Virtual and Physical Prototyping

ISSN: 1745-2759 (Print) 1745-2767 (Online) Journal homepage: [Link]

Extrusion-based additive manufacturing of PEEK

for biomedical applications

Mohammad Vaezi & Shoufeng Yang

To cite this article: Mohammad Vaezi & Shoufeng Yang (2015): Extrusion-based additive
manufacturing of PEEK for biomedical applications, Virtual and Physical Prototyping, DOI:
10.1080/17452759.2015.1097053

To link to this article: [Link]

Published online: 13 Oct 2015.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

[Link]

Download by: [University of Southampton Highfield] Date: 17 October 2015, At: 13:18
 VIRTUAL AND PHYSICAL PROTOTYPING, 2015
[Link]

Extrusion-based additive manufacturing of PEEK for biomedical applications

Mohammad Vaezi and Shoufeng Yang
Faculty of Engineering & the Environment, University of Southampton, Southampton, UK

ABSTRACT ARTICLE HISTORY

There has been a trend in recent years to develop polyetheretherketone (PEEK)-based medical Received 12 September 2015
devices due to the excellent cell biocompatibility and desirable mechanical properties of PEEK, Accepted 18 September 2015
which has elastic modulus comparable to cortical bone. Different manufacturing techniques such
KEYWORDS
as injection moulding, particulate leaching, compression moulding, and selective laser sintering Additive manufacturing (AM);
(SLS) have been used to produce porous PEEK for biomedical applications. Despite a large Polyetheretherketone (PEEK);
number of publications on extrusion-based additive manufacturing (AM) of porous structures
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

3D printing; extrusion
using various materials, there have been very few general reports on extrusion AM of low quality freeforming
small PEEK parts without defects such as warpage and delamination and no further assessment
of mechanical properties. Successful low-cost 3D printing of PEEK structures using filament-
based extrusion AM process is reported in this paper for the first time. Hot extrusion head
design, extrusion temperature, and ambient temperature were identified as important factors
need to be taken into consideration for printing PEEK structures without warpage, delamination,
and polymer degradation. Compression and tensile tests were conducted to investigate
mechanical properties of these new 3D printed PEEK structures. The air gap between infill
pattern and entrapped micro-bubbles inside filaments were identified as the main source of
mechanical properties reduction. In addition, three-point flexural test was performed on the 3D
printed PEEK and compared with flexural specimens printed using other AM materials and
techniques.

1. Introduction
et al. 2014), while it adds more complexity from manufac-
Polyetheretherketone (PEEK) is an efficient semi-crystalline turing prospective. Porous PEEK becomes a physical form
thermoplastic alternative to implantable metal materials as that more closely replicates the natural tissue and allows
it has excellent biocompatibility and combines good orientation of cells in configuration closer to nature, and
strength and stiffness (Kurtz and Devine 2007, Edwards enhanced fixation of device (Jarman-Smith et al. 2012).
et al. 2010, Toth 2012). Elastic modulus of PEEK is similar Manufacturing of porous PEEK materials has been
to cortical bone, which reduced stress shielding after widely investigated for use in different industries but
implantation (Sobieraj and Rimnac 2012). It is also radiolu- has been limited for medical application by either the
cent which permits radiographic assessment (Kurtz 2012c). type of manufacturing technology or its ability to fulfil
The combination of these properties has made PEEK of the key end-user requirements. The main end-user
great potential for orthopaedic application. Medical- requirements for an orthopaedic porous material are
grade PEEK-OPTIMA has been developed by Invibio that (i) the pore interconnectivity needs to be sufficient
([Link]) to meet with Food and Drug Adminis- to allow cells and nutrients to pass into and through
tration (FDA) requirements and has been used in multiple the material; (ii) the pore size must be suitable (∼100–
clinical applications such as spinal cage fusion, total join 600 µm) to permit cells and vascularisation of the bioma-
replacement, and craniomaxillofacial reconstruction (Toth terial, and yet able to retain desirable structural integrity
et al. 2006, Lovald and Kurtz 2012, Kurtz et al. 2012, Rao to resist the mechanical forces (Jarman-Smith et al.
et al. 2014). 2012). This is not easily achievable by direct utilisation
Despite possessing favourable biomaterial properties or modification of the current available industrial
described, PEEK is relatively bioinert, demonstrating porous material manufacturing technologies, due to
insufficient osseointegration following implantation. their limitations such as unsuitable producible geome-
Incorporating porosity into PEEK implants has been tries and process contamination. A few industrial tech-
identified as an effective approach to improve osseointe- niques have been identified as transferable into
gration (Jarman-Smith et al. 2012, Zhao et al. 2013, Wang medical, including particulate leaching, heat and

CONTACT Shoufeng Yang [Link]@[Link]

© 2015 Taylor & Francis
 2 M. VAEZI AND S. YANG

compression sintering, micromachining, and selective design with minimum die land, unique bioceramic
laser sintering (SLS) process (Jarman-Smith et al. 2012). paste formulation, and the equipment built in-house
Particulate leaching is a flexible and economic platform (Yang et al. 2008a, 2008b, 2008c).
among currently available porous PEEK manufacturing Although there has been a large number of publi-
techniques but has less control on micro/macrostructure cations in the field of extrusion freeforming, so far
(e.g. porosity and architecture), lacks versatility and there have been very few preliminary technical (Valentan
suffers from limitations such as manual intervention, and et al. 2013) and commercial reports (INDMATEC GmbH,
inconsistency. Compression moulding can be integrated [Link]) on extrusion AM of PEEK. It is
with particulate leaching for production of porous PEEK quite challenging to process PEEK through extrusion
by the addition of a fugitive particle (e.g. sodium chloride) freeforming due to its very high melting temperature
into the compound that is subsequently leached out by in comparison with other thermoplastics or biopolymers
soaking into a solvent, post moulding (Roeder and such as acrylonitrile butadiene styrene (ABS) and poly
Conrad 2012, Siddiq and Kennedy 2015). SLS, a type of (lactic acid) (PLA). The preliminary report on extrusion
powder-based additive manufacturing (AM) technology, AM of PEEK is promising (Valentan et al. 2013).
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

