July 2, 2008

Porous Ceramic Scaffolds With Complex Architectures

By Munch, E Franco, J; Deville, S; Hunger, P; Saiz, E; Tomsia, A P

This work compares Two novel techniques for the fabrication of ceramic scaffolds for bone tissue engineering with complex porosity: robocasting and freeze casting. Both techniques are based on the preparation of concentrated ceramic suspensions with suitable properties for the process, in robocasting, the computer-guided deposition of the suspensions is used to build porous materials with designed three dimensional geometries and microstructures. Freeze casting uses ice crystals as a template to form porous lamellar ceramic materials. Preliminary results on the compressive strengths of the materials are also reported. INTRODUCTION

Most strategies to regenerate bone depend on three-dimensional (3- D) porous structures (scaffolds) to support cell attachment, proliferation, and differentiation. Perhaps one of the most critical challenges in bone tissue engineering is to develop custom-designed scaffolds, tailored to mimic natural biological templates and with mechanical properties matching those of the surrounding tissue. The ideal scaffold for bone regeneration should be formed using an osteoconductive or osteoinductive porous material that will form a secure bond with the tissues by allowing and encouraging invasion of new cells. The porosity must be interconnected to allow the ingrowth of bone, vascularization, and diffusion of nutrients. The scaffold must provide the necessary support for cells to proliferate and slowly degrade and resorb as the tissue structures grow in vitro and/ or in vivo.1 In conventional porous scaffold fabrication methods2- 10 it is difficult to precisely control pore size, geometry, and spatial distribution; to create internal channels within the scaffold for vascularization; and to produce scaffolds with arbitrary and complex 3-D anatomical shapes.

This work describes two new techniques for the fabrication of scaffolds with complex pore distributions: robotic-assisted deposition and freeze casting. Both techniques are based on the preparation of water-based ceramic suspensions. Robotic-assisted deposition (robocasting) allows the fabrication of polymer and ceramic scaffolds without the need for a sacrificial support material or mold.15-19 This technique consists of the robotic deposition of inks capable of fully supporting their own weight during assembly, thanks to their carefully tailored composition and viscoelastic properties. In freeze casting, porous lamellar ceramic or polymeric materials are built through the controlled directional freezing of concentrated suspensions.15-19 This article will show how these techniques can be used to prepare porous hydroxyapatite (HA) ceramics. Hydroxyapatite is an osteoconductive ceramic closely related to the inorganic component of bone and one of the leading candidates for the fabrication of porous inorganic scaffolds for bone tissue engineering. However, both technologies can be easily extended to the fabrication of scaffolds using different polymer and ceramic materials (e.g., other calcium phosphates such as tri- calcium phosphate or bioactive glasses).

See the sidebar for experimental procedures.



Colloidal inks developed for robocasting must satisfy two important criteria:13 first, their viscoelastic properties must allow them to flow through a deposition nozzle and then "set" immediately so that shape is retained as additional layers are deposited or when they span gaps in the underlying structure (Figures 1 and 2). second, the suspensions must have a high solid volume concentration to minimize shrinkage during drying so that the particle network is able to resist the involved capillary stresses. The stability of these high-solids-loading suspensions requires high dispersive forces between particles, and therefore the role of dispersant is critical. The amount of dispersant has to be adjusted to efficiently coat the panicles in the suspension, but an excess will cause flocculation of the particles due to the depletion effect.13 It was possible to determine that for the average ceramic particle size used in this work (-1-2 [mu]m) the optimal concentration of dispersant is around 0.5% relative to powder weight.

Robocasting requires a pseudoplastic suspension with a yield stress high enough to prevent shape changes on the printed structure under its own weight. Therefore, the well-dispersed highsolids- loading suspensions are not yet functional for robocasting. Their rheology must be altered to form partially flocculated suspensions where a loosely bound particle network is formed.13 In the first approach polyethylenimine (PEI) was used as flocculant to create the desired links between particles through interaction between its positively charged amine groups and the negalively charged carboxylic acid groups of the dispersant covering the HA particles. The optimal amount of PEI was estimated to be around 0.4 wt.% relative to water content. The final tuning of the viscoelastic properties of the suspension is achieved by slightly adjusting its pH with HNO^sub 3^ or NH^sub 4^OH as needed. The pH of the suspension controls the activity of both dispersant and flocculam and, therefore, the strength of the bonds between the particles and its ability to form the desired network. In this system, the optimum pH is ~9. As an alternative to the use of a flocculant to achieve an ink with the suitable rheological properties, the authors have dispersed the particles in a hydrogel (Pluronic(R)). Pluronic is soluble in water at low temperatures and allows the dispersion of the ceramic particles when cold. At room temperature, it forms a gel that can be used in the robocasting process. By controlling the amount of Pluronic and the solid content, it is possible to tailor the properties of the suspension for the printing process. The optimum solid content of the inks varies between 35 vol.% and 40 vol.% for the inks prepared using Pluronic and is -45 vol.% for those prepared using PEI. In both cases, inks prepared with higher solid loading exhibited very limited printability and lower concentrations led to poor shape retention.

