of pore sizes.
It is clear from the results that both strand diameter and quality have a great effect on the
mechanical performance and morphology of the structures. It is possible to manufacture
samples with a wide range of strand diameters however in the case described here, that of
the production of bone in-growth structures, unit cell size and porosity combine to confine
strand diameter to the bounds of the overall pore structures being produced.
As seen in Figure , it is the case that the strand diameter increases logarithmically with beam
energy to a maximum strand diameter of 420 μm. The maximum strand diameter is directly linked
to the diameter of the melt pool, which, in turn is governed by the diameter of the melt isotherm
that is produced by the beam. The size of the isotherms can be used to successfully predict
the diameter of the fully melted and the sintered sections of the strands at specific beam energies. 38
It can also be seen that there is a difference between the strands produced in unit cell structures
and single strands built vertically and away from neighboring strands. This is due to the following:
1. The strands that comprise the unit cell structures described here are built at angles conforming to
the octahedral structure (45°). When building at an angle to the vertical, each subsequent laser
spot is offset step-wise relative to the previous one resulting in a larger strand diameter.
2. Angled strands also exhibit a greater amount of sintered plaque as the laser is forced to fire on
parts of the powder bed where the heat can only be dissipated through the powder thereby
resulting in a larger number of powder particles being partially melted on to the surface. This leads
to a decrease in the porosity of the structures without the associated increase in the
compressive strength.
3. Latent heat from neighboring strands within close proximity can heat the powder bed sufficiently
to reduce the amount of energy required to melt the powder and hence lead to an increase in
strand diameter. This is evident in Figure where at high laser energy a decrease in cell size
and corresponding increase in proximity of neighboring strands causes an increase in
strand diameter.

Paper | Sheng Yue
It can be postulated from points one and two and referring to the graph in Figure 7, that
the separation of these two curve lines is relatively constant and is predominantly related to
the differences seen when building vertical and angled strands.
With reference to Figure 8, it can be seen that at low beam energies, the melt spots are joined
by small diameter necks, whereas at high beam energies smooth, continuous strands are
formed. Therefore, the quality of the links within the structure will also have an effect on
its properties.
The range of unit cell sizes selected in this study ranged from 600 to 1400 μm, with the upper
range probably being too large for bone in-growth applications. It may have been preferable
to produce samples at lower unit cell sizes than 600 μm, however, within the bounds of
the specification and limits of the machine and machine set-up, the lower levels,
although manufacturable were outside the specification regarding porosity. Figure shows the
extents of all samples manufactured in this study. Several interesting points can be deduced:
1. Small unit cell sizes with low porosities approach the performance of solid material. Further work
on this aspect is under way. It should be noted however that preliminary results show that the
15% porosity samples have the elemental properties of CpTi grade 2, and therefore the
structures here out perform conventionally produced CpTi by almost a factor of two in terms of
their compression yield strength.
2. The use of a large unit cell size generally results in more porous materials and
correspondingly weaker structures.
3. If we consider an application where bone in-growth, at optimum unit cell size, together with
porosity in the range 60–65%, and compressive strength of 30 Mpa, then the processing window
is relatively small. However, if a more general structure is required, where either the unit cell size
or porosity are not the main concerns then the process offers a rapid and flexible
manufacturing method for producing strong structures.
Indeed, as mentioned earlier, the whole aspect of unit cell size is somewhat arbitrary. It can be
seen when studying images of trabecular bone, an SLM manufactured unit cell, and trabecular
metal that there is a substantial variation in overall pore size.
The porosity of trabecular (cancellous) bone can lie anywhere in the range of 30% to more than
90% and is dependant upon bone location, sex, and state of health. 39 The porosities selected for
our study covered the whole range of porosities achievable but centered on those values pertinent
to the human pelvic and knee region.

Paper | Sheng Yue
The strength of the structures is clearly related to the porosity as is shown in Figure 11. It is our
belief that these graphs can be split into three main regions; the first region is termed
“the homologisation zone” and is where the structures have little strength due to the strands
not being fully formed. This is due to the lack of full melting within the strands of the structure.
The second region is termed the “structural development zone” where strength rapidly increases
as fully melted strands are formed and the structure develops. The third region termed
“the densification region” is where the structure thickens sufficiently for the strands to merge
and eventually lead to full densification. Such a process that offers ease of porosity variation
is obviously of appeal to a variety of engineering structures.
The compressive strength of trabecular bone is dependant upon its location within the body and
can lie anywhere within the range 0.2–25 Mpa. This large variation is primarily caused by
the difference in loading requirements on the bone, which are determined by its location in
the body. 40–42 Consequently, it is necessary to be able to adjust the strength of the porous
devices to match the specific requirements for its intended application. The SLM process
would appear to offer this flexibility. Figure 11 shows that the compressive strength can be
varied within the useful range of 20–350 Mpa at the corresponding levels of porosity between 75
and 15%.
The quality of the strand has a profound influence on the failure mode of the structures when
tested under compression. For example, samples with a unit cell size of 600 μm and built with
low beam energy (<0.03 J) showed necking, and were observed to form stress concentrations at
these necks that resulted in the structure failing catastrophically in shear as cracks
propagated through the strands. For the structures manufactured at high beam energy (>0.07
J), these stress concentration points were less apparent and the links tended to bend within
the structure and therefore deform rather than fail catastrophically. Clearly, the energy values
quoted are for that specific cell size and will vary as these changes.
APPLICATION OF THE TECHNOLOGY TO A SPECIFIC

DESIGN REQUEST
From the data obtained in this study it is possible to select process parameters to manufacture
a structure with requested characteristics. For example, a typical specification for structures
suitable for application in orthopedic surgery could be;
1. Pore size range: 450–600 μm.

