- Email: [email protected]
Synthesis, microstructure, and mechanical behaviour of a unique porous PHBV scaffold manufactured using selective laser sintering
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
Contents lists available at ScienceDirect
Journal of the Mechanical Behavior of Biomedical Materials journal homepage: www.elsevier.com/locate/jmbbm
Synthesis, microstructure, and mechanical behaviour of a unique porous PHBV scaffold manufactured using selective laser sintering
T
Sven H. Diermanna, Mingyuan Lua, Yitian Zhaoa, Luigi-Jules Vandib, Matthew Darguschc, ⁎ Han Huanga, a
School of Mechanical and Mining Engineering, The University of Queensland, St Lucia, QLD 4072, Australia School of Chemical Engineering, The University of Queensland, QLD 4072, Australia c Centre for Advanced Materials Processing and Manufacturing (AMPAM), School of Mechanical and Mining Engineering, The University of Queensland, QLD 4072, Australia b
A R T I C LE I N FO
A B S T R A C T
Keywords: Selective laser sintering Poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) Mechanical properties Scaffold Bone tissue engineering
Selective Laser Sintering (SLS) is a promising technique for manufacturing bio-polymer scaffolds used in bone tissue engineering applications. Conventional scaffolds made using SLS have complex engineered architectures to introduce adequate porosity and pore interconnectivity. This study presents an alternative approach to manufacture scaffolds via SLS without using pre-designed architectures. In this work, a SLS process was developed for fabricating interconnected porous biodegradable poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) scaffolds with large surface areas and relative porosities of up to 80%. These characteristics provide great potential to enhance cell attachment inside the scaffolds. The scaffold microstructure was dependent on the laser energy density (LED) during the SLS process. An increase in LED led to scaffolds with higher relative densities, stronger inter-layer connections, and a reduced quantity of residual powder trapped inside the pores. An increase in relative density from 20.3% to 41.1% resulted in a higher maximum compressive modulus and strength of 36.4 MPa and 6.7 MPa, respectively.
1. Introduction Repair of complex bone fractures, including scenarios involving large bone loss, continues to pose great challenges for clinical management. Autologous and allogeneic bone grafting are commonly applied surgical procedures which come with inherent limitations and risks, including graft availability, harvest site pain, morbidity, rejection reactions, and risk of disease transmission (Dimitriou et al., 2011). Scaffold-assisted tissue engineering (TE) is regarded as a promising bone regeneration technique to replace the current bone grafting methods (Ramay and Zhang, 2004; Dean et al., 2003; Chua et al., 2004; Wu et al., 2013). In scaffold-assisted bone TE, a porous scaffold is implanted to provide support for cell migration, colonization, growth, and differentiation to promote tissue generation at the fracture site. Therefore, scaffolds are usually made of biocompatible and biodegradable materials and have an interconnected porous network (Cao and Hench, 1996; O'Brien, 2011; Yang et al., 2011; Hutmacher et al., 2014; Dutta et al., 2017). The fabrication of porous TE scaffolds have employed a variety of manufacturing techniques, such as thermally induced phase separation
⁎
Corresponding author. E-mail address: [email protected] (H. Huang).
https://doi.org/10.1016/j.jmbbm.2018.05.007 Received 13 March 2018; Received in revised form 21 April 2018; Accepted 2 May 2018 1751-6161/ © 2018 Elsevier Ltd. All rights reserved.
(TIPS) (Nam and Park, 1999; Jack et al., 2009), solvent casting and particulate leaching (Thomson et al., 1996), electrospinning (Sombatmankhong et al., 2007), and sintering (Wu et al., 2007). While these methods have their unique benefits, existing technical issues including manual intervention, inconsistency and inflexibility in processing, and exclusion of toxic solvents or porogen still need to be addressed. Most significantly, customised scaffold geometries often cannot be achieved due to the inherent restrictions of the conventional manufacturing processes. Over the last decade, additive manufacturing (AM) has emerged as a promising method for the synthesis of TE scaffolds and has opened a new field of intriguing bioengineering possibilities. AM techniques are ideally suited for biomedical applications due to their ability to make customised parts with complex shapes (Leong et al., 2003; Mazzoli, 2013; Chia and Wu, 2015; Chua et al., 2017; Caulfield et al., 2007). Among various AM techniques, selective laser sintering (SLS) was extensively used to make bone scaffolds due to its capability to make porous structures using a large variety of materials in powder form. SLS does not require support structures during the fabrication process and the existing porosity between particles can be preserved (Chia and Wu, 2015). In previous studies, SLS manufactured
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
scaffolds typically used a computer designed three-dimensional (3D) architecture to achieve the desired porosity and interconnectivity (Hutmacher et al., 2014; Duan et al., 2010; Eosoly et al., 2010; Shirazi et al., 2014; Du et al., 2017). However, the use of a pore architecture is potentially accompanied with drawbacks. First, the manufacture of a scaffold structure with small features via SLS has limited resolution due to relatively large laser spot sizes (Lohfeld et al., 2010). Second, scaffolds made using SLS with a pre-designed pore architecture have lower specific surface areas than scaffolds manufactured using alternative techniques such as e.g. TIPS, (Nam and Park, 1999, Jack et al., 2009), solvent casting and particulate leaching (Thomson et al., 1996), or electrospinning (Sombatmankhong et al., 2007). Requirements on biocompatibility, bioactivity and biodegradability have limited the choice of materials for scaffolds used in bone TE. Till now, ceramics (Shuai et al., 2017; No et al., 2017) and biodegradable polymers are frequently used for TE strategies in AM. For example, poly (ϵ) caprolactone (PCL) (Williams et al., 2005; Partee et al., 2005; Lohfeld et al., 2010), poly-L-lactide (PLLA) (Duan et al., 2010), poly-D-lactide (PDLA) (Bukharova et al., 2010), polyhydroxybutyrate (PHB) (Pereira et al., 2012), and poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) (Duan et al., 2010) are representative polymers that were used to manufacture scaffolds via SLS. Among these materials, PHBV offers favourable surface chemistry for cell attachment and proliferation (Kumarasuriyar et al., 2005), while its degradation by-products are biocompatible and can be metabolized. Previous in vitro and in vivo biocompatibility investigations of PHBV in TE applications have confirmed its inert nature (Sultana and Wang, 2007; Sultana and Khan, 2012). PHBV elicits minimal inflammatory responses as it degrades into products, normally found as constituents of human blood (Sultana and Wang, 2007; Sultana and Khan, 2012). In addition, PHBV degrades over a long period of time which allows the polymeric scaffold to maintain its mechanical integrity until an adequate portion of new bone grows into the scaffold (Sultana and Khan, 2012; Gogolewski et al., 1993). The objective of this study is to develop a SLS process to manufacture PHBV scaffolds with interconnected porous networks without engineered architectures, addressing the issues of process resolution and scaffold surface area. Using the developed SLS process, the manufactured scaffolds would have much smaller strut sizes and larger surface areas than those with pre-designed architectures, which is in favour of enhancing cell attachment inside the scaffold (O'Brien et al., 2005). In this study, the effect of SLS processing parameters on the microstructural and mechanical properties was systematically investigated, with a particular emphasis placed on the scaffolds' compressive deformation behaviour in both the normal and lateral direction.
Table 1 SLS processing conditions. Test
Laser power /P (W)
Scan spacing/SS (mm)
Laser energy density/ LED (J/ mm3)
T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11
3 3 3 4 4 4 5 5 5 6 6
0.2 0.15 0.1 0.2 0.15 0.1 0.2 0.15 0.1 0.2 0.15
0.030 0.040 0.060 0.040 0.053 0.080 0.050 0.067 0.100 0.060 0.080
defined as the metastable area between two phases. 2.2. Manufacture of scaffolds Porous scaffolds were fabricated using a SLS system (DTM Sinterstation 2500 Plus, 3D Systems, USA) equipped with a CO2 laser with an inherent wavelength of 10.6 μm and a spot size of 420 μm. To minimize the usage of powder in building small-scale specimens, a powder feeding system was designed and adapted into the SLS system. Both the volume of the powder feeding cartridges and the build area were substantially reduced. The area of the feed bed and build area were reduced to 125 × 105 mm2 and 85 × 85 mm2 respectively. Cubes (L × W × H=10 × 10 × 10 mm3) and cylinders (D = 6 mm, H = 10 mm) were manufactured in eleven tests, using different processing conditions, as shown in Table 1. The laser power (P) and scan spacing (ss ) varied from 3 to 7 W and 0.1–0.2 mm respectively. The laser scan speed was fixed at 5000 mm/s. The powder was spread by a roller system with a traverse speed of 177.8 mm/s. The layer thickness of each powder layer was 0.1 mm. The SLS chamber was filled with nitrogen gas to maintain an inert environment with an oxygen content below 5%. The processing temperature was 125 °C. Loose powder particles were removed from the manufactured scaffolds after the SLS process using a fine brush. The altered SLS processing parameters were characterized using laser energy density (LED). The LED (J/mm3) describes the intensity of the laser irradiation and is the quotient of P divided by ss , scan speed (v), and layer thickness (l) (Shahzad et al., 2013; Hagedorn, 2013):
LED =
P ss vl
(1)
The LED used in each test was calculated using Eq. (1) and is also shown in Table 1.
