Materials Science and Engineering C 33 (2013) 453–460
Contents lists available at SciVerse ScienceDirect
Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec
The effect of processing parameters and solid concentration on the mechanical and microstructural properties of freeze-casted macroporous hydroxyapatite scaffolds S. Farhangdoust ⁎, A. Zamanian, M. Yasaei, M. Khorami Nanotechnology and Advance Materials Department, Materials and Energy Research Center, Karaj, Alborz, Iran
a r t i c l e
i n f o
Article history: Received 2 July 2012 Received in revised form 28 July 2012 Accepted 17 September 2012 Available online 22 September 2012 Keywords: Hydroxyapatite Scaffold Freeze-casting Unidirectional Sintering Cooling rate
a b s t r a c t The design and fabrication of macroporous hydroxyapatite scaffolds, which could overcome current bone tissue engineering limitations, have been considered in recent years. In the current study, controlled unidirectional freeze-casting at different cooling rates was investigated. In the first step, different slurries with initial hydroxyapatite concentrations of 7–37.5 vol.% were prepared. In the next step, different cooling rates from 2 to 14 °C/min were applied to synthesize the porous scaffold. Additionally, a sintering temperature of 1350 °C was chosen as an optimum temperature. Finally, the phase composition (by XRD), microstructure (by SEM), mechanical characteristics, and the porosity of sintered samples were assessed. The porosity of the sintered samples was in a range of 45–87% and the compressive strengths varied from 0.4 MPa to 60 MPa. The mechanical strength of the scaffolds increased as a function of initial concentration, cooling rate, and sintering temperature. With regards to mechanical strength and pore size, the samples with the initial concentration and the cooling rate of 15 vol.% and 5 °C/min, respectively, showed better results. © 2012 Elsevier B.V. All rights reserved.
1. Introduction One of the key components for being successful in bone tissue engineering is developing a scaffold with an appropriate architecture for inducing cell proliferation, migration and differentiation. However, the proliferation of osteoblasts, vascular ingrowth and bone formation requires an interconnected pore system and a controlled pore size, as well as a desirable shape and orientation [1–3]. In recent years, several techniques have been employed to fabricate porous scaffolds, i.e., solvent casting/particulate leaching [4–7], gas foaming [8–10], rapid prototyping [11–13], extrusion [14,15], phase separation [16–19] and freeze-casting [3,20–22]. In the pursuit of such processing routes, freeze-casting has attracted considerably more focus in the last few years. Freeze-casting is environmentally friendly and cost effective because of its use of a minimum concentration of organic additive and its use of water as a suspension liquid. In this method, controlling the parameters, such as initial concentration and cooling rate, leads to an interconnected, porous structure. Moreover, the pore size, shape and orientation are controllable and possess the potential to be repeated, which are characteristics not observed with other methods. The freeze-casting technique seems to be rather versatile; furthermore, the use of a liquid solvent (mostly water) as a pore-forming agent is a promising asset [20,23]. Calcium phosphate ceramics and macroporous hydroxyapatite (HA, Ca10(PO4)6(OH)2), in particular, could be used to manufacture ideal scaffolds due to their biocompatibility, as well as their similarity ⁎ Corresponding author. Tel.: +98 912 3167022; fax: +98 26 36204139. E-mail address: [email protected] (S. Farhangdoust). 0928-4931/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.msec.2012.09.013
to the mineral component of bone. While HA shows desirable properties for applications in tissue engineering, it cannot be used in loadbearing applications due to its poor mechanical properties when compared with other materials, such as titanium [24–26]. During recent years, unidirectional freeze-casting has emerged to solve this problem. Deville et al. [3] prepared highly porous HA materials that exhibited extraordinarily high compressive strengths, up to 145 MPa, with a porosity that reached 47%. They suggested that these HA-based materials could be used in load-bearing applications, such as artificial bone. With this technique, the greatest influence on the porosity and the pore size distribution of the ceramic body was exerted over the thickness of the samples by the solid concentration of the suspension, the temperature gradient and the ice crystal growth rate. Little research has been conducted to determine the effective parameters on the process of freeze-casting HA scaffolds. Furthermore, current research hardly covers the entirety of the parameters, i.e., the initial concentration of the slurry, the cooling rate and the sintering temperature. Therefore, having a good understanding about the aforementioned parameters is essential in realizing the microstructure and the mechanical properties of freeze-casted HA scaffolds. The present investigation is focused on the influence of the solid concentration, cooling rates and sintering temperature on the quality of the synthetic scaffolds. 2. Materials and methods 2.1. Fabrication procedure Controlled, unidirectional freeze-casting was used to create porous HA scaffolds. In the first step, different slurries with initial HA concentrations
454
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
of 7–37.5 vol.% were prepared. To prepare stable slurries, a small amount (4 wt.% of the HA content) of commercially available dispersant (Dolapix CE 64, Zschimmer & Schwarz, Lahnstein, Germany) was added to distilled water. Then, HA powder (2196, Merck KGaA, Darmstadt, Germany) with a median particle size (d50) and specific surface area of 1.69 μm and 11.06 g/m2, respectively, was gradually added to the distilled water to prevent agglomeration. After that, polyvinyl alcohol (PVA, Mw = 15000, Merck, Darmstadt, Germany) was added as a binder at 4 wt.% of the HA content and followed by vigorous stirring at 1000 rpm for 1 h. Subsequently, to further stabilize the slurry, the pH of the slurry was determined and adjusted to 10 by the gradual addition of a 1 M NaOH solution. To remove air bubbles before casting, the prepared slurry was placed in a vacuum oven for 30 min at a pressure of 0.02 MPa. Freeze-casting of the prepared slurries was performed by pouring them into a PTFE mold with an inner diameter of 20 mm. The mold was placed on copper, where the temperature was controlled by using liquid nitrogen and a ring heater connected to a PID controller. The cooling rates utilized in this study ranged from 2 to 14 °C/min. After careful removal of the samples from the mold, the frozen samples were dried in the freeze-dryer (Beta 1-2LD plus, Martin Christ GmbH, Germany) at a low temperature of − 55 °C and pressure of 2.1 Pa. Subsequently, green bodies were sintered by double step sintering to prevent the grain growth [27] for 2 h with a heating rate of 3 °C/min; the first step of sintering was conducted at 600 °C, and the second step was conducted at different temperatures in the range of 1250–1350 °C. 2.2. Characterization 2.2.1. Phase analysis Phase composition of the sintered samples was evaluated using an automated X-ray diffractometer (XRD, Philips PW3710). Cu-Kα radiation was used under the operating conditions of 40 kV and 30 mA. XRD diagrams were constructed from 2θ data in the range of 20–60°. 2.2.2. Microstructure Microstructure analysis of the samples was determined by using a scanning electron microscope (SEM, Stereoscan S 360-Leica Cambridge, England). Due to the poor electrical conductivity of the samples, a thin layer of gold was coated onto the surfaces of the scaffolds before testing. 2.2.3. Mechanical properties For the compressive strength test, samples with a diameter of ~ 15 mm and a height of ~ 20 mm were loaded onto a crosshead and pulled at a speed of 1 mm/min using a screw-driven load frame (Instron 5565, Instron Corp., Canton, MA). During the compressive strength tests, the stress and strain responses of the samples were monitored. Five samples were tested to obtain an average value and its standard deviation.