2. Porosity range: 65% ± 5%.
3. Compressive strength: 40 Mpa.
Paper | Sheng Yue
Reference to Figure 11, however, shows that this is probably near the limit under these conditions
as the processing window is relatively small. Using an 800 μm unit cell, manufactured with a
beam energy of 0.0768 J a hip augment was produced that exceeded these requirements and can
be seen in Figure 12 . The structure exhibited a major pore diameter of 500 μm, a
fully interconnecting porosity of 65%, and a compressive strength (in the build direction) of 53 Mpa.
Figure 12. Photograph of the hip augment produced by SLM.
CONCLUSIONS
A unit cell approach has been devised and implemented for the manufacture of porous bone
in-growth structures by SLM. The SLM technique has been progressed to not only create
solid geometries, but so that structures can be produced that exhibit a range of cell sizes,
porosities, and mechanical properties that can be tailored to suit a variety of applications.
The novelty of this technology lies in the fact that orthopedic structures can be produced
with predefined physical and mechanical properties, so that their strength and bone in-
growth characteristics can be optimized for specific locations within the body.
Paper | Sheng Yue
The technique shows great promise for the manufacture of porous in-growth structures,
however further work is required to enlarge the currently narrow processing window and to refine
the technology before it can be considered to be a viable manufacturing alternative.
Acknowledgements
The authors thank Stryker Orthopaedics, New Jersey for their continued support and
financial backing of the project.
REFERENCES
1. Ryan G,Pandit A,Apatsidis DP. Fabrication methods of porous metals for use in
orthopaedic applications. Biomaterials 2006; 27: 2651–2670.
2. Huiskes R,Weinans H,Van Rietbergen B. The relationship between stress shielding and
bone resorption around total hip stems and the effects of flexible materials. Clin Orthop Relat
Res 1992; 274: 124–134.
3. Sumner DR,Galante JO. Determinants of stress shielding: Design versus materials versus
interface. Clin Orthop Relat Res 1992; 274: 202–212.
4. Van Lenthe GH,De Waal Malefijt MC,Huiskes R. Stress shielding after total knee replacement
may cause bone resorption in the distal femur. J Bone Joint Surg B 1997; 79: 117–122.
5. Buehler KO,Venn-Watson E,D’Lima DD,Colwell CWJr. The Press-Fit Condylar total knee system:
8- to 10-year results with a posterior cruciate-retaining design. J Arthroplasty 2000; 15: 698–701.
6. Schai PA,Thornhill TS,Scott RD. Total knee arthroplasty with the PFC system. J Bone Joint Surg
B 1998; 80: 850–858.
7. Lopez-Heredia MA,Goyenvalle E,Aguado E,Pilet P,Leroux C,Dorget M,Weiss P,Layrolle P.
Bone growth in rapid prototyped porous titanium implants. J Biomed Mater Res A 2008; 85:
664–673.
8. Imwinkelried T. Mechanical properties of open-pore titanium foam. J Biomed Mater Res A 2007;
81: 964–970.
9. Engh CA,Bobyn JD,Glassman AH. Porous-coated hip replacement. The factors governing
bone ingrowth, stress shielding, and clinical results. J Bone Joint Surg B 1987; 69: 45–55.
10. Bobyn JD,Stackpool GJ,Hacking SA,Tanzer M,Krygier JJ. Characteristics of bone ingrowth
and interface mechanics of a new porous tantalum biomaterial. J Bone Joint Surg B 1999; 81:
907–914.
11. Kienapfel H,Sprey C,Wilke A,Griss P. Implant fixation by bone ingrowth. J Arthroplasty 1999;
14: 355–368.
12. Oh IH,Nomura N,Masahashi N,Hanada S. Mechanical properties of porous titanium
compacts prepared by powder sintering. Scr Mater 2003; 49: 1197–1202.
13. Jasty M,Bragdon CR,Haire T,Mulroy JrRD,Harris WH. Comparison of bone ingrowth into
Paper | Sheng Yue
cobalt chrome sphere and titanium fiber mesh porous coated cementless canine
acetabular components. J Biomed Mater Res 1993; 27: 639–644.
14. Bobyn JD,Toh KK,Hacking SA,Tanzer M,Krygier JJ. Tissue response to porous tantalum
acetabular cups: A canine model. J Arthroplasty 1999; 14: 347–354.
15. Zardiackas LD,Parsell DE,Dillon LD,Mitchell DW,Nunnery LA,Poggie R. Structure, metallurgy,
and mechanical properties of a porous tantalum foam. J Biomed Mater Res 2001; 58: 180–187.
16. Gruen TA,Poggie RA,Lewallen DG,Hanssen AD,Lewis RJ,O’Keefe TJ,Stulberg SD,Sutherland
CJ. Radiographic evaluation of a monoblock acetabular component: A multicenter study with 2- to
5-year results. J Arthroplasty 2005; 20: 369–378.
17. Hacking SA,Bobyn JD,Toh KK,Tanzer M,Krygier JJ. Fibrous tissue ingrowth and attachment
to porous tantalum. JBiomed Mater Res 2000; 52: 631–638.
18. Hulbert SF,Young FA,Mathews RS,Klawitter JJ,Talbert CD,Stelling FH. Potential of
ceramic materials as permanently implantable skeletal prostheses. J Biomed Mater Res 1970; 4:
433–456.
19. Nilles JL,Coletti JM Jr,Wilson C. Biomechanical evaluation of bone porous material interfaces.
J Biomed Mater Res 1973; 7: 231–251.
20. Bobyn JD,Pilliar RM,Cameron HU,Weatherly GC. The optimum pore size for the fixation of
porous surfaced metal implants by the ingrowth of bone. Clin Orthop Relat Res 1980; 150: 263–270.
21. Dennis Bobyn J. Fixation and bearing surfaces for the next millennium. Orthopedics 1999; 22:
810–812.
22. Assad M,Jarzem P,Leroux MA,Coillard C,Chernyshov AV,Charette S,Rivard CH. Porous
titanium-nickel for intervertebral fusion in a sheep model, Part 1: Histomorphometric
and radiological analysis. J Biomed Mater Res B Appl Biomater 2003; 64: 107–120.
23. Kusakabe H,Sakamaki T,Nihei K,Oyama Y,Yanagimoto S,Ichimiya M,Kimura J,Toyama
Y. Osseointegration of a hydroxyapatite-coated multilayered mesh stem. Biomaterials 2004;
25: 2957–2969.
24. Widmer KH,Zurfluh B,Morscher EW. Load transfer and fixation mode of press-fit acetabular
sockets. J Arthroplasty 2002; 17: 926–935.
25. Banhart J. Manufacture, characterization and application of cellular metals and metal foams.
Prog Mater Sci 2001; 46: 559–632.
26. Hayashi T,Maekawa K,Tamura M,Hanyu K. Selective laser sintering method using titanium
powder sheet toward fabrication of porous bone substitutes. Int J A Solid Mech Mater Eng 2006;
48: 369–375.
27. Jones E,Sutcliffe C,Stamp R. Laser-produced porous surface. 2007.Patent no. EP1683593.
28. Abe F,Osakada K,Shiomi M,Uematsu K,Matsumoto M. The manufacturing of hard tools
from metallic powders by selective laser melting. J Mater Process Technol 2001; 111: 210–213.
29. Gerbhardt A. Rapid Prototyping. Munich: Hanser; 2003.
30. Santos EC,Shiomi M,Osakada K,Laoui T. Rapid manufacturing of metal components by
laser forming. Int J Mach Tools Manuf 2006; 46: 1459–1468.
31. Lu L,Fuh J,Wong Y. Laser-Induced Materials and Processes for Rapid Prototyping. Norwell,
MA: Kluwer Academic Publishers; 2001.
32. Tolochko NK,Laoui T,Khlopkov YV,Mozzharov SE,Titov VI,Ignatiev MB. Absorptance of
powder materials suitable for laser sintering. Rapid Prototyping J 2000; 6: 155–160.
33. Dimano N,An Y,Jiang Y,Ngo C,Kulesha G,Stamp R,Zhang R. The osseointegration of
Paper | Sheng Yue
porous materials using a rabbit femoral defect model (poster). Chicago; 2006.
34. Gibson L,Ashby M. Cellular solids: Structure and properties. New York: Cambridge University
Press; 1999.
35. Ashby M,Evans A,Fleck N,Gibson L,Hutchinson J,Wadley H. Metal Foams: A Design
Guide. Burlington, MA: Butterworth Heinemann; 2000.
36. Freeman R. The Flowability of Powders—An empirical approach. International Conference
on Powder and Bulk Solids Handling. London, UK; 2000.
37. Turner TM,Sumner DR,Urban RM. A comparative study of porous coatings in a weight-bearing
total hip-arthroplasty model. J Bone Joint Surg A 1986; 68: 1396–1409.
38. Tolochko NK,Arshinov MK,Gusarov AV,Titov VI,Laoui T,Froyen L. Mechanisms of selective
laser sintering and heat transfer in Ti powder. Rapid Prototyping J2003; 9: 314–326.
39. Carte DR,Hayes WC. The compressive behavior of bone as a two-phase porous structure. J
Bone Joint Surg A 1977; 59: 954–962.
40. Goldstein SA. The mechanical properties of trabecular bone: Dependence on anatomic location
and function. J Biomech 1987; 20: 1055–1061.
41. Rho JY,Kuhn-Spearing L,Zioupos P. Mechanical properties and the hierarchical structure of
bone. Med Eng Phys 1998; 20: 92–102.
42. Galante J,Rostoker W,Ray RD. Physical properties of trabecular bone. Calcif Tissue Res 1970; 5:
236–246.
Posted in: Paper, Publication |
Evaluation of 3-D bioactive glass scaffolds