2. Experimental details 2.1. PHBV powder
2.3. Characterization Poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) powder (ENMATTM , Y1000) with a 3-hydroxyvalerate (HV) content of 2–5% and average molecular weight of 1,187,000 g/mol, was used (supplied by TianAn Biologic Materials Ltd, Ningbo, China). The as-received powder consists of fine spherical particles that agglomerate to form large irregular clusters. The powder was sieved to below 250 μm using a vibratory sieve shaker (Retsch, Germany). The particle size distribution (PSD) of the powder was measured by means of laser diffraction analysis of a water-based dispersion using a Mastersizer Hydro (Malvern Instruments, England). The cumulative particle size distribution values D10 , D50 , D90 , based on a volume distribution, were 11.3 μm, 73.7 μm, and 188 μm, respectively. Differential Scanning Calorimetry (DSC) analysis on the as-received PHBV powder revealed a melting temperature Tm and crystallisation temperature Tc of 172 °C and 114 °C respectively. The onset temperature of Tm and Tc of 161 °C and 120 °C was measured respectively, resulting in a SLS processing window of 41 °C,
2.3.1. Microstructural characterization The microstructure and pore architecture of the fabricated scaffolds were characterized using scanning electron microscopy (SEM, FEI, Orgegon, USA) and 3D measuring laser microscope (OM, Olympus LEXT OLS4100, Japan). For SEM observation, the scaffolds were crosssectioned in different directions using a scalpel, and then coated with a 20 nm conductive film. For OM observation, the PHBV scaffolds were mounted in epoxy resin, sectioned, ground, and polished. The entire area of the polished scaffold was imaged to provide a holistic view of the sectioned plane. A minimum of 67 OM images were taken for each specimen and then stitched together to form a large area image mosaic. The stitched images of cross-sections along the x-y (normal) and y-z (lateral) plane were used to quantify the areal size of the skeleton (material, lm ) and pores (lp ) using a line inter-section method, which is detailed in the supplementary material S1. 152
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
plane and the x-z plane. Three typical microstructural features were identified and marked in the images, which include pores, trapped powder, and fused material in the forms of islands and bridges. Fused islands were connected by bridges which were both made of solidified PHBV. Note that the black regions are air bubbles in the resin that failed to escape from inside the porous structure during the vacuum impregnation process. They were therefore counted as pores. Evidently, the use of different SLS process parameters resulted in variations in the microstructure. Small fused islands connected by slender bridges are observed in scaffolds produced using low LED, as shown in Fig. 1(a). A considerable amount of unfused powder particles were trapped inside the pores. The use of higher LEDs led to a significant reduction in the quantity of trapped powder and an increase of fused material in the forms of larger islands and thicker bridges (Figs. 1(b, c, e, f)). Scaffolds made using low LED had partially fused islands and bridges which were much more prevalent in the lateral direction. These partially fused features were likely due to insufficient energy input for completely fusing the deposited powder layer in build direction. The use of high LED resulted in fully-fused bridges and islands in the x-z plane. Fig. 2 shows representative cross-sectional SEM images in the x-y and x-z plane of scaffolds manufactured using LEDs of 0.03, 0.067, and 0.10 J/mm3, respectively. Figs. 2(a-b) show a cross-section of a scaffold manufactured using a LED of 0.03 J/mm3 in the x-y plane. The figures present a well connected porous network that consists of numerous islands connected by thin bridges and large amounts of unfused powder trapped inside the pores. In the x-z plane, the porous network was not as well connected. A density gradient of fused material was observed within each layer deposited during the SLS process and will be further referred to as the intra-layer. These intra-layers were poorly fused to each other by the laser during SLS process. Thus, resulted in a small amount of inter-layer connections. These inter-layer connections are the connections created between the deposited powder layers in the SLS process (Fig. 2(c)). Figs. 2(d–e) show a cross-section of a scaffold manufactured using a LED of 0.067 J/mm3 in the x-y plane. The figures present a well connected porous network with moderate-sized islands and bridges, and a reduced amount of unfused powder. The cross-section in the x-z plane shown in Fig. 2(f) presents the porous reticulation in the x-z plane that consists of fused material stacks with small amounts of loose powder
2.3.2. Relative density estimation Weights and dimensions of the scaffolds were measured to calculate their densities. Six replicates were analysed for each SLS process condition. The relative density value of a scaffolds was calculated as
ρ* = ρscaffold / ρsolid ,
(2) 3
where ρsolid is the PHBV solid density of 1.25 g/cm . 2.3.3. Mechanical testing Compression tests were carried out at room temperature on a 5543 Instron universal testing machine using a 2 kN load cell. Uniaxial compressive load was applied to the scaffold specimens in two directions. Scaffolds that were loaded perpendicular to the build direction (see supplementary material S3 for schematic of planes and directions) will be referred to as compression in the normal direction (subscript symbol: n). Scaffolds which were loaded parallel to the build direction (y-z plane) will be referred to as compression in the lateral direction (subscript symbol: l) . Cylindrical shaped scaffolds were used for testing in the normal direction (D = 7 mm, H = 10 mm), and cubes (equal side length L = 10 mm) were used to investigate compression in the lateral direction. For each test condition, six replicates were loaded at 5 mm/ min until failure occurred. The elastic modulus E (MPa) for compression was calculated from the initial linear regions of the σ − ε curves. The critical load when the scaffold lost its load bearing capacity at the first inflection point on the σ − ε curve is denoted as compressive strength in the normal direction (σn ). Further densification resulted in higher compressive strengths, but this data was disregarded as the scaffolds lost their original shape. Scaffolds were tested in the lateral direction until a strain level of 70% was reached. The 1% offset (Zein et al., 2002) stress is deemed as the compressive strength in the lateral direction (σl ). A noise cancelling filter was applied with a moving averaging algorithm of seven data points for the illustration of the σ − ε curve. 3. Results 3.1. Microstructural characterization Fig. 1 shows the optical images of scaffold microstructures, produced using different process conditions, cross-sectioned along the x-y
Fig. 1. Optical images along the x-y plane (a-c) and x-z-plane (d-f) of selected scaffolds. T1 (a & d); T8 (b & e); T9 (c & f). 153
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
Fig. 2. SEM images of selected scaffolds along the x-y and x-z cross-sectional plane. T1 (a-c); T8 (d-f); T9 (g-i)).
inside the pores. Layers were seen connected by short but sizable bridges and some large islands from adjacent layers were connected. The cross-section of a scaffold fabricated using a LED of 0.10 J/mm3 in x-y plane contains a well connected porous network that consists of islands connected by large bridges and very little amounts of unfused powder, as it can be seen in Figs. 2(g–h). Fig. 2(i) shows a cross-section in the x-z plane. The amount of trapped powder was significantly reduced and a large number of inter-layer connections was found. These appeared in the form of bridges and island to island connections from adjacent layers. The SEM observation demonstrated that the scaffolds have characteristic microstructures in the form of islands, bridges, and pores which were dependent on the LED used. The use of higher LEDs lead to an increase in the volume and size of islands and bridges, better connections between layers, and less powder trapped inside the scaffold. All scaffolds appeared to be highly porous and interconnected. To confirm the pore interconnectivity of the manufactured scaffolds, a simple water uptake test was conducted. In this test, a scaffold was put into a petri dish with blue coloured water, as shown in Fig. 3(a). The scaffolds were submerged in the water to approximately 10% of their height to test whether the water can be absorbed and travel to non-submerged regions through capillary action. As shown in Fig. 3(b), after 5 to 10 minutes water was absorbed and transmitted from the bottom to the top of all tested scaffolds. This clearly indicates that the pores in the scaffolds are interconnected and the scaffolds have a good water uptake. The areal size of the skeleton (lm ) and pores (lp ) were measured using the line inter-section method. Fig. 4 presents four contour plots that illustrate the dependence of lm and lp on the of LED viewed in both the normal and lateral direction in detail. It appeared that the lm viewed in the normal (Fig. 4(a)) and lateral direction (Fig. 4(b)) increased as the relative frequency increased when using higher LED. The relative
Fig. 3. Water uptake test to investigate interconnectivity. Scaffold partially submerged in water (a); Water was absorbed and transmitted from bottom to top of scaffold by capillary action (b).
frequency is understood as the relative number of lm and lp in the data set which was divided by the total number of lm and lp . The most common lm in both directions was found to be approximately 0–100 μm with a relative frequency of approximately 12–22%. The spectrum of lp was found to be much larger. It ranged from 0 to 680 μm in the normal direction (Fig. 4(c)) and 0–620 μm in the lateral direction (Fig. 4(d)). The highest relative frequency around 6–8% was measured to be between 50 and 150 μm. The use of medium to high LED generally resulted in smaller lp in both directions. Fig. 5 displays the mean values of the areal size of the skeletons (lm ) and pore (lp ) measurement in the normal (Fig. 5(a)) and lateral planes (Fig. 5(b)). A general trend of increasing lm appeared with increasing LED from 69.8 μm (0.03 J/mm3) to 197.4 μm (0.10 J/mm3) in the normal planes and from 61.8 μm (0.03 J/mm3) to 158.3 μm (0.10 J/ mm3) in the lateral planes. In the normal planes, lp decreased from 341.8 μm (LED of 0.03 J/mm3) to 208.8 μm (LED of 0.10 J/mm3) and in 154
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
specimens printed using moderate LEDs (0.067–0.08 J/mm3), the typical σ − ε curve (Fig. 6 (b)) rises linearly with no significant change of slope until an abrupt load drop occurs that indicates the brittle collapse of the scaffold. The representative σ − ε curve of specimens fabricated using relatively high LEDs (0.08–0.10 J/mm3) is shown in Fig. 6 (c). The curve displays two linear regions with the second region having a lower gradient than the first. These observations indicate significantly different deformation behaviour of scaffolds when subjected to compression in the normal direction. In comparison, scaffolds tested in the lateral direction showed different σ − ε curves which can be divided into three distinct regions. Following an initial linear elastic region, the curves transitioned into a plateau at which the compressive strength in the lateral direction (σl ) was measured. After the plateau, a region with increased stress was observed. It appeared that scaffolds made using lower LED showed more pronounced plateaus (Fig. 6 (e)) than scaffolds fabricated using higher LED (Fig. 6 (f–g)). The relative density ρ*, elastic modulus E (∇I ), and compressive strength σ of the scaffolds are plotted as a function of LED in Fig. 7. It is apparent that ρ*, E, and σ increased monotonically with the increase of LED. More specifically, when the magnitude of LED increased from 0.03 to 0.10 J/mm3, ρ* increased from 20.3% to 41.1%; En increased from 0.9 to 36.4 MPa; El increased from 4.6 to 31.8 MPa; σn exhibits an increase from 0.9 to 6.7 MPa; and σl rose from 0.1 to 4.1 MPa. The microstructure of the scaffolds was examined by SEM after the compression testing in order to investigate the mechanisms of failure. Fig. 8 shows three major failure modes, which include surface cracks (Fig. 8 (a)), fracture of bridge connections (Fig. 8 (b)) and ductile ruptures (Fig. 8 (c)). Among these failure modes, fracture of the bridges was the most common.
Fig. 4. Contour plots showing the relative frequency of different skeleton sizes (lm ) in the normal (a) and lateral plane (b), and pore sizes (lp ) in the normal (c) and lateral plane (d), respectively, on LED.
4. Discussion 4.1. Microstructure formation The manufactured scaffolds had a unique interconnected porous microstructure, as shown in Fig. 2. The unique microstructure might be formed by the powder arrangement on the powder bed during the SLS process and also the agglomeration of fine particles into larger clusters in the starting powder. Fig. 9 illustrates the principle of the microstructural formation by coalescence for two particle clusters that were irradiated with the laser beam (Fig. 9(a)). These clusters melted partially or completely in the course of the SLS process, depending on the location of the powder and the energy input from the incident laser beam (Fig. 9(b)). Once the particle clusters were molten, coalescence of the particles occurred and shaped the PHBV into islands. The islands tended to form spheres to minimize surface area due to the in-process surface tension, but this spherodisation process was retarded by the high in-process viscosity (Fig. 9(b)). However, as the available time above the melting temperature was insufficient, the material started to solidify before it was able to completely form a sphere. While the material was cooling down, it contracted which resulted in bridges that connected islands and inter-layers in locations in which islands were partially joined together (Fig. 9(c)). It is hypothesized that bridges also formed by the random arrangement of powder particles in the powder bed. Overall, the high in-process viscosity and surface tension of the molten PHBV, which was affected by the magnitude of the supplied LED, greatly contributed to the unique scaffold microstructure.
Fig. 5. The size distribution of lm and lp based on different LED were summarized in a weighted mean value lm and lp along the normal (a) and lateral (b) plane, respectively.
the lateral planes, a decline from 307.5 μm (LED of 0.03 J/mm3) to 178.7 μm (LED of 0.10 J/mm3) was found.
3.2. Mechanical characterization Fig. 6 presents three characteristic stress strain (σ − ε ) curves of scaffolds made using low, medium, and high LEDs in the normal (Fig. 6(a–c)) and lateral direction (Fig. 6 (e–g)). Fig. 6 (a–c) shows three representative stress-strain (σ − ε ) curves of scaffolds tested in the normal direction which were fabricated using different LEDs. The typical σ − ε curves of scaffolds manufactured using relatively low LEDs (0.03–0.065 J/mm3) is shown in Fig. 6 (a). The curve exhibits a relatively short linear region at the initial stage of loading, followed by the gradually increased slope until reaching a second linear region. With the progress of loading, the stress reached a critical value at which the σ − ε curve was disrupted by a drop in stress. At this point, the scaffold collapsed in a brittle manner, named as compressive strength (σn ). For
4.2. Relation between relative density, microstructure, and mechanical properties Scaffolds made for bone tissue repair can be implanted at different locations in the human body. Thus, customised shapes are crucial for fitting a scaffold implant into the anatomical defect which are likely to experience loads from different directions. Therefore, loading the 155
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
Fig. 6. Compressive stress-strain curves of PHBV scaffolds tested in the normal (a-d) and lateral direction (e-h): T1 (a & e); T8 (b & f); T9 (c & g). Combined compressive stress-strain curves in the normal (d) and lateral direction (h).
Fig. 8. Occurring failure modes: Microcracks ((a) T9: LED = 0.100 J/mm3 (P = 5 W, ss = 0.1 mm)); Brittle failure ((b) T3: LED = 0.060 J/mm3 (P = 3 W, ss = 0.1 mm), and ductile failure ((c) T3: LED = 0.060 J/mm3 (P = 3 W, ss = 0.1 mm).
