Embed Size (px)
Citation preview
Stereolithography printing of PCL-based bone graft substitutes
Department of Materials
Imperial College London
2021
MPhil Thesis
Oystein Ovrebo
CID: 01598727
Supervisors:
Molly M. Stevens
Jonathan RT. Jeffers
2
Stereolithography printing of PCL-based bone graft substitutes
Abstract: Stereolithography printing stands out as a promising additive manufacturing technique
for the production of bone graft substitutes and other tissue-engineered constructs due to a high spatial
resolution, generation of internal microstructures, and high productivity. A current obstacle for the
application of this manufacturing method is the limited library of biocompatible materials and non-
standard characterization methods. In this work, a low molecular weight polycaprolactone (PCL)
triacrylate resin was synthesized for stereolithography printing, characterized using ISO standard testing
methods, and printed into complex lattice structures. The mechanical properties of the resin were
characterized in tension (E = 12.2 MPa, σbreak = 1.49 MPa, εbreak = 13 %) and in compression (E = 97.99 MPa,
σmax = 21.1 MPa, εmax = 30.1 %). Furthermore, the trade-off between scaffold stiffness and porosity was
investigated for two cellular lattice structures (octet-truss and tetradecahedron) in silico using Finite
Element Analysis (FEA). Initial FEA results indicate that the octet-truss lattice exhibits a higher stiffness at
a lower relative density (ρ’ < 20 %) than the tetradecahedron lattice.
3
Declaration of Originality
I, Øystein Øvrebø, declare that this is my original scientific work content reported within this
thesis, and was acquired and analysed by myself at Imperial College London, unless otherwise
stated. All collaborative efforts are noted by name.
The Copyright of this thesis rests with the author and is made available under a Creative Commons Attribution Non-Commercial No Derivatives license. Researchers are free to copy, distribute or transmit the thesis on the condition that they attribute it, that they do not use it for commercial purposes and that they do not alter, transform or build upon it. For any reuse or redistribution, researchers must make clear to others the license terms of this work.
4
Acknowledgements
This work was conducted as an MPhil project at the Department of Materials, Imperial College London.
The financial support was provided by Biotechnology and Biological Sciences Research Council,
Engineering and Physical Sciences Research Council, and Medical Research Council through the UKRMP2
Smart Materials Hub, and the Aker Scholarship.
I would like to thank Professor Molly Stevens and Dr Jonathan Jeffers for their support, supervision, and
for welcoming me into such an outstanding scientific and social environment. Their trust in me has
allowed me to thrive and has fostered my curiosity and appreciation for the complexity of the biomedical
world. They have given me the confidence that by getting the details right even the most complex system
can be engineered to our desire.
Further I want to express my appreciation to Dr. Axel Moore, Dr. Jonathan Wojciechowski, and Dr. Cécile
Echalier for their mentoring, thorough feedback, inspirational discussions, and forever lasting patience
with me. They have introduced me to everything from working in a chemistry lab to presenting data in a
clear manner. A field that previously seemed like a set of connected black boxes now looks like the finest
clockwork.
The two last years I have spent at Imperial would not have been the same without my cohort – Benge,
Chris, EJ, James, Jiaqing, Tabasom, and Tristan. They have always been to count on, either for small chats
in the lab, pints at the pub, or post-work climbing sessions, and have made me feel at home.
The environment in the Stevens Group and Biomechanics group has been magnificent. I am honoured to
have made so many friends and to have learned from so many smart people.
Lastly, I want to give a special thanks to Akemi, Brittany and Sirli, who’s extraordinary efforts made the
work in this MPhil possible.
With this I am ending a two-year chapter of my life. Two years that has helped me build character, taught
me how to think, and made me believe that through biomedical engineering I can make a difference. I will
look back with gratitude.
5
Contents Abbreviations ............................................................................................................................................... 6
1. Background ............................................................................................................................................... 7
1.1 Bone and bone repair ......................................................................................................................... 7
1.2 Demand for bone regeneration .......................................................................................................... 8
1.3 Bone grafts and substitutes for bone regeneration ........................................................................... 8
1.4 Scaffold manufacturing through stereolithography printing ............................................................ 10
1.5 Synthetic and biodegradable materials for SLA printing. ................................................................. 13
2. Project aims ............................................................................................................................................ 15
3. Experimental method ............................................................................................................................ 16
3.1 PCL functionalization (MRes) ............................................................................................................ 16
3.2 Resin optimisation ............................................................................................................................ 16
3.3 Custom SLA set-up ............................................................................................................................ 17
3.4 Lattice design .................................................................................................................................... 18
3.5 Mechanical testing ............................................................................................................................ 18
3.6 Accelerated degradation study ......................................................................................................... 20
3.7 FEA Modelling ................................................................................................................................... 20
3.8 Statistics and data handling .............................................................................................................. 22
4. Results..................................................................................................................................................... 23
4.1 Resin synthesis .................................................................................................................................. 23
4.2 Resin formulation .............................................................................................................................. 23
4.3 Printability ......................................................................................................................................... 24
4.4 Mechanical testing ............................................................................................................................ 25
4.5 FEA Modelling ................................................................................................................................... 27
4.6 Degradation ...................................................................................................................................... 30
5. Discussion ............................................................................................................................................... 31
5.1. Printing optimisation ....................................................................................................................... 31
5.2 Mechanical testing and FEA .............................................................................................................. 33
5.3 Biodegradation .................................................................................................................................. 39
5.4 Limitations ......................................................................................................................................... 39
6. Future work ............................................................................................................................................ 41
7. Conclusion .............................................................................................................................................. 42
8. References .............................................................................................................................................. 44
6
Supplementary Material ............................................................................................................................ 51
Abbreviations β-TCP β-Tricalcium Phosphate μCT Micro Computed Tomography BCC Body Centre Cubic BG Bone Graft BGS Bone Graft Substitute CL Caprolactone CLIP Continuous Liquid Interface Production CS Cell Size DCM Dichloromethane Dia. Diameter DLP Digital Light Projection DoF Degree of Functionalization FDM Fused Deposition Modeling FEA Finite Element Analysis HA Hydroxyapatite MW Molecular weight MEHQ 4-methoxyphenol MSC Mesenchymal Stem Cells NMR Nuclear Magnetic Resonance OB+ 2,5-Bis(5-tert-butyl-benzoxazol-2-yl)thiophene PCL Polycaprolactone PGSA Poly(Glycerol-Co-Sebacate) Acrylate PDLLA Poly(D,L-Lactic Acid) PLA Polylactic Acid PPF Poly(propylene fumarate) PTMC Poly(trimethylene carbonate) Rstrut Strut Radius RT Room Temperature SLA Stereolithography SLM Selective Laser Melting TMC Trimethylene Carbonate TPO Diphenyl(2,4,6-trimethylbenzoyl) Phosphine Oxide TPP Two Photon Polymerization UTS Ultimate Tensile Stress
7
Stereolithography printing of PCL-based bone graft substitutes
1. Background
1.1 Bone and bone repair Bone is a tissue with remarkable regenerative capabilities, particularly in young and healthy people. This
roots back to its highly dynamic nature and ability to constantly remodel according to need, such as upon
mechanical stimulation [1]. The regeneration process consists of a wide series of events and overlapping
phases. While many cell types are responsible for regulating bone mass (mesenchymal stem cells (MSCs),
macrophages and monocytes), the primary actors are osteoblasts, osteoclasts and osteocytes.
Osteoblasts deposit new bone tissue while osteoclasts control resorption. Osteocytes are derived from
osteoblast and are embedded inside the bone matrix. They are important in the signalling process during
bone absorption and formation and are stimulated by mechanical strain. Due to the involvement of
multiple cell types, the bone repair process is often described as an osteo-immunological phenomenon.
The cascades of this phenomenon are described in detail by Loi and co-workers [2], with the five main
stages of secondary fracture repair being described as: 1) Hematoma, where blood quickly coagulates at
injury site before forming a provisional fibrin matrix, and causes release of cytokines that recruit immune
cells such as neutrophils. Neutrophils then secrete macrophage- and monocyte-recruiting mediators
leading to the next stage. 2) Acute inflammation, in this stage, macrophages remove the fibrin matrix and
necrotic cells, and monocyte-derived osteoclasts resorb bone fragments and fracture edges. It is also the
stage where fibroblasts, MSCs and osteoprogenitors are recruited. The two first stages can last up to a
week before the fracture tissue is replaced with granulation tissue. 3) Granulation tissue, is an immature
fibrous extracellular matrix filled with MSCs and developing vasculature. The hypoxic environment of the
tissue causes differentiation into chondrocytes producing cartilage in the place of the granulation tissue.
4) Callus formation, describes the hypertrophic cartilage formed at the fracture site that eventually
becomes the scaffold for endochondral bone formation. The cartilage is replaced with vascularised bone,
allowing the bone to advance inward and substitute the cartilage. This new bone can be described as
woven bone. The woven bone combined with calcification of the cartilage core is referred to as hard callus.
5) Remodelling, occurs when the immature woven bone and cartilage is removed by osteoclasts and a
remodelling process occurs where normal bone integrity is obtained. The remodelling process can take
months or years.
8
1.2 Demand for bone regeneration Healthy bone is crucial to the musculoskeletal system and human mobility. However, bone is prone to
defects and damage, particularly with aging. This is part of the reason why orthopaedic surgery, which
includes treatment of bone defects, is the third largest spending group of the NHS [3]. Orthopaedics is
responsible for more than 1.2m surgical procedures per annum for the NHS [4], with knee (>102k per
annum) and hip (>91k per annum) replacement being the most conducted surgeries [5]. Considering
Europe and the US, the annual cost of bone defect repair exceeds US$3 billion [6]. In some cases, it is
necessary to replace bone with a permanent medical device, for instance with a titanium hip implant, but
in many cases the aim is to guide the bone’s natural regenerative potential.
In most cases the body can repair damages itself, but critically sized fractures and other larger defects are
not able to fully repair. There are many factors affecting this such as nature of damage, age, obesity [7],
genetics, activity level, and nutrition [8], among others. In these cases, it is necessary to assist the
regeneration, with bone grafts being a natural choice for treatment. Bone grafts (BG) and bone graft
substitutes (BGS) are generally used to stimulate and guide the formation of bone. They can be used in
trauma induced fractures where augmentation using bone grafts is required for stability [9]. In
osteosarcoma they can be used to replace excised tissue [10]. Another major problem in the western
world is lower backpain caused by compression of the spinal cord from degeneration of the intervertebral
disc, a condition that will affect two thirds of all adults at some point in their life [11]. To alleviate the pain
and regain vertebral spacing, a procedure called spinal fusion can be conducted where the degenerated
spinal disc is removed, a cage is inserted to obtain the correct spacing, and bone grafts are packed into
the space to generate new bone thereby fusing the vertebras together [11]. The fusion is conducted to
distract the intervertebral space to prevent further compression and damage of the spinal cord [11,12].
Spinal fusion stands out as the most common application for BG and BGS, and more than 151,000 lumbar
spinal fusions are performed in the US every year [11].
1.3 Bone grafts and substitutes for bone regeneration Autografts are the gold standard for bone regeneration. Autografts are preferred due to good
regeneration and no risk of immune rejection [13]. Typically, the bone is harvested from the pelvis,
however, the grafting quantity is limited, typically 30-40 mL, and often inadequate, and the donor site is
prone to serious morbidities such as infection and large hematomas [13]. In a study where bone was
harvested from the iliac crest of 75 patients, 4% suffered from persistent pain and 2.7% from persistent
sensory disturbance [14]. Alternatively, the bone can be obtained from other humans (allografts) or
animals (xenografts), primarily bovine bone [6]. To minimize the safety risk, these grafts must go through
9
heavy sterilization and decellularization processes which alter the mechanical and biological properties of
the bone [15,16]. An alternative strategy is to use hybrid grafts which are allo- or xenografts enhanced
using synthetic materials. With this approach the graft retains some bioactivity and its mechanics can be
tuned with the synthetic material. For instance, xeno-hybrids have been developed by reinforcing
decellularized bovine bone with a polymeric mixture. The bovine matrix mineral composition similar to
that of human bone, the poly(L-lactic-co-ε-caprolactone) improves the mechanical properties to meet
clinical requirements, and added collagen fragments improve cell attachment [17,18]. Furthermore, a
native-like porous structure with a porosity around 68.9 ± 2.8 % is maintained [19].
One concern with human and animal derived products is the inherent risk of pathogen transmission and
batch variability. This risk is non-existent for synthetic BGS where every step of the manufacturing can be
controlled. In addition, BGS allow for higher tunability and predictability of properties. BGS are formed
from a variety of materials: ceramics (e.g. bioactive glass [20], calcium phosphates [21], and titanium
dioxide [22]), polymers (e.g. polycaprolactone [23], polylactic acid, and polyglycolide [24]), and
composites utilizing both material groups [25]. Despite more than 25 years of research, hundreds of
millions of US dollars, and tens of thousands of papers on BGS, they have yet to be clinically adopted [26].