dissolution in a perfusion flow system with X-
ray microtomography
October 12, 2012 by Sheng Yue
Volume 7, Issue 6, June 2011, Pages 2637–2643
● Department of Materials, Imperial College London, London SW7 2AZ, UK
Abstract
Paper | Sheng Yue
Bioactive glass has high potential for bone regeneration due to its ability to bond to bone
and stimulate osteogenesis whilst dissolving in the body. Although three-dimensional (3-D)
bioactive glass scaffolds with favorable pore networks can be made from the sol–gel
process, compositional and structural evolutions in their porous structures during degradation
in vivo, or in vitro, have not been quantified. In this study, bioactive glass scaffolds were put in
a simulated body fluid flow environment through a perfusion bioreactor. X-ray
microtomography (μCT) was used to non-destructively image the scaffolds at different
degradation stages. A new 3-D image processing methodology was developed to quantify
the scaffold’s pore size, interconnect size and connectivity from μCT images. The
accurate measurement of individual interconnect size was made possible by a principal
component analysis-based algorithm. During 28 days of dissolution, the modal interconnect size
in the scaffold was reduced from 254 to 206 μm due to the deposition of mineral phases.
However, the pore size remained unchanged, with a mode of 682 μm. The data presented
are important for making bioactive glass scaffolds into clinical products. The technique described
for imaging and quantifying scaffold pore structures as a function of degradation time is applicable
to most scaffold systems.
Keywords
● Scaffold;
● Image analysis;
● Porosity;
● Bioactive glass;
● Dissolution
1. Introduction
Synthetic bone grafts (scaffolds) are needed that can regenerate diseased or damaged bone,
replacing the need for autografts. Autografts are the current gold standard whereby bone
is transplanted, usually from the pelvis. Their disadvantage is the lack of supply and donor
site morbidity, which can cause pain and other complications. Scaffolds are also vital components
in bone tissue engineering strategies, where the aim is to culture cells on scaffolds prior
to implantation. Scaffolds for bone regeneration are designed to act as temporary templates for
bone and blood vessel growth. To do this they require a three-dimensional (3-D) porous network
[1], [2] and [3] with high porosity and connectivity for solute transport; with suitable pore and,
more importantly, interconnect size distributions to enable cell migration, bone ingrowth
and vascularization. The minimum interconnect diameter for human bone ingrowth is
Paper | Sheng Yue
generally considered to be 100 μm [4] and [5]. The scaffold should also form a direct bond to the
host bone by enabling cell adhesion and activity. Ideally the scaffold would degrade as the host
bone forms, with the degradation products encouraging osteogenesis [2] and [6]. With all
these different criteria to match, a method for quantifying degradation in vitro is needed as a
first screening tool for new scaffolds.
Bioactive glasses are promising materials for bone tissue engineering scaffold fabrication [6] and
[7]. They can bond to bone, degrade in the body and release ions which stimulate bone
regeneration. When implanted in the body, bioactive glasses form a surface layer of
crystalline hydroxyl carbonate apatite (HCA) [8], [9] and [10], which is similar to bone mineral and
is responsible for glass bonding with bone. The osteogenic property of these glasses has
been attributed to the release of soluble silica and calcium ions during dissolution [6]. Using a sol–
gel foaming process, 3-D bioactive glass scaffolds can be fabricated from bioactive glasses
[11], achieving compressive strengths [2] and [12] and pore structures similar to human
trabecular bone, with macropore diameters of 300–600 μm and interconnect diameters in excess
of 100 μm [13]. Having achieved structural similarity, quantifying their dissolution behavior
is another essential step in their development.
X-ray microtomography (μCT, micro-CT or XMT) can non-destructively image scaffolds in

three dimensions [14] and [15]. Coupled with imaging processing, μCT has become an
established tool for scaffold quantification [16], [17], [18] and [19]. For example, the internal
pore structure can be characterized from μCT images, providing quantitative information such
as: percentage porosity, pore and interconnect size distributions, network connectivity, pore
shape and strut thickness [13], [20], [21] and [22]. The interaction between the scaffold material
and its host environment, including bone ingrowth, has also been observed and quantified with
μCT, both in vivo and ex vivo [20], [22], [23] and [24]. Furthermore, the non-destructive nature
of μCT can enable the evaluation of samples in situ, where experiments such as compression
and mineral deposition are performed in vitro, whilst non-invasively scanning the scaffold [25],
[26] and [27]. In addition, the 3-D scaffold images can be used to generate meshes for
computational fluid dynamic and mechanical simulations [13], [28] and [29]. Therefore,
many aspects of the performance of scaffolds can be evaluated and then improved in a non-
invasive and quantifiable manner, reducing product design costs and time.
Current bioactivity and degradation tests for bioceramics usually consist of immersing a sample in
a fixed volume of simulated body fluid (SBF), then placing it in an incubator and observing the
final extent of HCA formation. However, bioactive glasses, and many other bioceramics, are
dynamic materials which undergo dissolution before a HCA layer can deposit on their surface. A
test that quantifies this precursor stage is required. For example, a high glass concentration can
cause calcite to deposit on the glass surface instead of HCA [30]. The in vivo environment is not
Paper | Sheng Yue
a closed environment and saturation of ions is not expected. Therefore, the prescreening process
for new scaffolds needs improvement. Perfusion bioreactors are widely used in vitro in
tissue engineering applications because perfusion flow can enhance mass transport and
introduce shear stress, thereby improving tissue growth [31]. The aim of this work is to use
a perfusion bioreactor with a local inhomogeneous flow rate, combined with μCT imaging and
image analysis, to study the degradation behavior of bioactive glass foam scaffolds in an SBF
flow environment. In order to achieve this, new and improved algorithms were needed with
better accuracy and greater automation (using a principal component analysis (PCA)-
based approach) than those developed previously for scaffold quantification [13] and [21].
The hypothesis was that the porous scaffold would undergo dissolution over time, causing
the interconnects of the pores to open. Secondary objectives were to determine whether
the deposition of the HCA layer could be observed using this technique; whether changes in pore
size could be measured; and whether preferential flow within the scaffold would cause
heterogeneous dissolution.
2. Materials and methods