Fig. 9. Particle cluster coalescence. Laser beam irradiation of the starting PHBV particle cluster (a); formation of islands after laser irradiation and densification (b); further densification, bridging, and contraction (c).
scaffolds normal and lateral to the build direction can provide important information about the mechanical behaviour of scaffolds with anisotropic microstructures in compression. The combination of SLS process parameters resulted in different levels of LEDs and thus affected significantly the relative density of the scaffolds (see Fig. 7(a)). Models of open-cell foams constructed of beams with square cross-sections established the relationship between the compressive modulus and strength to the relative density of the scaffold by power-law relationships (Gibson and Ashby, 1999; Zein et al., 2002):
Fig. 7. Dependence of relative density ρ* , elastic modulus tested in the normal direction (En ), compressive strength tested in the normal direction (σn ), elastic modulus tested in the lateral direction (El ), compressive strength tested in the lateral direction (σl ) plotted on laser energy density (LED). Results are shown in mean ± standard deviation.
156
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
and σl . This indicates that in comparison to the ideal foams the microstructures of the fabricated scaffolds in this study appeared very dissimilar as it is very heterogeneous throughout the scaffold. Thus, an increase in compressive modulus and strength cannot solely related to increase in density. It is expected that with increasing ρ* also the change in microstructure contributes to the increase in mechanical properties (Woesz et al., 2004). 4.2.1. Intra-layer-density-gradient (ILDG ) It was deduced that the density of the scaffold varies over the layer thickness. The density profile in the z-direction is conceptually illustrated by the “intra-layer density gradient” (ILDG ) as shown in Fig. 11. The density gradient is dependent on how well particles were fused by the laser. It is seen in Fig. 11 that the wider and darker the scale, the more fused material or structure is available to bear loads. Furthermore, the ILDG model states the idea how well layers were connected to each other. Detailed correlations between microstructure and ILDG are described in this section. The first characteristic microstructure was formed when using relatively low LED, such as the case of 0.03 J/mm3. The manufactured scaffolds had a low relative density of 20% and showed alternating layers of high and low density in the x-z plane as previously as described in Section 3.1. These alternating layers were formed as the energy input from the incident laser beam was insufficient to completely melt the powder layer which led to large quantities of loose powder particles throughout the scaffold. A large ILDG was formed in scaffolds using a low LED which is illustrated in Fig. 11(a). The second characteristic microstructure was formed when using relatively high LED, such as the case of 0.1 J/mm3. The scaffolds had a higher relative density of 41% and were well connected in all directions as previously described in Section 3.1. The amount of trapped powder was significantly reduced and an increased number of inter-layer connections was observed as the higher laser energy input sufficiently melted a large fraction of the in-process deposited powder layers. Thus, a small ILDG was observed as illustrated in Fig. 11(c). Relative to the characteristic scaffolds, those fabricated using an intermediate LED consisted of an intermediate microstructure in regards to relative density, island and bridge size, and intra-layer density. For example, scaffolds fabricated using moderate LED (0.067 J/mm3) had a relative density of 30%, sizable islands and bridges, and a medium ILDG , as illustrated in Fig. 11(b). The microstructure can be used to deduce which mechanics may be attributed to their resultant deformation behaviour under compression and will be discussed in the next section.
Fig. 10. Experimental data fit to the model of brittle cellular solids: Compressive modulus tested in the normal direction over elastic modulus of solid PHBV (En/ Es ) (a), compressive strength tested in the normal direction over flexural strength of solid PHBV (σn/ σf ) (b), compressive modulus tested in the lateral direction over elastic modulus of solid PHBV (El/ Es ) (c), compressive strength tested in the lateral direction over flexural strength of solid PHBV (σl/ σf ) (d), plotted against relative density of the scaffold ρ* . Each graph shows the regression curve fitting of the experimental data following Eqs. (3) and (4). Results are shown in mean ± standard deviation.
E = Cρ*n Es
(3)
σ = Cρ*n σf
(4)
where Es is the elastic modulus of the cell wall material (3762 MPa; supplementary materials S2), σf is the flexural strength for the solid PHBV (75.7 MPa; supplementary materials S2), C is a constant, ρ* the relative density. For cellular solids with open pores made of materials that fail in a brittle manner, the flexural strength needs to be used to establish the relation between σ / σf . The constants C and n are strongly depended on the microstructure of the cellular solid, whereas n typically ranges between 1 and 4. The understanding how C and n relate to each other and how these variables relate to the deformation of the microstructure is not well understood. However, both values can give indications whether the microstructure is disordered or periodic. Roberts and Garboczi (2002). The calculated data points of E / Es and σ / σf , plotted on their corresponding relative density in the normal and lateral direction, are illustrated in Fig. 10 respectively. Best curve fits were calculated using Eqs. (3) and (4) in order to see whether the experimental data of E, and σ can be used to fit the theoretical values which are based on ideal open-cell foams that have a homogeneous microstructure that consists of arrays of cubes. All curves showed a coefficient of determination (R2 ) between 0.77 and 0.86, as shown in Fig. 10. The exponent n fell into the expected range of 1–4 for σn and El . However, large n values of 5.56 and 6.64 were found for En
4.2.2. ILDG and compressive properties When compressed in the normal direction, the scaffold with a large ILDG exhibited relatively low compressive modulus and strength as schematically shown in Fig. 11(a). Correspondingly, an initial linear region was observed in its σ − ϵ curve, which was attributed to the elastic deformation of the entire porous structure. A gradual increase in stiffness followed, possibly after failure of some connections of the scaffold's structure, such as the buckling or breaking of bridges. Consequently, densification of the scaffold structure occurred, in which the buckled or broken bridges touched each other, forming new connections that were less compliant. Scaffold failure could be due to cracking Fig. 11. Intra-layer density gradients (ILDG ) in correlation to the compressive deformation behaviour tested in the normal direction. Scaffolds manufactured using low LED showed a large ILDG (a), scaffolds manufactured using intermediate LED presented a medium ILDG (b), and scaffolds using a high LED appeared to have a small ILDG (c).
157
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
initiated between two layers, followed by crack propagation across the entire structure. Scaffolds with low ILDG , compared to scaffolds with high ILDG , behaved with higher stiffness when under compression in the normal direction, and showed higher compressive strengths (Fig. 11(c)). The typical compressive σ − ε curve of these scaffolds was characterized by an initial linear-elastic deformation of islands and bridges until the point where the gradient of the σ − ε curve decreased. The reduced gradient was likely caused by the start of bending and stretching of compliant bridges. As the deformation of these bridges became more dominant, the σ − ε curve transitioned into a second linear region. Simultaneously, an increase in the scaffold diameter was observed, likely caused by deformation of inter-layer and intra-layer bridges. Scaffold failure was likely caused by crack initiation and propagation across the entire structure. Scaffolds with a medium ILDG deformed via a combination of the mechanisms of the two characteristic scaffolds, resulting in relatively linear σ − ϵ curves, medium compressive elastic moduli and strengths (Fig. 11(b)). In comparison to the deformation behaviour of scaffolds loaded in the normal direction, the deformation in the lateral direction resulted in significantly different σ − ε curves. These σ − ε curves showed three characteristic regimes: elastic, delamination, and densification. The concept of ILDG can also be applied to explain these three regimes. Fig. 12 presents the concept of ILDG when scaffolds were exposed to compression in the lateral direction. Scaffolds with low relative densities showed a microstructure which consisted of an array of layers with a high ILDG , as shown in Fig. 12(a). Thus, the scaffold was constructed by layers which weakly adhered to each other. When the scaffolds were tested in the lateral direction, these higher density regions were taking large portion of the applied load. This resulted in a significantly higher El (5.2 MPa) than En (0.9 MPa). A lower En resulted as the low density regions were compressed more than the high density regions. For scaffolds with decreasing ILDG (Fig. 12(b–c)), the connections between layers became pronounced. With smaller ILDG , En and El increased and converged gradually as the scaffolds became more isotropic. The second region of the σ − ε curves, has a plateau in between two steep linear regions which is similar to a typical curve from testing cellular solids (Gibson and Ashby, 1999). These plateaus, found in
cellular solids, are generally associated with the collapse of cell walls in an elastic, plastic, or crushing manner. In this study, inter-layer delamination occurred when the σ − ε curve reached the plateau. The layer detachment was observed in-situ when the scaffolds were loaded laterally and was particularly prevalent among scaffolds made using high ILDG . The plateau in the σ − ε curves, following the initial linear region, was likely caused by the fact that the effective porosity decreased due to compression, and the stress could not increase because of the layer-by-layer failure. Delamination was likely caused by failure of weak connections in the low density regions as indicated in Fig. 12(a). The deformation of microstructures with high ILDG resulted in very low σl (0.15 MPa), compared to σn (0.9 MPa). Scaffolds with lower ILDG were less likely to experience inter-layer delamination as illustrated in Fig. 12(c). Thus, with increasing relative densities and decreasing ILDG (Fig. 12(b–c)), the compressive strength increased up to 4.1 MPa (σl ) and 6.4 MPa (σn ). Stronger connections supported the scaffold throughout which were able to withstand the applied stresses to a greater extent, resulting in smaller plateau regions. At the end of the plateau, scaffolds were densified to a level that they lost most of their porosity, leading to a last state in which the scaffolds were compressed almost in a ‘solid’ state. This is clearly shown by the steep increase of the σ − ε curves after the plateau. Overall, the scaffolds showed an orientation dependent mechanical behaviour under compressive loading, which was attributed to their anisotropic microstructure. It should be noted that the maximum compressive modulus comes close to the lower bound of reported moduli for trabecular bone of vertebra (67 – 344 MPa). The compressive strength is within the range of reported values of the trabecular bone of tibias (2.2 −5.3 MPa) and femurs (5.6 – 8.1 MPa) (Karim et al., 2013). For many applications in bone TE, scaffolds do not have to match the mechanical properties of the anatomical implantation site; rather mechanical properties are required that can support the scaffold throughout its lifetime (Jack et al., 2009). Furthermore, fixtures can be used to provide additional load bearing structures during the regeneration phase, in which the patient is normally expected to rest.
Fig. 12. Intra-layer density gradients (ILDG ) in correlation to the compressive deformation behaviour in the lateral direction. Scaffolds with low relative density showed a high ILDG (a), scaffolds with medium relative density led to a medium ILDG (b), and scaffolds with higher relative density presented a small ILDG (c).
4.4. Future works
4.3. Porosity and pore size Another important feature of the manufactured scaffolds in this study is their highly interconnected porous structure. Such a structure would ensure the transport of nutrients and gases, and removal of metabolic waste resulting from osteogenesis. For scaffolds used in bone TE applications, pore sizes in the range of 200 – 900 μm are typically suitable to emulate the microstructure of trabecular bone (Duan et al., 2010; Mygind et al., 2007). In this study, the average areal pore size of manufactured scaffolds was in the range from 209 to 342 μm in the normal direction and 179–308 μm in the lateral direction, which could be adjusted using different values of LED. Note that the line intersection method being used in this study measured multiple intersection lengths across a single pore, bridge, or island. For a hypothetical circular pore, most of those intersection lengths did not cross the centre of the pores, which therefore led to an underestimation of pore size. Although a weighting factor was introduced to minimize the effect, the method could still be considered conservative. The scaffolds also have an excellent interconnected pore network, as successfully demonstrated by the water absorption test and shown in Fig. 3. Overall, the scaffolds had a highly interconnected porous network with pore sizes that can typically be found in scaffolds made for bone tissue engineering applications.
The scaffolds manufactured in this study show characteristic features (microstructure, mechanical properties, and interconnectivity) which suggest a positive performance in clinical trials. However, two 158
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
possible limitations shall be considered. First, as a technologically inherent side effect of a powder bed based process, the fabricated scaffolds contained a portion of trapped powder particles, particularly present on the outer surface of the scaffold. Further studies should be carried out to refine the manufacturing process to decrease the amount of loose particles. Second, the manufactured scaffolds had a heterogeneous microstructure. Finite element analysis (FEA) could help to validate the findings in respect to the compressive deformation behaviour of the unique microstructure.