From a technical perspective the BGS performance tends to be insufficient in terms of simultaneously
exhibiting adequate mechanical, biodegradable and osteo-promotive properties, and from a practical
perspective the translation tends to be challenged by high throughput production, low cost, difficulty to
handle during surgery, and strict regulatory conditions [26]. For instance, many bioceramics have shown
promising pre-clinical results, but have failed clinically due to obstacles such as the material cost, ease of
production, and scalability [26]. It has also been pointed out that BGS tend to lack appropriate clinical
documentation demonstrating their success [6].
For BGS often described as bone scaffolds in literature, the material and scaffold design must be carefully
chosen to support bone cell migration, proliferation, and biocompatibility. Traditionally, biocompatibility
has been used to describe materials that stimulates minimal inflammatory response, a definition that
favours non-degradable inert materials [27]. However, this definition fails to account fortabl the
importance of inflammatory response for biomaterial success, such as how a brief acute inflammatory
response followed by a chronic response is vital for recruitment of MSCs and their differentiation towards
osteogenic lineages [28]. With the new generation of biomaterials consisting of bioactive and often
biodegradable materials, a more accurate definition would be materials that have a [27]: “strong affinity
for targeted cells to stimulate biochemical signalling pathways toward the neo-tissue formation”. This
10
highlights why a comprehensive physiochemical characterization is therefore required to predict and
understand the materials’ performance in vivo.
There has been extensive research on biomaterials for bone tissue regeneration, including on ceramics,
polymers, and diverse composites. These trends have been described in several reviews [6,20,25,29,30].
Some limitations worth noting are that bioceramics tend to fail in a brittle manner but have good
bioactivity and stiffness, meanwhile polymers tend to have low stiffness with poor bioactivity but are
highly flexible when it comes to manufacturing. One example of a polymer-based bone graft that has
made it to market is Osteopore. Osteopore is an FDM printed PCL scaffold intended for burr filling in
neurosurgery and in fractures such as orbital floor fracture [31]. Using extrusion-based printing this lattice
is produced by stacking layers of horizontal struts in the vertical direction. A downside of this technology
is a trade-off between resolution and productivity, which limits industrial scaling of this technology.
1.4 Scaffold manufacturing through stereolithography printing Recent trends within scaffold manufacturing have focused on additive manufacturing where the layer-by-
layer based manufacturing process allows for excellent control of scaffold architecture [32]. This involves
control of overall structure geometry, strut size, pore size, and interconnectivity. Conventional AM
methods include fused deposition modelling, where a material filament is extruded through a nozzle,
selective laser sintering/melting, where a laser is rastered over a powder bed to selectively sinter or melt
particles together, ink-jetting, where droplets of material are selectively deposed and solidified into a
structure, and stereolithography printing where a liquid polymer is selectively crosslinked using light [32].
These methods and their advantages/disadvantages are summarized in Table 1.1. Of these,
stereolithography stands out as a promising method as it supports the rapid fabrication of complex 3D
structures with high spatial resolution. This is very important as the feature size of bone such as pore size
and strut thickness tend to be in the scale of 100-800 μm, while the bone defect can be in the order of
centimetres.
11
Table 1.1: Summary of additive manufacturing processes frequently used for manufacturing of
engineered tissue.
Additive manufacturing (AM) processes
Process Laser Based
Extrusion (e.g. FDM) Inkjet printing SLS/SLM SLA1.
Cost (GBP - £)
Industrial
£400,000 One time
Commercial
£3,000 One time
Industrial
£42,500 Per annum
Commercial
£15,000 One time
Industrial
£200,000 One time
Commercial
£18,000 One time
Industrial
£100,0000 One time
Z-Resolution (μm)
40 25 1 200 100 16 10
XY-Resolution (μm)
>502. >1302. 50 200 100 200 150
Materials Polymers, metals,
ceramics, composites
Polymers, hydrogels,
composites
Polymers, hydrogels,
composites, metals Polymers
Advantages
• High resolution
• High strength
• No support required
• High accuracy
• Easy to achieve small
features
• High strength
• Multi-material
• Economical
• Multi-material
• High productivity
Limitations
• High investment
• High processing
temperatures
• Entrapped powder
• Photo-curable
materials only
• Single Resin
• Anisotropy
• Low accuracy
• Delamination
• Artifacts
• Low strength
• Poor surface finish
• Low 3D productivity
• Expensive materials
1. Multiphoton systems not considered. Industrial SLA refers to CLIP. 2. Based on laser spot size.
A typical SLA printer consists of a vat of a photo-curable polymer and a transparent bottom. A laser is
directed onto a galvanic mirror that reflects the laser through the transparent bottom into the vat causing
the polymer to polymerize at the interface of the build platform. When a layer is completed the build
platform moves up, delaminates the structure from the transparent silicon layer, and positions the build
platform for the next layer. The subsequent layer is crosslinked to the previous layer creating a 3D
12
structure. The layer thickness can be modified and there is usually a trade-off between a higher resolution
and shorter build times. The Formlab Form 2 printer has layer thickness options of 25, 50, and 100 μm.
Looking closer at the chemistry of the process: a reactive liquid resin is irradiated by a light source that
activates a photoinitiator to produce free radicals, which then trigger a free-radical polymerization
process. By controlling irradiation parameters (spot size, power, and exposure time) and resin
composition (initiator, absorber, and inhibitor) it is possible to obtain reasonable print times with high
spatial resolution. For commercial stereolithography printers, the resolution is typically down towards 20
μm [33]. A limitation with the conventional SLA method is that the productivity is limited to the raster
speed of the laser which is dictated by the speed of the galvanic mirrors and the activation time of the
resin. The development addressing this is called digital light projection (DLP), where a projector is used
to cure a full layer at the time [33]. This speeds up the productivity significantly, but the spatial resolution
is limited to the pixel size of the projector, which is not at the same level as a laser based system. The DLP
solutions are a newer innovation and the investment cost is still high (£15,000+). If a high-resolution is
required, the structure can be built using two-photon polymerization (TPP). In TPP two photons are
absorbed virtually simultaneously in the voxel by the photoinitiator, which allows a resolution down to
200 nm [33]. Unfortunately, the size of the printed structures tends to be as low as a few millimetres per
hour of build time, which limits the application for larger centimetre scale structures. Conventional SLA
stands out as an optimal balance between resolution and productivity for BGS manufacturing.
In conventional stereolithography, the structure is formed by curing together layers in an additive manner
into a three-dimensional structure. An inherent limitation of this is that the interface of the layers is weak.
To overcome this, the 3D structure must be manufactured in a continuous manner, which is feasible with
the newly developed Continuous Liquid Interface Production (CLIP). This method was developed and
patented in 2014 by the DeSimone group with the major difference being that they use an oxygen
permeable bottom window and that the build platform moves upwards in a continuous manner [34]. With
an oxygen permeable window, a dead zone can be created in which polymerization is inhibited. Further
away from the window, the oxygen concentration drops, and polymerization is possible. This means that
the polymer does not stick to the bottom window and does not need to be delaminated after each layer,
as with conventional methods. With continuous production, slicing artefacts are nearly eliminated. Since
conventional methods require delamination of the structure from the bottom window and then recoating
of the structure, the production speed is limited to only a few millimetres per hour, meanwhile DeSimone
13
and colleagues demonstrated a production speed of 500 mm/hour. CLIP is a very promising technology
for medical implants and will be considered when the resin is fully optimized.
1.5 Synthetic and biodegradable materials for SLA printing. Over the last few years additive manufacturing methods such as stereolithography printing have become
attractive methods to produce BGS due to high architectural freedom, print resolution down to a micro-
scale, and rapid production on-demand. However, there is a lack of commercially available biomaterials
for use in stereolithography printing. There are several biomaterials which have been developed for
stereolithography printing, including synthetic biomaterials such as polylactic acid [35], poly(propylene
fumurate) [36–38], poly(D,L-lactide) [39], poly(trimethylene carbonate) [40] and polycaprolactone [41,42]
and diverse modifications of these. These polymers, their modifications and mechanical properties are
summarized in Table 1.2. The limited information available on these materials might be a primary reason
they have received limited industrial adoption. Particularly, the mechanical characterization is limited. As
can be observed in the table, some polymers lack mechanical characterization all together. Those that
include characterization have only one mode of testing. For example, the poly(trimethylene carbonate)
(PTMC) with hydroxyapatite and β-TCP resins suggested for bone applications, were only characterized in
tension even though bone is subject to tension, compression, bending, and shear. Only two resins were
characterized in compression and were done so as porous scaffolds (70% porosity), therefore it was not
possible to compare their mechanical properties to that of the other material groups. Commercial
adoption of these resins has been slow and is at least partially linked to the lack of standards-based
materials characterisation.
14
Table 1.2: Comparisons of biocompatible resins with mechanical strength. Comparison to cancellous and
cortical bone at the bottom.
Polymer Variant Filler Mw kDa
Modulus MPa
Strength MPa
Strain1 %
Test mode Source
PTMC Trimethacrylated 0.7 801 25 28 Tensile [43]
10.3 2.6 2.5 558
Trimethacrylated 40wt% nHA 10.4 51.1 3.9 100 Tensile [40,44]
Trimethacrylated 60wt% β-TCP 11.6 353 4.0 10.0 Tensile [45]
P(CL-TMC) Diacrylated, 90:10
61 5.0 275 Tensile [46]
PCL Triacrylated
0.3 6.9 0.70 13 Tensile [42]
Trimethacrylated
1.5 15.4 2.55 19.3 Tensile [41]
Trimethacrylated 20wt% bioG 0.7
Compressive2 [47]
Trimethacrylated, depsipeptide-co-CL
0.75
Compressive2 [48]
Divinyl-fumarate-co-PCL
None [49]
PDLLA Dimethacrylated 0.6 2900 94 5.0 Flexural [50]
Dimethacrylated 20wt% nHA 5100 50.6 1.2 Flexural [51]
PPF Dissolved in DMF for printing
1.1 199 21 18 Tensile [38]
PGSA Acrylated
22 NA 45 Tensile [52]
Cancellous 100-5000 0.1-30.0 5-7 Compressive [26]
Cortical 3000-30000 70-200 1-3 Compressive [26]
1: Strain at break. 2: Reported only the modulus of porous scaffold and not the bulk material.
Abbreviations: nHA=nano-hydroxyapatite (Diameter < 200 μm), BioG=Bioactive glass, TCP=tricalcium phosphate
A promising polymer candidate for bone regeneration is PCL. PCL offers long biodegradation times,
typically 3 to 4 years, and is easy to manufacture into a diverse range of implants and medical devices
[53]. The stiffness of PCL has been reported to be in the 0.21-0.44 GPa range, depending on molecular
weight and degree of crystallinity [54]. When it comes to stereolithography printed PCL, the structure is
crosslinked so the properties will vary with the number of arms, molecular weight, and degree of
functionalization of the PCL precursor. For photo-crosslinkable materials the compressive Young’s
modulus of the material has been shown to linearly increase with the density of crosslinking [55].
There are a few cases where PCL has been used for SLA printing: Elomaa et al. [41] were able to synthesise
methacrylate-functionalized PCL-polymers with molecular weights of 1600, 2700, 4200 and 6000 Da.
Under tensile loading, these materials exhibited a Young’s modulus of 15.4, 10.0, 7.1 and 6.7 MPa, tensile
strength of 2.55, 2.49, 2.02 and 2.87 MPa, and elongation at break of 19.3, 30.8, 38.2 and 78.5%,
respectively. The compressive properties were not characterized. To demonstrate printability, they used
a mini multilens SLA machine to print a highly porous structure, obtaining a porosity of 70.5 ± 0.8 vol.%
(designed to be 70 %). As PCL has limited bioactivity and mechanical stiffness, they continued this work
by investigating two alternative tactics. First, they tried incorporating up to 20 wt.% bioactive glass. This
increased the scaffold compressive modulus by more than 140% and more than doubled initial cell
15
attachment [47]. This also improved mineralisation in simulated body fluid solution while the main
features were still captured in the print. Since it was a material suggested for bone application, they
measured the compressive properties of the scaffold both in dry and wet conditions, but they did not
report any mechanical properties for the bulk materials, nor the strain rate for mechanical testing. In an
alternative tactic a co-polymer solution of poly(depsipeptide-co-ε-caprolactone) was produced to
improve the biological performance while still maintaining good printability [48]. The co-polymer had
increased hydrophilicity, which accelerated the hydrolytic degradation, and increased compressive
strength. Compared to PCL, the yield stress increased from 0.50 to 1.78 MPa and Young’s modulus from
3.1 to 5.8 MPa. Mechanical testing was performed in compression, but again only for the scaffold
structure and not the material itself. Without reporting the bulk material properties, it is hard to predict
the materials performance in different structures. In a more recent paper Green et al. [42] functionalized
PCL-diol (MW = 530 and 1250 Da) and PCL-triol (MW = 300 and 900 Da), and investigated the effect of
changing the molecular weight (MW) and degree of functionalization (DoF). They showed that a higher
DoF and lower MW improves the print accuracy and yields a stiffer material. The latter conforms to the
conclusion of Nijst et al. [55], as a higher DoF and lower MW gives a higher density of crosslinking. They
illustrated biocompatibility in vitro using murine-induced pluripotent stem cells and in vivo when
implanting into pig retinas. While dynamic mechanical analysis was conducted, no results were provided
and when calculating the DoF of the acrylated polymers impurity peaks were found in their NMR
integration (as detected by Dr Cécile Echalier).