is capable of fabricating porous and/or bioactive PEEK- However, very small parts could be printed with insuffi-
based composites with very complex architecture (great cient quality (e.g. defects such as part warpage and/or
control on pore size/shape), permitting greater design delamination) and no further discussion on the main
freedom (Tan et al. 2003, 2005, Schmidt et al. 2007, Von Wil- challenges and obstacles, and mechanical properties
mowsky et al. 2008). has been presented. In this paper, successful extrusion
Extrusion freeforming, another type of AM technol- AM of PEEK structures is reported for the first time with
ogies, has been identified as the most efficient method in-depth discussion on extrusion system, the main chal-
to make porous 3D structures such as tissue engineering lenges, and the process parameters need to be adjusted.
(TE) scaffolds due to its ability of processing a wide range Furthermore, mechanical properties of the extrusion
of materials (ceramics, thermoplastics, and cells), high freeformed porous PEEK parts are investigated for the
reproducibility, great control on macro/microarchitec- first time in this paper.
ture, and more critically its low cost (Vaezi et al. 2013a,
2013b, Vaezi and Yang 2014). This technology can also
be used for 3D printing of non-scaffold constructs such 2. Materials and methods
as cells (Bhuthalingam et al. 2015), hydrogels (Wang
2.1. Experimental set-up
et al. 2015), and food (Sun et al. 2015). The authors
have reported the highest resolution low-temperature Two different high-temperature extrusion-based AM
bioceramic extrusion freeforming in which filaments systems with different extrusion head configurations
can be delivered with high precision with diameters were set up and tested: syringe-based (Figure 1(a)) and
down to 60 µm with the use of optimised nozzle filament-based extrusion head designs (Figure 1(b)). In

Figure 1. Schematic of (a) the syringe-based and (b) the filament-based device set-up for PEEK 3D printing.
 VIRTUAL AND PHYSICAL PROTOTYPING 3

both techniques, one nozzle was used for deposition of (that is suitable for printing with a build layer thickness
both part and support materials. of 0.2 mm) and an ambient temperature of 80°C to
As for the syringe-based method, the pre-set device define the extrusion temperature range with no nozzle
suitable for low-temperature printing of bioceramic scaf- clogging and polymer degradation.
folds (Vaezi and Yang 2015) was modified in order to
reach nozzle temperature up to 450°C for PEEK extrusion.
2.2. Mechanical tests
The extrusion syringe was made from two different metal
tubes coupled together: a brass tube with an internal The 3D printed porous samples were cut using a
diameter of 17 mm with good thermal conductivity diamond cutter (Mecatome T210, Presi, France) for
attached to a 500 µm nozzle (high-pressure sapphire microstructural analysis. The morphologies of the 3D
nozzle-type 60, Quick-ohm, Germany) and a stainless printed parts and porous scaffolds were examined
steel tube with high strength and lower thermal conduc- using optical microscopy (Olympus BH2-UMA, Japan)
tivity. Using the brass tube, PEEK can absorb sufficient and scanning electron microscopy (SEM) (JEOL JSM-
energy to get fully melted and easy to extrude through 6500F, Oxford Instruments, UK). In addition, the printed
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

a nozzle. In the other region (within stainless steel PEEK scaffolds were characterised in terms of porosity
tube), PEEK has lower temperature since there is less and mechanical properties. Porosity was measured by
thermal conductivity, and thus less chance to get over- correlating PEEK scaffolds’ geometrical dimensions and
heated and degraded. A self-adhesive etched foil material density as per ASTM F 2450-04:
silicone heater was also used to get the substrate

m
heated (up to 170°C) in order to transfer heat to the VT −
r
printed PEEK parts continuously to minimise thermal Porosity (%) = × 100, (1)
VT
stress. As for filament-based system, a driving gear
feeding system was used to extrude PEEK through a where VT is the total volume occupied by the network
0.4 mm nozzle and deposited layer by layer to build up of pores (determined from calliper measurements of
complex 3D structures. For this end, a UP 3D printer length, width, and height), ρ is the material density
([Link]) was modified so that nozzle tempera- (1.30 g/cm3 for Victrex® PEEK 450G), m is specimen’s
ture up to 460°C and heated build plate of 130°C could mass measured using a Mettler AE240 micro-balance
be achieved. In addition, heating lamps surrounding (LET, USA).
the build area were used to control ambient temperature Based on the standard compressive test method
(approx. 80°C) to avoid severe part warpage and specified by ASTM D695-02a, the stress–strain responses
delamination. were investigated on the printed closed-cell porous
PEEK-OPTIMA LT3 powder with a median particle size specimens with dimensions of 12.5 mm (diameter) by
of 10 µm was used as received from Invibio for the exper- 25 mm (height). The specimens had an average porosity
iments using the syringe-based device. PEEK-OPTIMA of 38% with a strut width of 600 µm, a strut height of
LT3 has a low average molecular weight of 83,000, 200 µm, and a pore size of 450 µm, 3D printed using a
which is an easy flow grade PEEK (with a melt index of nozzle temperature of 410°C, build plate heated to
36.4 at a condition of 2.16 kg load in 10 min) that was 100°C, and an ambient temperature of 80°C. The
used to minimise friction between PEEK and syringe’s samples were tested using the Instron 8032 test
wall, and thus reduction of extrusion pressure. Victrex® machine at a strain rate of 10−3 s−1. A 100 kN load cell
PEEK 450G filaments with a diameter of 1.75 mm (type was used and the test data were collected by Instron’s
8110, ZYEX, UK) were used for the experiments using Strainsmart 6200 software. Solid PEEK-OPTIMA LT1
the filament-based device. Victrex® PEEK 450G has a mol- samples (porosity 0%) were also tested to compare
ecular weight of 115,000 which is similar to medical- those with 38% porosity. Three samples were tested for
grade PEEK-OPTIMA LT1 which has been developed to each group for reproducibility. Due to the ductile
meet with FDA requirements. The materials were dried nature of PEEK in compression, samples were deformed
at 150°C for three hours before using in 3D printing to large strains and all strains referenced in this study are
process. true strains (logarithmic strains). Compressive yield
The filament-based device was identified more stable strength was defined as the stress after which the
and efficient than the syringe-based method. A series of initial linear region deviated from linearity; yield strain
experiments were conducted to define the optimum was defined as the strain associated with the compres-
nozzle temperature range in accordance with mass sive yield strength; and modulus of elasticity was calcu-
flow rate. Nozzle temperatures from 350°C to 450°C lated from data representing the slope of the initial
were tested with an extrusion flow rate of 2.2 mg/s linear region (Espalin et al. 2010).
 4 M. VAEZI AND S. YANG