2.3.5. Pore size With the HA scaffolds being so anisotropic, pore sizes were determined in both the long and short axes. Five samples were studied, with 50 measurements conducted for each sample. 2.2.6. Shrinkage Longitudinal and circumferential shrinkages were calculated using the following expressions [29]: Sc ¼ ðD0 −Df Þ=D0
ð3Þ
SL ¼ ðL0 −Lf Þ=L0
ð4Þ
where Sc and SL are the circumferential and longitudinal shrinkages, respectively, D0 and Df are the initial and final diameters, respectively, and L0 and Lf are the initial and final lengths, respectively. 3. Results and discussion 3.1. Phase analysis The XRD patterns of the initial powder and the sintered sample (sintered at 1350 °C) are shown in Fig. 1, which can be completely indexed with HA (JCPDS#76-0694). HA was the only phase in these patterns, and no secondary phase was found after sintering at 1350 °C. 3.2. Microstructure A lamellar HA scaffold with unidirectional aligned channels can be observed in Fig. 2. Macroscopic aligned pores were formed almost uniformly over the entire sample. In fact, the porosity of the sintered materials was a replica of the ice structure before sublimation. These pores were generated during the sublimation of the ice. The corresponding horizontal cross sections are shown in Fig. 2 and reveal the arrangement of the ice crystals in the direction parallel to the solidification direction. The crystals were lamellar but were arranged in domains with similar orientations. The orientation of each domain could be related to the original nucleation conditions. According to the microstructure of a vertical cross-section in Fig. 3, the scaffolds can be divided into three distinct zones in the direction of solidification, as reported by Deville et al. [3]. In zone 1, the closest zone to the initial cold finger, no microporosity was observed and the material was dense. In zone 2, the material was characterized by a cellular morphology. Finally, in the upper zone (zone 3), the ceramic was lamellar with long parallel pores aligned with the movement direction of the ice front. Parameters such as the particle size of
2.2.4. Porosity The total porosity (Pt) of the sintered samples was determined using the following expression [28]: Pt ¼ 100 ð1−Db =Dt Þ
ð1Þ
where Dt is the theoretical density of the powder and Db is the bulk density. Db was calculated using the following expression: Db ¼ m=v
ð2Þ
where m is the weight of the specimen and V is its volume. Five samples were measured to obtain an average value and its standard deviation.
Fig. 1. XRD pattern of the initial powder and the sintered sample sintered at 1350 °C.
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
455
Fig. 2. Lamellar scaffold with unidirectional aligned channels with an initial concentration of 15 vol.% HA.
hydroxyapatite affect the thickness of these zones which is being investigated in ongoing research. It is well known that to form unidirectional porous structures, the particles must be rejected from the solidification front and collected between the arms of the solidification front. At the very beginning
of solidification, the interface was planar (Zone 1) and needed to somehow undergo a transition towards an irregular morphology, i.e., cellular (zone 2) and lamellar (zone 3). This transition occurred due to the perturbation of advancing solidification behind. Until now, two different mechanisms have been proposed. One mechanism
Fig. 3. Microstructure of a vertical cross-section of scaffold with an initial concentration of 22.5 vol.% HA. a) Scaffolds can be divided into three distinct zones with respect to the direction of solidification, which is labeled by Greek numbers, and is magnified in the black squares. b) Zone 1, the closest to the initial cold finger. c) Zone 2, characterized by a cellular morphology. d) Zone 3, where the ceramic is lamellar with long parallel pores aligned with the movement direction of the ice front.
456
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
is the inherent thermodynamic instability of the interface, known as the Mullins–Sekerka instability [30]. In this mechanism, the development of the instability is based on supercooling effects building up with the solute rejection ahead of the interface. Another mechanism, proposed by Haji, is related to the presence of the particles [28]. In this case, the instability is attributed to the reversal of the thermal gradient in the liquid ahead of the interface and behind the particle. This instability occurs at pulling velocities below the threshold for the onset of the Mullins–Sekerka instability. Previously, it was suspected that the Mullins–Sekerka mechanism leads to progressive evolution of the interface morphology from flat to cellular and, finally, to lamellar. Recently, in situ X-ray radiography and tomography observations by Deville et al. [31,32] revealed that two types of crystals (Fig. 4) were initially present: a first population of lamellar crystals, with their main axis oriented along the cooling direction, and a second population of crystals, more or less lamellar, but oriented predominantly in the radial direction (perpendicular to the cooling direction). When the temperature at the copper surface reached the nucleation point, homogeneous (spatially speaking) nucleation of ice crystals occurred. Crystals with a random crystallographic orientation were nucleating and growing very fast so that all the particles were initially entrapped (creating zone 1) because they did not have enough time to be repelled by the ice crystals. As the crystals started to grow along the cooling axis, the interface velocity diminished, and the particle redistribution progressively took place as particles started to be repelled by the interface. While the crystals grow larger and
larger, the particle packing between the crystals becomes more and more efficient. Therefore, the entrapped particle fraction progressively diminishes. Crystals in all orientations grew along the cooling axis in zone 2, which can be called the transition zone. It is well known that the growth along the c-axis of hexagonal ice is theoretically two orders of magnitude slower than the growth along the a-axis or b-axis (anisotropy of crystal growth kinetics). Both populations grew faster in directions perpendicular to the c-axis of the hexagonal ice structure but with a different orientation with regards to the temperature gradient. As the conditions are dictated by the temperature gradient, the second population of crystals (horizontally growing) stopped growing, leaving only the first population of crystals that were growing into the suspension. In this stage, the crystals were growing with steadystate conditions that initiated and formed zone 3. In zone 3, the continuous crystals growing along the solidification direction were noticeable. The specimens with an initial concentration of 7% HA, which were frozen at a cooling rate of 2 °C/min, are shown in Fig. 5a. On the internal walls of the lamellae dendritic, a branch-like structure was observed. Some ceramic bridges and linking adjacent plates were also observed (Fig. 5b). The morphology of these numerous fine features is very different from the dendrites that cover the ceramic lamellae. This difference suggests another formation mechanism. Deville et al. [33] proposed that these features might be formed because of the specific conditions encountered during the slow freezing of highly concentrated solutions. However, the results of the current study are not in agreement with the hypothesis of Deville et al. because these bridges not only can be observed in low concentration slurries but can also be detected in
Fig. 4. a) Two types of crystals that are present from the beginning: a first population of lamellar crystals, with their main axis oriented along the cooling direction (left), and a second population of crystals, more or less lamellar, but oriented predominantly in the radial direction or perpendicular to the cooling direction (right). b) Hexagonal ice crystals: growth along the c-axis of hexagonal ice is theoretically two orders of magnitude slower than growth along either the a- or b-axis (anisotropy of crystal growth kinetics).
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
457
below and up to 1300 °C. In fact, at the sintering temperature of 1350 °C, the compressive strength is about three orders of magnitude larger than the compressive strength at 1300 °C. At 1350 °C, the porosity and shrinkage are ~75% and ~20%, respectively. It is worth mentioning that the enhancement in mechanical properties occurs because of porosity attenuation. Werner et al. [2] sintered HA at temperatures between 1250 and 1450 °C and reported formation of α-TCP at 1400 °C. Meanwhile, Prokopiev [42] reported sintering at a temperature of 1280 °C as the plateau stage, and Deville [3] chose 1325 °C as the optimum point. In the current study, the plateau stage at sintering temperatures up to 1350 °C was not reached. 1350 °C was chosen as the most favorable sintering temperature because of the desirable compressive strength and porosity. It can be observed that longitudinal shrinkage, parallel to solidification direction, is 1.05± 0.2%, which is less than the circumferential shrinkage (perpendicular to the cooling
Fig. 5. The internal walls of the lamellae of the scaffolds with an initial concentration of 7.5 vol.% HA. a) Dendritic and branch-like features. b) Ceramic bridges and linking adjacent plates.