The previous work on PCL for stereolithography lacks a consistent and comprehensive mechanical
characterization of both the material and the scaffolds under different modes of loading. From this work
it is unclear if polymerized PCL has potential as a load bearing BGS. The overarching goal of this project is
to evaluate if stereolithography printed PCL has the printability and mechanical performance necessary
to be a suitable material for BGS.
2. Project aims In this project I investigated the printing of PCL based bone graft substitutes, by looking at resin
formulation, printability, mechanical properties, lattice structures, and degradation properties. The
project goals can be summarized as following:
1. Develop a PCL-based resin adequate for stereolithography printing and optimise the printing
process.
2. Characterize the mechanical properties of the printed PCL (bulk).
16
3. Print PCL scaffolds with different geometries and high porosity (>70%).
4. Characterize the mechanical, morphological and degradation properties of the PCL scaffolds.
5. Identify which type of lattice structure provides the highest scaffold stiffness at a wanted porosity
level.
3. Experimental method
3.1 PCL functionalization (MRes) 3.1.1 Materials
Polycaprolactone-triol (average Mw = 300 Da), acryloyl chloride, methacryloyl chloride, 2-benzyl-2-
(dimethylamino)-4’-morpholinobutyrophenone and diphenyl(2,4,6-trimethylbenzoyl) phosphine oxide, 4-
methoxyphenol (MEHQ), and acetone were purchased from Sigma Aldrich and used as provided.
Magnesium sulphate, potassium carbonate, dichloromethane and tetrahydrofuran were purchased from
VWR Chemicals and used as provided, apart from dichloromethane that was passed through an alumina
column under inert atmosphere before use.
3.1.2 Resin Functionalization
PCL tri(meth)acrylate was synthesized by reacting (meth)acryloyl chloride with PCL-triol following a
procedure developed during my MRes [56]. The detailed protocol is available in the supplementary
material (S1).
3.1.3 Polymer Characterization
1H nuclear magnetic resonance (NMR – Bruker, MA, US) was used to confirm the functionalization and
calculate the degree of functionalization of the resins. The resins (A3, PCL-A, PCL-MA) were dissolved in
deuterated chloroform (CDCl3) at a 10 mg/mL concentration. 1H spectra were collected on a Bruker
Advance III HD 400 MHz spectrometer at 297 K. Chemical shifts are given in ppm calibrated with the
reference solvent (CDCl3, = 7.26 ppm). 1H NMR spectra were typically recorded using 64 scans and a
recovery delay of 10 s. Spectra were processed and visualized with MestReNova 12 (Mestrelab research,
Santiago, Spain).
3.2 Resin optimisation The resin was composed of polymer, photoinitiator (TPO – diphenyl(2,4,6-trimethylbenzoyl)phosphine
oxide, or Irgacure 369 – 2-benzyl-2-(dimethylamino)-1-[4-(morpholinyl) phenyl)]-1-butanone),
photoabsorber (OB+ – 2,5-bis(5-tert-butyl-benzoxazol-2-yl)thiophene), and photoinhibitor (MEHQ – 4-
Methoxyphenol). Concentrations of each component were optimized to obtain adequate cross-linking
17
kinetics for the Form 2 SLA printer. The cross-linking kinetics were investigated by observing the viscosity
change under constant UV exposure using an Anton-Paar MCR302 (Graz, Austria) rheometer. A
transparent acrylic plate was used as the bottom surface and a 25 mm stainless steel parallel plate, fitted
with 120 grit sandpaper, was used as the upper surface. The UV was provided from a fibre laser using an
Omnicure S2000. 150 μL of resin was added to the base of the rheometer and then the gap was adjusted
to 0.1 mm. The rheological properties (shear storage modulus and shear loss modulus) were measured at
a constant shear rate = 1% and frequency = 10 Hz. The viscosity of the unpolymerized resin was measured
for 40 s without UV. Following baseline viscosity, the UV source (P = 1.14 mW/cm2) was turned on for 100
s (150 s for slow reactions). Data was acquired at 0.1 samples/s.
3.3 Custom SLA set-up A practical limitation to commercial SLA systems is that they require very large resin quantities for
printing. The SLA system we purchased, Form 2 by Formlabs, required a minimum of 150 mL to print. This
is very material consuming and expensive, particularly when working with resin optimisation which
requires several iterations with different resins composition. Consequently, Dr Wojciechowski modified
the SLA set-up to reduce the resin need. This was done by cutting the build platform down from 145 by
145 mm to 52 by 52 mm and placing a smaller resin tank inside the existing tank, similar to what was done
by Smith et al. [57], successfully reducing the resin requirement from 150 mL to approximately 15 mL. The
smaller tank was printed using a Form 2 printer. We utilized the transparent acrylate layer of the original
tank and glued our new tank directly on top. This required us to remove the PDMS layer protecting the
18
original tank, so we had to make a new layer for the custom tank. The modification is illustrated in Figure
3.1.
Figure 3.1: Custom SLA setup. Grey tank is the resin tank inside the original tank. The reduced build
platform can be observed above.
3.4 Lattice design Cellular lattices were made in Rhinoceros 3D (Robert McNeel & Associate, US) using the octet-truss cell,
as a known stretch-dominated lattice, and the tetradecahedron cell, as a known bending-dominated
lattice [58]. They were designed as 4 cell repeats in the lateral and vertical planes (4x4x4), with a cell size
(CS) of 2, 3, 4, or 5 mm. In addition, the strut radius was set to 0.1, 0.2, 0.3, or 0.4 mm (also 0.5 mm for CS
= 2 mm tetradecahedron lattice). Only lattices with a CS > 3 mm and a strut radius > 0.3 mm were printed.
3.5 Mechanical testing The mechanical testing was done using a Bose ElectroForce 3200 (Minnesota, USA) uniaxial material
testing machine with a 100 lbf load-cell was used. The test substrates were dimensioned after the
preferred specimen dimensions defined in the relevant ISO standards summarized in Table 3.1.
19
Table 3.1: Relevant mechanical testing standards.
Test group Property Standard for Material Standard for Scaffold
Compressive Elastic modulus ISO 604:2002 ISO 844-2014
Yielding ISO 604:2002 ISO 844-2014
Failure ISO 604:2002 ISO 844-2014
Tensile Elastic modulus ISO 527-1:2012, ISO 527-2:2012
Yielding ISO 527-1:2012, ISO 527-2:2012
Failure ISO 527-1:2012, ISO 527-2:2013
Flexural Modulus ISO 178:2019 ISO 1209-1:2007, ISO 1209-2:2007
DMA ASTM D4065-12
DMA = Dynamic Mechanical Analysis
The dimensional conformity was controlled using a digital calliper with 0.01 mm resolution. For the tensile
test substrates, the samples were designed after the instructions in ISO 527, and are illustrated in Figure
3.2. It was assumed that only the narrow region deformed under tension, as by ISO 527, where the gauge
length is 12 mm, width is 2 mm, and thickness is 1 mm. This was used to calculate stress and strain.
Figure 3.2: Tensile test substrate as defined in ISO 527-2 2012.
Both for the tensile samples and the compressive sample, the young’s modulus was calculated using linear
regression in the linear region of on the stress-strain plots (up to ultimate tensile stress point (UTS for
tensile), and between 20-30% stress for compressive sample). The modulus of toughness was calculated
for the tensile samples from the start point up until the fracture point of the samples according to
equation 3.1 [59]:
𝑇𝑚 = ∫ 𝜎 𝑑𝜀𝜀𝑓
0
3.1
20
Where Tm is the modulus of toughness, σ the stress, ε the strain, and εf the fracture strain.
3.6 Accelerated degradation study To confirm the biodegradation of PCL-trimethacrylate (Mw = 300, DoF = 99%) an accelerated degradation
study using sodium hydroxide, similar to that of Lam et al. [60], was conducted. The PCL-trimethacrylate
was printed into a cylinder-shaped body-centre cubic unit structure (Diameter (Dia.) = 4 mm, height =
4mm) (Figure 3.3). The lattice was first washed five times in milli Q water, then twice in absolute ethanol,
wiped off with Kimtech precision wipes, then dried in a vacuum desiccator overnight, before the initial
weight (m0) of the cell cylinder was measured. The cell cylinders were then placed in individual 1.5 mL
Eppendorf tubes, before 1 mL of 2 M sodium hydroxide was added. The samples were then incubated at
37 °C for 1, 2, 4, 7, 14, or 21 days before they were taken out, washed and dried in a similar manner as
described above, then the new weight (m1) was measured. Three repeats were conducted for each time
point.
Figure 3.3: Modified body centre cubic unit cell (Dia. = 4 mm, h = 4 mm, Rstrut =0.5 mm).
3.7 FEA Modelling Finite element analysis (FEA) was used to investigate the apparent modulus of different lattice structures,
focusing on two different unit cells – octet-truss and tetradecahedron (Figure 3.4). The lattice designs, as
described in 3.4., were imported to the modelling software ABAQUS 2019 (Dassault Systems). The method
is described in detail by Galarreta et al. [61]. Briefly, the lattices were compressed between two rigid
planes with the applied force and displacement of the top plane being used to derive the apparent
modulus of the lattice using a linear regression in Excel (Microsoft, US). For all simulations the top and
bottom struts of the lattice were removed to expose flat surfaces to give a well-defined contact with the
rigid plates. Further, only a quarter of the lattice was considered but symmetric boundary conditions were
applied. A simulated isotropic material with an elastic modulus of 1000 MPa and a Poisson’s ratio of 0.3
21
was used to model the system. The lattices were meshed into quadratic tetrahedral elements (C3D10)
and were subject to stepwise loading in the negative vertical direction. For a strut radius of 0.3 and 0.4
mm the load steps were 5, 15, 30, 50, and 75 N, for a radius of 0.2 mm it was 5, 10, 15, 20 and 25 N, and
for a radius of 0.1 mm it was 1, 2, 3, 4, and 5 N.
Figure 3.4: Illustration of the octet-truss (left) and tetradecahedron (right) unit cells. Gray struts indicate
external elements, meanwhile orange indicate internal elements.
3.7.1 Mesh refinement study
The mesh refinement was done to determine the appropriate number of nodes per edge. For this a lattice
of 2x2x2 cells with cell size of 3 mm and a strut radius of 0.4 mm was used. The lattice was compressed
by the previously described loading sequence using a friction coefficient of 0.1 between the lattice and
the plates. The number of nodes per edge was investigated between 2 and 8 in increments of 1. The
relative modulus and the computational cost in seconds were extracted (Figure 4.6).
3.7.2 Cell repeat and frictional conditions
After identifying 4 nodes per edge as the optimal mesh density, the required number of cell repeats and
frictional condition at the interface was investigated. The lattice sizes investigated were 2x2x2, 4x4x2,
4x4x4, and 6x6x3 (LxWxH) cell repeats using a cell size of 3 mm and a strut radius of 0.4 mm. The
investigation was done for a frictionless and a 0.3 friction coefficient contact. The relative modulus was
extracted (Figure 4.7).
22
3.7.2 Cell size and strut radius
Having identified 4 nodes per edge and a 4x4x4 cell structure as the optimal design, the effect of the cell
size and strut radius on the relative modulus was investigated. The cell sizes investigated were 2, 3, 4, and
5 mm, and for each of these cell sizes the radius (Rstrut) of 0.1, 0.2, 0.3, and 0.4 mm was investigated. To
normalize the results between the different cell sizes the relative density was calculated by dividing the
solid volume of the lattice with the external dimensions of the lattice structure.
3.7.3 Comparison to a bending dominated structure
A similar study to that described in section 3.5.2 was conducted for a tetradecahedron lattice (Kelvin
foam), which is known for being a bending-dominated structure [62]. The CS investigated were 2, 3, 4,
and 5 mm, and for each of these Rstrut of 0.2, 0.3, and 0.4 mm was investigated. Additionally, a strut radius
of 0.5 mm was investigated for a cell size of 2 mm. The results were normalized by taking the relative
density and relative modulus.
3.7.6 Modelling verification
Solid cylinders (Dia. = 3 mm, height = 3 mm) and octet-truss lattices (4x4x4, CS = 4 mm, Rstrut = 0.4 mm)
were printed in the commercial Clear V4 resin on the Form 2 printer. The structures were loaded in 5
times between 25 and 250 N, before they were taken to 425 N (maximum test frame capacity) and finally
unloaded. The material and apparent compressive modulus was derived from the bulk material and
lattices, respectively, by fitting a linear regression to the 4th loading cycle. The relative modulus of the
lattices was obtained by dividing the apparent modulus by the material modulus. The experimental results
were then compared to the FEA results.