Simplified tensile test bars with rectangular cross injected moulded PEEK parts (unfilled PEEK-OPTIMA pro-
section were also printed using 100% infill rate with cessing guide, Invibio Ltd, UK): first, the samples were
the aim of achieving maximum density. The samples dried at 150°C for three hours; next, the samples were
were printed at size 60 × 4.5 × 3 mm3 in the X direction heated up at 10°C/h to reach a temperature of 200°C
(e.g. build height of 3 mm) with +45°/−45° criss-cross and they were held for four hours at the target tempera-
raster orientation. An Instron 4204 tensile test machine ture; then the samples were cooled down to below 140°C
with 50 kN load cell was used for tensile test at a strain with a rate of 10°C/h and the furnace was switched off
rate of 2 × 10−3s−1 and a room temperature of 23°C. In allowing the samples to cool down to room temperature.
addition, specimens with a lower infill rate of 80% were
printed at size 60 × 11 × 3.5 mm3 and were subjected
to tensile test for investigating the effect of incorporating 3. Results and discussion
porosity into PEEK.
3.1. Optimum condition for 3D printing of PEEK
The three-point flexural test was also conducted
according to the ISO 178 standard procedure using an The critical factors in PEEK extrusion freeforming are
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

Instron 5569 machine to compare bending behaviour high-temperature extrusion head design, and environ-
of the 3D printed PEEK with some other available AM mental heat management to achieve continuous print-
materials and techniques, as follows: ing without nozzle clogging or polymer degradation,
. Acrylate-based clear resin (as received from Formlabs, good layer bonding and minimised part warpage (in par-
USA) printed using a Form 1 stereolithography (SLA) ticular, detachment of first printed layer from build plate
machine. which yields failure of printing). According to the results
. ABS M30 and ABSPlus printed using Stratasys’ Fortus of the experiments, syringe-based extrusion head design
400mc and Dimension 1200es fused deposition mod- does not provide sufficient heat management over PEEK
elling (FDM) systems, respectively. feedstock inside a syringe. It was difficult to control vis-
. Polyamide (PA12) and alumide (aluminium/polyamide cosity of PEEK extrudate, and to avoid thermal degra-
composite called PA12-Al) printed by external 3D dation of the material inside a syringe. In addition, the
printing service company ([Link]) using syringe-based system evinces limitation on the volume
an EOS P730 SLS machine. of the part can be printed. Figure 2 depicts the prelimi-
. Victrex® PEEK 450G printed using the developed fila- nary results using the syringe-based extruder in which
ment-based 3D printing system. the printed three-layer PEEK scaffold detached from
The flexural test specimens were printed at size 10 × the heated build plate of 170°C (which is above glass
4 × 80 mm3 and tested in a similar build direction (build transition temperature of PEEK) during printing
layers were parallel to the loading span in all test speci- process. It should be noted that the printed three-layer
mens). All the compression, flexural, and tensile test scaffold was not uniform due to insufficient heat man-
specimens were annealed at 200°C to achieve a high agement on the PEEK inside the extrusion syringe. As
level of crystallinity using the normal protocol for can be seen in Figure 2(b), PEEK extrudates were too

Figure 2. PEEK extrusion freeforming using the syringe-based extrusion system and heated build plate (170°C), printing process failed
due to the first-layer detachment from build plate: (a) back side and (b) top side.
 VIRTUAL AND PHYSICAL PROTOTYPING 5

viscose in the second and third layers due to over range in accordance with extrusion rate, ambient temp-
heating and could not retain their shape upon depo- erature, and feedstock filament size so that the PEEK fila-
sition (red arrows). ment can retain its shape upon deposition, and there is
Conversely, using the filament-based extrusion no polymer degradation, nozzle clogging, and delamina-
system, it was possible to minimise the heat-affected tion. The melting temperature of PEEK-OPTIMA LT1 is
zone in the PEEK feedstock with the use of an efficient 337°C, however, it is normally processed through injec-
heat sink as illustrated schematically in Figure 1(b). tion moulding at temperatures of 360–400°C with vis-
High temperature of PEEK melt causes a remarkable cosity from ∼77 × 103 to ∼66 × 103 Pa s, respectively
thermal stress on PEEK extrudate upon deposition (Kurtz 2012a). To cover this processing range, extrusion
which results in defects such as delamination (insuffi- temperatures from 350°C to 450°C were tested in this
cient layer bonding) and part warpage. Current available research with a flow rate of 2.2 mg/s using PEEK feed-
material melting extrusion AM systems use either temp- stock filaments with a diameter of 1.75 mm, and an
erature-controlled build chamber or heated build plate ambient temperature of 80°C. PEEK degradation could
to avoid these defects in thermoplastics such as ABS be detected by either colour change or formation of
entrapped voids inside PEEK filaments after deposition
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