2.1. Bioactive glass scaffolds synthesis
Bioactive glass foams of 70S30C composition (70 mol.% SiO2, 30 mol.% CaO) were prepared using
a sol–gel foaming method as previously described [11] and [12]. The sol preparation began
with mixing 0.2 N nitric acid with deionized water using a magnetic stirrer, followed by addition
of tetraethyl orthosilicate (TEOS) and calcium nitrate (all Sigma–Aldrich) in order. The initial
molar ratio of water to TEOS (R ratio) was 12:1. Aliquots of 50 ml of the sol were combined with 3
ml of 5 vol.% hydrofluoric acid (HF, catalyst) and 0.35 ml of Teepol (surfactant, Thames Mead
Ltd, London), then foamed using vigorous agitation. As the foamed sol approached the
gelation point, it was cast into cylindrical Teflon moulds, sealed and aged at 60 °C for 72 h.
Previous work has shown that Teflon moulds are non-reactive and facilitate removal, as well
as improving the homogeneity of the glass composition [32]. The samples were then dried in air for
a total of 94 h at 60, 90 and 130 °C. Finally, the scaffolds were stabilized at 600 °C for 3 h and
then sintered at 800 °C for 2 h. The thermal processing was optimized in previous studies, with
the drying process designed to prevent cracking and the sintering used to improve
compressive strength [12].
2.2. Perfusion bioreactor system

An SBF solution, with ion concentrations similar to those of human extracellular fluid, was
prepared following the Kokubo method [33]. Cylindrical bioactive glass scaffolds, 6 mm in
diameter and 7 mm high and weighing 0.13 ± 0.2 g, together with a length of non-degradable
Paper | Sheng Yue
silica glass fibre, were wrapped with Teflon tape and inserted into a Teflon ring (termed a
scaffold assembly). (Note that the silica glass fibre was added to calibrate the X-ray attenuation
in the μCT, enabling the change in density of the scaffold to be monitored). The Teflon
ring containing the sample was then placed into the cylindrical chamber of a perfusion
bioreactor (Gradient Container, Minucells Minutissue, Weinheim, Germany [34]), as illustrated
in Fig. 1. In order to prevent flow from bypassing the scaffold, the outer diameter (12 mm) of
the Teflon ring matched the inner diameter of the chamber, while the Teflon ring was also
wrapped with Teflon tape to ensure a tight seal. Three samples were run in separate experiments
on separate occasions using fresh SBF, but using the same perfusion chamber.
Fig. 1. Schematic of the dissolution experimental setup. The arrows on the silicone tubing indicate
the SBF flow direction.
The SBF flow circulation, which was directed through the scaffold from bottom to top with a flow
rate of 1 ml min 1, was maintained with a peristaltic pump (Masterflex, model 07519–25, Cole-
Palmer Instrument Co., Niles, IL, USA). A 1 l reservoir of SBF was maintained at 37 °C in
an incubator.

Paper | Sheng Yue
2.3. Optical emission inductive coupled plasma spectroscopy

For each of the three samples, at dissolution time points of 0, 1, 8, 24 h, 7 days and 28 days, 50 ml
of the SBF solution was collected for analysis. The entire SBF was replaced after 7 days. The
ion concentrations of P, Si and Ca in the collected solutions were measured with optical
emission inductive coupled plasma (ICP) spectroscopy (Thermo Scientific ICP Spectrometer,
Model iCAP 6300 Series Duo).
2.4. X-ray microtomography

μCT was used to scan the scaffold assembly as a function of degradation time in SBF flow.
Each scaffold assembly was scanned before the dissolution study using a lab-based μCT unit
(Phoenix X-ray Systems and Services GmbH, Wunstorf, Germany) at 100 kV and 100 μA, and with
a voxel size of 8 μm. At 24 h, 7 days and 28 days, the dissolution experiment was paused and
the scaffold assembly was removed out from the bioreactor, dried at 37 °C overnight and
rescanned by μCT with the same setting as that of the 0 h scan. After μCT scanning, the assembly
was reinstalled and the dissolution experiment resumed.
2.5. Scanning electron microscopy and energy-dispersive X-

ray spectrometry
Two additional bioactive glass scaffolds were used for electron microscopic analysis as it
is destructive – one was scanned a 0 h and the other was immersed in SBF flow for 3 days and
then scanned, and one of the three repeats was also scanned after 28 days. Each was dried
and scanned using a scanning electron microscope (Leo 1525) equipped for energy-dispersive X-
ray spectrometry (EDX) to characterize the morphology as well as the composition. For the 28
days sample, a μCT scan was performed prior to scanning electron microscopy (SEM) and
EDX analysis to provide confirmation of the μCT data interpretation.
2.6. Three-dimensional image analysis

The initial stage of 3-D image processing builds on a previously published technique [13] and
[21]. The segmentation process steps are as follows.
1.
A 3 × 3 × 3 median filter was applied to all μCT images to remove noise.