fast-frozen samples. Fig. 5b shows the specimens with an initial concentration of 15% HA which were frozen at a cooling rate of 11 °C/min. It seems that the addition of PVA caused the formation of these bridges. Pekor et al. [34] investigated the effect of PVA on the microstructure of freeze-casted alumina, and they concluded that when 5 wt.% of PVA was added, due to the effect of constitutional supercooling the fine secondary dendrites appear to coarsen. It was previously shown that the particles themselves might induce morphological transitions, such as dendrite tip splitting or healing [35]. Moreover, entrapped particles in ice might induce such morphology [36]. In addition, the enhancement in the mechanical strength of the scaffolds was related to the morphological changes induced by these bridges. In this way, different mechanisms have been proposed. Munch et. al [37] suggested that the presence of stiff ceramic bridges between the grains with micrometer and sub-micrometer dimensions promotes controlled sliding and sliding interferences between the rough ceramic interlayers, thereby enhancing the toughness through extremely efficient energy dissipation. By limiting sliding, they provided very effective toughening mechanisms in natural and synthetic materials. Luz et al. [38] believed that the existence of such inorganic bridges reinforces the weak interfaces and that these interfaces become just suitable for the crack to extend itself. It is worth mentioning that evolving these bridges is one of the toughening mechanisms employed by nacre and synthetic bio-inspired materials [39–41]. 3.3. Effect of sintering temperature Sintering temperature plays a key role in the mechanical and physical characteristics of the scaffold. Fig. 6 shows the effect of sintering temperature on the compressive strength, total porosity and shrinkage of the samples with an initial concentration of 15 vol.% HA, which were solidified at a rate of 8 °C/min. The mechanical strength and shrinkage increased as a function of temperature, while temperature had an inverse effect on the total porosity. According to the compressive strength results, the samples did not sinter completely at temperatures
Fig. 6. Effect of sintering temperature on the compressive strength, total porosity and shrinkage of samples with an initial concentration of 15 vol.% HA, solidified at a rate of 8 °C/min.
458
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
direction) on average. It seems that there was more empty space (spatially speaking) in the direction of circumferential shrinkage. As a result, shrinkage occurred more readily in the circumferential direction rather than in the longitudinal direction. 3.4. Effect of initial concentration of slurry Fig. 7 shows the effect of the initial concentration of the slurry on the total porosity of the scaffolds. A wide range of porosity was obtained, ranging from 45 to 87%. Porosity has a linear relation with the initial concentration, which can suggest the final porosity. As a result, the strength is completely adjustable. The compressive strength (Fig. 7) increased sharply from 0.4 MPa to ~60 MPa, while the porosity decreased constantly. The compressive strengths that were measured were ~20 MPa with 68% porosity and ~60 MPa with 48% porosity. Although higher compressive strengths, up to 140 MPa, were reported in one study [3], our results were compared better with other studies [22,43,44]. It should be noted that the tested specimens in the aforementioned studies were smaller (3 mm× 4 mm × 4 mm) than those used in the current study (15 mm in diameter and 20 mm in height). Actually, the initial concentration has a key role in determining the lamella and the thickness of the pores in the samples. Higher initial concentration resulted in thicker lamella and thinner porosity. Fig. 7 shows the pore size of the scaffold in the long and short axis as a function of the initial concentration of the slurry. The reported sizes are the mean size of 3 samples with a confidence level of 95%, which was measured in 20 different places. The pore size in the long axis of the scaffold declined dramatically with increases in slurry concentration (falling from ~ 180 μm at 7.5% HA to ~ 8 μm at 37.5% HA). Meanwhile, the pore size in the short axis decreased monotonically from ~ 42 μm at 7.5% HA to ~ 5 μm at 37.5% HA. Fig. 8 shows three different samples with initial concentrations of 7%, 18% and 22% HA. Lamellar space and pore thickness decreased with increased slurry concentration. As a result, the compressive strength increased with some limitations. On one hand, with low initial concentrations (lower than 10 vol.%), the ceramic walls of the green body were weaker and more difficult to handle. In addition, it was difficult to achieve homogeneity through the entire length of the specimens. On the other hand, with high initial concentrations (larger than 35 vol.%), the particles could not be rejected by the ice; therefore, the lamellar structure and interconnectivity were not achieved [33]. 3.5. Effect of cooling rate Fig. 9 shows the compressive strength and the large and short axes of the samples versus the cooling rate. The compressive strength increased monotonically while the lamellar space decreased as the result of an increased cooling rate. The compressive strengths were 4.1, 6.4 and 9.5 MPa at the cooling rates of 2, 8 and 14 °C/min, respectively. With an increase in the freezing rate, larger temperature gradients resulted in smaller pore sizes, and as a result, the strength of the ceramic bodies increased. Actually, the thickness of the ice crystals is strongly dependent on the speed of the solidification front. Faster freezing velocities result in larger supercooling in front of the growing crystals that will influence the crystal thickness. In addition, as faster growth is imposed in the direction of the temperature gradient, lateral growth along the c-axis is increasingly limited, resulting in thinner lamellae. At slow solidification rates, the particles easily diffuse away from the interface and the temperature of the suspension, ahead of the interface, is always warmer than the freezing temperature. At faster solidification rates, the concentration and concentration gradient increase at the interface. When the concentration gradient at the interface is steep enough that the gradient in the freezing temperature is larger than the temperature gradient, the suspension ahead of the interface is below its freezing temperature (constitutionally supercooled). In analogy with binary alloys, constitutional supercooling is closely related
Fig. 7. Effect of the initial concentration of the slurry on the total porosity and the compressive strength.
to morphological instability. By controlling the temperature at the base of the suspension, it is possible to control the extension of crystals [33,45]. As mentioned previously, higher front velocities lead to thinner colloidal configurations because more particles are engulfed before any solid formation. Some of the engulfed particles form bridges that link neighboring colloidal domains, thus mimicking the inorganic bridges that are found in nacre. We found that higher front velocities result in rougher surfaces because the particles have less time to rearrange when they are being pushed. This effect is more pronounced in higher initial volume fractions.
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
459
Fig. 9. Compressive strength and the large and short axes of the scaffold versus the cooling rate.
Fig. 8. Three different samples with initial concentrations of a) 7%, b) 18% and c) 22% HA.
patterns, roughness, etc. The results show the beneficial effect of surface roughness, patterns and architecture, which seem to improve bone ingrowth and tissue formation [49–52]. It seems that by adjusting the initial concentration and the cooling rate of slurry, the size and shape of these features can be regulated to match more favorable dimensions of guidance patterns, improving the osteoconduction and osteoinduction characteristics of the scaffold.
3.6. Suitability as a bone substitute 4. Conclusions The porosity and pore size of biomaterial scaffolds play a critical role in bone formation both in vitro and in vivo. These morphological features influence osteogenesis as well as the mechanical properties of the scaffolds [46,47]. Lower porosity stimulates osteogenesis in vitro by suppressing cell proliferation and forcing cell aggregation. In contrast, greater bone ingrowth is achieved in vivo with higher porosity and larger pore size. However, this trend results in diminished mechanical properties, thereby setting an upper functional limit for pore size and porosity. Thus, a balance must be reached. Based on previous studies, the minimum requirement for pore size is considered 100 μm due to cell size [48]. Thus, the sample with an initial concentration of 15 vol.% HA and a cooling rate of 5 °C/min, along with a porosity of 76± 2.75%, a compressive strength of 5.26 ±0.52 MPa and a pore size of 88.4 ± 7.9 μm, was chosen as an optimum sample. It is worth mentioning that dendritic features on the internal surface of the walls might act as a guiding pattern for cell growth, which would improve the osteoconduction characteristics [2]. Considerable attention has been paid to the relationships between cellular response and morphological features, such as mesostructure,
Based on the experimental study of HA freeze-casted scaffolds with different initial concentration, cooling rates and sintering conditions, the following points are concluded: 1. Porous scaffolds with a total porosity of 45–87% and compressive strengths between 0.4 and 60 MPa are obtained by freeze-casting HA, followed by the sublimation and sintering of the green bodies. 2. The porosities are unidirectionally aligned along the entire lengths of the samples. The formation and morphology of these porosities can be controlled via the initial concentrations of the HA slurry and the cooling rate. 3. The bridges that connect the adjacent plates improve the mechanical strength of the scaffolds. 4. The synthesized scaffolds can be employed as bone-repairing material due to the lamellar structure of the macropores and the presence of dendritic branch-like features. 5. This method can be easily utilized to synthesize scaffolds with other calcium phosphates or any other materials.