and PLA with melt temperature below 250°C. As for

PEEK, the thermal stress effect is much more significant as illustrated in Figure 3. Delamination and filament
and necessary to be managed. Both heated build plate deformation could be detected by visual and optical
and heating lamps (instead of build chamber) were microscope inspection of the specimens.
used to minimise the thermal stress. The use of a According to the results, temperature of 400–430°C
custom-designed build plate heated up to 130°C with was identified as an applicable range so that tempera-
features to hold the first PEEK layer after deposition, tures below 400°C caused either nozzle clogging or
and heating lamps surrounding the build area (ambient delamination and above 430°C resulted in either con-
temperature of 80°C) was identified as a suitable plat- siderable filament deformation or material degradation
form for successful 3D printing of PEEK parts using an over a long time extrusion. This temperature range can
extrusion temperature of 410–430°C. be further narrowed by taking into consideration other
Also, extrusion temperature needs to be set appropri- factors such as dimensional accuracy, mechanical prop-
ately since it can affect the level of crystallinity, and thus erties, and biological performance. Further dimensional
mechanical properties of the final 3D printed PEEK struc- analysis, mechanical and biological experiments are
ture. Furthermore, if the extrusion temperature is too required to define the optimum temperature for printing
high the material may degrade or the extrudate cannot PEEK parts with high accuracy, maximised mechanical
retain its shape upon deposition which results in filament strength with sufficient biocompatibility.
deformation and dimensional inaccuracy. On the other Figure 4(a) depicts some 3D printed parts successfully
hand, if the extrusion temperature is not high enough using a filament-based device with a build layer
the material does not have enough time to get fully
melted and this can cause nozzle clogging. In particular,
nozzle clogging more likely occurs when the filament-
based extrusion head is used since the heat-affected
zone is much less than the syringe-based extrusion
head, and in turn the PEEK filament has less time to
absorb energy and get fully melted. A low extrusion
temperature can also result in delamination as the
material does not have enough energy to make a good
contact fusion bonding with the previous layer. Further-
more, there is correlation between nozzle temperature
and extrusion flow rate in the filament-based extrusion
system. For instance, if a high flow rate is required (for
shortening build time using thicker printing layers) the
material has a shorter time to absorb energy and conse-
quently nozzle clogging occurs. Therefore, a higher
nozzle temperature is required to avoid nozzle clogging
at high flow rates. Figure 3. 3D printed PEEK scaffold using the filament-based
As a consequence of what was discussed earlier, it was extrusion system with entrapped bubbles due to polymer
necessary to define the optimum extrusion temperature degradation.
 6 M. VAEZI AND S. YANG
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

Figure 4. (a) PEEK samples 3D printed in 0.2 mm layer thickness; (b) microscopic image from the surface of the 3D printed part,
showing a good layer bonding, scale bar 200 µm; (c) 3D printed PEEK scaffold with varying pore sizes (red: 150 µm, blue: 300 µm),
showing capability of the process to control pore size within layers, scale bar 200 µm.

thickness of 0.2 mm and a nozzle temperature of 410°C. but might not be optimal for bone cell attachment and
As can be seen, the 3D printed parts have very good proliferation where pore size ranging from 200 to 350
quality with no warpage and delamination due to their µm is normally suggested (Jarman-Smith et al. 2012).
strong layer bonding (Figure 4(b)). The use of extrusion The limitation in minimum pore size makes it difficult
freeforming in this research work permitted excellent to design an implant with desired strength as well. More-
control on pore size and interconnectivity which is over, the very low recycle rate of PEEK powder makes SLS
necessary for bone ingrowth. Conventional techniques, an extremely expensive and non-eco-friendly process. In
such as particulate leaching have poor control on poros- contrast, with the use of extrusion freeforming, a wide
ity, and suffer from limitations such as inconsistency and range of pore sizes is achievable. This makes the tech-
manual intervention. Using the extrusion freeforming nique a versatile approach, enabling formation of
method, porous PEEK implants could be rapidly printed porous PEEK which can be tailored to specific biomecha-
with a much higher control level on external shape and nical requirements of a variety of clinical applications.
porosity with enhanced reproducibility, a key require-
ment in the production of medical devices. Figure 4(c)
3.2. Crystallinity
illustrates the capability of the process to make porous
PEEK structures with controlled pore size down to 150 PEEK’s mechanical properties is influenced by the level of
µm using computer design. In Figure 4(c), PEEK filament crystallinity of the material, and it has erstwhile been
size is 300 µm and there are two different spacings, reported that increasing crystallinity can increase
namely, 150 µm (red arrow) and 300 µm (blue arrow). elastic modulus and yield strength, while conferring
Possibility of varying pore size through the part and con- lower toughness (Chivers and Moore 1994, Kurtz and
trolling of external shape by computer design are the Devine 2007, Jaekel et al. 2011). The level of crystallinity
main advantages of extrusion freeforming process over is affected by thermal history and material cooling rate.
other traditional techniques such as particulate leaching Hence, the level of crystallinity of PEEK is normally con-
and compression moulding. trolled by mould temperature in the injection moulding
Furthermore, extrusion freeforming is a low-cost process (Kurtz 2012b). In addition to mould temperature
process in comparison with the SLS process, while afford- control, injection-moulded part may be subjected to
ing a greater control of pore size. The minimum achiev- further annealing to achieve the optimal level of
able pore size in SLS is limited (currently is ∼400–500 crystallinity.
µm) as pore size gets smaller it gets more difficult to Similarly, the cooling rate needs to be controlled in
remove excess unsintered powders trapped inside extrusion freeforming process so that deposited
pores. This can meet the requirements recommended materials are cooled down slowly to allow creation of
in literatures for bone in-growth and vascularisation crystalline PEEK structure. Skin colour change occurred
 VIRTUAL AND PHYSICAL PROTOTYPING 7
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