Paper | Sheng Yue
2.
For each scaffold assembly, the μCT images at 24 h, 7 days and 28 days were numerically
registered to the appropriate image taken at 0 h. The registration process resampled the images
first (by an initial factor of 4 and a refined factor of 2), then transformed the moving image
rigidly (Rigid3-D transform, only rotation and translation), based on the Mean Squares
similarity metrics using ITK [35] (Insight Segmentation and Registration Toolkit, ITK).
3.
The Teflon ring and non-degradable silica glass fibre in the μCT images were thresholded and
then located using a connected component labeling algorithm [36]. The X-ray attenuation
properties of the Teflon and silica glass fibre were assumed to remain constant, allowing all the
μCT images to be normalized to each other.
4.
Global thresholding was applied to the normalized images to classify each voxel as either scaffold
or background according to its intensity. The thresholding values were determined from
histograms of the images, picking the value equidistant for the peak background and scaffold levels.
5.
A dilation-based distance transform was then applied to the binarized images. The background
phase in the images was filled step by step by dilating of the strut phase (26-connectivity kernel)
until all the voxels were filled. The number of steps required to reach each replaced voxel was
then recorded as the value of that position, forming a 3-D distance map.
6.
Using the pore centroid locations (local maxima in the distance map), a 3-D watershed
transform [37] was then applied to the distance map and the background voxels were segmented
and labeled as separated macropores according to the topological variation.
7.
Any background voxels that neighbored (with 26-connectivity) more than one labeled pore
were identified and grouped as interconnects.
Once the background phase in the μCT image was segmented and labeled as macropores
and interconnects, the following quantification procedures were applied:
1.
Paper | Sheng Yue
The pores and interconnects touching the border of the region of interest were not used for
property measurement because they may only be part of the pore/interconnect and the
following measuring method would not be appropriate to reflect the sizes of those partial
pores/interconnects.
2.
The volume, equivalent diameter (the diameter of a sphere with the same volume as a
pore), connectivity and many other characteristics of the pores were quantified.
To quantify the interconnect size and morphology, an improved method was developed and
applied that was significantly different from previous studies [13] and [21] (see Fig. 2). This
new methodology first located the principal plane of each interconnect (disc shape in Fig. 2c) by
a PCA method. The steps used to find the principal plane and then measure the interconnect size are:
1.
For each interconnect, the PCA was started by calculating the covariance matrix of its
voxels coordinates.
2.
The eigenvector which corresponded to the smallest eigenvalue of the covariance matrix is
the normal to the interconnect’s principal plane. The principal plane also passes the centroid of
the interconnect voxels. Therefore, for each interconnect in the 3-D space, there is only one
unique principal plane, which can be located by its normal and the centroid of the interconnect.
3.
In order to measure the interconnect size, all voxels of a given interconnect were projected to
its principal plane. This allows the 3-D disc-like shape to be reduced to a 2-D projection. A
convex hull algorithm was then used to determine the effective area of the interconnect’s
projection in two dimensions. An equivalent diameter was also determined equal to the diameter of
a circle with the same area as the interconnect.
4.
The distributions of the pore size and interconnect size for each scaffold were produced, weighted
by volume fraction and equivalent area fraction, respectively.

Paper | Sheng Yue
Fig. 2. 3-D rendering of μCT images show: (a) a 3-D porous bioactive glass foam scaffold; (b) a
sub-volume of the scaffold containing two color-labeled pores (scaffold struts rendered with
50% transparency); (c) two pores from (b) with their neighboring interconnects; (d) equivalent
pores and interconnects of (c) from the quantification algorithms; (e) a single interconnect with
its coordinate bounding box. The bounding box diagonal that was used as an equivalent diameter
by Atwood et al. [21] was 413.8 μm in length; (f) the same interconnect in (e) which was rotated to
its principal plane. The length of the equivalent diameter, which was obtained by the PCA
method, was 273.6 μm. In (c) and (d), pores were rendered with 50% transparency, and each
cone indicates the normal of the principal plane for every interconnect.

Paper | Sheng Yue
This new method of performing a PCA of the interconnects and then measuring them in the
principal plane significantly improved the accuracy of the dimensions obtained, as shown
by comparing Fig. 2e and f. In this example the bounding box diagonal of the single interconnect
was 414 μm (using the method from Ref. [21]), where the equivalent diameter of the
same interconnect obtained by the PCA method was 274 μm, showing a 51% improvement over
the old method.
3. Results
High-resolution μCT allowed the non-destructive scanning of the same bioactive glass scaffold at
a series of dissolution time points and, by registering these scans with respect to each other, a
highly accurate quantification of the dissolution kinetics was determined. The changes in one
scaffold at four dissolution time points is shown in Fig. 3a–d as a 3-D rendering and as 2-D slices
in Fig. 3e–h (perpendicular to the flow direction). The regions with higher intensity values in
the volume image indicate where the material has a higher X-ray attenuation, and are rendered
with a warmer color in Fig. 3. The scaffold phase had a higher intensity level than the
background phase, which was easily distinguished. The amount of attenuation was a combination
of both the composition (higher atomic number atoms attenuate more) and density;
correlations between the μCT scan and SEM-EDX illustrated that the red regions contained a
greater percentage of calcium. Fig. 3a and e shows the initial pore network prior to dissolution (0
h). The color variation of the struts reflects the inhomogeneous initial composition within
the amorphous scaffold; note that the high X-ray attenuating material was observed more
frequently close to the surface of the struts.

Paper | Sheng Yue
Fig. 3. 3-D sub-volumes and transverse slices (perpendicular to the flow direction) from
the normalized μCT images of the scaffold at time points of 0 h (a, e), 24 h (b, f), 7 days (c, g) and
28 days (d, h). (Note, the colors indicate relative X-ray attenuation and are all normalized to
each other. Warmer color indicates higher attenuation material).
After 24 h in SBF flow, there was a slight reduction in overall image intensity level (compare Fig.
3b and f to a and e). However, the reduction was quite heterogeneous: the voxel intensity
reduced much more in some regions than in others, whilst in some regions the intensity
increased. These observations suggest that in SBF flow conditions material leached out from
the struts of the scaffold unevenly, while redepositing at other locations.

Paper | Sheng Yue
After 7 days of immersion (Fig. 3c and g), the voxel intensity level of the volume image
significantly reduced compared to the 0 and 24 h images, indicating dissolution of calcium.
In addition, there was a significant decrease in the size of many interconnects, with only a
few increasing in size. After 7 days there were also a number of regions of localized deposition
of highly attenuating species at the periphery of a number of pores in the top left-hand corner of
the scaffold (red and yellow regions), which were not present at 0 and 24 h. As confirmed below
using EDX, these were depositions of Ca-rich mineral phases. After 28 days of immersion (Fig.
3d and f), almost all areas of the scaffold showed reduced attenuation. The exceptions were the
upper left region in the 3-D image and the top and bottom regions in the 2-D slice, which had
highly attenuating (red) regions in the walls and around the periphery of many of the pores.
The deposition of Ca-rich phases that began at 7 days seemed to have increased at 28 days, shown
by an increase in red intensity at the pore surface in Fig. 3d compared to Fig. 3c.
The quantification illustrates that all pores were well connected, with a mean of five disc-
like interconnects to neighboring pores, 3.6 (72%) of which, on average, had an equivalent
diameter greater than 100 μm (Fig. 4b). The modal equivalent pore diameter remained unchanged
at 682 μm for the four time points; however, there was a slight broadening of the pore
size distribution at 28 days towards smaller pores, as deposition on the insides of the pores
occurs. Prior to dissolution, 95% (in area) of interconnects had an equivalent diameter larger
than 100 μm, with a modal value of 254 μm. After 24 h, the distribution became skewed to
slightly smaller interconnect sizes (modal value of 254 μm), which had reduced further to 206
μm after 7 days. The reduction in average interconnect size halted after 7 days, with the
mode remaining at 206 μm at 28 days. Note that although the average reduced, there were
locations where the interconnect size enlarged, but this was less common. Note also that
throughout the dissolution process 95% of the interconnects remained larger than 100 μm,
providing an easy transport mechanism of nutrients and cells between macropores.