Fisher, J.P., Cooke, M., Rimnac, C., 2003. Poly (propylene fumarate) and poly (dllactic-co-glycolic acid) as scaffold materials for solid and foam-coated composite tissue-engineered constructs for cranial reconstruction. Tissue Eng. 9 (3), 495–504. Dimitriou, R., Jones, E., McGonagle, D., Giannoudis, P.V., 2011. Bone regeneration: current concepts and future directions. BMC Med. 9 (66–66, 1741-7015-9-66)(PII) (21627784)(pmid) (BMC Med.). Du, Y., Liu, H., Yang, Q., Wang, S., Wang, J., Ma, J., Noh, I., Mikos, A.G., Zhang, S., 2017. Selective laser sintering scaffold with hierarchical architecture and gradient composition for osteochondral repair in rabbits. Biomaterials 137, 37–48. Duan, B., Wang, M., Zhou, W.Y., Cheung, W.L., Li, Z.Y., Lu, W.W., 2010. Three-dimensional nanocomposite scaffolds fabricated via selective laser sintering for bone tissue engineering. Acta Biomater. 6 (12), 4495–4505. Dutta, R.C., Dey, M., Dutta, A.K., Basu, B., 2017. Competent processing techniques for scaffolds in tissue engineering. Biotechnol. Adv. 35 (2), 240–250. Eosoly, S., Brabazon, D., Lohfeld, S., Looney, L., 2010. Selective laser sintering of hydroxyapatite/poly-epsilon-caprolactone scaffolds. Acta Biomater. 6 (7), 2511–2517. Gibson, L.J., Ashby, M.F., 1999. Cellular Solids: Structure and Properties. Cambridge university press. Gogolewski, S., Jovanovic, M., Perren, S., Dillon, J., Hughes, M., 1993. Tissue response and in vivo degradation of selected polyhydroxyacids: polylactides (PLA), poly (3hydroxybutyrate)(PHB), and poly (3-hydroxybutyrate-co-3-hydroxyvalerate)(PHB/ VA). J. Biomed. Mater. Res. Part A 27 (9), 1135–1148. Hagedorn, Y.-C., 2013. Additive Manufacturing of High Performance Oxide Ceramics via Selective Laser Melting (Dissertation). Hutmacher, D.W., Woodfield, T.B.F., Dalton, P.D., 2014. Scaffold Design and Fabrication. Academic Press, Oxford, pp. 311–346 (Chapter 10). Jack, K.S., Velayudhan, S., Luckman, P., Trau, M., Grondahl, L., Cooper-White, J., 2009. The fabrication and characterization of biodegradable HA/PHBV nanoparticlepolymer composite scaffolds. Acta Biomater. 5 (7), 2657–2667. Karim, L., Hussein, A.I., Morgan, E.F., Bouxsein, M.L., 2013. The Mechanical Behavior of Bone, book Section 19 - The Mechanical Behavior of Bone, 4th ed. Academic Press, San Diego, pp. 431–452. Kumarasuriyar, A., Jackson, R., Grondahl, L., Trau, M., Nurcombe, V., Cool, S., 2005. Poly (beta-hydroxybutyrate-co-beta-hydroxyvalerate) supports in vitro osteogenesis. Tissue Eng. 11 (7–8), 1281–1295. Leong, K.F., Cheah, C.M., Chua, C.K., 2003. Solid freeform fabrication of three-dimensional scaffolds for engineering replacement tissues and organs. Biomaterials 24 (13), 2363–2378. Lohfeld, S., Tyndyk, M.A., Cahill, S., Flaherty, N., Barron, V., McHugh, P.E., 2010. A method to fabricate small features on scaffolds for tissue engineering via selective laser sintering. J. Biomed. Sci. Eng. 03 (02), 10. Mazzoli, A., 2013. Selective laser sintering in biomedical engineering. Med. Biol. Eng. Comput. 51 (3), 245–256. Mygind, T., Stiehler, M., Baatrup, A., Li, H., Zou, X., Flyvbjerg, A., Kassem, M., Buenger, C., 2007. Mesenchymal stem cell ingrowth and differentiation on coralline hydroxyapatite scaffolds. Biomaterials 28 (6), 1036–1047. Nam, Y.S., Park, T.G., 1999. Porous biodegradable polymeric scaffolds prepared by thermally induced phase separation. J. Biomed. Mater. Res. 47 (1), 8–17. No, Y.J., Li, J.J., Zreiqat, H., 2017. Doped calcium silicate ceramics: a new class of candidates for synthetic bone substitutes. Materials 10 (2), 153. O'Brien, F.J., Harley, B.A., Yannas, I.V., Gibson, L.J., 2005. The effect of pore size on cell adhesion in collagen-gag scaffolds. Biomaterials 26 (4), 433–441. O'Brien, F.J., 2011. Biomaterials and scaffolds for tissue engineering. Mater. Today 14 (3), 88–95. Partee, B., Hollister, S.J., Das, S., 2005. Selective laser sintering process optimization for layered manufacturing of CAPA 6501 polycaprolactone bone tissue engineering scaffolds. J. Manuf. Sci. Eng. 128 (2), 531–540 (10.1115/1.2162589). Pereira, T.F., Silva, M.A.C., Oliveira, M.F., Maia, I.A., Silva, J.V.L., Costa, M.F., Thire, R.M.S.M., 2012. Effect of process parameters on the properties of selective laser sintered poly(3-hydroxybutyrate) scaffolds for bone tissue engineering. Virtual Phys. Prototyp. 7 (4), 275–285. Ramay, H.R.R., Zhang, M., 2004. Biphasic calcium phosphate nanocomposite porous scaffolds for load-bearing bone tissue engineering. Biomaterials 25 (21), 5171–5180. Roberts, A.P., Garboczi, E.J., 2002. Elastic properties of model random three-dimensional open-cell solids. J. Mech. Phys. Solids 50 (1), 33–55. Shahzad, K., Deckers, J., Kruth, J.-P., Vleugels, J., 2013. Additive manufacturing of alumina parts by indirect selective laser sintering and post processing. J. Mater. Process. Technol. 213 (9), 1484–1494. Shirazi, F.S., Mehrali, M., Oshkour, A.A., Metselaar, H.S.C., Kadri, N.A., Abu Osman, N.A., 2014. Mechanical and physical properties of calcium silicate/alumina composite for biomedical engineering applications. J. Mech. Behav. Biomed. Mater. 30, 168–175. Shuai, C., Sun, H., Gao, C., Feng, P., Guo, W., Yang, Y., Zhao, M., Yang, S., Yuan, F., Peng, S., 2017. Mechanical reinforcement of bioceramics scaffolds via fracture energy dissipation induced by sliding action of MoS2 nanoplatelets. J. Mech. Behav. Biomed. Mater. 75, 423–433. Sombatmankhong, K., Sanchavanakit, N., Pavasant, P., Supaphol, P., 2007. Bone scaffolds from electrospun fiber mats of poly(3-hydroxybutyrate), poly(3-hydroxybutyrate-co3-hydroxyvalerate) and their blend. Polymer 48 (5), 1419–1427. Sultana, N., Khan, T.H., 2012. In vitro degradation of phbv scaffolds and nHA/PHBV composite scaffolds containing hydroxyapatite nanoparticles for bone tissue engineering. J. Nanomater. 2012, 12. Sultana, N., Wang, M., 2007. Fabrication of HA/PHBV composite scaffolds through the emulsion freezing/freeze-drying process and characterisation of the scaffolds. J. Mater. Sci.: Mater. Med. 19 (7), 2555. Thomson, R.C., Yaszemski, M.J., Powers, J.M., Mikos, A.G., 1996. Fabrication of biodegradable polymer scaffolds to engineer trabecular bone. J. Biomater. Sci. Polym. Ed.
5. Conclusions In this study, interconnected and highly porous (59–80%) PHBV scaffolds were successfully manufactured using SLS without utilizing pre-designed internal pore architectures. The manufactured scaffolds had a unique microstructure that consisted of pores, islands, and bridges. The average weighted areal pore size of the scaffolds ranged between 209 and 342 μm in the normal direction and 179 – 307μm in the lateral direction. The maximum areal pore size was measured at 680 μm. The formation of such porous networks was attributed to the agglomeration of particles in the starting powder, the in-process surface tension and viscosity of the molten PHBV, and the SLS process parameters. The use of greater LED resulted in denser scaffold microstructures with improved mechanical properties. The compressive elastic modulus of scaffolds could reach up to 36.4 and 31.8 MPa with a relative porosity of 59% when tested in normal and lateral direction respectively. Scaffolds with the same relative porosity showed a compressive strength of 6.7 MPa (normal direction) and 4.1 MPa (lateral direction). The porosity and mechanical properties of the scaffold could be adjusted by varying the level of LED used within the SLS process. Overall, the manufactured scaffolds with their unique microstructure present great potential for further in-vitro and in-vivo testing. Acknowledgments S.H.D. thanks the University of Queensland (UQ) and the School of Mechanical & Mining Engineering for the tuition fee waiver and living allowance scholarship. The authors would like to thank the Enginnering @ UQ SEED grants for financial support and acknowledge the facilities, and the scientific and technical assistance of the Australian Microscopy & Microanalysis Research Facility at the Centre for Microscopy and Microanalysis (CMM), UQ. M. Dargusch would like to acknowledge the support of the ARC Research Hub for Advanced Manufacturing of Medical Devices (IH150100024). Appendix A. Supplementary data Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jmbbm.2018.05.007. References Bukharova, T., Antonov, E., Popov, V., Fatkhudinov, T.K., Popova, A., Volkov, A., Bochkova, S., Bagratashvili, V., Gol' Dshtein, D., 2010. Biocompatibility of tissue engineering constructions from porous polylactide carriers obtained by the method of selective laser sintering and bone marrow-derived multipotent stromal cells. Bull. Exp. Biol. Med. 149 (1), 148–153. Cao, W., Hench, L.L., 1996. Bioactive materials. Ceram. Int. 22 (6), 493–507. Caulfield, B., McHugh, P.E., Lohfeld, S., 2007. Dependence of mechanical properties of polyamide components on build parameters in the SLS process. J. Mater. Process. Technol. 182 (1), 477–488. Chia, H.N., Wu, B.M., 2015. Recent advances in 3D printing of biomaterials. J. Biol. Eng. 9 (1), 4. Chua, C.K., Leong, K.F., Tan, K.H., Wiria, F.E., Cheah, C.M., 2004. Development of tissue scaffolds using selective laser sintering of polyvinyl alcohol/hydroxyapatite biocomposite for craniofacial and joint defects. J. Mater. Sci.: Mater. Med. 15 (10), 1113–1121. Chua, C.K., Yeong, W.Y., An, J., 2017. Special issue: 3D printing for biomedical engineering. Materials. Dean, D., Topham, N.S., Meneghetti, S.C., Wolfe, M.S., Jepsen, K., He, S., Chen, J.E.-K.,
159
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
Wu, J., Xue, K., Li, H., Sun, J., Liu, K., 2013. Improvement of PHBV scaffolds with bioglass for cartilage tissue engineering. PLoS ONE 8 (8), e71563. Yang, Y., Kang, Y., Sen, M., Park, S., 2011. Biomaterials for Tissue Engineering Applications. A Review of the Past and Future Trends., Book Section Bioceramics in Tissue Engineering. Springer, pp. 179–207. Zein, I., Hutmacher, D.W., Tan, K.C., Teoh, S.H., 2002. Fused deposition modeling of novel scaffold architectures for tissue engineering applications. Biomaterials 23 (4), 1169–1185.
7 (1), 23–38. Williams, J.M., Adewunmi, A., Schek, R.M., Flanagan, C.L., Krebsbach, P.H., Feinberg, S.E., Hollister, S.J., Das, S., 2005. Bone tissue engineering using polycaprolactone scaffolds fabricated via selective laser sintering. Biomaterials 26 (23), 4817–4827. Woesz, A., Stampfl, J., Fratzl, P., 2004. Cellular solids beyond the apparent density - an experimental assessment of mechanical properties. Adv. Eng. Mater. 6 (3), 134–138. Wu, C., Chang, J., Zhai, W., Ni, S., 2007. A novel bioactive porous bredigite (Ca7MgSi4O16) scaffold with biomimetic apatite layer for bone tissue engineering. J. Mater. Sci.: Mater. Med. 18 (5), 857–864.