Figure 5. (a) A typical 3D printed human bone structure showing surface colour change in different layers due to a different level of
crystallinity; (b)–(e) a 3D printed sample with colour change in the same layers. (b) and (c) Front view with a high level of amorphous
structure in dark brown colour, (d) and (e) side view with a high level of crystallinity in normal PEEK beige colour.

either in different layers (Figure 5(a)) or in the same layers with 38% porosity, respectively. Average compressive
but different regions (Figure 5(b)–(e)) in the 3D printed modulus was determined as 1.82 GPa for the solid
large human bone structures with non-uniform cooling PEEK samples which is very close to the value that has
rate. These colour changes should not be confused been reported in the literatures for a similar compression
with degraded material and can be interpreted as differ- condition (Rae et al. 2007). Both curves show work hard-
ent levels of crystallinity: darker parts have more amor- ening by plastic deformation; after a strain of 0.45 for the
phous structure, whereas the brighter areas have a solid PEEK and after a strain of 0.10 for the sample
higher level of crystallinity (Kurtz 2012b). The brown with 38% porosity. From the stress–strain plots, by incor-
amorphous skins could be eliminated by further poration of 38% porosity into PEEK, the yield strength
annealing. decreases by approx. 71%. It was not possible to directly
compare this value with porous PEEK prepared by other
methods; however, it can be concluded from the test
3.3. Mechanical properties
results that incorporating a high level of porosity into
Figure 6(a) shows microstructure of a typical 3D printed the PEEK necessary for gaining early bone apposition
Victrex PEEK 450G compression test specimen with a can yield considerable reduction of mechanical
strut width of 600 µm, a strut height of 200 µm, and a properties.
pore size of 450 µm. As can be seen, the deposited The 3D printed porous PEEK can be used for either
struts have relatively uniform shape which insures inter- load-bearing or non-load-bearing applications such as
connectivity of the pores suitable for cell attachment and cranial maxillo-facial plate. Solid PEEK has matching com-
in-growth. Figure 6(b) depicts compressive response of pressive properties with cortical bone which is in the
the 3D printed closed-cell porous Victrex PEEK 450G range of 106–215 MPa (Lawson and Czernuszka 1998),
samples with 38% porosity and solid PEEK-OPTIMA LT1 however, incorporation of a high level of porosity using
samples (porosity 0%) compressed at a stain rate of extrusion-based 3D printing needs to be implemented
10−3 s−1. Yield strength of 102.38 MPa and 29.34 MPa with care. This can limit applications of these porous
and yield strain of 0.056 and 0.044 were determined PEEK structures as load-bearing bone implants since
through the curves for solid samples and the sample materials without sufficient strength are considered
 8 M. VAEZI AND S. YANG
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

Figure 6. (a) SEM image of the compression test specimen with 38% porosity; (b) compressive stress–strain plots for PEEK-OPTIMA and
Victrex 450G with 0% and 38% porosity, respectively. Strain rate of 10−3 s−1 at 25°C.

risky to use as they increase chance of baleful failure of 9085 and injection-moulded PEEK. The printed sample
implants. with 14% porosity had ultimate tensile strength (UTS)
The 3D printed PEEK tensile test specimens were at break of 75.06 MPa which is approx. 33% less that
investigated through macro/microscopic observations injection-moulded PEEK that is 113 MPa at the same
and porosity measurement. The samples printed with strain rate (Rae et al. 2007). Conversely, the 3D printed
100% and 80% infill rates had average porosity of 14% PEEK has remarkably higher UTS than Ultem 9085,
and 31%, respectively. The 14% porosity in samples although it has less elongation at break. It should be
with 100% raster infill rate is mainly due to the formation noted that random formation of air gaps and micro-
of air gap between deposited filaments within layers. bubbles in the PEEK test specimens and inconsistency
Different types of air gap are normally produced in extru- in solidification of the material upon deposition had
sion freeformed parts. The air gaps can be formed led to relatively inconsistent test results for UTS with a
between layers due to inherent limitation of this standard deviation of 4.1 MPa.
process in filament’s geometry. Also, the filament’s geo- The UTS determined for the extrusion freeformed
metry limits materials filling in each layer so that there PEEK is within the range reported for cortical bone that
are undesirable gaps between layer’s perimeter and is 50–150MPa (Abu Bakar et al. 2003b). The printed
infill filaments. These air gaps tend to be formed more
in the first printing layers where the part bonds to the
base material. The other types of air gaps are those pro-
duced between adjacent infill filaments caused by either
variation in feedstock filament diameter or temporary
change in proportion between XY table speed and extru-
sion rate. In addition to air gaps, micro-air bubbles and
voids can be entrapped inside extrudate in order to
high processing temperature, air bubble inside feedstock
filament, and moisture. All these defects induce porosity
into the extrusion freeformed PEEK parts and sub-
sequently reduction of mechanical properties.
Figure 7 represents the tensile stress–strain curve of
the 3D printed samples using 420°C nozzle temperature
in comparison with injection-moulded PEEK and 3D
printed polyetherimide with the trade name Ultem
9085 which is the highest strength material currently
Figure 7. Tensile stress–strain curve of different materials includ-
available for use in commercial Stratasys’ FDM machines. ing injection-moulded PEEK (Rae et al. 2007), 3D printed Ultem
The 3D printed PEEK samples with 14% and 31% porosity 9085 (Bagsik et al. 2010), and 3Dprinted PEEK with 14% and
had a smaller strain to failure in comparison with Ultem 31% porosity.
 VIRTUAL AND PHYSICAL PROTOTYPING 9