3.8 Statistics and data handling For mechanical testing and UV Rheology, the data were loaded using Jupyter Notebook and was processed
with custom Python scripts. Using the library SciKit Learn, linear regression was used to calculate the
compressive and tensile modulus of the materials. The SciPy library was used for fitting the logistic curves
to the UV Rheology data. For the tensile testing, as the sample group was too small to determine
normality, a non-parametric two-sided Mann-Whitney rank test was used to determine statistical
significance.
23
4. Results
4.1 Resin synthesis PCL-triol (Mw=300) was successfully acrylated to a DoF = 60% using acryloyl chloride and potassium
carbonate. This polymer was used for all the characterization described below, with the exception of the
accelerated degradation study. In a newer functionalization protocol, developed by Dr Echalier, where the
base (potassium carbonate) was replaced with triethylamine a DoF above 90% was achieved both for
methacrylation and acrylation. A PCL-trimethacrylate with DoF = 99 % was used for printing the substrates
used in the degradation study. The DoF was calculated by integrating of the vinyl protons and methyl
protons (Figure S3.1)
4.2 Resin formulation A limitation of the print set-up is that the Form 2 (Formlabs) is locked to the manufacturer’s settings for
exposure time and laser power. Therefore, we needed to develop a resin mimicking the properties of the
commercial resin in terms of polymerization kinetics. We measured reactivity by UV-rheology, where the
resin was loaded onto the rheometer and the change in viscoelastic properties were measured while
simultaneously being irradiated with UV light. The temporal rheology (storage modulus) data was
characterized by fitting a logistic curve described by eq. 4.1:
𝐺′(𝑡) =𝐿
1 + 𝑒−𝑘(𝑡−𝑡0)4.1
Where G’ is the storage modulus, t is time, L is the plateau value of G’ when time goes towards infinity, k
is the exponential growth rate, and t0 is the inflection point. We started by measuring the behaviour of a
commercial resin, Clear V4 by Formlabs, to benchmark the custom resin against. For this we performed
measurements both at room temperature (25°C) and at 35°C. It was observed that a higher temperature
increased the rate of crosslinking (Figure S4.1). The average of the benchmark resin gave values L = 78000
Pa, k = 0.18 s-1, t0 = 93.6 s (Figure S4.2). We then used this plot to compare to acrylated (DoF=60%) and
methacrylated (DoF=99%) PCL with 0.40 wt.% photoinitiator (here TPO). The data collected for these two
groups (n=3) had a high degree of discrepancy. However, the results indicated that the acrylated resin had
faster kinetics than the methacrylated, and that the commercial resin kinetics fell between them (Figure.
4.1).
24
Figure 4.1: UV rheology of PCL-triacrylate (Red; DoF=60%), PCL-trimethacrylate (Green; DoF=99%), and
commercial Clear V4 (Blue – Average of 2 repeats; Formlabs, MA, USA) at RT. The blue dashed line
indicates inflection point of the Clear V4 resin.
The effect of initiator type and concentration were also investigated. When comparing TPO to Irgacure
369, it was observed that TPO had faster kinetics, but Irgacure gave less discrepancy in the results (Figure
S4.4). As expected, increasing the initiator concentration increased the rate of polymerization. When
adding an inhibitor to the resin (200 ppm MEHQ) in an attempt to improve the shelf-life, the crosslinking
rate was largely reduced (Figure S4.5). The decrease in rate is due to the inhibitor quenching the free
radicals, thereby reducing the concentration of radicals, which reduces the rate of polymerization.
4.3 Printability A resin for stereolithography printing needs to consist of three components in addition to the
functionalized polymer. The resin needs to have an initiator that forms radicals upon UV irradiation,
thereby starting the polymerization, an absorber that limits the light penetration into to the resin,
preventing irradiation beyond the designed layer height, and finally, an inhibitor that quenches the
radicals, preventing polymerization during storage. Based on the results described in section 4.2, we
decided upon the addition of 0.40 wt.% TPO-L (initiator), 0.16 wt.% OB+ (absorber), and 200 ppm MEHQ
(inhibitor) to our PCL-triacrylate (Mw=300 Da, DoF=60%) polymer, which proved itself adequate for the
printing.
25
The printability of this resin was illustrated by first printing the modified body centre cubic (BCC) unit cell
cylinder, shown in Figure 3.1. After successful printing of the individual cell, the structure was scaled into
a full scaffold based on the modified BCC cell, identical to that used by Reznikov and colleagues in their
sheep model [63]. The printed structure is shown in Figure 4.2.
Figure 4.2: Stereolithography printed a) modified body centre cubic cell (Dia. = 4 mm, height = 4mm,
strut diameter = 1 mm) and b) lattice (Dia. = 15 mm, h = 15 mm) based on the same cell.
4.4 Mechanical testing PCL-triacrylate (Mw=300, DoF=60%) was tested in tension and compression.
4.4.1 Tensile testing
The tensile testing was conducted in accordance with the testing standard ISO 527-2, using the tensile
test substrates and strain rate described. To investigate the impact of print direction, half of the substrates
(n=5 in each direction) were printed standing vertically and the other half laying horizontally (Figure 4.3).
All samples seem to deform linear elastically under tensile loading before fracturing with limited yielding.
In one of the vertically printed samples, there is indications of yielding as it at 7.5% strain deviated from
the trend observed for the remaining sample of the two groups. For the other samples, rapture seems to
occur before or shortly after yielding. The horizontally printed substrates reaching a mean ultimate tensile
stress level of 1.64 ± 0.11 MPa and the vertically printed a UTS level of 1.34 ± 0.16 MPa which is statistically
significant (P = 0.036, α=0.05) based on a two-sided Mann-Whitney U test.
26
Figure 4.3: Left: Stress-strain plot for tensile testing of PCL-triacrylate (Mw=300 Da, DoF=60%). Right:
Ultimate tensile strength. Samples are printed in the vertical (blue) and horizontal (red) direction
relative to the build platform. Results: Mean ± SD, n=5.
4.4.2 Compressive testing
A single run of a cylinder (Dia. = h = 5 mm) under compression was conducted (Figure 4.4). The specimen
did not failure under the limits of the test frame and reached a compressive stress of 21.1 MPa and
compressive strain of 30.1%. At 20-30% strain the material had an apparent modulus of 97.9 MPa.
27
Figure 4.4: Single repeat compressive loading of PCL-triacrylate (Mw=300, DoF=60%) sample up to a load
of 425 N before unloading.
4.5 FEA Modelling 4.5.1 Mesh refinement
First an appropriate mesh refinement for convergence of results was identified. This was done by
considering the change in relative modulus of a 2x2x2 cell octet-truss lattice when changing the number
of nodes per edge. It was observed that the between 2 and 3 nodes per edge there was a 3.7% change in
the respective relative modulus, thereafter there was less than 1% change when increasing the number
of nodes by one (Figure 4.5). However, the computational cost increased exponentially. Already at 3 the
results seemed to have converged, but we progressed using 4 as an extra precaution. The increase from
3 to 4 did not add much computational cost.
Figure 4.5: The relative modulus (black, left axis) and computational cost (in seconds – red, right axis) as
functions of nodes per edge in FEA modelling of an octet-truss lattice with 2x2x2 cell repeats.
4.5.2 Cell repeat and frictional condition
The required number of cell repeats in the lateral and vertical directions was investigated under two
frictional conditions (Figure 4.6). For the frictionless condition, there was a stable decrease of 0.8-1.5%
in the relative modulus when increasing the number of cell repeats. This trend was not observed for the
0.3 friction coefficient condition, where the modulus seems to be increasing when the number of lateral
28
cells increases. The frictionless condition was chosen for further investigations as it was more stable.
Further, 4x4x4 was identified as an appropriate cell repeat number, as this was where the difference
between the frictionless and the 0.3 friction coefficient condition was at its lowest.
Figure 4.6: The effect of cell repeats in octet-truss lattice on relative modulus using a frictionless (blue)
interface or a frictional coefficient of 0.3 (red) based on Abaqus FEA modelling. Refers to repeat in
length x width x height.
4.5.3. Cell size and strut radius
The effect on both the cell size and the strut radius on the relative modulus of the octet-truss structure
was investigated. It was observed that with a larger strut radius the relative modulus increased. The effect
of the radius increase was greater with a smaller cell size. For a more accurate comparison between the
different cell sizes the results were normalized by calculating the relative density of the lattice and plotting
the relative modulus against the relative density (Figure 4.7). It was then observed that all the cell sizes
followed the same trend. At a relative density of 28% all the lattices, independent of cell size, had a relative
modulus of 10%, and at a relative density of 8.25% both the 2 mm and 4 mm cell size lattices had a relative
modulus of 1.7%.
29
Figure 4.7: Relative modulus as a function of relative density for the octet-truss lattice (4x4x4) with a cell
size of 2 (black), 3 (red), and 4 (blue) mm. Results based on Abaqus FEA modelling.
4.5.4 Comparison to bending dominated structure
Since the octet-truss has a known stretch-dominated deformation, a bending dominated unit cell called
tetradecahedron was used as a comparison (Figure. 4.8). At a high relative density, the tetradecahedron
lattice seems to have a higher stiffness than the octet-truss, meanwhile at a lower density the octet-truss
seems to be stiffer.
30
Figure 4.8: Relative modulus versus relative density for a 4x4x4 lattice of either the octet-truss (red) or
tetradecahedron (blue) unit cell modelled in FEA. The relative density was modulated by varying the cell
size (2-5 mm) and strut radius (0.2-0.4 mm).
4.5.5 Model validation
Both a cylinder (Dia. = 3 mm, h = 3 mm) and an octet-truss lattice (CS = 4 mm, Rstrut = 0.4 mm, 4x4x4
repeats) were printed in the commercial Clear V4 and tested in compression to validate the results. The
results (n = 3 for both groups) suggests a material modulus of 450.9 ± 38.8 MPa and an apparent modulus
of 54.7 ± 3.4 MPa, yielding a relative modulus of 12.13 ± 1.81 %. In comparison, the FEA yielded a relative
modulus of 10.57 % which is within the standard deviation of the experimental mean.
4.6 Degradation A accelerated degradation study in sodium hydroxide, similar to that of Lam and colleagues [60], was
conducted to predict if the material would be degradable under physiological conditions (Figure 4.9). At
the start of the experiment the average weight of the samples (n=18) was 22.31 ± 2.63 mg. After the first
day the mass drop was on average 5.7 ± 3.13 %, thereafter the mean remaining mass dropped gradually
by approximately 1% every day. However, at 14 days the mass loss is similar to that after 7 days, 11.8 ±
0.2 % versus 11.6 ± 2.4 % respectively. After 21 days the mass reduction has increased to 20.3 ± 3.2 %.
This could be initial indications of the degradation rate increasing, which is expected as PCL degrades
through hydrolysis where ester-bond cleavage causes the formation of carboxylic-acid, giving an
31
autocatalytic effect that boosts the degradation rate [60]. Nevertheless, there are indications of a gradual
degradation, which is promising as this can limit the build-up of the carboxylic acid and reduce the
autocatalytic effect.
Figure 4.9: Mass remaining of PCL-trimethacrylate (DoF = 93%) unit cell (Modified BCC, Dia. = 4 mm, h =
4 mm, Rstrut = 0.5 mm) against time in days after being incubated in an alkaline environment (2M NaOH).
Mean ± SD, n=3.
5. Discussion
5.1. Printing optimisation PCL was functionalized with (meth)acrylate functional groups to make it photo-crosslinkable, a
prerequisite for stereolithography printing. It was then formulated into a resin by mixing in a
photoinitiator, an absorber and an inhibitor. A limitation with the print set-up is that the Form 2 is locked
to manufacturer’s settings for exposure time and laser power. Therefore, we needed to develop a resin
mimicking the properties of the commercial resin in terms of polymerization kinetics. The reactivity of the
resin depends on the number of polymer arms, the functional endgroups, temperature [64] in addition to
photoinitiator, photoinhibitor, and photoabsorber, and their concentration. We measured reactivity by
UV-rheology, where the resin was loaded onto the rheometer and the change in viscoelastic properties
were measured while simultaneously being irradiated with UV light. The temporal rheology (storage
modulus) displayed a classic logistic growth curve which coincides with the transition from an
32
uncrosslinked liquid polymer to a crosslinked solid polymer. This observation agrees semi-empirically with
the trends observed in literature, where it is explained that acrylate resins undergo an autoacceleration
process under UV irradiation where the rate quickly accelerates, before it flattens out when the resin
approaches full crosslinking [65].