Paper | Sheng Yue
Fig. 4. Size distributions of the equivalent diameter of the (a) pores and (b) interconnects.
The 0 h struts had a uniform surface (Fig. 5a) and the EDX analysis (Fig. 5d) gave a
composition similar to the nominal composition of the bioactive glass, i.e. 70 mol.% of SiO2 and
30 mol.% of CaO. After 3 days in SBF flow the scaffold’s surface had become roughened in

Paper | Sheng Yue
some regions (Fig. 5b), perhaps due to partial dissolution. EDX (Fig. 5e) of these regions showed
a significant reduction in Ca, supporting the calcium dissolution hypothesis. After 28 days (Fig. 5c),
a material with a crystalline appearance covered most scaffold surface areas. EDX analysis (Fig. 5f)
of this deposit indicated high calcium and phosphorous levels, suggesting that an HCA layer
had formed. The presence of this crystalline structure was observed on all the scaffold
surfaces imaged in the scanning electron microscope for the 28 days sample.

Paper | Sheng Yue
Fig. 5. SEM images (a–c) and associated EDX (d–f) analysis of the 70S30C bioactive scaffolds at 0
h (a, d), 3 days (b, e) and 28 days (c, f) time points.
Changes in the SBF Si, P and Ca levels with time relative to the initial composition are plotted in
Fig. 6, as determined by ICP analysis. During the first 7 days, the Si level increased by 8.8 ppm
(Fig. 6a). The P levels decreased after 7 days, probably due to phosphate depositing on the
glass surface as calcium phosphate-rich phases (Fig. 6b) [26]. Ca levels increased slightly in the
first 24 h (Fig. 6c), due to an initial burst of Ca dissolution, exceeding the rate of redeposition. After
7 days, the Ca level decreased to 5.7 ppm lower than fresh SBF after 7 days, indicating that
the deposition of Ca-rich phases dominates over dissolution. After sampling at 7 days, the
SBF solution was completely replaced with fresh medium. Over the next 21 days, the Ca and P
levels dropped considerably, indicating further deposition of calcium phosphate-rich phases
(as confirmed via EDX above and the decrease in pore size in Fig. 4a).

Paper | Sheng Yue
Fig. 6. ICP analysis for the (a) Si, (b) P and (c) Ca ion concentrations in SBF solution at each
time point.
4. Discussion
The non-destructive and non-invasive nature of μCT enabled the capture of the changing structure
of bioactive glass scaffolds during dissolution for the first time. Quantification of the μCT scans of
the sol–gel-derived 70S30C bioactive glass foam showed that throughout 28 days of degradation
it retained its interconnected porous network suitable for bone regeneration applications both
in terms of pore size and interconnect diameter. The pore size had a log-normal distribution, with
a mode of 682 μm, which did not change over 28 days of immersion. The macropores within
Paper | Sheng Yue
the network were well connected (with an average of 3.6 large interconnects per pore).
More importantly, the majority of interconnects (95% of their total area) had equivalent diameters
in excess of 100 μm, meaning that they are potentially large enough for vascularized bone
growth throughout the scaffold.
When characterizing the porous structure of a scaffold, a transition from qualitative interpretation
to quantitative measurement was made possible with the novel 3-D image analysis techniques. The
3-D scan of the scaffold was segmented using a sequence of image processing algorithms as
described in Materials and methods above. The measurement of the interconnect sizes has
been made more accurate compared to previous work (51% for the example given in the
Methods Section), and fully automated by introducing a PCA-based algorithm. The PCA
algorithm allows objects that are randomly oriented in 3-D space to be aligned such that an
optimal projection onto a 2-D plane is achieved and a more accurate measurement
obtained. Although this method was demonstrated using interconnects in this study,
the methodology is applicable to a wide range of features, from bone nodules to fibrous features
in composite scaffolds. The improvement obtained by this method is limited to objects that
are elongated in either one or two directions (i.e. that have one or two strong principal components).
The μCT quantification showed that calcium distribution in the unreacted scaffolds was
not homogeneous. The sol–gel process was previously thought to be a method of choice for
producing homogeneous compositions. However, previous work has shown that the use of
calcium nitrate as a calcium precursor, which is the traditional choice, does not give a
homogeneous calcium distribution [32]. Here, the μCT showed this to be true in foams.
The inhomogeneity was due to the calcium nitrate being soluble in water and calcium not going
into the silica network until a temperature of 400 °C was reached. After drying, the struts consist
of silica nanoparticles coated with calcium nitrate. During stabilization and sintering,
the nanoparticles fused together, the calcium nitrate dissociated and calcium diffused into the
struts. The foams have greater homogeneity than monoliths [32] but cannot be considered to
be completely homogeneous.
During dissolution there were two dominant events: dissolution of the scaffold followed by
mineral deposition. Over the first 24 h, calcium content was depleted near the surface of the
scaffold due to exchange of calcium ions with H+ from the solution. Soluble silica was also lost to
the solution. μCT quantification showed that the majority of interconnects shrunk slightly during
the first 7 days period, but the pore size remained similar. The interconnects shrunk as
calcium phosphate was deposited on the surface of the scaffolds, growing into the
interconnect region. This was confirmed both by ICP, where a reduction of Ca and P was measured
in SBF, and via SEM–EDX, where deposits were visually observed and Ca and P were quantified
by EDX. In addition, μCT showed a decrease in calcium (shown by X-ray attenuation) within
the scaffold and an increase in dense/high atomic number deposits on the pore surfaces,
including near the interconnects. Note that a few interconnects enlarged, indicating preferential
Paper | Sheng Yue
flow channels formed in the perfusion bioreactor, causing dissolution of the glass at edges of
these interconnects. Unfortunately, since a laboratory source μCT machine was used
with polychromatic X-rays, only the relative changes in calcium level, rather than the absolute
levels, could be determined.
The 3-D μCT images and image analysis indicated that there was little change in the
strut morphology between 7 and 28 days, which implies that the silica did not continue to
degrade. The ICP results agree with this, as the silica content of the SBF increased less between 7
and 10 days than between 0 and 7 days. The local pH must be greater than nine for SiOSi bonds
to break and for soluble silica to be released, and this is quite likely in flow conditions in the first
7 days as the calcium was released by the glass. The flow then removed the calcium ions from close
to the glass, causing a buffering effect. The HCA deposition will also reduce the surface area of
the glass that is in contact with the SBF. Bioactive glasses have been found to degrade in
vivo; therefore long-term degradation must be due to the action of osteoclasts. This is preferable
to continual degradation by aqueous dissolution as degradation by continuous dissolution may be
too fast to allow bone remodelling to occur, which can take weeks or months, depending on the
defect site.
5. Conclusions
High-resolution μCT was used for the first time to non-destructively image 3-D sol–gel-
derived bioactive glass scaffolds during the dissolution process in SBF flow. The 3-D image
analysis techniques required significant improvement from previous work to provide
accurate quantification. This was done by introducing a PCA-based method to fully
automatically measure the interconnect sizes within the scaffold, providing an increased
accuracy. The quantification shows that the sol–gel-derived 70S30C bioactive glass foam has
a suitable porous structure for use as a bone tissue engineering scaffold. The pore size
remained relatively unchanged, with a constant mode of 682 μm, whereas the interconnects
reduced slightly in size (from 254 to 206 μm) due to the deposition of mineral phases. A
crystalline phase matching the composition of hydroxyl carbonate apatite was found to form over
the majority of the scaffold after 28 days in an SBF flow environment.
Acknowledgements
J.R.J. was a Royal Academy of Engineering/EPSRC Research Fellow. The authors
acknowledge financial support from the Philip Leverhulme Prize and EPSRC (EP/F001452).
They also thank Dr. R.C. Atwood and R.W. Hamilton for assistance with μCT imaging and Dr.
M Ardakani for SEM. S.Y. gratefully acknowledges a bursary from Stryker Orthopaedics.