460
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
References [1] J.S. Son, M. Appleford, J.L. Ong, J.C. Wenke, J.M. Kim, S.H. Choi, D.S. Oh, J. Control. Release 153 (2011) 133–140. [2] J. Werner, B. Linner-Krčmar, W. Friess, P. Greil, Biomaterials 23 (2002) 4285–4294. [3] S. Deville, E. Saiz, A.P. Tomsia, Biomaterials 27 (2006) 5480–5489. [4] D. Sin, X. Miao, G. Liu, F. Wei, G. Chadwick, C. Yan, T. Friis, Mater. Sci. Eng. C 30 (2010) 78–85. [5] J. Wei, F. Chen, J.-W. Shin, H. Hong, C. Dai, J. Su, C. Liu, Biomaterials 30 (2009) 1080–1088. [6] S. Zadegan, M. Hosainalipour, H.R. Rezaie, H. Ghassai, M.a. Shokrgozar, Mater. Sci. Eng. C 31 (2011) 954–961. [7] S.K. Misra, T.I. Ansari, S.P. Valappil, D. Mohn, S.E. Philip, W.J. Stark, I. Roy, J.C. Knowles, V. Salih, A.R. Boccaccini, Biomaterials 31 (2010) 2806–2815. [8] Y.-M. Lim, H.-J. Gwon, J. Shin, J.P. Jeun, Y.C. Nho, Ind. Eng. Chem. Res. 14 (2008) 436–441. [9] C. Ji, N. Annabi, M. Hosseinkhani, S. Sivaloganathan, F. Dehghani, Acta Biomater. 8 (2011) 570–578. [10] A. Salerno, S. Zeppetelli, E.D. Maio, S. Iannace, P.a. Netti, Compos. Sci. Technol. 70 (2010) 1838–1846. [11] I.-K. Jun, Y.-H. Koh, S.-H. Lee, H.-E. Kim, J. Mater. Sci. Mater. Med. 18 (2007) 1071–1077. [12] C.K. Chua, K.F. Leong, K.H. Tan, F.E. Wiria, C.M. Cheah, Design 5 (2004) 1113–1121. [13] M. Schumacher, F. Uhl, R. Detsch, U. Deisinger, G. Ziegler, J. Mater. Sci. Mater. Med. 21 (2010) 3039–3048. [14] H.Y. Yang, I. Thompson, S.F. Yang, X.P. Chi, J.R.G. Evans, R.J. Cook, J. Mater. Sci. Mater. Med. 19 (2008) 3345–3353. [15] H.-J. Yen, C.-S. Tseng, S.-H. Hsu, C.-L. Tsai, Biomed. Microdevices 11 (2009) 615–624. [16] J.M. Holzwarth, P.X. Ma, Biomaterials 32 (2011) 9622–9629. [17] J.P. Kennedy, S.P. McCandless, R. a Lasher, R.W. Hitchcock, Biomaterials 31 (2010) 1126–1132. [18] a S. Rowlands, S. a Lim, D. Martin, J.J. Cooper-White, Biomaterials 28 (2007) 2109–2121. [19] K. Shin, B. Kim, J. Kim, T. Park, J. Nam, D. Lee, Polymer 46 (2005) 3801–3808. [20] T. Moritz, H.-J. Richter, J. Eur. Ceram. Soc. 27 (2007) 4595–4601. [21] K. Araki, J.W. Halloran, J. Am. Ceram. Soc. 87 (2005) 2014–2019. [22] E. Landi, F. Valentini, A. Tampieri, Acta Biomater. 4 (2008) 1620–1626.
[23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52]
S. Deville, Adv. Eng. Mater. 10 (2008) 155–169. C.V. Ragel, L.M. Rodr, Biomaterials 23 (2002) 1865–1872. F.N. Oktar, G. Göller, Ceram. Int. 28 (2002) 617–621. K. So, S. Fujibayashi, M. Neo, Y. Anan, T. Ogawa, T. Kokubo, T. Nakamura, Biomaterials 27 (2006) 4738–4744. M.E. Launey, et al., Acta Mater. 57 (2009) 2919–2932. M. Khorami, S. Hesaraki, T. Ebadzadeh, S. Farhangdoust, A. Zamanian, Key Eng. Mater. 493–494 (2011) 231–235. A. Heunisch, A. Dellert, A. Roosen, J. Eur. Ceram. Soc. 30 (2010) 3397–3406. W.W. Mullins, R.F. Sekerka, J. Appl. Phys. 35 (1964) 444. L. Hadji, Eur. Phys. J. 37 (2003) 85–89. S. Deville, E. Maire, A. Lasalle, A. Bogner, C. Gauthier, J. Leloup, C. Guizard, J. Am. Ceram. Soc. 92 (2009) 2497–2503. S. Deville, E. Saiz, A.P. Tomsia, Acta Mater. 55 (2007) 1965–1974. K. Maca, V. Pouchly, P. Zalud, J. Eur. Ceram. Soc. 30 (2010) 583–589. J.A. Sekhar, R. Trivedi, Mater. Sci. Eng. A 147 (1991) 9–21. S.a. Barr, E. Luijten, Acta Mater. 58 (2010) 709–715. E. Munch, M.E. Launey, D.H. Alsem, E. Saiz, a P. Tomsia, R.O. Ritchie, Science 322 (2008) 1516–1520. G.M. Luz, J.F. Mano, Philos. Transact. A Math. Phys. Eng. Sci. 367 (2009) 1587–1605. F. Barthelat, Bioinspir. Biomim. 5 (2010) 035001. R. Rabiei, S. Bekah, F. Barthelat, Acta Biomater. 6 (2010) 4081–4089. Z. Tang, N. a Kotov, S. Magonov, B. Ozturk, Nat. Mater. 2 (2003) 413–418. O. Prokopiev, I. Sevostianov, Mater. Sci. Eng. A 431 (2006) 218–227. K.H. Zuo, Y. Zhang, Y.-ping Zeng, D. Jiang, Ceram. Int. 37 (2011) 407–410. E. Munch, J. Franco, S. Deville, P. Hunger, E. Saiz, A.P. Tomsia, JOM 60 (2008) 54–58. A. Bareggi, E. Maire, A. Lasalle, S. Deville, J. Am. Ceram. Soc. 94 (2011) 3570–3578. F. Tavangarian, R. Emadi, Mater. Lett. 65 (2011) 740–743. F. Tavangarian, R. Emadi, Mater. Lett. 65 (2011) 2354–2356. V. Karageorgiou, D. Kaplan, Biomaterials 26 (2005) 5474–5491. J.B.J.B. Recknor, J.C.J.C. Recknor, D.S.D.S. Sakaguchi, Biomaterials 25 (2004) 2753–2767. T.T. Yu, M.S. Shoichet, Biomaterials 26 (2005) 1507–1514. C.M. Patist, M.B. Mulder, S.E. Gautier, V. Maquet, R. Jérôme, M. Oudega, Biomaterials 25 (2004) 1569–1582. B. Zhu, Q. Zhang, Q. Lu, Y. Xu, J. Yin, J. Hu, Z. Wang, Biomaterials 25 (2004) 4215–4223.