Initially, there was poor consistency in UV rheology traces. We hypothesized this was due to TPO only
having a moderate solubility in the resin. To improve this, we allowed more time for mixing of the TPO
but unfortunately this did not improve the kinetics (Figure S4.3). We then substituted TPO with another
photoinitiator, Irgacure 369 and investigated the crosslinking behaviour with different concentrations
(Figure S4.4). The behaviour with this photoinitiator was much more reproducible, supporting our theory
of TPO’s poor solubility. With the new initiator we could clearly observe a relation between the resin
initiator concentration and the reaction kinetics. The downside was that this initiator seemed to yield
slower kinetics than TPO, so we decided to focus on the TPO solubility. We considered grinding the crystals
into a finer powder and dissolving it in solvent (e.g. DCM) before combining it with the resin. However,
preliminary results using the liquid version of TPO, TPO-L, indicate that the solubility issue has been solved
using this TPO version.
Another problem we encountered was that the resin seemed to crosslink under ambient storage
conditions leading to a short shelf-life. To resolve this, we added 200 ppm of the photoinhibitor MEHQ,
which slowed down the crosslinking rate significantly (Figure S4.5), requiring an increased concentration
of initiator. Currently, the resin has surpassed a shelf-life of 6 months while being stored in a dark freezer
(-18 °C). The shelf-life at room temperature will need to be investigated in the future.
Using a PCL-triacrylate (Mw = 300 Da, DoF = 60%) with 0.40 wt.% TPO and 0.16 wt.% OB+, we were
successful in replicating the modified body centre cubic unit cell (Dia. = 4 mm, height = 4 mm, strut radius
= 0.5 mm), and when printing a full scaffold structure (Dia. = 15 mm, height = 15 mm), demonstrating the
ability to print larger structures with complex 3D design with high resolution. For structures with smaller
strut radius or smaller unit cell size, we did struggle with failed prints or the structure fusing into a solid,
respectively. This is however believed to be due to the SLA printer’s, Form 2, XY-resolution which is limited
by a laser spot size of 130 μm. The laser diameter sets the limit to the smallest features that can be printed
[64]. With newer DLP or CLIP systems using projectors with pixel sizes of typically 50-75 μm, structures
with finer features should be printable. We also experienced issues with overcuring, a phenomenon where
the light penetrates past the designated layer, causing unwanted polymerization [64]. The light
penetration depth can be controlled by introducing a photoabsorber.
33
5.2 Mechanical testing and FEA For a material to be applied, particularly toward loading application such as a bone scaffold, it is important
to understand the material’s performance, under diverse loading modes. This refers to compression,
tensile and flexural testing, as well as time dependent loading tests such as fatigue and creep. When
conducting material development, it is important to recognise that the scaffold geometry tested might
not be the same used for a final product design. Therefore, it is very important to test the material itself,
and any promising lattice structures based on this material, using readily reproducible testing standard
such as those developed by the ISO (International Organizations of Standardization) or ASTM (American
Society for Testing and Materials). In Table 3.1 I have identified the relevant standards (with preference
of ISO) to characterize different material and scaffold properties. These standards were applied for the
tensile testing presented in this work. One thing worth noting when using these standards is that the test
substrates should be manufactured the same way as a final product will (e.g., print direction).
One of the first thing investigated was the material’s tensile properties and how print direction might
influence the modulus. Dog bone-like test substrates were printed in two different directions (vertically
and horizontally) and tested under tensile loading. It became obvious that the SLA printed PCL was
anisotropic with the vertically printed sample (print layers were orthogonal to the loading direction) being
the weaker sample group. The most probably source of this anisotropy were the weak interfaces formed
during layered SLA printing [66].
Ideally the SLA printed material would have isotropic properties. However, as it is produced in a layer-by-
layer manner where physical delamination is required between the part and the polymer vat bottom the
material is anisotropic. It has however been illustrated that this effect is not observed when using CLIP
printing due to the continuous nature of the process, and it has also been illustrated that CLIP prevents
the stair-casing effect for inclined details which can be observed for SLA produced parts [67]. In addition,
the increased productivity of CLIP suggests it will be an important technology for translation.
A similar PCL-based resin to ours was developed by Green et al. [42]. Even though they were able to print
complex structures with high accuracy, they chose to cast their tensile test substrates, meaning that the
above discussed anisotropy or other artefacts from SLA printing is not represented in their results. With
that in mind, when comparing our results (results averaged for both print directions, n=10) to that of
Green et al., we observed a 2-fold increase in modulus, equivalent strain at breakage, 2-fold increase in
ultimate tensile strength, and 2.5-fold increase in modulus of toughness (Figure 5.1). A high material
stiffness, UTS, and strain level is generally important to be able to perform adequately in musculoskeletal
34
systems. Besides that, a high modulus of toughness is critical. The modulus of toughness measures the
amount of energy the material can absorb through deformation before rupture. The lower it is, the higher
the risk of material rupture, which would lead to implant failure.
Compared to Green and colleagues, our standard deviation is generally higher. This is likely because we
used samples printed in two different directions, meanwhile Green and colleagues casted their tensile
samples. Thereby will the printing artefact with weak layer interfaces affect our results. In conventional
linear polycaprolactone, the stiffness and strength of the material is linked to the molecular weight of the
polymer, as a higher molecular weight would give a more entangled polymer, thereby raising the stiffness.
For cross-linked polymers, the stiffness and strength are governed by the density of cross-linking. The
density is dependent on the degree of functionalization, but at a given DoF it can be increased by reducing
the molecular weight length, thereby reducing the distance between the functional end groups, or by
increasing the number of arms in a branched polymer [42]. This suggests that our resin’s DoF (60%) is
higher than that of Green and colleagues’ resin, even though they reported a higher DoF (77%). This agrees
with Dr. Echalier’s observation of impurities being included in their NMR integration. In a newer version
of our resins, we have obtained a higher DoF (> 90 %). With a higher DoF we expect increased material
stiffness and UTS, but decreased strain at rupture.
35
Figure 5.1: Comparison of tensile properties for PCL-TA300 (DoF=60%), with a comparative polymer
from literature (DoF=77%). Modulus refers to Young’s Modulus, UTS to ultimate tensile strength, and
toughness to modulus of toughness. Green’s results derived from [42], where samples were done in
triplicates. Results: Mean ± SD, n=10.
The PCL developed is intended for bone application, which is a tissue which is typically loaded under
compression. Therefore, the compressive properties are of great interest. In Table 5.1 the compressive
properties of PCL-triacrylate (section 4.4 – Figure 4.4) have been put in context to the two bone groups,
cancellous and cortical bone. It should be kept in mind that the PCL was not brought to failure before the
load cell reached its maximum, so some of the reported properties are likely higher if the material was
brought to failure. Based on these results, however, it can be observed that the PCL’s stiffness is just
outside the lower region of cancellous bone and more than a magnitude away from the cortical region.
The stress level reached is within the mid-range of cancellous bone, while the strain level reached is more
than a magnitude higher than the two bone groups. Given the lower modulus, high stress carrying capacity
and greater elongation to failure there is a potential for this type of implant to promote strain induced
osteogenesis. The strain induced osteogenesis is not yet fully understood, but is believed to be a
synergetic effect from matrix deformation, load-induced flow of canalicular fluids, and electrical
36
streaming from ionic fluids flowing past, that stimulates osteocytes [68]. The osteocytes then coordinate
the osteoblast and osteoclast activity by altering its expression of biochemical cues, thereby regulating
bone formation and resorption [69]. At bone locations with low loading apoptotic osteocytes are
observed, that is linked to stimulation of osteoclastogenesis through RANKL (Receptor activator of nuclear
factor Kappa-B Ligand) secretion [69].
With conventional biomaterials, for instance a titanium hip implant, a problem is that the titanium is much
stiffer than the surrounding bone (ETi6Al4V = 114 GPa [70]), which causes stress shielding and local bone
resorption. With a more compliant yet high load carrying structure, the load will be transferred from the
implant to the surrounding bone, avoiding stress shielding and inducing osteogenesis. The effect may be
similar to that observed for a nylon scaffold in the work by Reznikov and colleagues [63]. A limitation to
this comparison is that the printed PCL sample describe here is solid, whereas cortical and cancellous bone
are porous structures. Introducing porosity to the PCL will decrease the apparent modulus of the scaffold
and this will have to be optimized to support the local mechanics of the defect site.
37
Table 5.1: Comparison of the compressive properties of PCL-triacrylate (Mw=300, DoF=60%, bulk material)
to cancellous and cortical bone. Bone values are taken from a recent review by Schickert and colleagues
[26]. σmax and εmax refers to the maximum reached for the PCL during testing, and for the value at fracture
for the bone groups. n=1.
Property PCL-TA300 Cancellous Cortical
E – MPa 97.9 100-5,000 3,000-30,000
σmax- MPa 21.1 0.1-30.0 70.0-200.0
εmax - % 30.1 5.0-7.0 1.0-3.0
Porosity - % 0 50-95 5-30
FEA modelling was used to understand how the introduction of porosity will affect the mechanical
properties, particularly the stiffness. Two lattice structures were investigated: (1) octet-truss, which is a
stretch dominated cell, and (2) the tetradecahedron, which is a bending dominated cell [58]. In the density
range below 0.3, which is the upper limit for true cellular solids [71], a perfectly stretch dominating
structure should have a slope raised to the power of 1, and a bending dominated should have a slope
raised to the power of 2 [72]. Through best fit of a power function for the results in Figure 5.2 the octet-
truss had a slope of 1.40, meanwhile the tetradecahedron has a slope of 2.17. Suggesting that the octet-
truss’ deformation is dominated by stretching, meanwhile the tetradecahedron’s deformation is
dominated by bending. This means that a octet-truss lattice should maintain its stiffness-to-mass ratio
better than the tetradecahedron structure at a low relative density (ρ’ < 0.3), making it interesting for a
scaffold design for a BGS, as BGS needs a low relative density.
Ideally a bone graft should have a high porosity (> 90%) if it has a slow degradation rate [73], which tends
to be the case for PCL. At a porosity of 90% (equivalent to a relative density of 10%), a stretch dominated
lattice is expected to be ten times as stiff as a bending dominated lattice [74]. Our FEA results indicates
that the octet-truss lattice is indeed stiffer than the tetradecahedron structure at this density (E’tet = 0.91%
versus E’oct = 1.46% - Extrapolation from Figure 5.2), but not by a ten-fold as suggested in literature. FEA
yield numerical results based on the conditions applied. In our analysis it is assumed an isotropic material,
frictionless conditions, and an assumed Poisson’s ratio, therefore experimental verifications must be
conducted before anything can be concluded. For a preliminary verification for an octet-truss lattice (CS
= 4 mm, Rstrut = 0.4 mm) printed in Clear V4 resin the relative stiffness of the FEA model (E’ = 10.6 %) was
within the standard deviation of the experimental results (E’ = 12.1 ± 1.8 %), supporting the credibility of
38
the FEA model. A more through validation across relative densities should be conducted before the model
can be considered valid.
Figure 5.2: FEA results for octet-truss (red) and tetradecahedron (blue) lattice for a relative density lower
than 0.3, which is the relative density limit for true cellular solids [71]. Solid lines represent best fit power
laws, where the octet-truss is raised to the power of 1.40 and the tetradecahedron to the power of 2.17.
Based on the FEA results, a scaffold using our low DoF (≈60%) and with a porosity of 90%, will have an
apparent modulus of less than 1 MPa. This is far below the stiffness of bone. Since this is unlikely to
provide the necessary mechanical integrity for load bearing applications, the material needs to be
reinforced. With a higher DoF, that we have obtained with the newer resin versions, we expect the
modulus to increase, but not in orders of magnitude. Another option is to include ceramic particles in the
resin, similar to what Elomaa and colleagues did with bioactive glass in their PCL resin [47]. In addition to
improving the stiffness, PCL composites with bioactive glass and other ceramics such as hydroxyapatite
and β-tri-calcium has been demonstrated improved cell adhesion and proliferation [47,75]. Thus, a double
positive effect can be obtained. However, if the increased DoF and ceramic inclusion is enough to provide
mechanical integrity is not feasible to conclude. Another option to consider is the addition of a stiffer
material, such as a low molecular, multiarmed PDLLA which has a modulus of up to 3.6 GPa [76]. The
downside of mixing with the PDLLA is that the material is likely to become far more brittle.
39
5.3 Biodegradation The trend observed is an initial slow, but steady degradation during the first seven day. Thereafter the
trend seems to plateau between day seven and fourteen, before it increases again up until day twenty-
one. The first day there is observed a higher mass drop in mean mass per day than for the next six days.
Some possible explanations could be uncured resins which diffused out in the alkaline environment, or
humidity absorbed by the sample prior to the first measurement.