Paper | Sheng Yue
References
❍ [1]
❍ R. Langer, J.P. Vacanti
❍ Tissue engineering
❍ Science, 260 (1993), pp. 920–926
❍ [2]
❍ J.R. Jones, P.D. Lee, L.L. Hench
❍ Hierarchical porous materials for tissue engineering
❍ Phil Trans R Soc A, 364 (2006), pp. 263–281
❍ [3]
❍ R. Emadi, F. Tavangarian, S.I.R. Esfahani, A. Sheikhhosseini, M. Kharaziha
❍ Nanostructured forsterite coating strengthens porous hydroxyapatite for bone tissue engineering
❍ J Am Ceram Soc, 93 (2010), pp. 2679–2683
❍ [4]
❍ S.F. Hulbert, S.J. Morrison, J.J. Klawitter
❍ Tissue reaction to three ceramics of porous and non-porous structures
❍ J Biomed Mater Res, 6 (1972), pp. 347–374
❍ [5]
❍ O. Gauthier, J.-M. Bouler, E. Aguado, P. Pilet, G. Daculsi
❍ Macroporous biphasic calcium phosphate ceramics: influence of macropore diameter
and macroporosity percentage on bone ingrowth
❍ Biomaterials, 19 (1998), pp. 133–139
❍ [6]
❍ L.L. Hench, J.M. Polak
❍ Third-generation biomedical materials
❍ Science, 295 (2002), pp. 1014–1017
❍ [7]
❍ I.D. Xynos, A.J. Edgar, L.D.K. Buttery, L.L. Hench, J.M. Polak
❍ Gene-expression profiling of human osteoblasts following treatment with the ionic products
of Bioglass 45 S 5 dissolution
❍ [8]
❍ L.L. Hench, H.A. Paschall
❍ Direct chemical bond of bioactive glass–ceramic materials to bone and muscle
❍ [9]
❍ T. Kokubo, H. Takadama
❍ How useful is SBF in predicting in vivo bone bioactivity
❍ Biomaterials, 27 (2006), pp. 2907–2915
❍ [10]
❍ R. Emadi, F. Tavangarian, S.I.R. Esfahani
❍ Biodegradable and bioactive properties of a novel bone scaffold coated with nanocrystalline
Paper | Sheng Yue
bioactive glass for bone tissue engineering

❍ Mater Lett, 64 (2010), pp. 1528–1531
❍ [11]
❍ P. Sepulveda, J.R. Jones, L.L. Hench
❍ Bioactive sol–gel foams for tissue repair
❍ [12]
❍ J.R. Jones, L.M. Ehrenfried, L.L. Hench
❍ Optimising bioactive glass scaffolds for bone tissue engineering
❍ Biomaterials, 27 (2006), pp. 964–973
❍ [13]
❍ J.R. Jones, G. Poologasundarampillai, R.C. Atwood, D. Bernard, P.D. Lee
❍ Non-destructive quantitative 3D analysis for the optimisation of tissue scaffolds
❍ Biomaterials, 28 (2007), pp. 1404–1413
❍ [14]
❍ S.R. Stock
❍ X-ray microtomography of materials
❍ Int Mater Rev, 44 (1999), pp. 141–164
❍ [15]
❍ S.R. Stock
❍ Recent advances in X-ray microtomography applied to materials
❍ Int Mater Rev, 53 (2008), pp. 129–181
❍ [16]
❍ G.H. van Lenthe, H. Hagenmuller, M. Bohner, S.J. Hollister, L. Meinel, R. Müller
❍ Nondestructive micro-computed tomography for biological imaging and quantification of
scaffold-bone interaction in vivo
❍ Biomaterials, 28 (2007), pp. 2479–2490
❍ [17]
❍ R.E. Guldberg, C.L. Duvall, A. Peister, M.E. Oest, A.S.P. Lin, A.W. Palmer et al.
❍ 3D imaging of tissue integration with porous biomaterials
❍ Biomaterials, 29 (2008), pp. 3757–3761
❍ [18]
❍ J.R. Jones, R.C. Atwood, G. Poologasundarampillai, S. Yue, P.D. Lee
❍ Quantifying the 3D macrostructure of tissue scaffolds
❍ J Mater Sci Mater Med, 20 (2009), pp. 463–471
❍ [19]
❍ C. Renghini, V. Komlev, F. Fiori, E. Verné, F. Baino, C. Vitale-Brovarone
❍ Micro-CT studies on 3-D bioactive glass-ceramic scaffolds for bone regeneration
❍ Acta Biomater, 5 (2009), pp. 1328–1337
❍ [20]
❍ M. Bohner, G.H. Van Lenthe, S. Grünenfelder, W. Hirsiger, R. Evison, R. Müller
❍ Synthesis and characterization of porous [beta]-tricalcium phosphate blocks
❍ Biomaterials, 26 (2005), pp. 6099–6105
❍ [21]
Paper | Sheng Yue
❍ R.C. Atwood, J.R. Jones, P.D. Lee, L.L. Hench