Contents lists available at SciVerse ScienceDirect
Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec
The effect of processing parameters and solid concentration on the mechanical and microstructural properties of freeze-casted macroporous hydroxyapatite scaffolds S. Farhangdoust ⁎, A. Zamanian, M. Yasaei, M. Khorami Nanotechnology and Advance Materials Department, Materials and Energy Research Center, Karaj, Alborz, Iran
a r t i c l e
i n f o
Article history: Received 2 July 2012 Received in revised form 28 July 2012 Accepted 17 September 2012 Available online 22 September 2012 Keywords: Hydroxyapatite Scaffold Freeze-casting Unidirectional Sintering Cooling rate
a b s t r a c t The design and fabrication of macroporous hydroxyapatite scaffolds, which could overcome current bone tissue engineering limitations, have been considered in recent years. In the current study, controlled unidirectional freeze-casting at different cooling rates was investigated. In the first step, different slurries with initial hydroxyapatite concentrations of 7–37.5 vol.% were prepared. In the next step, different cooling rates from 2 to 14 °C/min were applied to synthesize the porous scaffold. Additionally, a sintering temperature of 1350 °C was chosen as an optimum temperature. Finally, the phase composition (by XRD), microstructure (by SEM), mechanical characteristics, and the porosity of sintered samples were assessed. The porosity of the sintered samples was in a range of 45–87% and the compressive strengths varied from 0.4 MPa to 60 MPa. The mechanical strength of the scaffolds increased as a function of initial concentration, cooling rate, and sintering temperature. With regards to mechanical strength and pore size, the samples with the initial concentration and the cooling rate of 15 vol.% and 5 °C/min, respectively, showed better results. © 2012 Elsevier B.V. All rights reserved.
1. Introduction One of the key components for being successful in bone tissue engineering is developing a scaffold with an appropriate architecture for inducing cell proliferation, migration and differentiation. However, the proliferation of osteoblasts, vascular ingrowth and bone formation requires an interconnected pore system and a controlled pore size, as well as a desirable shape and orientation [1–3]. In recent years, several techniques have been employed to fabricate porous scaffolds, i.e., solvent casting/particulate leaching [4–7], gas foaming [8–10], rapid prototyping [11–13], extrusion [14,15], phase separation [16–19] and freeze-casting [3,20–22]. In the pursuit of such processing routes, freeze-casting has attracted considerably more focus in the last few years. Freeze-casting is environmentally friendly and cost effective because of its use of a minimum concentration of organic additive and its use of water as a suspension liquid. In this method, controlling the parameters, such as initial concentration and cooling rate, leads to an interconnected, porous structure. Moreover, the pore size, shape and orientation are controllable and possess the potential to be repeated, which are characteristics not observed with other methods. The freeze-casting technique seems to be rather versatile; furthermore, the use of a liquid solvent (mostly water) as a pore-forming agent is a promising asset [20,23]. Calcium phosphate ceramics and macroporous hydroxyapatite (HA, Ca10(PO4)6(OH)2), in particular, could be used to manufacture ideal scaffolds due to their biocompatibility, as well as their similarity ⁎ Corresponding author. Tel.: +98 912 3167022; fax: +98 26 36204139. E-mail address: [email protected] (S. Farhangdoust). 0928-4931/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.msec.2012.09.013
to the mineral component of bone. While HA shows desirable properties for applications in tissue engineering, it cannot be used in loadbearing applications due to its poor mechanical properties when compared with other materials, such as titanium [24–26]. During recent years, unidirectional freeze-casting has emerged to solve this problem. Deville et al. [3] prepared highly porous HA materials that exhibited extraordinarily high compressive strengths, up to 145 MPa, with a porosity that reached 47%. They suggested that these HA-based materials could be used in load-bearing applications, such as artificial bone. With this technique, the greatest influence on the porosity and the pore size distribution of the ceramic body was exerted over the thickness of the samples by the solid concentration of the suspension, the temperature gradient and the ice crystal growth rate. Little research has been conducted to determine the effective parameters on the process of freeze-casting HA scaffolds. Furthermore, current research hardly covers the entirety of the parameters, i.e., the initial concentration of the slurry, the cooling rate and the sintering temperature. Therefore, having a good understanding about the aforementioned parameters is essential in realizing the microstructure and the mechanical properties of freeze-casted HA scaffolds. The present investigation is focused on the influence of the solid concentration, cooling rates and sintering temperature on the quality of the synthetic scaffolds. 2. Materials and methods 2.1. Fabrication procedure Controlled, unidirectional freeze-casting was used to create porous HA scaffolds. In the first step, different slurries with initial HA concentrations
454
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
of 7–37.5 vol.% were prepared. To prepare stable slurries, a small amount (4 wt.% of the HA content) of commercially available dispersant (Dolapix CE 64, Zschimmer & Schwarz, Lahnstein, Germany) was added to distilled water. Then, HA powder (2196, Merck KGaA, Darmstadt, Germany) with a median particle size (d50) and specific surface area of 1.69 μm and 11.06 g/m2, respectively, was gradually added to the distilled water to prevent agglomeration. After that, polyvinyl alcohol (PVA, Mw = 15000, Merck, Darmstadt, Germany) was added as a binder at 4 wt.% of the HA content and followed by vigorous stirring at 1000 rpm for 1 h. Subsequently, to further stabilize the slurry, the pH of the slurry was determined and adjusted to 10 by the gradual addition of a 1 M NaOH solution. To remove air bubbles before casting, the prepared slurry was placed in a vacuum oven for 30 min at a pressure of 0.02 MPa. Freeze-casting of the prepared slurries was performed by pouring them into a PTFE mold with an inner diameter of 20 mm. The mold was placed on copper, where the temperature was controlled by using liquid nitrogen and a ring heater connected to a PID controller. The cooling rates utilized in this study ranged from 2 to 14 °C/min. After careful removal of the samples from the mold, the frozen samples were dried in the freeze-dryer (Beta 1-2LD plus, Martin Christ GmbH, Germany) at a low temperature of − 55 °C and pressure of 2.1 Pa. Subsequently, green bodies were sintered by double step sintering to prevent the grain growth [27] for 2 h with a heating rate of 3 °C/min; the first step of sintering was conducted at 600 °C, and the second step was conducted at different temperatures in the range of 1250–1350 °C. 2.2. Characterization 2.2.1. Phase analysis Phase composition of the sintered samples was evaluated using an automated X-ray diffractometer (XRD, Philips PW3710). Cu-Kα radiation was used under the operating conditions of 40 kV and 30 mA. XRD diagrams were constructed from 2θ data in the range of 20–60°. 2.2.2. Microstructure Microstructure analysis of the samples was determined by using a scanning electron microscope (SEM, Stereoscan S 360-Leica Cambridge, England). Due to the poor electrical conductivity of the samples, a thin layer of gold was coated onto the surfaces of the scaffolds before testing. 2.2.3. Mechanical properties For the compressive strength test, samples with a diameter of ~ 15 mm and a height of ~ 20 mm were loaded onto a crosshead and pulled at a speed of 1 mm/min using a screw-driven load frame (Instron 5565, Instron Corp., Canton, MA). During the compressive strength tests, the stress and strain responses of the samples were monitored. Five samples were tested to obtain an average value and its standard deviation.