160
Contents lists available at ScienceDirect
Journal of the Mechanical Behavior of Biomedical Materials journal homepage: www.elsevier.com/locate/jmbbm
Synthesis, microstructure, and mechanical behaviour of a unique porous PHBV scaffold manufactured using selective laser sintering
T
Sven H. Diermanna, Mingyuan Lua, Yitian Zhaoa, Luigi-Jules Vandib, Matthew Darguschc, ⁎ Han Huanga, a
School of Mechanical and Mining Engineering, The University of Queensland, St Lucia, QLD 4072, Australia School of Chemical Engineering, The University of Queensland, QLD 4072, Australia c Centre for Advanced Materials Processing and Manufacturing (AMPAM), School of Mechanical and Mining Engineering, The University of Queensland, QLD 4072, Australia b
A R T I C LE I N FO
A B S T R A C T
Keywords: Selective laser sintering Poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) Mechanical properties Scaffold Bone tissue engineering
Selective Laser Sintering (SLS) is a promising technique for manufacturing bio-polymer scaffolds used in bone tissue engineering applications. Conventional scaffolds made using SLS have complex engineered architectures to introduce adequate porosity and pore interconnectivity. This study presents an alternative approach to manufacture scaffolds via SLS without using pre-designed architectures. In this work, a SLS process was developed for fabricating interconnected porous biodegradable poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) scaffolds with large surface areas and relative porosities of up to 80%. These characteristics provide great potential to enhance cell attachment inside the scaffolds. The scaffold microstructure was dependent on the laser energy density (LED) during the SLS process. An increase in LED led to scaffolds with higher relative densities, stronger inter-layer connections, and a reduced quantity of residual powder trapped inside the pores. An increase in relative density from 20.3% to 41.1% resulted in a higher maximum compressive modulus and strength of 36.4 MPa and 6.7 MPa, respectively.
1. Introduction Repair of complex bone fractures, including scenarios involving large bone loss, continues to pose great challenges for clinical management. Autologous and allogeneic bone grafting are commonly applied surgical procedures which come with inherent limitations and risks, including graft availability, harvest site pain, morbidity, rejection reactions, and risk of disease transmission (Dimitriou et al., 2011). Scaffold-assisted tissue engineering (TE) is regarded as a promising bone regeneration technique to replace the current bone grafting methods (Ramay and Zhang, 2004; Dean et al., 2003; Chua et al., 2004; Wu et al., 2013). In scaffold-assisted bone TE, a porous scaffold is implanted to provide support for cell migration, colonization, growth, and differentiation to promote tissue generation at the fracture site. Therefore, scaffolds are usually made of biocompatible and biodegradable materials and have an interconnected porous network (Cao and Hench, 1996; O'Brien, 2011; Yang et al., 2011; Hutmacher et al., 2014; Dutta et al., 2017). The fabrication of porous TE scaffolds have employed a variety of manufacturing techniques, such as thermally induced phase separation
⁎
Corresponding author. E-mail address: [email protected] (H. Huang).
https://doi.org/10.1016/j.jmbbm.2018.05.007 Received 13 March 2018; Received in revised form 21 April 2018; Accepted 2 May 2018 1751-6161/ © 2018 Elsevier Ltd. All rights reserved.
(TIPS) (Nam and Park, 1999; Jack et al., 2009), solvent casting and particulate leaching (Thomson et al., 1996), electrospinning (Sombatmankhong et al., 2007), and sintering (Wu et al., 2007). While these methods have their unique benefits, existing technical issues including manual intervention, inconsistency and inflexibility in processing, and exclusion of toxic solvents or porogen still need to be addressed. Most significantly, customised scaffold geometries often cannot be achieved due to the inherent restrictions of the conventional manufacturing processes. Over the last decade, additive manufacturing (AM) has emerged as a promising method for the synthesis of TE scaffolds and has opened a new field of intriguing bioengineering possibilities. AM techniques are ideally suited for biomedical applications due to their ability to make customised parts with complex shapes (Leong et al., 2003; Mazzoli, 2013; Chia and Wu, 2015; Chua et al., 2017; Caulfield et al., 2007). Among various AM techniques, selective laser sintering (SLS) was extensively used to make bone scaffolds due to its capability to make porous structures using a large variety of materials in powder form. SLS does not require support structures during the fabrication process and the existing porosity between particles can be preserved (Chia and Wu, 2015). In previous studies, SLS manufactured
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
scaffolds typically used a computer designed three-dimensional (3D) architecture to achieve the desired porosity and interconnectivity (Hutmacher et al., 2014; Duan et al., 2010; Eosoly et al., 2010; Shirazi et al., 2014; Du et al., 2017). However, the use of a pore architecture is potentially accompanied with drawbacks. First, the manufacture of a scaffold structure with small features via SLS has limited resolution due to relatively large laser spot sizes (Lohfeld et al., 2010). Second, scaffolds made using SLS with a pre-designed pore architecture have lower specific surface areas than scaffolds manufactured using alternative techniques such as e.g. TIPS, (Nam and Park, 1999, Jack et al., 2009), solvent casting and particulate leaching (Thomson et al., 1996), or electrospinning (Sombatmankhong et al., 2007). Requirements on biocompatibility, bioactivity and biodegradability have limited the choice of materials for scaffolds used in bone TE. Till now, ceramics (Shuai et al., 2017; No et al., 2017) and biodegradable polymers are frequently used for TE strategies in AM. For example, poly (ϵ) caprolactone (PCL) (Williams et al., 2005; Partee et al., 2005; Lohfeld et al., 2010), poly-L-lactide (PLLA) (Duan et al., 2010), poly-D-lactide (PDLA) (Bukharova et al., 2010), polyhydroxybutyrate (PHB) (Pereira et al., 2012), and poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) (Duan et al., 2010) are representative polymers that were used to manufacture scaffolds via SLS. Among these materials, PHBV offers favourable surface chemistry for cell attachment and proliferation (Kumarasuriyar et al., 2005), while its degradation by-products are biocompatible and can be metabolized. Previous in vitro and in vivo biocompatibility investigations of PHBV in TE applications have confirmed its inert nature (Sultana and Wang, 2007; Sultana and Khan, 2012). PHBV elicits minimal inflammatory responses as it degrades into products, normally found as constituents of human blood (Sultana and Wang, 2007; Sultana and Khan, 2012). In addition, PHBV degrades over a long period of time which allows the polymeric scaffold to maintain its mechanical integrity until an adequate portion of new bone grows into the scaffold (Sultana and Khan, 2012; Gogolewski et al., 1993). The objective of this study is to develop a SLS process to manufacture PHBV scaffolds with interconnected porous networks without engineered architectures, addressing the issues of process resolution and scaffold surface area. Using the developed SLS process, the manufactured scaffolds would have much smaller strut sizes and larger surface areas than those with pre-designed architectures, which is in favour of enhancing cell attachment inside the scaffold (O'Brien et al., 2005). In this study, the effect of SLS processing parameters on the microstructural and mechanical properties was systematically investigated, with a particular emphasis placed on the scaffolds' compressive deformation behaviour in both the normal and lateral direction.
Table 1 SLS processing conditions. Test
Laser power /P (W)
Scan spacing/SS (mm)
Laser energy density/ LED (J/ mm3)
T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11
3 3 3 4 4 4 5 5 5 6 6
0.2 0.15 0.1 0.2 0.15 0.1 0.2 0.15 0.1 0.2 0.15
0.030 0.040 0.060 0.040 0.053 0.080 0.050 0.067 0.100 0.060 0.080
defined as the metastable area between two phases. 2.2. Manufacture of scaffolds Porous scaffolds were fabricated using a SLS system (DTM Sinterstation 2500 Plus, 3D Systems, USA) equipped with a CO2 laser with an inherent wavelength of 10.6 μm and a spot size of 420 μm. To minimize the usage of powder in building small-scale specimens, a powder feeding system was designed and adapted into the SLS system. Both the volume of the powder feeding cartridges and the build area were substantially reduced. The area of the feed bed and build area were reduced to 125 × 105 mm2 and 85 × 85 mm2 respectively. Cubes (L × W × H=10 × 10 × 10 mm3) and cylinders (D = 6 mm, H = 10 mm) were manufactured in eleven tests, using different processing conditions, as shown in Table 1. The laser power (P) and scan spacing (ss ) varied from 3 to 7 W and 0.1–0.2 mm respectively. The laser scan speed was fixed at 5000 mm/s. The powder was spread by a roller system with a traverse speed of 177.8 mm/s. The layer thickness of each powder layer was 0.1 mm. The SLS chamber was filled with nitrogen gas to maintain an inert environment with an oxygen content below 5%. The processing temperature was 125 °C. Loose powder particles were removed from the manufactured scaffolds after the SLS process using a fine brush. The altered SLS processing parameters were characterized using laser energy density (LED). The LED (J/mm3) describes the intensity of the laser irradiation and is the quotient of P divided by ss , scan speed (v), and layer thickness (l) (Shahzad et al., 2013; Hagedorn, 2013):
LED =
P ss vl
(1)
The LED used in each test was calculated using Eq. (1) and is also shown in Table 1.