PEEK with 31% porosity had a UTS of 49.22 MPa and a Figure 9 depicts fracture surface of a typical extrusion
strain to failure of 0.31. Hence, incorporation of 17% por- freeformed PEEK sample with +45°/−45° raster orien-
osity (increasing porosity from 14% to 31%) into 3D tation printed using a high extrusion temperature of
printed samples resulted in approx. 35% and 13% 430°C. As shown in Figure 9(b), no welding line
reduction in UTS and strain to failure, respectively. It between infill and perimeter filaments is evident that
should also be noted that the UTS of the printed proves excellent filament-to-filament bonding in each
porous PEEK samples are comparable to those reported layer. Welding line between layers (the yellow arrow)
for PEEK/hydroxyapatite (HA) composites with a similar could be realised, yet there was sufficient interlayer
HA content. For instance, the UTS of 78.10–80 MPa bonding. This strong filament-to-filament bonding in
with a strain to failure of 0.15 has been reported for each layer influences failure behaviour of the extrusion
annealed PEEK/HA composites with 15% HA (Abu Bakar freeformed PEEK. The PEEK specimens printed in low
et al. 2003a, Converse et al. 2007) that is comparable to temperature, having evident welding line between infill
the 3D printed samples with 14% porosity. filaments, exhibit multiple failures of infill filaments in
The noticeable difference in the stress–strain curve both shear and tension. Failure occurs via pulling and
probabilistic fracturing of individual filaments whereby
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

of the 3D printed and injection-moulded PEEK parts

can be attributed to their different fracture mechanisms. the filaments detached at +45° and −45° angle relative
Figure 8 depicts failure mode of a typical specimen with to the tensile load. In contrast, the samples processed
+45°/−45° criss-cross raster orientation with air gaps in in high temperatures, with ideally no clear welding
the first layer (Figure 8(a)) discussed earlier. The extru- lines, can be analysed as a laminated material. Hence,
sion freeformed parts consist of many individual fila- the materials fracture at 90° angle relative to the
ments that each one can be treated as a single tensile load as can be seen in Figure 9(a) since no
component for fracture analysis. Thus, it is important shear stress on infill filaments is applied.
to form and deposit the PEEK filaments with minimum The air gaps influence the strength of the extrusion
entrapped micro-air bubbles that are considered as freeformed samples by locally decreasing physical
the main reason of decrease in filaments’ strength. In cross-sectional area of material specimens and micro-
addition to the strength of individual filaments, a crack formation under tension (Ziemian et al. 2012).
good filament-to-filament bonding is critical in extru- A typical micro-crack initiated from the existing air gap
sion freeforming to have maximum structural integrity can be seen in Figure 9(c). The micro-crack was formed
and energy absorption under load. The tensile strength and propagated in the direction where two filaments
of extrusion freeformed parts is influenced much more get welded upon deposition. The failure mechanism of
by the filament-to-filament bonding rather than the the specimen can be explained with eventual rupturing
strength of the filaments themselves (Sung-Hoon et al. of perimeter filaments (left area in Figure 9(a)) and
2002, Ziemian et al. 2012). Using a high processing the weak region in the incipient build layers (down
temperature is an asset to achieve sufficient filament- area in Figure 9(a)). The air gaps close to perimeter fila-
to-filament bonding. This can also minimise the for- ments contributed in initiation of micro-cracks and con-
mation of air gap between infill filaments since the sequently propagation and rupture of more infill
PEEK extrudate has lower viscosity and thus spread filaments from left to right (i.e. centre of the specimen).
more in the lateral direction upon deposition, although Meanwhile, more material has ruptured from down
it may affect dimensional accuracy. toward the centre of the specimen. The centre of the

Figure 8. Fracture mode of a typical tensile specimen printed with +45°/−45° criss-cross raster orientation: (a) bottom view showing air
gaps in the first build layer, (b) side view with a narrow showing crack propagation direction, and (c) top view.
 10 M. VAEZI AND S. YANG
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

Figure 9. (a) Fracture surface of PEEK tensile test sample showing excellent infill filaments bonding in each layer; (b) magnified view of
the green solid inset box; (c) magnified view of the red dash inset box; and (d) magnified view of the blue dash dot inset box.

specimen (the blue box in Figure 9(a)) is the last region As can be seen from the error bars (determined
has torn due to its strong material bonding and less from 10 records for each material), there was quite a
amount of air gaps. As seen in Figure 9(d), this region spread in both modulus and the maximum strength
possesses an excellent filament-to-filament and inter- of the samples printed using SLA and SLS processes
layer bonding, although it possess micro-bubbles than those printed using FDM. The variations in SLA
inside filaments. samples might be due to varying degrees of post-
Figure 10 represents the results of three-point flexural curing for the specimens, since inadequate control of
test on the 3D printed PEEK and other AM materials and the post-curing process can lead to highly varying
processes. The 3D printed PEEK had the highest flexural properties. As for SLS samples, the alumide (PA12-Al)
and failure strength of 132.37 MPa and 117.09 MPa, showed a remarkable variation in both strength and
respectively (approx. twice the samples printed from modulus than the polyamide samples. This can be
ABS) and flexural modulus of 2.43 GPa. The difference explained as a wide range of physical properties of alu-
between the failure strength and the maximum strength minium and polyamide particles (e.g. particle size,
is how much plastic deformation the specimen can shape, and density) negates efficient and consistent
sustain. If there is less difference, that means the speci- mixing, and thus non-uniform distribution of these
mens snapped cleanly, without any plastic deformation. materials in each layer results in variation in flexural
In a same way, the specimen can withstand more stress properties of different specimens. The variation in
before it breaks if the gap between the two values is strength and modulus of the 3D printed PEEK
bigger. samples can also be described by inconsistency in
 VIRTUAL AND PHYSICAL PROTOTYPING 11
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