The degradation study was conducted under sodium hydroxide accelerated condition. The strongly
alkaline environment the sodium hydroxide creates is not similar to what we expect under physiological
conditions but was used to get information of the degradation properties within a shorter time span. This
accelerated condition is commonly used in literature for accelerated degradation of PCL [77,78]. An
important aspect to the difference in environment is that under body stimulated conditions linear PCL
degrades through bulk degradation, meanwhile under alkaline accelerated condition surface erosion
becomes more dominant [60]. An alternative would have been to use acidic accelerated conditions, where
bulk degradation is dominant. It has recently been suggested that acidic accelerated conditions are more
representative than alkaline conditions [79]. Despite this potential difference, the degradation mode is
still hydrolysis [60], where water penetrates into the polymer causing random hydrolytic chain scission
and reduces the molecular weight of the polymer [53]. Lam et al. [60], used thermoplastic PCL (Mn = 80
kDa) FDM printed into a porous 0/90° lattice and a 5M NaOH solution, saw roughly 15% mass loss after 7
days meanwhile we saw 12 ± 2 % (n = 3). It is difficult to draw firm conclusions regarding the degradation
rate of stereolithography printed PCL with respect to linear PCL due to differences in initial molecular
weight, NaOH concentration, and specimen geometry. Regardless, our results demonstrate that the
material is indeed biodegradable through hydrolysis. It also shows indications of a gradual degradation of
the material, which is beneficial in terms of limiting the build-up of carboxylic degradation products.
5.4 Limitations Based on the current characterization, a crucial limitation to the material is the limited stiffness, where
the modulus of the bulk material seems to be around 100 MPa. Moreover, our FEA results indicates that
with the introduction of porosity, already at a lower porosity level of 60% the apparent modulus of the
scaffold will have dropped to less than 10% of the material modulus. Although the material is likely to
become stiffer if we increase the DoF, mix the resin with PDLLA, or include ceramic particles, we cannot
conclude whether the increase will be sufficient for loading application. In the current state, the SLA
40
printable PCL is unlikely to be used for bone grafts applied to high loading areas such as in spinal fusion
and fracture repair in the hip and legs. Instead, this material might be suitable for non-loaded bone repair
such as cranial defect repair. Nevertheless, to fully answer these questions the mechanical
characterization of the material must be comprehensive. We have identified the relevant ISO standards
for characterization of the material and scaffold properties but have only completed characterization of
the tensile properties for PCL-triacrylate (DoF = 60%). Although compressive testing was conducted, only
a single specimen was tested, and the test frame was unable to bring it to failure. Other important
properties such as flexural and fatigue performance have yet to be characterized, in addition to the
performance of the printed scaffold structures.
For printing we used the Form 2 SLA printer. Although it should have a resolution close to the laser spot
size of 130 μm, when printing small features (< 400 μm) we either experienced failed printing for solid
features or that gaps and pores fused together. This using a commercial resin already optimized by the
manufacturer. This could be due to the spot size being too large, but also be related to that the resin used
is old, or that window of print tank has been worn thereby degrading the resolution. For a bone graft,
with features and pores on the size of a few hundred micrometers, we need to be able to print these finer
features. If resolution is indeed the problem, newer projector based DLP and CLIP systems typically has a
resolution down towards 50-75 μm, which could solve the problem. A practical obstacle for the
commercialization of a stereolithography printed bone graft is the production time of the SLA system. But
again, a DLP system where a full layer is solidified at the time, or a CLIP system in which the manufacturing
occurs in a continuous manner, is likely to bring the productivity to an industrially acceptable level. Since
all systems build on the same principle of initiating a selective radical polymerization reaction, it is
expected that the material developed in this work can be translated from the Form 2 SLA system to a DLP
or CLIP system.
The biggest limitation with this project is the lack of characterization of the bioactivity of the material.
Currently only an accelerated degradation has been conducted indicating a slow degradation for a high
DoF resin, with the acceleration being mediated by an alkaline environment. Alkaline conditions are
established in literature for accelerated degradation model for PCL as bases improves catalysis of ester
hydrolysis [79]. The problem with alkaline acceleration is that the degradation mechanism is dominated
by surface erosion, meanwhile PCL is known for degrading through bulk degradation in vivo. Therefore it
has recently been suggested that acidic conditions are better acceleration models as the degradation
mechanism, bulk degradation, is replicated [79]. Still there is limited literature addressing the ability of
41
these models to replicate in vivo conditions accurately. Therefore, in vivo investigations are still required
to understand the materials degradation behaviour, both locally at the implantation site, but also
systematic effect.
Other important aspects such as cytotoxicity and the osteo-immuno response have yet to be
characterized. Previous work with a similar PCL system demonstrated biocompatibility with murine-
induced pluripotent stem cells and when implanted into pig retinas [42], and with fibroblast [41], but the
ability to induce osteogenesis or prevent a foreign body response and chronic inflammation has not been
proven. These aspects are crucial to understand before the material can be translated into clinical
application. In a similar work to ours, an SLA resin based on poly(trimethylene carbonate) was used in a
sheep model [80]. With the polymer only, they observed fibrous encapsulation of the material and
reported the material to lack osteoconductivity. When they introduced 40 wt.% hydroxyapatite to the
composition they saw some improvement in bone growth in their scaffold, but in five out of six samples
fibrous tissue ingrowth dominated. To the best of my any research directly comparing stereolithography
printed poly(trimethylene carbonate) to stereolithography PCL in vivo, there is well established in
literature that synthetic polymers has limited osteoconductivity compared to other biomaterials for bone
regeneration [6]. This support the concern of that stereolithography PCL might lack adequate bioactivity
to support tissue regeneration.
6. Future work Recently we have developed a protocol for (meth)acrylation of PCL which give us a high DoF (> 90%). The
material should then be optimized as a resin for our newly purchased DLP system (Prusa SL1). The material
properties of the material and lattice structures (Octet-truss or other promising alternatives) can then be
characterized according to the ISO/ASTM standards highlighted in Table 3.1. Thereafter the mixing with
other crosslinkable polymers, such as PDLLA, or inclusion of ceramic particles such as bioactive glass or
hydroxyapatite should be considered in order to increase the stiffness of the material. Lastly a thorough
characterization of the biological properties should be conducted, to build on the current work presented
in literature, the material’s osteo-immuno response should be evaluated. This includes investigating the
materials osteogenic capabilities, similar to what has been described by Ferreira et al. [81]. In addition, a
study investigating macrophage polarization, similar to the investigation by Goodman and colleagues [82],
would help to identify the immune response to the material system.
42
7. Conclusion In this work a three-armed PCL (meth)acrylate resin was successfully printed into a complex three-
dimensional lattice, and the material and scaffold properties were characterized.
PCL-triol was reacted with (meth)acryloyl chloride to form PCL-tri(meth)acrylate. The molar equivalence
between the PCL and the (meth)acryloyl chloride and the type of base (potassium
carbonate/triethylamine) governed the DoF. When crosslinked the potassium-based resin was
translucently colourless and fragile at handling, meanwhile the triethylamine-based resin had a yellow
translucent colour and seemed stronger when handling. The functionalized polymer was then formulated
into a photopolymerizable resin by introducing a photoinitiator, absorber, and inhibitor, and optimizing
their concentration to obtain suitable cross-linking kinetics. The initiator was used to initiate the radical
polymerization process, the absorber controlled the light penetration depth, and the inhibitor was added
to improve the shelf-life of the polymer. It was observed that without the inhibitor, the resin was unstable
and prone to self-polymerization during storage. In situ photorheology was used to quantify the reaction
kinetics by monitoring the change in storage modulus during UV irradiation. Faster kinetics were observed
with PCL-triacrylate compared to PCL-trimethacrylate. Furthermore, faster kinetics were observed with
increased photoinitiator concentration, meanwhile increased inhibitor concentration slowed down the
curing rate. This gave a working range, which in tandem with controlling the printer variables, can be used
for optimizing the polymerization process.
The ability to optimize the polymerization kinetics was demonstrated on a Form 2 SLA printer. A scaffold
on the centimetre scale with features on the submillimetre scale was successfully produced. An identical
scaffold structure has previously been printed by the group in nylon using SLS and in titanium using SLM.
The Form 2 is limited by that the settings is pre-defined by the manufacturer, which limits the tuning of
curing profile for custom-made resins. Therefore, further work will be conducted on a DLP system, which
allows for the control of curing time and the power density of the incident light. Another advantage with
switching to a projector-based system, such as DLP or CLIP, is that both the resolution and productivity
are expected to improve.
In sodium hydroxide accelerated conditions high DoF (> 90%) PCL-trimethacrylate had more than 20%
weight reduction in a 3-week period. This suggests the material exhibits a slow degradation through
hydrolysis in an aqueous medium. Under physiological conditions the degradation rate is likely to be
significantly slower. Ideally for tissue regeneration the degradation should occur in tandem with the
growth of the native tissue, but the high DoF PCL is likely to degrade slower than the tissue regrowth.
43
Some strategies worth investigating to improve the degradation rate is to use a lower DoF, replacing the
(meth)acrylate groups with anhydrides, or introducing a different polymer structure (e.g. PDLLA) The
accelerated degradation study allowed us to quickly evaluate if the material was degradable, a more
representable model would be non-accelerated in saline conditions, but this was outside our time scope
and it still would not give the full picture. To fully evaluate whether the degradation kinetics are
appropriate for a given tissue engineering application an in vivo model should be used to understand the
interplay between implant degradation and tissue regrowth.
The mechanical performance of the scaffold is dependent on the geometry of the scaffold, the material
type and the manufacturing method. Acknowledging this, it is important to use a readily reproduceable
and universally accepted method for characterization of the mechanical properties of both the porous
scaffold and the printed material. Therefore, ISO and ASTM standards describing the testing methodology
were identified and used for mechanical characterization. An interesting aspect from the tensile testing is
that the printed material exhibited anisotropic properties dependent on the print direction. Specifically,
the tensile substrates were weaker when loaded orthogonal to the print layers, which is suspected to be
due to weak binding between the layers. This highlights how also the manufacturing method affects the
mechanical performance of a material and its structure. If switching to CLIP, where the structure is printed
in a continuous manner, the printed material is expected and has been shown to behave isotropically as
there is no layering effect.
A preliminary trial was done for compressive testing, indicating that the material’s compressive modulus
is subphysiological. When introducing porosity into the scaffold, our FEA modelling suggests that the
relative modulus will be insufficient to support critical defects in load-bearing regions; however, there is
a need for bone graft substitutes in non-load-bearing regions such as cranial defect repair. Alternatively,
this material may find uses in other tissue engineering applications such as cartilage repair in the
craniofacial region.
At the same time, the material withstood stresses and strains that were higher than normal healthy bone,
suggesting that with appropriate reinforcement, the material may be applicable for bone repair in load-
bearing regions. Some promising strategies are the use of a second stiffer polymer such as PDLLA or use
of a reinforcing filler, such as a calcium phosphate or bioactive glass. A second advantage with including
an inorganic phase is that bioceramics tend to have a higher bioactivity than synthetic polymers and can
improve the initial cell attachment compared to PCL alone.
44
The (meth)acrylated PCL is a promising candidate material for tissue engineering, having demonstrated
excellent print resolution on a consumer grade stereolithographic printer. With regard to bone tissue
engineering, it is a promising material candidate for non-loading bearing applications, such as cranial
repair, but may require modifications through inclusion of ceramic particles or surface modification to
obtain adequate bioactivity. To fully evaluate this material, comprehensive mechanical testing must be
conducted according to the identified ISO/ASTM standards. When it comes to biodegradation, the kinetics
are likely to be too slow for ideal bone regeneration and strategies for increasing the degradation rate
should be investigated. To identify the severity of the slow degradation an in vivo model should be used.
8. References [1] C.H. Turner, Three rules for bone adaptation to mechanical stimuli, Bone. (1998).
doi:10.1016/S8756-3282(98)00118-5.
[2] F. Loi, L.A. Córdova, J. Pajarinen, T. hua Lin, Z. Yao, S.B. Goodman, Inflammation, fracture and bone repair, Bone. (2016). doi:10.1016/j.bone.2016.02.020.
[3] T. Briggs, A national review of adult elective orthopaedic services in England - GETTING IT RIGHT FIRST TIME, 2017.
[4] Royal College of Surgeons of England, Surgery and the NHS in numbers, (2014). https://www.rcseng.ac.uk/news-and-events/media-centre/media-background-briefings-and-statistics/surgery-and-the-nhs-in-numbers/ (accessed June 9, 2020).
[5] National Joint Registry, 15th annual report, 2018.
[6] H.J. Haugen, S.P. Lyngstadaas, F. Rossi, G. Perale, Bone grafts: which is the ideal biomaterial?, J. Clin. Periodontol. (2019). doi:10.1111/jcpe.13058.
[7] T.H. Ambrosi, A. Scialdone, A. Graja, S. Gohlke, A.M. Jank, C. Bocian, L. Woelk, H. Fan, D.W. Logan, A. Schürmann, L.R. Saraiva, T.J. Schulz, Adipocyte accumulation in the bone marrow during obesity and aging impairs stem cell-based hematopoietic and bone regeneration, Cell Stem Cell. (2017). doi:10.1016/j.stem.2017.02.009.