2. Experimental details 2.1. PHBV powder
2.3. Characterization Poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) powder (ENMATTM , Y1000) with a 3-hydroxyvalerate (HV) content of 2–5% and average molecular weight of 1,187,000 g/mol, was used (supplied by TianAn Biologic Materials Ltd, Ningbo, China). The as-received powder consists of fine spherical particles that agglomerate to form large irregular clusters. The powder was sieved to below 250 μm using a vibratory sieve shaker (Retsch, Germany). The particle size distribution (PSD) of the powder was measured by means of laser diffraction analysis of a water-based dispersion using a Mastersizer Hydro (Malvern Instruments, England). The cumulative particle size distribution values D10 , D50 , D90 , based on a volume distribution, were 11.3 μm, 73.7 μm, and 188 μm, respectively. Differential Scanning Calorimetry (DSC) analysis on the as-received PHBV powder revealed a melting temperature Tm and crystallisation temperature Tc of 172 °C and 114 °C respectively. The onset temperature of Tm and Tc of 161 °C and 120 °C was measured respectively, resulting in a SLS processing window of 41 °C,
2.3.1. Microstructural characterization The microstructure and pore architecture of the fabricated scaffolds were characterized using scanning electron microscopy (SEM, FEI, Orgegon, USA) and 3D measuring laser microscope (OM, Olympus LEXT OLS4100, Japan). For SEM observation, the scaffolds were crosssectioned in different directions using a scalpel, and then coated with a 20 nm conductive film. For OM observation, the PHBV scaffolds were mounted in epoxy resin, sectioned, ground, and polished. The entire area of the polished scaffold was imaged to provide a holistic view of the sectioned plane. A minimum of 67 OM images were taken for each specimen and then stitched together to form a large area image mosaic. The stitched images of cross-sections along the x-y (normal) and y-z (lateral) plane were used to quantify the areal size of the skeleton (material, lm ) and pores (lp ) using a line inter-section method, which is detailed in the supplementary material S1. 152
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
plane and the x-z plane. Three typical microstructural features were identified and marked in the images, which include pores, trapped powder, and fused material in the forms of islands and bridges. Fused islands were connected by bridges which were both made of solidified PHBV. Note that the black regions are air bubbles in the resin that failed to escape from inside the porous structure during the vacuum impregnation process. They were therefore counted as pores. Evidently, the use of different SLS process parameters resulted in variations in the microstructure. Small fused islands connected by slender bridges are observed in scaffolds produced using low LED, as shown in Fig. 1(a). A considerable amount of unfused powder particles were trapped inside the pores. The use of higher LEDs led to a significant reduction in the quantity of trapped powder and an increase of fused material in the forms of larger islands and thicker bridges (Figs. 1(b, c, e, f)). Scaffolds made using low LED had partially fused islands and bridges which were much more prevalent in the lateral direction. These partially fused features were likely due to insufficient energy input for completely fusing the deposited powder layer in build direction. The use of high LED resulted in fully-fused bridges and islands in the x-z plane. Fig. 2 shows representative cross-sectional SEM images in the x-y and x-z plane of scaffolds manufactured using LEDs of 0.03, 0.067, and 0.10 J/mm3, respectively. Figs. 2(a-b) show a cross-section of a scaffold manufactured using a LED of 0.03 J/mm3 in the x-y plane. The figures present a well connected porous network that consists of numerous islands connected by thin bridges and large amounts of unfused powder trapped inside the pores. In the x-z plane, the porous network was not as well connected. A density gradient of fused material was observed within each layer deposited during the SLS process and will be further referred to as the intra-layer. These intra-layers were poorly fused to each other by the laser during SLS process. Thus, resulted in a small amount of inter-layer connections. These inter-layer connections are the connections created between the deposited powder layers in the SLS process (Fig. 2(c)). Figs. 2(d–e) show a cross-section of a scaffold manufactured using a LED of 0.067 J/mm3 in the x-y plane. The figures present a well connected porous network with moderate-sized islands and bridges, and a reduced amount of unfused powder. The cross-section in the x-z plane shown in Fig. 2(f) presents the porous reticulation in the x-z plane that consists of fused material stacks with small amounts of loose powder
2.3.2. Relative density estimation Weights and dimensions of the scaffolds were measured to calculate their densities. Six replicates were analysed for each SLS process condition. The relative density value of a scaffolds was calculated as
ρ* = ρscaffold / ρsolid ,
(2) 3
where ρsolid is the PHBV solid density of 1.25 g/cm . 2.3.3. Mechanical testing Compression tests were carried out at room temperature on a 5543 Instron universal testing machine using a 2 kN load cell. Uniaxial compressive load was applied to the scaffold specimens in two directions. Scaffolds that were loaded perpendicular to the build direction (see supplementary material S3 for schematic of planes and directions) will be referred to as compression in the normal direction (subscript symbol: n). Scaffolds which were loaded parallel to the build direction (y-z plane) will be referred to as compression in the lateral direction (subscript symbol: l) . Cylindrical shaped scaffolds were used for testing in the normal direction (D = 7 mm, H = 10 mm), and cubes (equal side length L = 10 mm) were used to investigate compression in the lateral direction. For each test condition, six replicates were loaded at 5 mm/ min until failure occurred. The elastic modulus E (MPa) for compression was calculated from the initial linear regions of the σ − ε curves. The critical load when the scaffold lost its load bearing capacity at the first inflection point on the σ − ε curve is denoted as compressive strength in the normal direction (σn ). Further densification resulted in higher compressive strengths, but this data was disregarded as the scaffolds lost their original shape. Scaffolds were tested in the lateral direction until a strain level of 70% was reached. The 1% offset (Zein et al., 2002) stress is deemed as the compressive strength in the lateral direction (σl ). A noise cancelling filter was applied with a moving averaging algorithm of seven data points for the illustration of the σ − ε curve. 3. Results 3.1. Microstructural characterization Fig. 1 shows the optical images of scaffold microstructures, produced using different process conditions, cross-sectioned along the x-y
Fig. 1. Optical images along the x-y plane (a-c) and x-z-plane (d-f) of selected scaffolds. T1 (a & d); T8 (b & e); T9 (c & f). 153
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
Fig. 2. SEM images of selected scaffolds along the x-y and x-z cross-sectional plane. T1 (a-c); T8 (d-f); T9 (g-i)).
inside the pores. Layers were seen connected by short but sizable bridges and some large islands from adjacent layers were connected. The cross-section of a scaffold fabricated using a LED of 0.10 J/mm3 in x-y plane contains a well connected porous network that consists of islands connected by large bridges and very little amounts of unfused powder, as it can be seen in Figs. 2(g–h). Fig. 2(i) shows a cross-section in the x-z plane. The amount of trapped powder was significantly reduced and a large number of inter-layer connections was found. These appeared in the form of bridges and island to island connections from adjacent layers. The SEM observation demonstrated that the scaffolds have characteristic microstructures in the form of islands, bridges, and pores which were dependent on the LED used. The use of higher LEDs lead to an increase in the volume and size of islands and bridges, better connections between layers, and less powder trapped inside the scaffold. All scaffolds appeared to be highly porous and interconnected. To confirm the pore interconnectivity of the manufactured scaffolds, a simple water uptake test was conducted. In this test, a scaffold was put into a petri dish with blue coloured water, as shown in Fig. 3(a). The scaffolds were submerged in the water to approximately 10% of their height to test whether the water can be absorbed and travel to non-submerged regions through capillary action. As shown in Fig. 3(b), after 5 to 10 minutes water was absorbed and transmitted from the bottom to the top of all tested scaffolds. This clearly indicates that the pores in the scaffolds are interconnected and the scaffolds have a good water uptake. The areal size of the skeleton (lm ) and pores (lp ) were measured using the line inter-section method. Fig. 4 presents four contour plots that illustrate the dependence of lm and lp on the of LED viewed in both the normal and lateral direction in detail. It appeared that the lm viewed in the normal (Fig. 4(a)) and lateral direction (Fig. 4(b)) increased as the relative frequency increased when using higher LED. The relative
Fig. 3. Water uptake test to investigate interconnectivity. Scaffold partially submerged in water (a); Water was absorbed and transmitted from bottom to top of scaffold by capillary action (b).
frequency is understood as the relative number of lm and lp in the data set which was divided by the total number of lm and lp . The most common lm in both directions was found to be approximately 0–100 μm with a relative frequency of approximately 12–22%. The spectrum of lp was found to be much larger. It ranged from 0 to 680 μm in the normal direction (Fig. 4(c)) and 0–620 μm in the lateral direction (Fig. 4(d)). The highest relative frequency around 6–8% was measured to be between 50 and 150 μm. The use of medium to high LED generally resulted in smaller lp in both directions. Fig. 5 displays the mean values of the areal size of the skeletons (lm ) and pore (lp ) measurement in the normal (Fig. 5(a)) and lateral planes (Fig. 5(b)). A general trend of increasing lm appeared with increasing LED from 69.8 μm (0.03 J/mm3) to 197.4 μm (0.10 J/mm3) in the normal planes and from 61.8 μm (0.03 J/mm3) to 158.3 μm (0.10 J/ mm3) in the lateral planes. In the normal planes, lp decreased from 341.8 μm (LED of 0.03 J/mm3) to 208.8 μm (LED of 0.10 J/mm3) and in 154
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
specimens printed using moderate LEDs (0.067–0.08 J/mm3), the typical σ − ε curve (Fig. 6 (b)) rises linearly with no significant change of slope until an abrupt load drop occurs that indicates the brittle collapse of the scaffold. The representative σ − ε curve of specimens fabricated using relatively high LEDs (0.08–0.10 J/mm3) is shown in Fig. 6 (c). The curve displays two linear regions with the second region having a lower gradient than the first. These observations indicate significantly different deformation behaviour of scaffolds when subjected to compression in the normal direction. In comparison, scaffolds tested in the lateral direction showed different σ − ε curves which can be divided into three distinct regions. Following an initial linear elastic region, the curves transitioned into a plateau at which the compressive strength in the lateral direction (σl ) was measured. After the plateau, a region with increased stress was observed. It appeared that scaffolds made using lower LED showed more pronounced plateaus (Fig. 6 (e)) than scaffolds fabricated using higher LED (Fig. 6 (f–g)). The relative density ρ*, elastic modulus E (∇I ), and compressive strength σ of the scaffolds are plotted as a function of LED in Fig. 7. It is apparent that ρ*, E, and σ increased monotonically with the increase of LED. More specifically, when the magnitude of LED increased from 0.03 to 0.10 J/mm3, ρ* increased from 20.3% to 41.1%; En increased from 0.9 to 36.4 MPa; El increased from 4.6 to 31.8 MPa; σn exhibits an increase from 0.9 to 6.7 MPa; and σl rose from 0.1 to 4.1 MPa. The microstructure of the scaffolds was examined by SEM after the compression testing in order to investigate the mechanisms of failure. Fig. 8 shows three major failure modes, which include surface cracks (Fig. 8 (a)), fracture of bridge connections (Fig. 8 (b)) and ductile ruptures (Fig. 8 (c)). Among these failure modes, fracture of the bridges was the most common.
Fig. 4. Contour plots showing the relative frequency of different skeleton sizes (lm ) in the normal (a) and lateral plane (b), and pore sizes (lp ) in the normal (c) and lateral plane (d), respectively, on LED.
4. Discussion 4.1. Microstructure formation The manufactured scaffolds had a unique interconnected porous microstructure, as shown in Fig. 2. The unique microstructure might be formed by the powder arrangement on the powder bed during the SLS process and also the agglomeration of fine particles into larger clusters in the starting powder. Fig. 9 illustrates the principle of the microstructural formation by coalescence for two particle clusters that were irradiated with the laser beam (Fig. 9(a)). These clusters melted partially or completely in the course of the SLS process, depending on the location of the powder and the energy input from the incident laser beam (Fig. 9(b)). Once the particle clusters were molten, coalescence of the particles occurred and shaped the PHBV into islands. The islands tended to form spheres to minimize surface area due to the in-process surface tension, but this spherodisation process was retarded by the high in-process viscosity (Fig. 9(b)). However, as the available time above the melting temperature was insufficient, the material started to solidify before it was able to completely form a sphere. While the material was cooling down, it contracted which resulted in bridges that connected islands and inter-layers in locations in which islands were partially joined together (Fig. 9(c)). It is hypothesized that bridges also formed by the random arrangement of powder particles in the powder bed. Overall, the high in-process viscosity and surface tension of the molten PHBV, which was affected by the magnitude of the supplied LED, greatly contributed to the unique scaffold microstructure.
Fig. 5. The size distribution of lm and lp based on different LED were summarized in a weighted mean value lm and lp along the normal (a) and lateral (b) plane, respectively.
the lateral planes, a decline from 307.5 μm (LED of 0.03 J/mm3) to 178.7 μm (LED of 0.10 J/mm3) was found.
3.2. Mechanical characterization Fig. 6 presents three characteristic stress strain (σ − ε ) curves of scaffolds made using low, medium, and high LEDs in the normal (Fig. 6(a–c)) and lateral direction (Fig. 6 (e–g)). Fig. 6 (a–c) shows three representative stress-strain (σ − ε ) curves of scaffolds tested in the normal direction which were fabricated using different LEDs. The typical σ − ε curves of scaffolds manufactured using relatively low LEDs (0.03–0.065 J/mm3) is shown in Fig. 6 (a). The curve exhibits a relatively short linear region at the initial stage of loading, followed by the gradually increased slope until reaching a second linear region. With the progress of loading, the stress reached a critical value at which the σ − ε curve was disrupted by a drop in stress. At this point, the scaffold collapsed in a brittle manner, named as compressive strength (σn ). For
4.2. Relation between relative density, microstructure, and mechanical properties Scaffolds made for bone tissue repair can be implanted at different locations in the human body. Thus, customised shapes are crucial for fitting a scaffold implant into the anatomical defect which are likely to experience loads from different directions. Therefore, loading the 155
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
Fig. 6. Compressive stress-strain curves of PHBV scaffolds tested in the normal (a-d) and lateral direction (e-h): T1 (a & e); T8 (b & f); T9 (c & g). Combined compressive stress-strain curves in the normal (d) and lateral direction (h).
Fig. 8. Occurring failure modes: Microcracks ((a) T9: LED = 0.100 J/mm3 (P = 5 W, ss = 0.1 mm)); Brittle failure ((b) T3: LED = 0.060 J/mm3 (P = 3 W, ss = 0.1 mm), and ductile failure ((c) T3: LED = 0.060 J/mm3 (P = 3 W, ss = 0.1 mm).