Figure 10. Comparison graph of (a) the maximum and failure strength and (b) flexural moduli.

solidification of the material upon deposition. As the environment surrounding the part could affect the
ambient temperature was controlled using lamps sur- level of crystallinity in the 3D printed PEEK structure.
rounding printed part there might be less control in Nozzle temperature within the range of 400–430°C,
the cooling rate as discussed earlier. The random for- ambient temperature of 80°C, and heated build plate
mation of the air gaps and micro-bubbles could be up to 130°C was identified as applicable conditions for
an alternative source for variation. The samples extrusion rate of 2.2 mg/s. The findings of this research
printed from ABS using commercial FDM systems had work suggest that 3D printed porous PEEK have promis-
the least variation as they were printed in a heat-con- ing compressive properties with potential for both load-
trolled chamber. bearing and non-load-bearing application. Compressive
Quality control standards stipulating the testing of yield strength of 102.38 MPa and 29.34MPa and yield
medical device biocompatibility require extensive inves- strain of 0.056 and 0.044 were determined for the solid
tigation to confirm the true readiness for clinical appli- samples with 0% and 38% porosity, respectively. The
cation. As a pre-cursor to future development, the goal 3D printed samples with 100% infill rate had 14% poros-
of this study was set out to show preliminary evidence ity and UTS of 75.06MPa that is 33% reduction in injec-
supporting feasibility of successful 3D printing of tion-moulded PEEK. The printed PEEK samples had a
medical-grade PEEK using extrusion freeforming and flexural modulus of 2.43 GPa and a flexural strength of
the authors acknowledge the depth of work yet to be 132.37 MPa that was significantly higher than flexural
completed to exhaustively document biocompatibility strength of the specimens printed from acrylate-based
of these novel materials. resin, ABS, polyamide, and alumide using other AM
techniques.

4. Conclusions
Acknowledgements
High-temperature extrusion freeforming of PEEK for bio-
medical applications is reported in this paper. Heat man- The authors are grateful to Invibio Ltd., and Faculty of Engineer-
ing and the Environments, University of Southampton for finan-
agement during print process was identified as an
cial support of this research project. The authors acknowledge
important parameter to assure a good PEEK/substrate the TSRL mechanical test lab at the University of Southampton,
and PEEK interlayer bonding, and to avoid warpage and Mr. Gjermund Holm for conducting flexural tests on the
and delamination. In addition, heat distribution in the printed samples.
 12 M. VAEZI AND S. YANG

Disclosure statement Lovald, S. and Kurtz, S.M., 2012. Chapter 15 – applications of

polyetheretherketone in trauma, arthroscopy, and cranial
No potential conflict of interest was reported by the authors. defect repair. In: S.M. Kurtz, ed. PEEK biomaterials handbook.
Oxford: William Andrew Publishing, 243–260.
Rae, P.J., Brown, E.N., and Orler, E.B., 2007. The mechanical
properties of poly(ether-ether-ketone) (PEEK) with emphasis
References on the large compressive strain response. Polymer, 48,
598–615.
Abu Bakar, M.S., Cheang, P., and Khor, K.A., 2003a. Mechanical Rao, P.J., et al., 2014. Spine interbody implants: Material selec-
properties of injection molded hydroxyapatite-polyether- tion and modification, functionalization and bioactivation
etherketone biocomposites. Composites Science and of surfaces to improve osseointegration. Orthopaedic
Technology, 63, 421–425. Surgery, 6, 81–89.
Abu Bakar, M.S., Cheang, P., and Khor, K.A., 2003b. Tensile prop- Roeder, R.K. and Conrad, T.L., 2012. Chapter 11 – bioactive poly-
erties and microstructural analysis of spheroidized hydroxya- aryletherketone composites. In: S.M. Kurtz, ed. PEEK bioma-
patite–poly (etheretherketone) biocomposites. Materials terials handbook. Oxford: William Andrew Publishing, 163–
Science and Engineering: A, 345, 55–63. 179.
Bagsik, A., Schöppner, V., and Klemp, E., 2010. FDM part quality Schmidt, M., Pohle, D., and Rechtenwald, T., 2007. Selective
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015