[8] I. Hernández-Gil, M.A.A. Gracia, M. Pingarrón, L. Jerez, Physiological bases of bone regeneration II. The remodeling process, Med Oral Patol Oral Cir Bucal. 11 (2006) E151-215.
[9] R. Ferracini, A. Bistolfi, R. Garibaldi, V. Furfaro, A. Battista, G. Perale, Composite xenohybrid bovine bone-derived scaffold as bone substitute for the treatment of tibial plateau fractures, Appl. Sci. 9 (2019) 2675.
[10] A. Tiwari, Current concepts in surgical treatment of osteosarcoma, J. Clin. Orthop. Trauma. 3 (2012) 4–9. doi:10.1016/j.jcot.2012.04.004.
[11] S.J. Lipson, Spinal-fusion surgery — Advances and concerns, N. Engl. J. Med. 350 (2004) 643–644. doi:10.1056/NEJMp038162.
45
[12] A.E. Ropper, A.H. Ropper, Acute spinal cord compression, N. Engl. J. Med. 376 (2017) 1358–1369. doi:10.1056/NEJMra1516539.
[13] P. Baldwin, D.J. Li, D.A. Auston, H.S. Mir, R.S. Yoon, K.J. Koval, Autograft, allograft, and bone graft substitutes: clinical evidence and indications for use in the setting of orthopaedic trauma surgery, J. Orthop. Trauma. 33 (2019) 203–213. doi:10.1097/BOT.0000000000001420.
[14] H. Schaaf, S. Lendeckel, H.-P. Howaldt, P. Streckbein, Donor site morbidity after bone harvesting from the anterior iliac crest, Oral Surgery, Oral Med. Oral Pathol. Oral Radiol. Endodontology. 109 (2010) 52–58. doi:https://doi.org/10.1016/j.tripleo.2009.08.023.
[15] L. Vastel, A. Meunier, H. Siney, L. Sedel, J.P. Courpied, Effect of different sterilization processing methods on the mechanical properties of human cancellous bone allografts, Biomaterials. (2004). doi:10.1016/j.biomaterials.2003.08.067.
[16] M.F. Moreau, Y. Gallois, M.F. Baslé, D. Chappard, Gamma irradiation of human bone allografts alters medullary lipids and releases toxic compounds for osteoblast-like cells, Biomaterials. (2000). doi:10.1016/S0142-9612(99)00193-3.
[17] G. Pertici, F. Rossi, T. Casalini, G. Perale, Composite polymer-coated mineral grafts for bone regeneration: material characterisation and model study, Ann. Oral Maxillofac. Surg. 2 (2014).
[18] G. Pertici, F. Carinci, G. Carusi, D. Epistatus, T. Villa, F. Crivelli, F. Rossi, G. Perale, Composite polymer-coated mineral scaffolds for bone regeneration: from material characterization to human studies, J. Biol. Regul. Homeost. Agents. 29 (2015) 136–148.
[19] G. Perale, M. Monjo, J.M. Ramis, Ø. Øvrebø, F. Betge, P. Lyngstadaas, H.J. Haugen, Biomimetic biomolecules in next generation xeno-hybrid bone graft material show enhanced in vitro bone cells response, J. Clin. Med. (2019). doi:10.3390/jcm8122159.
[20] J.R. Jones, Review of bioactive glass: from Hench to hybrids, Acta Biomater. 9 (2013) 4457–4486. doi:10.1016/j.actbio.2012.08.023.
[21] D. Tadic, M. Epple, A thorough physicochemical characterisation of 14 calcium phosphate-based bone substitution materials in comparison to natural bone, Biomaterials. (2004). doi:10.1016/S0142-9612(03)00621-5.
[22] H. Haugen, J. Will, A. Kohler, J. Aigner, E. Wintermantel, Processing and characterisation of a potential TiO2 scaffold, Bioceram. Vol 16. 254–2 (2004) 941–944.
[23] T. Patrício, M. Domingos, A. Gloria, P. Bártolo, Characterisation of PCL and PCL/PLA scaffolds for tissue engineering, 2013. doi:10.1016/j.procir.2013.01.022.
[24] H. Xu, D. Han, J.-S. Dong, G.-X. Shen, G. Chai, Z.-Y. Yu, W.-J. Lang, S.-T. Ai, Rapid prototyped PGA/PLA scaffolds in the reconstruction of mandibular condyle bone defects, Int. J. Med. Robot. Comput. Assist. Surg. 6 (2010) 66–72. doi:10.1002/rcs.290.
[25] K. Rezwan, Q.Z. Chen, J.J. Blaker, A.R. Boccaccini, Biodegradable and bioactive porous polymer/inorganic composite scaffolds for bone tissue engineering, 2006. doi:10.1016/j.biomaterials.2006.01.039.
[26] S. de Lacerda Schickert, J.J.J.P. van den Beucken, S.C.G. Leeuwenburgh, J.A. Jansen, Pre-clinical evaluation of biological bone substitute materials for application in highly loaded skeletal sites,
46
Biomolecules. 10 (2020) 883. doi:10.3390/biom10060883.
[27] M. Rahmati, E.A. Silva, J.E. Reseland, C.A. Heyward, H.J. Haugen, Biological responses to physicochemical properties of biomaterial surface, Chem. Soc. Rev. (2020). doi:10.1039/D0CS00103A.
[28] J. Pajarinen, T. Lin, E. Gibon, Y. Kohno, M. Maruyama, K. Nathan, L. Lu, Z. Yao, S.B. Goodman, Mesenchymal stem cell-macrophage crosstalk and bone healing, Biomaterials. (2019). doi:10.1016/j.biomaterials.2017.12.025.
[29] M.M. Stevens, Biomaterials for bone tissue engineering, Mater. Today. 11 (2008) 18–25. doi:10.1016/S1369-7021(08)70086-5.
[30] E. Garcia-Gareta, M.J. Coathup, G.W. Blunn, Osteoinduction of bone grafting materials for bone repair and regeneration, Bone. 81 (2015) 112–121. doi:10.1016/j.bone.2015.07.007.
[31] S.H. Teoh, D.W. Hutmacher, K.C. Tan, K.F. Tam, I. Zein, Three-dimensional bioresorbable scaffolds for tissue engineering applications, US 8,071,007 B1, 2011.
[32] F.P.W. Melchels, M.A.N. Domingos, T.J. Klein, J. Malda, P.J. Bartolo, D.W. Hutmacher, Additive manufacturing of tissues and organs, 2012. doi:10.1016/j.progpolymsci.2011.11.007.
[33] F.P.W. Melchels, J. Feijen, D.W. Grijpma, A review on stereolithography and its applications in biomedical engineering, Biomaterials. 31 (2010) 6121–6130. doi:10.1016/j.biomaterials.2010.04.050.
[34] J.R. Tumbleston, D. Shirvanyants, N. Ermoshkin, R. Janusziewicz, A.R. Johnson, D. Kelly, K. Chen, R. Pinschmidt, J.P. Rolland, A. Ermoshkin, Continuous liquid interface production of 3D objects, Science (80-. ). 347 (2015) 1349–1352. doi:10.1126/science.aaa2397.
[35] A. Koroleva, A.A. Gill, I. Ortega, J.W. Haycock, S. Schlie, S.D. Gittard, B.N. Chichkov, F. Claeyssens, Two-photon polymerization-generated and micromolding-replicated 3D scaffolds for peripheral neural tissue engineering applications, Biofabrication. 4 (2012) 25005. doi:10.1088/1758-5082/4/2/025005.
[36] P.X. Lan, J.W. Lee, Y.-J. Seol, D.-W. Cho, Development of 3D PPF/DEF scaffolds using micro-stereolithography and surface modification, J. Mater. Sci. Mater. Med. 20 (2009) 271–279. doi:10.1007/s10856-008-3567-2.
[37] J.W. Lee, G.S. Ahn, D.S. Kim, D.W. Cho, Development of nano- and microscale composite 3D scaffolds using PPF/DEF-HA and micro-stereolithography, Microelectron. Eng. (2009). doi:10.1016/j.mee.2008.12.038.
[38] Y. Luo, G. Le Fer, D. Dean, M.L. Becker, 3D Printing of Poly(propylene fumarate) Oligomers: Evaluation of Resin Viscosity, Printing Characteristics and Mechanical Properties, Biomacromolecules. 20 (2019) 1699–1708. doi:10.1021/acs.biomac.9b00076.
[39] F.P.W. Melchels, J. Feijen, D.W. Grijpma, A poly(d,l-lactide) resin for the preparation of tissue engineering scaffolds by stereolithography, Biomaterials. 30 (2009) 3801–3809. doi:10.1016/J.BIOMATERIALS.2009.03.055.
[40] O. Guillaume, M.A. Geven, C.M. Sprecher, V.A. Stadelmann, D.W. Grijpma, T.T. Tang, L. Qin, Y. Lai, M. Alini, J.D. de Bruijn, Surface-enrichment with hydroxyapatite nanoparticles in
47
stereolithography-fabricated composite polymer scaffolds promotes bone repair, Acta Biomater. 54 (2017) 386–398. doi:10.1016/j.actbio.2017.03.006.
[41] L. Elomaa, S. Teixeira, R. Hakala, H. Korhonen, D.W. Grijpma, J. V Seppälä, Preparation of poly (ε-caprolactone)-based tissue engineering scaffolds by stereolithography, Acta Biomater. 7 (2011) 3850–3856. doi:10.1016/j.actbio.2011.06.039.
[42] B.J. Green, K.S. Worthington, J.R. Thompson, S.J. Bunn, M. Rethwisch, E.E. Kaalberg, C. Jiao, L.A. Wiley, R.F. Mullins, E.M. Stone, E.H. Sohn, B.A. Tucker, C.A. Guymon, Effect of molecular weight and functionality on acrylated poly(caprolactone) for stereolithography and biomedical applications, Biomacromolecules. 19 (2018) 3682–3692. doi:10.1021/acs.biomac.8b00784.
[43] B. van Bochove, D.W. Grijpma, Photo-crosslinked synthetic biodegradable polymer networks for biomedical applications, J. Biomater. Sci. Polym. Ed. 30 (2019) 77–106. doi:10.1080/09205063.2018.1553105.
[44] M.A. Geven, D. Barbieri, H. Yuan, J.D. de Bruijn, D.W. Grijpma, Preparation and mechanical properties of photo-crosslinked poly(trimethylene carbonate) and nano-hydroxyapatite composites, Clin. Hemorheol. Microcirc. 60 (2015) 3–11. doi:10.3233/CH-151936.
[45] K.E.G. Dienel, B. van Bochove, J. V Seppälä, Additive manufacturing of bioactive poly(trimethylene carbonate)/β-tricalcium phosphate composites for bone regeneration, Biomacromolecules. 21 (2020) 366–375. doi:10.1021/acs.biomac.9b01272.
[46] T. Kuhnt, R.M. García, S. Camarero-Espinosa, A. Dias, A.T. ten Cate, C.A. van Blitterswijk, L. Moroni, M.B. Baker, Poly (caprolactone-co-trimethylenecarbonate) urethane acrylate resins for digital light processing of bioresorbable tissue engineering implants, Biomater. Sci. 7 (2019) 4984–4989.
[47] L. Elomaa, A. Kokkari, T. Närhi, J. V. Seppälä, Porous 3D modeled scaffolds of bioactive glass and photocrosslinkable poly(ε-caprolactone) by stereolithography, Compos. Sci. Technol. (2013). doi:10.1016/j.compscitech.2012.10.014.
[48] L. Elomaa, Y. Kang, J. V Seppälä, Y. Yang, Biodegradable photocrosslinkable poly(depsipeptide-co-ε-caprolactone) for tissue engineering: Synthesis, characterization, and In vitro evaluation, J. Polym. Sci. Part A Polym. Chem. 52 (2014) 3307–3315. doi:10.1002/pola.27400.
[49] A. Ronca, S. Ronca, G. Forte, S. Zeppetelli, A. Gloria, R. De Santis, L. Ambrosio, Synthesis and characterization of divinyl-fumarate poly-ε-caprolactone for scaffolds with controlled architectures, J. Tissue Eng. Regen. Med. 12 (2018) e523–e531. doi:10.1002/term.2322.
[50] F.P.W. Melchels, J. Feijen, D.W. Grijpma, A poly(d,l-lactide) resin for the preparation of tissue engineering scaffolds by stereolithography, Biomaterials. (2009). doi:10.1016/j.biomaterials.2009.03.055.
[51] A. Ronca, L. Ambrosio, D.W. Grijpma, Preparation of designed poly(d,l-lactide)/nanosized hydroxyapatite composite structures by stereolithography, Acta Biomater. 9 (2013) 5989–5996. doi:https://doi.org/10.1016/j.actbio.2012.12.004.
[52] E. Collin, M. Pereira, J.R. Maia, V. Cirillo, J. Lamazouade, S. Keaveney, E. O’Cearbhaill, 3D PRINTING COMPOSITION FOR BIOMATERIALS, WO2019180208A1, 2018.