Fig. 9. Particle cluster coalescence. Laser beam irradiation of the starting PHBV particle cluster (a); formation of islands after laser irradiation and densification (b); further densification, bridging, and contraction (c).
scaffolds normal and lateral to the build direction can provide important information about the mechanical behaviour of scaffolds with anisotropic microstructures in compression. The combination of SLS process parameters resulted in different levels of LEDs and thus affected significantly the relative density of the scaffolds (see Fig. 7(a)). Models of open-cell foams constructed of beams with square cross-sections established the relationship between the compressive modulus and strength to the relative density of the scaffold by power-law relationships (Gibson and Ashby, 1999; Zein et al., 2002):
Fig. 7. Dependence of relative density ρ* , elastic modulus tested in the normal direction (En ), compressive strength tested in the normal direction (σn ), elastic modulus tested in the lateral direction (El ), compressive strength tested in the lateral direction (σl ) plotted on laser energy density (LED). Results are shown in mean ± standard deviation.
156
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
and σl . This indicates that in comparison to the ideal foams the microstructures of the fabricated scaffolds in this study appeared very dissimilar as it is very heterogeneous throughout the scaffold. Thus, an increase in compressive modulus and strength cannot solely related to increase in density. It is expected that with increasing ρ* also the change in microstructure contributes to the increase in mechanical properties (Woesz et al., 2004). 4.2.1. Intra-layer-density-gradient (ILDG ) It was deduced that the density of the scaffold varies over the layer thickness. The density profile in the z-direction is conceptually illustrated by the “intra-layer density gradient” (ILDG ) as shown in Fig. 11. The density gradient is dependent on how well particles were fused by the laser. It is seen in Fig. 11 that the wider and darker the scale, the more fused material or structure is available to bear loads. Furthermore, the ILDG model states the idea how well layers were connected to each other. Detailed correlations between microstructure and ILDG are described in this section. The first characteristic microstructure was formed when using relatively low LED, such as the case of 0.03 J/mm3. The manufactured scaffolds had a low relative density of 20% and showed alternating layers of high and low density in the x-z plane as previously as described in Section 3.1. These alternating layers were formed as the energy input from the incident laser beam was insufficient to completely melt the powder layer which led to large quantities of loose powder particles throughout the scaffold. A large ILDG was formed in scaffolds using a low LED which is illustrated in Fig. 11(a). The second characteristic microstructure was formed when using relatively high LED, such as the case of 0.1 J/mm3. The scaffolds had a higher relative density of 41% and were well connected in all directions as previously described in Section 3.1. The amount of trapped powder was significantly reduced and an increased number of inter-layer connections was observed as the higher laser energy input sufficiently melted a large fraction of the in-process deposited powder layers. Thus, a small ILDG was observed as illustrated in Fig. 11(c). Relative to the characteristic scaffolds, those fabricated using an intermediate LED consisted of an intermediate microstructure in regards to relative density, island and bridge size, and intra-layer density. For example, scaffolds fabricated using moderate LED (0.067 J/mm3) had a relative density of 30%, sizable islands and bridges, and a medium ILDG , as illustrated in Fig. 11(b). The microstructure can be used to deduce which mechanics may be attributed to their resultant deformation behaviour under compression and will be discussed in the next section.
Fig. 10. Experimental data fit to the model of brittle cellular solids: Compressive modulus tested in the normal direction over elastic modulus of solid PHBV (En/ Es ) (a), compressive strength tested in the normal direction over flexural strength of solid PHBV (σn/ σf ) (b), compressive modulus tested in the lateral direction over elastic modulus of solid PHBV (El/ Es ) (c), compressive strength tested in the lateral direction over flexural strength of solid PHBV (σl/ σf ) (d), plotted against relative density of the scaffold ρ* . Each graph shows the regression curve fitting of the experimental data following Eqs. (3) and (4). Results are shown in mean ± standard deviation.
E = Cρ*n Es
(3)
σ = Cρ*n σf
(4)
where Es is the elastic modulus of the cell wall material (3762 MPa; supplementary materials S2), σf is the flexural strength for the solid PHBV (75.7 MPa; supplementary materials S2), C is a constant, ρ* the relative density. For cellular solids with open pores made of materials that fail in a brittle manner, the flexural strength needs to be used to establish the relation between σ / σf . The constants C and n are strongly depended on the microstructure of the cellular solid, whereas n typically ranges between 1 and 4. The understanding how C and n relate to each other and how these variables relate to the deformation of the microstructure is not well understood. However, both values can give indications whether the microstructure is disordered or periodic. Roberts and Garboczi (2002). The calculated data points of E / Es and σ / σf , plotted on their corresponding relative density in the normal and lateral direction, are illustrated in Fig. 10 respectively. Best curve fits were calculated using Eqs. (3) and (4) in order to see whether the experimental data of E, and σ can be used to fit the theoretical values which are based on ideal open-cell foams that have a homogeneous microstructure that consists of arrays of cubes. All curves showed a coefficient of determination (R2 ) between 0.77 and 0.86, as shown in Fig. 10. The exponent n fell into the expected range of 1–4 for σn and El . However, large n values of 5.56 and 6.64 were found for En
4.2.2. ILDG and compressive properties When compressed in the normal direction, the scaffold with a large ILDG exhibited relatively low compressive modulus and strength as schematically shown in Fig. 11(a). Correspondingly, an initial linear region was observed in its σ − ϵ curve, which was attributed to the elastic deformation of the entire porous structure. A gradual increase in stiffness followed, possibly after failure of some connections of the scaffold's structure, such as the buckling or breaking of bridges. Consequently, densification of the scaffold structure occurred, in which the buckled or broken bridges touched each other, forming new connections that were less compliant. Scaffold failure could be due to cracking Fig. 11. Intra-layer density gradients (ILDG ) in correlation to the compressive deformation behaviour tested in the normal direction. Scaffolds manufactured using low LED showed a large ILDG (a), scaffolds manufactured using intermediate LED presented a medium ILDG (b), and scaffolds using a high LED appeared to have a small ILDG (c).
157
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
initiated between two layers, followed by crack propagation across the entire structure. Scaffolds with low ILDG , compared to scaffolds with high ILDG , behaved with higher stiffness when under compression in the normal direction, and showed higher compressive strengths (Fig. 11(c)). The typical compressive σ − ε curve of these scaffolds was characterized by an initial linear-elastic deformation of islands and bridges until the point where the gradient of the σ − ε curve decreased. The reduced gradient was likely caused by the start of bending and stretching of compliant bridges. As the deformation of these bridges became more dominant, the σ − ε curve transitioned into a second linear region. Simultaneously, an increase in the scaffold diameter was observed, likely caused by deformation of inter-layer and intra-layer bridges. Scaffold failure was likely caused by crack initiation and propagation across the entire structure. Scaffolds with a medium ILDG deformed via a combination of the mechanisms of the two characteristic scaffolds, resulting in relatively linear σ − ϵ curves, medium compressive elastic moduli and strengths (Fig. 11(b)). In comparison to the deformation behaviour of scaffolds loaded in the normal direction, the deformation in the lateral direction resulted in significantly different σ − ε curves. These σ − ε curves showed three characteristic regimes: elastic, delamination, and densification. The concept of ILDG can also be applied to explain these three regimes. Fig. 12 presents the concept of ILDG when scaffolds were exposed to compression in the lateral direction. Scaffolds with low relative densities showed a microstructure which consisted of an array of layers with a high ILDG , as shown in Fig. 12(a). Thus, the scaffold was constructed by layers which weakly adhered to each other. When the scaffolds were tested in the lateral direction, these higher density regions were taking large portion of the applied load. This resulted in a significantly higher El (5.2 MPa) than En (0.9 MPa). A lower En resulted as the low density regions were compressed more than the high density regions. For scaffolds with decreasing ILDG (Fig. 12(b–c)), the connections between layers became pronounced. With smaller ILDG , En and El increased and converged gradually as the scaffolds became more isotropic. The second region of the σ − ε curves, has a plateau in between two steep linear regions which is similar to a typical curve from testing cellular solids (Gibson and Ashby, 1999). These plateaus, found in
cellular solids, are generally associated with the collapse of cell walls in an elastic, plastic, or crushing manner. In this study, inter-layer delamination occurred when the σ − ε curve reached the plateau. The layer detachment was observed in-situ when the scaffolds were loaded laterally and was particularly prevalent among scaffolds made using high ILDG . The plateau in the σ − ε curves, following the initial linear region, was likely caused by the fact that the effective porosity decreased due to compression, and the stress could not increase because of the layer-by-layer failure. Delamination was likely caused by failure of weak connections in the low density regions as indicated in Fig. 12(a). The deformation of microstructures with high ILDG resulted in very low σl (0.15 MPa), compared to σn (0.9 MPa). Scaffolds with lower ILDG were less likely to experience inter-layer delamination as illustrated in Fig. 12(c). Thus, with increasing relative densities and decreasing ILDG (Fig. 12(b–c)), the compressive strength increased up to 4.1 MPa (σl ) and 6.4 MPa (σn ). Stronger connections supported the scaffold throughout which were able to withstand the applied stresses to a greater extent, resulting in smaller plateau regions. At the end of the plateau, scaffolds were densified to a level that they lost most of their porosity, leading to a last state in which the scaffolds were compressed almost in a ‘solid’ state. This is clearly shown by the steep increase of the σ − ε curves after the plateau. Overall, the scaffolds showed an orientation dependent mechanical behaviour under compressive loading, which was attributed to their anisotropic microstructure. It should be noted that the maximum compressive modulus comes close to the lower bound of reported moduli for trabecular bone of vertebra (67 – 344 MPa). The compressive strength is within the range of reported values of the trabecular bone of tibias (2.2 −5.3 MPa) and femurs (5.6 – 8.1 MPa) (Karim et al., 2013). For many applications in bone TE, scaffolds do not have to match the mechanical properties of the anatomical implantation site; rather mechanical properties are required that can support the scaffold throughout its lifetime (Jack et al., 2009). Furthermore, fixtures can be used to provide additional load bearing structures during the regeneration phase, in which the patient is normally expected to rest.
Fig. 12. Intra-layer density gradients (ILDG ) in correlation to the compressive deformation behaviour in the lateral direction. Scaffolds with low relative density showed a high ILDG (a), scaffolds with medium relative density led to a medium ILDG (b), and scaffolds with higher relative density presented a small ILDG (c).
4.4. Future works
4.3. Porosity and pore size Another important feature of the manufactured scaffolds in this study is their highly interconnected porous structure. Such a structure would ensure the transport of nutrients and gases, and removal of metabolic waste resulting from osteogenesis. For scaffolds used in bone TE applications, pore sizes in the range of 200 – 900 μm are typically suitable to emulate the microstructure of trabecular bone (Duan et al., 2010; Mygind et al., 2007). In this study, the average areal pore size of manufactured scaffolds was in the range from 209 to 342 μm in the normal direction and 179–308 μm in the lateral direction, which could be adjusted using different values of LED. Note that the line intersection method being used in this study measured multiple intersection lengths across a single pore, bridge, or island. For a hypothetical circular pore, most of those intersection lengths did not cross the centre of the pores, which therefore led to an underestimation of pore size. Although a weighting factor was introduced to minimize the effect, the method could still be considered conservative. The scaffolds also have an excellent interconnected pore network, as successfully demonstrated by the water absorption test and shown in Fig. 3. Overall, the scaffolds had a highly interconnected porous network with pore sizes that can typically be found in scaffolds made for bone tissue engineering applications.
The scaffolds manufactured in this study show characteristic features (microstructure, mechanical properties, and interconnectivity) which suggest a positive performance in clinical trials. However, two 158
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
possible limitations shall be considered. First, as a technologically inherent side effect of a powder bed based process, the fabricated scaffolds contained a portion of trapped powder particles, particularly present on the outer surface of the scaffold. Further studies should be carried out to refine the manufacturing process to decrease the amount of loose particles. Second, the manufactured scaffolds had a heterogeneous microstructure. Finite element analysis (FEA) could help to validate the findings in respect to the compressive deformation behaviour of the unique microstructure.