2.3.5. Pore size With the HA scaffolds being so anisotropic, pore sizes were determined in both the long and short axes. Five samples were studied, with 50 measurements conducted for each sample. 2.2.6. Shrinkage Longitudinal and circumferential shrinkages were calculated using the following expressions [29]: Sc ¼ ðD0 −Df Þ=D0
ð3Þ
SL ¼ ðL0 −Lf Þ=L0
ð4Þ
where Sc and SL are the circumferential and longitudinal shrinkages, respectively, D0 and Df are the initial and final diameters, respectively, and L0 and Lf are the initial and final lengths, respectively. 3. Results and discussion 3.1. Phase analysis The XRD patterns of the initial powder and the sintered sample (sintered at 1350 °C) are shown in Fig. 1, which can be completely indexed with HA (JCPDS#76-0694). HA was the only phase in these patterns, and no secondary phase was found after sintering at 1350 °C. 3.2. Microstructure A lamellar HA scaffold with unidirectional aligned channels can be observed in Fig. 2. Macroscopic aligned pores were formed almost uniformly over the entire sample. In fact, the porosity of the sintered materials was a replica of the ice structure before sublimation. These pores were generated during the sublimation of the ice. The corresponding horizontal cross sections are shown in Fig. 2 and reveal the arrangement of the ice crystals in the direction parallel to the solidification direction. The crystals were lamellar but were arranged in domains with similar orientations. The orientation of each domain could be related to the original nucleation conditions. According to the microstructure of a vertical cross-section in Fig. 3, the scaffolds can be divided into three distinct zones in the direction of solidification, as reported by Deville et al. [3]. In zone 1, the closest zone to the initial cold finger, no microporosity was observed and the material was dense. In zone 2, the material was characterized by a cellular morphology. Finally, in the upper zone (zone 3), the ceramic was lamellar with long parallel pores aligned with the movement direction of the ice front. Parameters such as the particle size of
2.2.4. Porosity The total porosity (Pt) of the sintered samples was determined using the following expression [28]: Pt ¼ 100 ð1−Db =Dt Þ
ð1Þ
where Dt is the theoretical density of the powder and Db is the bulk density. Db was calculated using the following expression: Db ¼ m=v
ð2Þ
where m is the weight of the specimen and V is its volume. Five samples were measured to obtain an average value and its standard deviation.
Fig. 1. XRD pattern of the initial powder and the sintered sample sintered at 1350 °C.
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
455
Fig. 2. Lamellar scaffold with unidirectional aligned channels with an initial concentration of 15 vol.% HA.
hydroxyapatite affect the thickness of these zones which is being investigated in ongoing research. It is well known that to form unidirectional porous structures, the particles must be rejected from the solidification front and collected between the arms of the solidification front. At the very beginning
of solidification, the interface was planar (Zone 1) and needed to somehow undergo a transition towards an irregular morphology, i.e., cellular (zone 2) and lamellar (zone 3). This transition occurred due to the perturbation of advancing solidification behind. Until now, two different mechanisms have been proposed. One mechanism
Fig. 3. Microstructure of a vertical cross-section of scaffold with an initial concentration of 22.5 vol.% HA. a) Scaffolds can be divided into three distinct zones with respect to the direction of solidification, which is labeled by Greek numbers, and is magnified in the black squares. b) Zone 1, the closest to the initial cold finger. c) Zone 2, characterized by a cellular morphology. d) Zone 3, where the ceramic is lamellar with long parallel pores aligned with the movement direction of the ice front.
456
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
is the inherent thermodynamic instability of the interface, known as the Mullins–Sekerka instability [30]. In this mechanism, the development of the instability is based on supercooling effects building up with the solute rejection ahead of the interface. Another mechanism, proposed by Haji, is related to the presence of the particles [28]. In this case, the instability is attributed to the reversal of the thermal gradient in the liquid ahead of the interface and behind the particle. This instability occurs at pulling velocities below the threshold for the onset of the Mullins–Sekerka instability. Previously, it was suspected that the Mullins–Sekerka mechanism leads to progressive evolution of the interface morphology from flat to cellular and, finally, to lamellar. Recently, in situ X-ray radiography and tomography observations by Deville et al. [31,32] revealed that two types of crystals (Fig. 4) were initially present: a first population of lamellar crystals, with their main axis oriented along the cooling direction, and a second population of crystals, more or less lamellar, but oriented predominantly in the radial direction (perpendicular to the cooling direction). When the temperature at the copper surface reached the nucleation point, homogeneous (spatially speaking) nucleation of ice crystals occurred. Crystals with a random crystallographic orientation were nucleating and growing very fast so that all the particles were initially entrapped (creating zone 1) because they did not have enough time to be repelled by the ice crystals. As the crystals started to grow along the cooling axis, the interface velocity diminished, and the particle redistribution progressively took place as particles started to be repelled by the interface. While the crystals grow larger and
larger, the particle packing between the crystals becomes more and more efficient. Therefore, the entrapped particle fraction progressively diminishes. Crystals in all orientations grew along the cooling axis in zone 2, which can be called the transition zone. It is well known that the growth along the c-axis of hexagonal ice is theoretically two orders of magnitude slower than the growth along the a-axis or b-axis (anisotropy of crystal growth kinetics). Both populations grew faster in directions perpendicular to the c-axis of the hexagonal ice structure but with a different orientation with regards to the temperature gradient. As the conditions are dictated by the temperature gradient, the second population of crystals (horizontally growing) stopped growing, leaving only the first population of crystals that were growing into the suspension. In this stage, the crystals were growing with steadystate conditions that initiated and formed zone 3. In zone 3, the continuous crystals growing along the solidification direction were noticeable. The specimens with an initial concentration of 7% HA, which were frozen at a cooling rate of 2 °C/min, are shown in Fig. 5a. On the internal walls of the lamellae dendritic, a branch-like structure was observed. Some ceramic bridges and linking adjacent plates were also observed (Fig. 5b). The morphology of these numerous fine features is very different from the dendrites that cover the ceramic lamellae. This difference suggests another formation mechanism. Deville et al. [33] proposed that these features might be formed because of the specific conditions encountered during the slow freezing of highly concentrated solutions. However, the results of the current study are not in agreement with the hypothesis of Deville et al. because these bridges not only can be observed in low concentration slurries but can also be detected in
Fig. 4. a) Two types of crystals that are present from the beginning: a first population of lamellar crystals, with their main axis oriented along the cooling direction (left), and a second population of crystals, more or less lamellar, but oriented predominantly in the radial direction or perpendicular to the cooling direction (right). b) Hexagonal ice crystals: growth along the c-axis of hexagonal ice is theoretically two orders of magnitude slower than growth along either the a- or b-axis (anisotropy of crystal growth kinetics).
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
457
below and up to 1300 °C. In fact, at the sintering temperature of 1350 °C, the compressive strength is about three orders of magnitude larger than the compressive strength at 1300 °C. At 1350 °C, the porosity and shrinkage are ~75% and ~20%, respectively. It is worth mentioning that the enhancement in mechanical properties occurs because of porosity attenuation. Werner et al. [2] sintered HA at temperatures between 1250 and 1450 °C and reported formation of α-TCP at 1400 °C. Meanwhile, Prokopiev [42] reported sintering at a temperature of 1280 °C as the plateau stage, and Deville [3] chose 1325 °C as the optimum point. In the current study, the plateau stage at sintering temperatures up to 1350 °C was not reached. 1350 °C was chosen as the most favorable sintering temperature because of the desirable compressive strength and porosity. It can be observed that longitudinal shrinkage, parallel to solidification direction, is 1.05± 0.2%, which is less than the circumferential shrinkage (perpendicular to the cooling
Fig. 5. The internal walls of the lamellae of the scaffolds with an initial concentration of 7.5 vol.% HA. a) Dendritic and branch-like features. b) Ceramic bridges and linking adjacent plates.