❍ Analysis of pore interconnectivity in bioactive glass foams using X-ray microtomography
❍ Scripta Mater, 51 (2004), pp. 1029–1033
❍ [22]
❍ A.C. Jones, C.H. Arns, D.W. Hutmacher, B.K. Milthorpe, A.P. Sheppard, M.A. Knackstedt
❍ The correlation of pore morphology, interconnectivity and physical properties of 3D ceramic
scaffolds with bone ingrowth
❍ Biomaterials, 30 (2009), pp. 1440–1451
❍ [23]
❍ S. Cartmell, K. Huynh, A. Lin, S. Nagaraja, R. Guldberg
❍ Quantitative microcomputed tomography analysis of mineralization within three-
dimensional scaffolds in vitro
❍ J Biomed Mater Res A, 69A (2004), pp. 97–104
❍ [24]
❍ A.C. Jones, C.H. Arns, A.P. Sheppard, D.W. Hutmacher, B.K. Milthorpe, M.A. Knackstedt
❍ Assessment of bone ingrowth into porous biomaterials using MICRO-CT
❍ Biomaterials, 28 (2007), pp. 2491–2504
❍ [25]
❍ T. Ohgaki, H. Toda, M. Kobayashi, K. Uesugi, T. Kobayashi, M. Niinomi et al.
❍ In-situ high-resolution X-ray CT observation of compressive and damage behaviour of
aluminium foams by local tomography technique
❍ Adv Eng Mater, 8 (2006), pp. 473–475
❍ [26]
❍ S. Yue, P.D. Lee, G. Poologasundarampillai, Z. Yao, P. Rockett, A. Devlin et al.
❍ Synchrotron X-ray microtomography for assessment of bone tissue scaffolds
❍ J Mater Sci Mater Med, 21 (2010), pp. 847–853
❍ [27]
❍ H. Hagenmüller, S. Hofmann, T. Kohler, H. Merkle, D. Kaplan, G. Vunjak-Novakovic et al.
❍ Non-invasive time-lapsed monitoring and quantification of engineered bone-like tissue
❍ Ann Biomed Eng, 35 (2007), pp. 1657–1667
❍ [28]
❍ B. Porter, R. Zauel, H. Stockman, R. Guldberg, D. Fyhrie
❍ 3-D Computational modeling of media flow through scaffolds in a perfusion bioreactor
❍ J Biomech, 38 (2005), pp. 543–549
❍ [29]
❍ R. Singh, P.D. Lee, T.C. Lindley, R.J. Dashwood, E. Ferrie, T. Imwinkelried
❍ Characterization of the structure and permeability of titanium foams for spinal fusion devices
❍ Acta Biomater, 5 (2009), pp. 477–487
❍ [30]
❍ J.R. Jones, P. Sepulveda, L.L. Hench
❍ Dose-dependent behavior of bioactive glass dissolution
❍ [31]
❍ B.D. Porter, A.S.P. Lin, A. Peister, D. Hutmacher, R.E. Guldberg
Paper | Sheng Yue
❍ Noninvasive image analysis of 3D construct mineralization in a perfusion bioreactor

❍ Biomaterials, 28 (2007), pp. 2525–2533
❍ [32]
❍ S. Lin, C. Ionescu, S. Baker, M.E. Smith, J.R. Jones
❍ Characterisation of the inhomogeneity of sol–gel-derived SiO2–CaO bioactive glass and a strategy
for its improvement
❍ J Sol–Gel Sci Technol, 53 (2010), pp. 255–262
❍ [33]
❍ T. Kokubo
❍ Bioactive glass ceramics: properties and applications
❍ Biomaterials, 12 (1991), pp. 155–163
❍ [34]
❍ S. Seitz, K. Ern, G. Lamper, D. Docheva, I. Drosse, S. Milz et al.
❍ Influence of in vitro cultivation on the integration of cell–matrix constructs after
subcutaneous implantation
❍ Tissue Eng, 13 (2007), pp. 1059–1067
❍ [35]
❍ Ibá ez L, Schroeder W, Ng L, Cates J, Consortium IS. The ITK Software Guide. Second
edition, updated for ITK version 2.4, 2005.
❍ [36]
❍ R. Haralick, L. Shapiro
❍ Computer and robot vision
❍ Addison-Wesley Longman Publishing Co., Boston, MA (1992)
❍ [37]
❍ A.P. Mangan, R.T. Whitaker
❍ Partitioning 3-D surface meshes using watershed segmentation
❍ IEEE Trans Visual Comput Graphics, 5 (1999), pp. 308–321
Corresponding author. Tel.: +44 20 75946749; fax: +44 20 75946757.
Copyright 2011 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Posted in: Paper, Publication |
Post
2014 Sheng Yue | Powered by WordPress | Theme Mon Cahier by Bluelime Media
navigation Last updated 20 September

2012

Crecimiento de Hueso

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Crecimiento de Hueso

Hochgeladen von

Copyright:

Verfügbare Formate

Paper | Sheng Yue

PhD, MPhil, BSc(Eng), BEng

❍ Pore network quantification

● Honours and Awards

Category Archives: Paper

Article first published online: 6 OCT 2008

Email: Lewis Mullen (lewis.mullen@liv.ac.uk)

*Department of Engineering, The University of Liverpool, Liverpool, United Kingdom

Conventional manufacturing methods 25 are either difficult to employ or are unsuccessful

http://www.yue.cn.com/category/publication/paper/ (4 de 44) [13/01/2014 02:06:13 p.m.]

MATERIALS AND METHODS

TableI.The Variables that were Altered During the Study

http://www.yue.cn.com/category/publication/paper/ (6 de 44) [13/01/2014 02:06:13 p.m.]

THE UNIT CELL APPROACH

and CAD geometries.

http://www.yue.cn.com/category/publication/paper/ (9 de 44) [13/01/2014 02:06:13 p.m.]

Figure 5. Powder characterization graph showing the particle size distribution.

Figure 6. SEM image of Sumitomo <45 μm gas atomized titanium powder.

Unit Cell Properties

http://www.yue.cn.com/category/publication/paper/ (11 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (12 de 44) [13/01/2014 02:06:13 p.m.]

Unit Cell Size

http://www.yue.cn.com/category/publication/paper/ (13 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (14 de 44) [13/01/2014 02:06:13 p.m.]

Compressive Strength of Porous Structures

http://www.yue.cn.com/category/publication/paper/ (15 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (16 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (17 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (18 de 44) [13/01/2014 02:06:13 p.m.]

APPLICATION OF THE TECHNOLOGY TO A SPECIFIC

1. Pore size range: 450–600 μm.

Figure 12. Photograph of the hip augment produced by SLM.

Posted in: Paper, Publication |

Evaluation of 3-D bioactive glass scaffolds

Volume 7, Issue 6, June 2011, Pages 2637–2643

● Department of Materials, Imperial College London, London SW7 2AZ, UK

X-ray microtomography (μCT, micro-CT or XMT) can non-destructively image scaffolds in

2. Materials and methods

2.2. Perfusion bioreactor system

http://www.yue.cn.com/category/publication/paper/ (27 de 44) [13/01/2014 02:06:13 p.m.]

2.3. Optical emission inductive coupled plasma spectroscopy

2.4. X-ray microtomography

2.5. Scanning electron microscopy and energy-dispersive X-

2.6. Three-dimensional image analysis

http://www.yue.cn.com/category/publication/paper/ (28 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (30 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (31 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (32 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (33 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (34 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (35 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (36 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (37 de 44) [13/01/2014 02:06:13 p.m.]

http://www.yue.cn.com/category/publication/paper/ (40 de 44) [13/01/2014 02:06:13 p.m.]

bioactive glass for bone tissue engineering

❍ R.C. Atwood, J.R. Jones, P.D. Lee, L.L. Hench

❍ Noninvasive image analysis of 3D construct mineralization in a perfusion bioreactor

Corresponding author. Tel.: +44 20 75946749; fax: +44 20 75946757.

Posted in: Paper, Publication |

navigation Last updated 20 September

http://www.yue.cn.com/category/publication/paper/ (44 de 44) [13/01/2014 02:06:13 p.m.]

Das könnte Ihnen auch gefallen