Fisher, J.P., Cooke, M., Rimnac, C., 2003. Poly (propylene fumarate) and poly (dllactic-co-glycolic acid) as scaffold materials for solid and foam-coated composite tissue-engineered constructs for cranial reconstruction. Tissue Eng. 9 (3), 495–504. Dimitriou, R., Jones, E., McGonagle, D., Giannoudis, P.V., 2011. Bone regeneration: current concepts and future directions. BMC Med. 9 (66–66, 1741-7015-9-66)(PII) (21627784)(pmid) (BMC Med.). Du, Y., Liu, H., Yang, Q., Wang, S., Wang, J., Ma, J., Noh, I., Mikos, A.G., Zhang, S., 2017. Selective laser sintering scaffold with hierarchical architecture and gradient composition for osteochondral repair in rabbits. Biomaterials 137, 37–48. Duan, B., Wang, M., Zhou, W.Y., Cheung, W.L., Li, Z.Y., Lu, W.W., 2010. Three-dimensional nanocomposite scaffolds fabricated via selective laser sintering for bone tissue engineering. Acta Biomater. 6 (12), 4495–4505. Dutta, R.C., Dey, M., Dutta, A.K., Basu, B., 2017. Competent processing techniques for scaffolds in tissue engineering. Biotechnol. Adv. 35 (2), 240–250. Eosoly, S., Brabazon, D., Lohfeld, S., Looney, L., 2010. Selective laser sintering of hydroxyapatite/poly-epsilon-caprolactone scaffolds. Acta Biomater. 6 (7), 2511–2517. Gibson, L.J., Ashby, M.F., 1999. Cellular Solids: Structure and Properties. Cambridge university press. Gogolewski, S., Jovanovic, M., Perren, S., Dillon, J., Hughes, M., 1993. Tissue response and in vivo degradation of selected polyhydroxyacids: polylactides (PLA), poly (3hydroxybutyrate)(PHB), and poly (3-hydroxybutyrate-co-3-hydroxyvalerate)(PHB/ VA). J. Biomed. Mater. Res. Part A 27 (9), 1135–1148. Hagedorn, Y.-C., 2013. Additive Manufacturing of High Performance Oxide Ceramics via Selective Laser Melting (Dissertation). Hutmacher, D.W., Woodfield, T.B.F., Dalton, P.D., 2014. Scaffold Design and Fabrication. Academic Press, Oxford, pp. 311–346 (Chapter 10). Jack, K.S., Velayudhan, S., Luckman, P., Trau, M., Grondahl, L., Cooper-White, J., 2009. The fabrication and characterization of biodegradable HA/PHBV nanoparticlepolymer composite scaffolds. Acta Biomater. 5 (7), 2657–2667. Karim, L., Hussein, A.I., Morgan, E.F., Bouxsein, M.L., 2013. The Mechanical Behavior of Bone, book Section 19 - The Mechanical Behavior of Bone, 4th ed. Academic Press, San Diego, pp. 431–452. Kumarasuriyar, A., Jackson, R., Grondahl, L., Trau, M., Nurcombe, V., Cool, S., 2005. Poly (beta-hydroxybutyrate-co-beta-hydroxyvalerate) supports in vitro osteogenesis. Tissue Eng. 11 (7–8), 1281–1295. Leong, K.F., Cheah, C.M., Chua, C.K., 2003. Solid freeform fabrication of three-dimensional scaffolds for engineering replacement tissues and organs. Biomaterials 24 (13), 2363–2378. Lohfeld, S., Tyndyk, M.A., Cahill, S., Flaherty, N., Barron, V., McHugh, P.E., 2010. A method to fabricate small features on scaffolds for tissue engineering via selective laser sintering. J. Biomed. Sci. Eng. 03 (02), 10. Mazzoli, A., 2013. Selective laser sintering in biomedical engineering. Med. Biol. Eng. Comput. 51 (3), 245–256. Mygind, T., Stiehler, M., Baatrup, A., Li, H., Zou, X., Flyvbjerg, A., Kassem, M., Buenger, C., 2007. Mesenchymal stem cell ingrowth and differentiation on coralline hydroxyapatite scaffolds. Biomaterials 28 (6), 1036–1047. Nam, Y.S., Park, T.G., 1999. Porous biodegradable polymeric scaffolds prepared by thermally induced phase separation. J. Biomed. Mater. Res. 47 (1), 8–17. No, Y.J., Li, J.J., Zreiqat, H., 2017. Doped calcium silicate ceramics: a new class of candidates for synthetic bone substitutes. Materials 10 (2), 153. O'Brien, F.J., Harley, B.A., Yannas, I.V., Gibson, L.J., 2005. The effect of pore size on cell adhesion in collagen-gag scaffolds. Biomaterials 26 (4), 433–441. O'Brien, F.J., 2011. Biomaterials and scaffolds for tissue engineering. Mater. Today 14 (3), 88–95. Partee, B., Hollister, S.J., Das, S., 2005. Selective laser sintering process optimization for layered manufacturing of CAPA 6501 polycaprolactone bone tissue engineering scaffolds. J. Manuf. Sci. Eng. 128 (2), 531–540 (10.1115/1.2162589). Pereira, T.F., Silva, M.A.C., Oliveira, M.F., Maia, I.A., Silva, J.V.L., Costa, M.F., Thire, R.M.S.M., 2012. Effect of process parameters on the properties of selective laser sintered poly(3-hydroxybutyrate) scaffolds for bone tissue engineering. Virtual Phys. Prototyp. 7 (4), 275–285. Ramay, H.R.R., Zhang, M., 2004. Biphasic calcium phosphate nanocomposite porous scaffolds for load-bearing bone tissue engineering. Biomaterials 25 (21), 5171–5180. Roberts, A.P., Garboczi, E.J., 2002. Elastic properties of model random three-dimensional open-cell solids. J. Mech. Phys. Solids 50 (1), 33–55. Shahzad, K., Deckers, J., Kruth, J.-P., Vleugels, J., 2013. Additive manufacturing of alumina parts by indirect selective laser sintering and post processing. J. Mater. Process. Technol. 213 (9), 1484–1494. Shirazi, F.S., Mehrali, M., Oshkour, A.A., Metselaar, H.S.C., Kadri, N.A., Abu Osman, N.A., 2014. Mechanical and physical properties of calcium silicate/alumina composite for biomedical engineering applications. J. Mech. Behav. Biomed. Mater. 30, 168–175. Shuai, C., Sun, H., Gao, C., Feng, P., Guo, W., Yang, Y., Zhao, M., Yang, S., Yuan, F., Peng, S., 2017. Mechanical reinforcement of bioceramics scaffolds via fracture energy dissipation induced by sliding action of MoS2 nanoplatelets. J. Mech. Behav. Biomed. Mater. 75, 423–433. Sombatmankhong, K., Sanchavanakit, N., Pavasant, P., Supaphol, P., 2007. Bone scaffolds from electrospun fiber mats of poly(3-hydroxybutyrate), poly(3-hydroxybutyrate-co3-hydroxyvalerate) and their blend. Polymer 48 (5), 1419–1427. Sultana, N., Khan, T.H., 2012. In vitro degradation of phbv scaffolds and nHA/PHBV composite scaffolds containing hydroxyapatite nanoparticles for bone tissue engineering. J. Nanomater. 2012, 12. Sultana, N., Wang, M., 2007. Fabrication of HA/PHBV composite scaffolds through the emulsion freezing/freeze-drying process and characterisation of the scaffolds. J. Mater. Sci.: Mater. Med. 19 (7), 2555. Thomson, R.C., Yaszemski, M.J., Powers, J.M., Mikos, A.G., 1996. Fabrication of biodegradable polymer scaffolds to engineer trabecular bone. J. Biomater. Sci. Polym. Ed.
5. Conclusions In this study, interconnected and highly porous (59–80%) PHBV scaffolds were successfully manufactured using SLS without utilizing pre-designed internal pore architectures. The manufactured scaffolds had a unique microstructure that consisted of pores, islands, and bridges. The average weighted areal pore size of the scaffolds ranged between 209 and 342 μm in the normal direction and 179 – 307μm in the lateral direction. The maximum areal pore size was measured at 680 μm. The formation of such porous networks was attributed to the agglomeration of particles in the starting powder, the in-process surface tension and viscosity of the molten PHBV, and the SLS process parameters. The use of greater LED resulted in denser scaffold microstructures with improved mechanical properties. The compressive elastic modulus of scaffolds could reach up to 36.4 and 31.8 MPa with a relative porosity of 59% when tested in normal and lateral direction respectively. Scaffolds with the same relative porosity showed a compressive strength of 6.7 MPa (normal direction) and 4.1 MPa (lateral direction). The porosity and mechanical properties of the scaffold could be adjusted by varying the level of LED used within the SLS process. Overall, the manufactured scaffolds with their unique microstructure present great potential for further in-vitro and in-vivo testing. Acknowledgments S.H.D. thanks the University of Queensland (UQ) and the School of Mechanical & Mining Engineering for the tuition fee waiver and living allowance scholarship. The authors would like to thank the Enginnering @ UQ SEED grants for financial support and acknowledge the facilities, and the scientific and technical assistance of the Australian Microscopy & Microanalysis Research Facility at the Centre for Microscopy and Microanalysis (CMM), UQ. M. Dargusch would like to acknowledge the support of the ARC Research Hub for Advanced Manufacturing of Medical Devices (IH150100024). Appendix A. Supplementary data Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jmbbm.2018.05.007. References Bukharova, T., Antonov, E., Popov, V., Fatkhudinov, T.K., Popova, A., Volkov, A., Bochkova, S., Bagratashvili, V., Gol' Dshtein, D., 2010. Biocompatibility of tissue engineering constructions from porous polylactide carriers obtained by the method of selective laser sintering and bone marrow-derived multipotent stromal cells. Bull. Exp. Biol. Med. 149 (1), 148–153. Cao, W., Hench, L.L., 1996. Bioactive materials. Ceram. Int. 22 (6), 493–507. Caulfield, B., McHugh, P.E., Lohfeld, S., 2007. Dependence of mechanical properties of polyamide components on build parameters in the SLS process. J. Mater. Process. Technol. 182 (1), 477–488. Chia, H.N., Wu, B.M., 2015. Recent advances in 3D printing of biomaterials. J. Biol. Eng. 9 (1), 4. Chua, C.K., Leong, K.F., Tan, K.H., Wiria, F.E., Cheah, C.M., 2004. Development of tissue scaffolds using selective laser sintering of polyvinyl alcohol/hydroxyapatite biocomposite for craniofacial and joint defects. J. Mater. Sci.: Mater. Med. 15 (10), 1113–1121. Chua, C.K., Yeong, W.Y., An, J., 2017. Special issue: 3D printing for biomedical engineering. Materials. Dean, D., Topham, N.S., Meneghetti, S.C., Wolfe, M.S., Jepsen, K., He, S., Chen, J.E.-K.,
159
Journal of the Mechanical Behavior of Biomedical Materials 84 (2018) 151–160
S.H. Diermann et al.
Wu, J., Xue, K., Li, H., Sun, J., Liu, K., 2013. Improvement of PHBV scaffolds with bioglass for cartilage tissue engineering. PLoS ONE 8 (8), e71563. Yang, Y., Kang, Y., Sen, M., Park, S., 2011. Biomaterials for Tissue Engineering Applications. A Review of the Past and Future Trends., Book Section Bioceramics in Tissue Engineering. Springer, pp. 179–207. Zein, I., Hutmacher, D.W., Tan, K.C., Teoh, S.H., 2002. Fused deposition modeling of novel scaffold architectures for tissue engineering applications. Biomaterials 23 (4), 1169–1185.
7 (1), 23–38. Williams, J.M., Adewunmi, A., Schek, R.M., Flanagan, C.L., Krebsbach, P.H., Feinberg, S.E., Hollister, S.J., Das, S., 2005. Bone tissue engineering using polycaprolactone scaffolds fabricated via selective laser sintering. Biomaterials 26 (23), 4817–4827. Woesz, A., Stampfl, J., Fratzl, P., 2004. Cellular solids beyond the apparent density - an experimental assessment of mechanical properties. Adv. Eng. Mater. 6 (3), 134–138. Wu, C., Chang, J., Zhai, W., Ni, S., 2007. A novel bioactive porous bredigite (Ca7MgSi4O16) scaffold with biomimetic apatite layer for bone tissue engineering. J. Mater. Sci.: Mater. Med. 18 (5), 857–864.
160