fast-frozen samples. Fig. 5b shows the specimens with an initial concentration of 15% HA which were frozen at a cooling rate of 11 °C/min. It seems that the addition of PVA caused the formation of these bridges. Pekor et al. [34] investigated the effect of PVA on the microstructure of freeze-casted alumina, and they concluded that when 5 wt.% of PVA was added, due to the effect of constitutional supercooling the fine secondary dendrites appear to coarsen. It was previously shown that the particles themselves might induce morphological transitions, such as dendrite tip splitting or healing [35]. Moreover, entrapped particles in ice might induce such morphology [36]. In addition, the enhancement in the mechanical strength of the scaffolds was related to the morphological changes induced by these bridges. In this way, different mechanisms have been proposed. Munch et. al [37] suggested that the presence of stiff ceramic bridges between the grains with micrometer and sub-micrometer dimensions promotes controlled sliding and sliding interferences between the rough ceramic interlayers, thereby enhancing the toughness through extremely efficient energy dissipation. By limiting sliding, they provided very effective toughening mechanisms in natural and synthetic materials. Luz et al. [38] believed that the existence of such inorganic bridges reinforces the weak interfaces and that these interfaces become just suitable for the crack to extend itself. It is worth mentioning that evolving these bridges is one of the toughening mechanisms employed by nacre and synthetic bio-inspired materials [39–41]. 3.3. Effect of sintering temperature Sintering temperature plays a key role in the mechanical and physical characteristics of the scaffold. Fig. 6 shows the effect of sintering temperature on the compressive strength, total porosity and shrinkage of the samples with an initial concentration of 15 vol.% HA, which were solidified at a rate of 8 °C/min. The mechanical strength and shrinkage increased as a function of temperature, while temperature had an inverse effect on the total porosity. According to the compressive strength results, the samples did not sinter completely at temperatures
Fig. 6. Effect of sintering temperature on the compressive strength, total porosity and shrinkage of samples with an initial concentration of 15 vol.% HA, solidified at a rate of 8 °C/min.
458
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
direction) on average. It seems that there was more empty space (spatially speaking) in the direction of circumferential shrinkage. As a result, shrinkage occurred more readily in the circumferential direction rather than in the longitudinal direction. 3.4. Effect of initial concentration of slurry Fig. 7 shows the effect of the initial concentration of the slurry on the total porosity of the scaffolds. A wide range of porosity was obtained, ranging from 45 to 87%. Porosity has a linear relation with the initial concentration, which can suggest the final porosity. As a result, the strength is completely adjustable. The compressive strength (Fig. 7) increased sharply from 0.4 MPa to ~60 MPa, while the porosity decreased constantly. The compressive strengths that were measured were ~20 MPa with 68% porosity and ~60 MPa with 48% porosity. Although higher compressive strengths, up to 140 MPa, were reported in one study [3], our results were compared better with other studies [22,43,44]. It should be noted that the tested specimens in the aforementioned studies were smaller (3 mm× 4 mm × 4 mm) than those used in the current study (15 mm in diameter and 20 mm in height). Actually, the initial concentration has a key role in determining the lamella and the thickness of the pores in the samples. Higher initial concentration resulted in thicker lamella and thinner porosity. Fig. 7 shows the pore size of the scaffold in the long and short axis as a function of the initial concentration of the slurry. The reported sizes are the mean size of 3 samples with a confidence level of 95%, which was measured in 20 different places. The pore size in the long axis of the scaffold declined dramatically with increases in slurry concentration (falling from ~ 180 μm at 7.5% HA to ~ 8 μm at 37.5% HA). Meanwhile, the pore size in the short axis decreased monotonically from ~ 42 μm at 7.5% HA to ~ 5 μm at 37.5% HA. Fig. 8 shows three different samples with initial concentrations of 7%, 18% and 22% HA. Lamellar space and pore thickness decreased with increased slurry concentration. As a result, the compressive strength increased with some limitations. On one hand, with low initial concentrations (lower than 10 vol.%), the ceramic walls of the green body were weaker and more difficult to handle. In addition, it was difficult to achieve homogeneity through the entire length of the specimens. On the other hand, with high initial concentrations (larger than 35 vol.%), the particles could not be rejected by the ice; therefore, the lamellar structure and interconnectivity were not achieved [33]. 3.5. Effect of cooling rate Fig. 9 shows the compressive strength and the large and short axes of the samples versus the cooling rate. The compressive strength increased monotonically while the lamellar space decreased as the result of an increased cooling rate. The compressive strengths were 4.1, 6.4 and 9.5 MPa at the cooling rates of 2, 8 and 14 °C/min, respectively. With an increase in the freezing rate, larger temperature gradients resulted in smaller pore sizes, and as a result, the strength of the ceramic bodies increased. Actually, the thickness of the ice crystals is strongly dependent on the speed of the solidification front. Faster freezing velocities result in larger supercooling in front of the growing crystals that will influence the crystal thickness. In addition, as faster growth is imposed in the direction of the temperature gradient, lateral growth along the c-axis is increasingly limited, resulting in thinner lamellae. At slow solidification rates, the particles easily diffuse away from the interface and the temperature of the suspension, ahead of the interface, is always warmer than the freezing temperature. At faster solidification rates, the concentration and concentration gradient increase at the interface. When the concentration gradient at the interface is steep enough that the gradient in the freezing temperature is larger than the temperature gradient, the suspension ahead of the interface is below its freezing temperature (constitutionally supercooled). In analogy with binary alloys, constitutional supercooling is closely related
Fig. 7. Effect of the initial concentration of the slurry on the total porosity and the compressive strength.
to morphological instability. By controlling the temperature at the base of the suspension, it is possible to control the extension of crystals [33,45]. As mentioned previously, higher front velocities lead to thinner colloidal configurations because more particles are engulfed before any solid formation. Some of the engulfed particles form bridges that link neighboring colloidal domains, thus mimicking the inorganic bridges that are found in nacre. We found that higher front velocities result in rougher surfaces because the particles have less time to rearrange when they are being pushed. This effect is more pronounced in higher initial volume fractions.
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
459
Fig. 9. Compressive strength and the large and short axes of the scaffold versus the cooling rate.
Fig. 8. Three different samples with initial concentrations of a) 7%, b) 18% and c) 22% HA.
patterns, roughness, etc. The results show the beneficial effect of surface roughness, patterns and architecture, which seem to improve bone ingrowth and tissue formation [49–52]. It seems that by adjusting the initial concentration and the cooling rate of slurry, the size and shape of these features can be regulated to match more favorable dimensions of guidance patterns, improving the osteoconduction and osteoinduction characteristics of the scaffold.
3.6. Suitability as a bone substitute 4. Conclusions The porosity and pore size of biomaterial scaffolds play a critical role in bone formation both in vitro and in vivo. These morphological features influence osteogenesis as well as the mechanical properties of the scaffolds [46,47]. Lower porosity stimulates osteogenesis in vitro by suppressing cell proliferation and forcing cell aggregation. In contrast, greater bone ingrowth is achieved in vivo with higher porosity and larger pore size. However, this trend results in diminished mechanical properties, thereby setting an upper functional limit for pore size and porosity. Thus, a balance must be reached. Based on previous studies, the minimum requirement for pore size is considered 100 μm due to cell size [48]. Thus, the sample with an initial concentration of 15 vol.% HA and a cooling rate of 5 °C/min, along with a porosity of 76± 2.75%, a compressive strength of 5.26 ±0.52 MPa and a pore size of 88.4 ± 7.9 μm, was chosen as an optimum sample. It is worth mentioning that dendritic features on the internal surface of the walls might act as a guiding pattern for cell growth, which would improve the osteoconduction characteristics [2]. Considerable attention has been paid to the relationships between cellular response and morphological features, such as mesostructure,
Based on the experimental study of HA freeze-casted scaffolds with different initial concentration, cooling rates and sintering conditions, the following points are concluded: 1. Porous scaffolds with a total porosity of 45–87% and compressive strengths between 0.4 and 60 MPa are obtained by freeze-casting HA, followed by the sublimation and sintering of the green bodies. 2. The porosities are unidirectionally aligned along the entire lengths of the samples. The formation and morphology of these porosities can be controlled via the initial concentrations of the HA slurry and the cooling rate. 3. The bridges that connect the adjacent plates improve the mechanical strength of the scaffolds. 4. The synthesized scaffolds can be employed as bone-repairing material due to the lamellar structure of the macropores and the presence of dendritic branch-like features. 5. This method can be easily utilized to synthesize scaffolds with other calcium phosphates or any other materials.