[53] M.A. Woodruff, D.W. Hutmacher, The return of a forgotten polymer—polycaprolactone in the
48
21st century, Prog. Polym. Sci. 35 (2010) 1217–1256. doi:10.1016/j.progpolymsci.2010.04.002.
[54] M. Labet, W. Thielemans, Synthesis of polycaprolactone: a review, Chem. Soc. Rev. 38 (2009) 3484–3504. doi:10.1039/B820162P.
[55] C.L.E. Nijst, J.P. Bruggeman, J.M. Karp, L. Ferreira, A. Zumbuehl, C.J. Bettinger, R. Langer, Synthesis and characterization of photocurable elastomers from poly (glycerol-co-sebacate), Biomacromolecules. 8 (2007) 3067–3073. doi:10.1021/bm070423u.
[56] Ø. Øvrebø, “Development of a biodegradable polymer resin for stereolithography printing of tissue engineering constructs” - MRes Thesis, Imperial College London, London, 2019.
[57] P.T. Smith, B. Narupai, J.H. Tsui, S.C. Millik, R.T. Shafranek, D.-H. Kim, A. Nelson, Additive manufacturing of bovine serum albumin-based hydrogels and bioplastics, Biomacromolecules. 21 (2020) 484–492. doi:10.1021/acs.biomac.9b01236.
[58] V.S. Deshpande, N.A. Fleck, M.F. Ashby, Effective properties of the octet-truss lattice material, J. Mech. Phys. Solids. 49 (2001) 1747–1769. doi:10.1016/S0022-5096(01)00010-2.
[59] S.-J. Wang, M.-C. Yip, Y.-S. Hsu, K.-G. Lai, S.-Y. Wang, The Modulus of Toughness of Urinary Calculi , J. Biomech. Eng. 124 (2001) 133–134. doi:10.1115/1.1431264.
[60] C.X.F. Lam, M.M. Savalani, S.-H. Teoh, D.W. Hutmacher, Dynamics of in vitro polymer degradation of polycaprolactone-based scaffolds: accelerated versus simulated physiological conditions, Biomed. Mater. 3 (2008) 34108. doi:10.1088/1748-6041/3/3/034108.
[61] S. Ruiz de Galarreta, J.R.T. Jeffers, S. Ghouse, A validated finite element analysis procedure for porous structures, Mater. Des. (2020). doi:10.1016/j.matdes.2020.108546.
[62] X. Zheng, H. Lee, T.H. Weisgraber, M. Shusteff, J. DeOtte, E.B. Duoss, J.D. Kuntz, M.M. Biener, Q. Ge, J.A. Jackson, S.O. Kucheyev, N.X. Fang, C.M. Spadaccini, Ultralight, ultrastiff mechanical metamaterials, Science (80-. ). 344 (2014) 1373 LP – 1377. doi:10.1126/science.1252291.
[63] N. Reznikov, O.R. Boughton, S. Ghouse, A.E. Weston, L. Collinson, G.W. Blunn, J.R.T.T. Jeffers, J.P. Cobb, M.M. Stevens, Individual response variations in scaffold-guided bone regeneration are determined by independent strain-and injury-induced mechanisms, Biomaterials. 194 (2019) 183–194. doi:10.1016/J.BIOMATERIALS.2018.11.026.
[64] P.J. Bártolo, Stereolithography: materials, processes and applications, Springer Science & Business Media, 2011.
[65] V. Jašo, D. Stoiljković, R. Radičević, O. Bera, Kinetic modeling of bulk free-radical polymerization of methyl methacrylate, Polym. J. 45 (2013) 631–636. doi:10.1038/pj.2013.6.
[66] C.M. Cheah, J.Y.H. Fuh, A.Y.C. Nee, L. Lu, Y.S. Choo, T. Miyazawa, Characteristics of photopolymeric material used in rapid prototypes Part II. Mechanical properties at post-cured state, J. Mater. Process. Technol. 67 (1997) 46–49. doi:10.1016/S0924-0136(96)02816-6.
[67] R. Janusziewicz, J.R. Tumbleston, A.L. Quintanilla, S.J. Mecham, J.M. DeSimone, Layerless fabrication with continuous liquid interface production, Proc. Natl. Acad. Sci. 113 (2016) 11703–11708. doi:10.1073/PNAS.1605271113.
[68] R. Oftadeh, M. Perez-Viloria, J.C. Villa-Camacho, A. Vaziri, A. Nazarian, Biomechanics and mechanobiology of trabecular bone: a review, J. Biomech. Eng. 137 (2015) 10802.
49
doi:10.1115/1.4029176.
[69] T. Bellido, Osteocyte-driven bone remodeling, Calcif. Tissue Int. 94 (2014) 25–34. doi:10.1007/s00223-013-9774-y.
[70] R. Wauthle, S.M. Ahmadi, S. Amin Yavari, M. Mulier, A.A. Zadpoor, H. Weinans, J. Van Humbeeck, J.-P. Kruth, J. Schrooten, Revival of pure titanium for dynamically loaded porous implants using additive manufacturing, Mater. Sci. Eng. C. 54 (2015) 94–100. doi:https://doi.org/10.1016/j.msec.2015.05.001.
[71] L.J. Gibson, M.F. Ashby, Cellular solids: structure and properties, Cambridge university press, 1999.
[72] M.F. Ashby, The properties of foams and lattices, Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 364 (2006) 15–30. doi:10.1098/rsta.2005.1678.
[73] V. Karageorgiou, D. Kaplan, Porosity of 3D biomaterial scaffolds and osteogenesis, Biomaterials. 26 (2005) 5474–5491. doi:10.1016/j.biomaterials.2005.02.002.
[74] V.S. Deshpande, M.F. Ashby, N.A. Fleck, Foam topology: bending versus stretching dominated architectures, Acta Mater. 49 (2001) 1035–1040. doi:10.1016/S1359-6454(00)00379-7.
[75] B. Huang, G. Caetano, C. Vyas, J.J. Blaker, C. Diver, P. Bártolo, Polymer-ceramic composite scaffolds: the effect of hydroxyapatite and β-tri-calcium phosphate, Materials (Basel). 11 (2018) 129. doi:10.3390/ma11010129.
[76] F.P.W. Melchels, J. Feijen, D.W. Grijpma, A poly (D, L-lactide) resin for the preparation of tissue engineering scaffolds by stereolithography, Biomaterials. 30 (2009) 3801–3809. doi:10.1016/j.biomaterials.2009.03.055.
[77] S.-H. Lee, J.H. Lee, Y.-S. Cho, Analysis of degradation rate for dimensionless surface area of well-interconnected PCL scaffold via in-vitro accelerated degradation experiment, Tissue Eng. Regen. Med. 11 (2014) 446–452. doi:10.1007/s13770-014-0067-y.
[78] C.X.F. Lam, S.H. Teoh, D.W. Hutmacher, Comparison of the degradation of polycaprolactone and polycaprolactone–(β‐tricalcium phosphate) scaffolds in alkaline medium, Polym. Int. 56 (2007) 718–728. doi:10.1002/pi.2195.
[79] M. Bartnikowski, T.R. Dargaville, S. Ivanovski, D.W. Hutmacher, Degradation mechanisms of polycaprolactone in the context of chemistry, geometry and environment, Prog. Polym. Sci. 96 (2019) 1–20. doi:10.1016/j.progpolymsci.2019.05.004.
[80] O. Guillaume, M.A. Geven, V. Varjas, P. Varga, D. Gehweiler, V.A. Stadelmann, T. Smidt, S. Zeiter, C. Sprecher, R.R.M. Bos, D.W. Grijpma, M. Alini, H. Yuan, G.R. Richards, T. Tang, L. Qin, L. Yuxiao, P. Jiang, D. Eglin, Orbital floor repair using patient specific osteoinductive implant made by stereolithography, Biomaterials. 233 (2020) 119721. doi:https://doi.org/10.1016/j.biomaterials.2019.119721.
[81] S.A. Ferreira, G. Young, J.R. Jones, S. Rankin, Bioglass/carbonate apatite/collagen composite scaffold dissolution products promote human osteoblast differentiation, Mater. Sci. Eng. C. 118 (2021) 111393. doi:https://doi.org/10.1016/j.msec.2020.111393.
[82] J.K. Antonios, Z. Yao, C. Li, A.J. Rao, S.B. Goodman, Macrophage polarization in response to wear
50
particles in vitro, Cell. Mol. Immunol. 10 (2013) 471–482. doi:10.1038/cmi.2013.39.
51
Supplementary Material S1. Resin Functionalization
Methodology for resin functionalization similar to that developed during MRes and presented in MRes
thesis [56]:
PCL-triol (Mw=300 Da, 50 g, 0.17 mol, 1 equivalent) was introduced in a two-necked round bottom flask
and was melted using a heat gun at 70 °C. Potassium carbonate (103.65 g, 0.75 mol, 4.5 equivalents) was
added and the suspension was then dried under high vacuum for 15 min. 400 mL of anhydrous
dichloromethane was added. The mixture was stirred vigorously until the polymer was fully dissolved.
Acryloyl chloride (60.6 mL, 0.75 mol, 4.5 equivalent) was dropwise over 30 min under constant stirring.
The reaction was covered with aluminium foil and left at room temperature under N2 atmosphere to react
overnight. 30 mL methanol was added to neutralize unreacted acryloyl chloride, and the solution was
stirred under N2 for 30 mins. Potassium carbonate was separated from the solution by centrifugation (5
min, 5000 RCF, 10 °C). The supernatant was poured into a round bottom flask and concentrated in vacuo.
The residue was then dissolved in a minimal amount of tetrahydrofuran and precipitated by dropwise
addition into cold methanol under vigorous stirring. The methanol was cooled to -80 °C using a dry
ice/acetone bath. The PCL precipitate was recovered after centrifugation for 1 min at 7000 RCF, -11 °C,
then any remaining solvent was removed from the PCL using rotary evaporation. The resin was moved to
a dark glass container. Finally, the resin was left under vacuum overnight to remove any volatile impurities
such as methyl acrylate. The resin was stored in a freezer at -20 °C. 200 ppm MEHQ inhibitor was added
to improve storage. For PCL-trimethacrylate, the acryloyl chloride was replaced by methacryloyl chloride,
but the molar equivalences were kept the same.
Resins with a lower DoF (<70%) were synthesized under the direction of Dr. Jonathan Wojciechowski
following the protocol above. Resins with a high DoF (>90%) were synthesized by Dr. Cécile Echalier
following an optimized protocol that uses triethylamine instead of potassium carbonate as a base. When
potassium carbonate was used the resin had a transparent look, meanwhile with triethylamine the resin
obtained a yellow tint.
The DoF was calculated from NMR spectra by integrating the vinyl groups with referenced to the
integration of the PCL-triol initiator methyl protons (Figure 3.1) The NMR was done as following ([56]):
“1H nuclear magnetic resonance (NMR – Bruker, MA, US) was used to confirm the functionalization and
calculate the degree of functionalization of the resins. The resins were dissolved in deuterated chloroform
(CDCl3) at a 10 mg/mL concentration. 1H spectra were collected on a Bruker Advance III HD 400 MHz
52
spectrometer at 297 K equipped with a 5 mm PABBO BB sensor. Chemical shifts are given in ppm
calibrated with the residual solvent (CDCl3 – δ= 7.28 ppm). 1H NMR spectra were typically recorded using
64 scans with a spectral width of 6000 Hz and a recovery delay of 10 s. Spectra were processed and
visualized with MestReNova 6.0.2 (Mestrelab research, Santiago, Spain).
Figure S3.1: Top: Reaction schematic for acrylate functionalization of PCL-triol. Bottom left: sample NMR
spectra for PCL-triacrylate resin. Bottom right: sample NMR spectra for PCL-trimethacrylate resin. In b)
and c) m refers to the methyl proton peak, a, b, and c to the vinyl protons of the acrylate, and d and e to
the vinyl protons of the methacrylate. Reproduced from MRes thesis [56].
53
S4.1 – UV Rheology, Clear V4, Red=35 °C, blue=RT
S4.2 UV Rheology, mean plot for Clear V4 at RT (n=2):
54
S4.3 UV Rheology mixing time points – MA300 DoF=99% with 200 ppm inhibitor. Blue = Clear V4 at RT
(Average of 2 repeats).
S4.4 UV Rheology, different concentrations Irgacure 369 photoinitiatior, PCL-MA300 w. 200ppm MEHQ.
(Blue: Clear V4 – Average of 2 repeats).
55
S4.5 UV Rheology, PCL-TMA300 with (green) and without (red) 200 ppm MEHQ.
(Blue: Clear V4 – Average of 2 repeats).
![PTMC: MICROFABRICATION & STEREOLITHOGRAPHY · Stereolithography is a form of prototyping that has been shown to be very versatile with highest accuracy and precision.[11] Stereolithography](https://img.pdfslide.us/doc/110x75/605ef4b2b0307a40e8391640/ptmc-microfabrication-stereolithography-stereolithography-is-a-form-of-prototyping.jpg)