manufactured with Ultem*9085. 14th International scientific Laser Sintering of PEEK. CIRP Annals – Manufacturing
conference, 15–17 September, Halle. Technology, 56, 205–208.
Bhuthalingam, R., et al., 2015. A novel 3D printing method for Siddiq, A.R. and Kennedy, A.R., 2015. Porous poly-ether ether
cell alignment and differentiation. International Journal of ketone (PEEK) manufactured by a novel powder route
Bioprinting, 1, 57–65. using near-spherical salt bead porogens: Characterisation
Chivers, R.A. and Moore, D.R., 1994. The effect of molecular and mechanical properties. Materials Science and
weight and crystallinity on the mechanical properties of Engineering: C, 47, 180–188.
injection moulded poly(aryl-ether-ether-ketone) resin. Sobieraj, M.C. and Rimnac, C.M., 2012. Chapter 5 – fracture,
Polymer, 35, 110–116. fatigue, and notch behavior of PEEK. In: S.M. Kurtz, ed. PEEK
Converse, G.L., Yue, W., and Roeder, R.K., 2007. Processing and biomaterials handbook. Oxford: William Andrew Publishing,
tensile properties of hydroxyapatite-whisker-reinforced poly- 61–73.
etheretherketone. Biomaterials, 28, 927–935. Sun, J., et al., 2015. 3D food printing – An innovative way of
Edwards, S., et al., 2010. Polyetheretherketone multifilament and mass customization in food fabrication. International
monofilament woven tissue engineering scaffolds. transactions Journal of Bioprinting, 1, 27–38.
society for biomaterials. Seattle, Washington, DC: Society for Sung-Hoon, A., et al., 2002. Anisotropic material properties of
Biomaterials. fused deposition modeling ABS. Rapid Prototyping Journal,
Espalin, D., et al., 2010. Fused deposition modeling of patient- 8, 248–257.
specific polymethylmethacrylate implants. Rapid Tan, K.H., et al., 2003. Scaffold development using selective laser
Prototyping Journal, 16, 164–173. sintering of polyetheretherketone–hydroxyapatite biocom-
Jaekel, D.J., Macdonald, D.W., and Kurtz, S.M., 2011. posite blends. Biomaterials, 24, 3115–3123.
Characterization of PEEK biomaterials using the small Tan, K.H., et al., 2005. Fabrication and characterization of three-
punch test. Journal of the Mechanical Behavior of dimensional poly(ether-ether-ketone)/-hydroxyapatite bio-
Biomedical Materials, 4, 1275–1282. composite scaffolds using laser sintering. Proceedings of the
Jarman-Smith, M., et al., 2012. Chapter 12 – porosity in polyar- Institution of Mechanical Engineers, Part H: Journal of
yletheretherketone. In: S.M. Kurtz, ed. PEEK biomaterials Engineering in Medicine, 219, 183–194.
handbook. Oxford: William Andrew Publishing, 181–199. Toth, J.M., 2012. Chapter 7 – biocompatibility of polyarylether-
Kurtz, S.M., 2012a. Chapter 1 – an overview of PEEK biomater- etherketone polymers. In: S.M. Kurtz, ed. PEEK biomaterials
ials. In: S.M. Kurtz, ed. PEEK biomaterials handbook. Oxford: handbook. Oxford: William Andrew Publishing, 81–92.
William Andrew Publishing, 1–7. Toth, J.M., et al., 2006. Polyetheretherketone as a biomaterial for
Kurtz, S.M., 2012b. Chapter 2 – synthesis and processing of PEEK spinal applications. Biomaterials, 27, 324–334.
for surgical implants. In: S.M. Kurtz, ed. PEEK biomaterials Vaezi, M., et al., 2013a. Multiple material additive manufacturing
handbook. Oxford: William Andrew Publishing, 9–22. – Part 1: a review. Virtual and Physical Prototyping, 8, 19–50.
Kurtz, S.M., 2012c. Chapter 6 – chemical and radiation stability Vaezi, M., Seitz, H., and Yang, S., 2013b. A review on 3D micro-addi-
of PEEK. In: S.M. Kurtz, ed. PEEK biomaterials handbook. tive manufacturing technologies. The International Journal of
Oxford: William Andrew Publishing, 75–79. Advanced Manufacturing Technology, 67, 1721–1754.
Kurtz, S.M. and Devine, J.N., 2007. PEEK biomaterials in trauma, Vaezi, M. and Yang, S., 2014. Chapter 2 – Freeform fabrication of
orthopedic, and spinal implants. Biomaterials, 28, 4845–4869. nanobiomaterials using 3D printing. In: R. Narayan, ed. Rapid
Kurtz, S.M., Day, J., and Ong, K., 2012. Chapter 14 – isoelastic prototyping of biomaterials. Cambridge: Woodhead
polyaryletheretherketone implants for total joint replace- Publishing, 16–74.
ment. In: S.M. Kurtz, ed. PEEK biomaterials handbook. Vaezi, M. and Yang, S.F., 2015. A novel bioactive PEEK/HA com-
Oxford: William Andrew Publishing, 221–242. posite with controlled 3D interconnected HA network.
Lawson, A.C. and Czernuszka, J.T., 1998. Collagen-calcium phos- International Journal of Bioprinting, 1, 66–76.
phate composites. Proceedings of the Institution of Valentan, B., et al., 2013. Processing poly(Ether Etherketone) On
Mechanical Engineers, Part H: Journal of Engineering in A 3d printer for thermoplastic modelling. Materials and
Medicine, 212, 413–425. Technology, 47, 715–721.
 VIRTUAL AND PHYSICAL PROTOTYPING 13

Von Wilmowsky, C., et al., 2008. Effects of bioactive glass and β- Yang, H.Y., et al., 2008b. Sintering behaviour of calcium phos-
TCP containing three-dimensional laser sintered polyether- phate filaments for use as hard tissue scaffolds. Journal of
etherketone composites on osteoblasts in vitro. Journal of the European Ceramic Society, 28, 159–167.
Biomedical Materials Research Part A, 87A, 896–902. Yang, S.F., et al., 2008c. Rapid prototyping of ceramic lattices for
Wang, L., et al., 2014. Polyetheretherketone/nano-fluorohydrox- hard tissue scaffolds. Materials & Design, 29, 1802–1809.
yapatite composite with antimicrobial activity and osseointe- Zhao, Y., et al., 2013. Cytocompatibility, osseointegration, and
gration properties. Biomaterials, 35, 6758–6775. bioactivity of three-dimensional porous and nanostructured
Wang, S., Lee, J.M., and Yeong, W.Y., 2015. Smart hydrogels for network on polyetheretherketone. Biomaterials, 34, 9264–
3D bioprinting. International Journal of Bioprinting, 1, 3–14. 9277.
Yang, H.Y., et al., 2008a. Dissolution characteristics of extrusion Ziemian, C., Sharma, M., and Ziemian, S., 2012. Chapter 7 –
freeformed hydroxyapatite-tricalcium phosphate scaffolds. Anisotropic mechanical properties of ABS parts fabricated
Journal of Materials Science: Materials in Medicine, 19, 3345– by fused deposition modelling. In: M. Gokcek, ed.
3353. Mechanical engineering. Rijeka: InTech, 159–180.
Downloaded by [University of Southampton Highfield] at 13:18 17 October 2015