460
S. Farhangdoust et al. / Materials Science and Engineering C 33 (2013) 453–460
References [1] J.S. Son, M. Appleford, J.L. Ong, J.C. Wenke, J.M. Kim, S.H. Choi, D.S. Oh, J. Control. Release 153 (2011) 133–140. [2] J. Werner, B. Linner-Krčmar, W. Friess, P. Greil, Biomaterials 23 (2002) 4285–4294. [3] S. Deville, E. Saiz, A.P. Tomsia, Biomaterials 27 (2006) 5480–5489. [4] D. Sin, X. Miao, G. Liu, F. Wei, G. Chadwick, C. Yan, T. Friis, Mater. Sci. Eng. C 30 (2010) 78–85. [5] J. Wei, F. Chen, J.-W. Shin, H. Hong, C. Dai, J. Su, C. Liu, Biomaterials 30 (2009) 1080–1088. [6] S. Zadegan, M. Hosainalipour, H.R. Rezaie, H. Ghassai, M.a. Shokrgozar, Mater. Sci. Eng. C 31 (2011) 954–961. [7] S.K. Misra, T.I. Ansari, S.P. Valappil, D. Mohn, S.E. Philip, W.J. Stark, I. Roy, J.C. Knowles, V. Salih, A.R. Boccaccini, Biomaterials 31 (2010) 2806–2815. [8] Y.-M. Lim, H.-J. Gwon, J. Shin, J.P. Jeun, Y.C. Nho, Ind. Eng. Chem. Res. 14 (2008) 436–441. [9] C. Ji, N. Annabi, M. Hosseinkhani, S. Sivaloganathan, F. Dehghani, Acta Biomater. 8 (2011) 570–578. [10] A. Salerno, S. Zeppetelli, E.D. Maio, S. Iannace, P.a. Netti, Compos. Sci. Technol. 70 (2010) 1838–1846. [11] I.-K. Jun, Y.-H. Koh, S.-H. Lee, H.-E. Kim, J. Mater. Sci. Mater. Med. 18 (2007) 1071–1077. [12] C.K. Chua, K.F. Leong, K.H. Tan, F.E. Wiria, C.M. Cheah, Design 5 (2004) 1113–1121. [13] M. Schumacher, F. Uhl, R. Detsch, U. Deisinger, G. Ziegler, J. Mater. Sci. Mater. Med. 21 (2010) 3039–3048. [14] H.Y. Yang, I. Thompson, S.F. Yang, X.P. Chi, J.R.G. Evans, R.J. Cook, J. Mater. Sci. Mater. Med. 19 (2008) 3345–3353. [15] H.-J. Yen, C.-S. Tseng, S.-H. Hsu, C.-L. Tsai, Biomed. Microdevices 11 (2009) 615–624. [16] J.M. Holzwarth, P.X. Ma, Biomaterials 32 (2011) 9622–9629. [17] J.P. Kennedy, S.P. McCandless, R. a Lasher, R.W. Hitchcock, Biomaterials 31 (2010) 1126–1132. [18] a S. Rowlands, S. a Lim, D. Martin, J.J. Cooper-White, Biomaterials 28 (2007) 2109–2121. [19] K. Shin, B. Kim, J. Kim, T. Park, J. Nam, D. Lee, Polymer 46 (2005) 3801–3808. [20] T. Moritz, H.-J. Richter, J. Eur. Ceram. Soc. 27 (2007) 4595–4601. [21] K. Araki, J.W. Halloran, J. Am. Ceram. Soc. 87 (2005) 2014–2019. [22] E. Landi, F. Valentini, A. Tampieri, Acta Biomater. 4 (2008) 1620–1626.
[23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52]
S. Deville, Adv. Eng. Mater. 10 (2008) 155–169. C.V. Ragel, L.M. Rodr, Biomaterials 23 (2002) 1865–1872. F.N. Oktar, G. Göller, Ceram. Int. 28 (2002) 617–621. K. So, S. Fujibayashi, M. Neo, Y. Anan, T. Ogawa, T. Kokubo, T. Nakamura, Biomaterials 27 (2006) 4738–4744. M.E. Launey, et al., Acta Mater. 57 (2009) 2919–2932. M. Khorami, S. Hesaraki, T. Ebadzadeh, S. Farhangdoust, A. Zamanian, Key Eng. Mater. 493–494 (2011) 231–235. A. Heunisch, A. Dellert, A. Roosen, J. Eur. Ceram. Soc. 30 (2010) 3397–3406. W.W. Mullins, R.F. Sekerka, J. Appl. Phys. 35 (1964) 444. L. Hadji, Eur. Phys. J. 37 (2003) 85–89. S. Deville, E. Maire, A. Lasalle, A. Bogner, C. Gauthier, J. Leloup, C. Guizard, J. Am. Ceram. Soc. 92 (2009) 2497–2503. S. Deville, E. Saiz, A.P. Tomsia, Acta Mater. 55 (2007) 1965–1974. K. Maca, V. Pouchly, P. Zalud, J. Eur. Ceram. Soc. 30 (2010) 583–589. J.A. Sekhar, R. Trivedi, Mater. Sci. Eng. A 147 (1991) 9–21. S.a. Barr, E. Luijten, Acta Mater. 58 (2010) 709–715. E. Munch, M.E. Launey, D.H. Alsem, E. Saiz, a P. Tomsia, R.O. Ritchie, Science 322 (2008) 1516–1520. G.M. Luz, J.F. Mano, Philos. Transact. A Math. Phys. Eng. Sci. 367 (2009) 1587–1605. F. Barthelat, Bioinspir. Biomim. 5 (2010) 035001. R. Rabiei, S. Bekah, F. Barthelat, Acta Biomater. 6 (2010) 4081–4089. Z. Tang, N. a Kotov, S. Magonov, B. Ozturk, Nat. Mater. 2 (2003) 413–418. O. Prokopiev, I. Sevostianov, Mater. Sci. Eng. A 431 (2006) 218–227. K.H. Zuo, Y. Zhang, Y.-ping Zeng, D. Jiang, Ceram. Int. 37 (2011) 407–410. E. Munch, J. Franco, S. Deville, P. Hunger, E. Saiz, A.P. Tomsia, JOM 60 (2008) 54–58. A. Bareggi, E. Maire, A. Lasalle, S. Deville, J. Am. Ceram. Soc. 94 (2011) 3570–3578. F. Tavangarian, R. Emadi, Mater. Lett. 65 (2011) 740–743. F. Tavangarian, R. Emadi, Mater. Lett. 65 (2011) 2354–2356. V. Karageorgiou, D. Kaplan, Biomaterials 26 (2005) 5474–5491. J.B.J.B. Recknor, J.C.J.C. Recknor, D.S.D.S. Sakaguchi, Biomaterials 25 (2004) 2753–2767. T.T. Yu, M.S. Shoichet, Biomaterials 26 (2005) 1507–1514. C.M. Patist, M.B. Mulder, S.E. Gautier, V. Maquet, R. Jérôme, M. Oudega, Biomaterials 25 (2004) 1569–1582. B. Zhu, Q. Zhang, Q. Lu, Y. Xu, J. Yin, J. Hu, Z. Wang, Biomaterials 25 (2004) 4215–4223.