Using these inks, scaffolds have been printed with strut thicknesses ranging between 200 [mu]m and 500 [mu]m and line spacing ranging from 75 [mu]m to 500 [mu]m (Figure 3). Preliminary tests indicate that the compressive strength of the robocasted grids ranges from 25 MPa to 40 MPa. This is of the order of porous calcium phosphates with similar porosity (~45%) fabricated by conventional techniques2-10 and slightly higher than the corresponding value for cancellous bone (7-10 MPa).:(t No de-bonding of the lines was observed during testing, indicating excellent adhesion between the printed lines. The structures prepared using Pluronic-based inks usually have a larger volume of microporosity in the printed line (Figure 3). This is due to the larger organic content of the inks (up to IO wt.%). The microporosity can affect negatively the compressive strength but several studies have shown that it can significantly enhance bone formation.21

Freeze Casting

Freeze casting is a simple technique to produce porous complex- shaped ceramic or polymeric parts (Figure 4). In freeze casting, a ceramic slurry is poured into a mold and then frozen. Under steady- state conditions, it is possible to grow ice crystals in the form of platelets, with a very high aspect ratio. The ice thus formed will have a lamellar microstructure, with the lamellae thickness depending mainly on the speed of the freezing front and the solid content of the ceramic suspension.15-17 The frozen solvent acts temporarily as a binder to hold the part together for demolding. Subsequently, the part is subject to freeze drying to sublimate the solvent under vacuum, avoiding the drying stresses and shrinkage that may lead to cracks and warping during normal drying. During freezing of the ceramic suspensions, the ceramic particles concentrate in the space between the ice crystals, creating a layered material (Figure 4). After the ice is removed via sublimation and the green body is sintered, the result is a porous ceramic scaffold that exhibits striking similarities to the inorganic component of nacre's multilayered structure across a wide range of length scales (Figure 5).

A critical step during processing is to achieve a stable suspension. If there is a significant amount of sedimentation during the process, the solid content (and therefore the microstructure) will not be homogeneous along the freezing direction. The analysis of sedimentation of HA slurries (Figure 6) has been used to identify optimum compositions. Low dispersant (0.5-1 wt.%) and high binder (- 3-4 wt.%) concentrations result in more stable dispersions suitable for the freeze-casting process. However, if a very low dispersant concentration is used (

The pores are highly anisotropic and the pore channels can be characterized by two-dimensional parameters (Figure 5): (a) the long axis and (b) the short axis. Depending on the cooling rates, the width of the long axis typically ranges from 100 [mu]m to 600 [mu]m and the short axis from 5 [mu]m to 50 [mu]m. Like nacre, some bridges form between the layers, which are believed to increase fracture resistance. These characteristics contribute to the scaffold's mechanical strength, resulting in a material that can be up to four times stronger than the porous hydroxyapatite scaffolds with similar porosities currently used in bone substitutes (Figure 7). The final scaffold microstructure is a replica of the ice. Modifying the amount of water in the slurry as well as the shape of the solvent crystals will modify the final amount, shape, and size of the porosity.15-17 In particular, during the steady freezing regime, the ice crystals exhibit a homogeneous morphology throughout the sample resulting in a very homogenous lamellae thickness. This also means that the ratio between the thickness of the pore and the thickness of the ceramic lamellae is primarily determined by the initial slurry concentration. For instance, if the volume fraction of ice and ceramic particles is the same, layer thickness (e.g., 20 [mu]m) will be the same as the porosity (20 [mu]m in their smallest dimension). For low slurry concentration, the porosity becomes predominant and layer thickness decreases. The final porosity of the scaffolds ranges from 45 to 75 vol.%.

The microstructural features in freeze casting can be controlled by applying the physics of ice formation. In particular, the ice- tip radius (and as a consequence the thickness of the ice crystals), which is physically determined by the magnitude of supercooling ahead of the freezing front,22 can be modified by increasing or decreasing the cooling rate during freezing. For fast cooling rates, supercooling becomes larger and the width of the pores and the ceramic layers can be scaled down. On the other hand, under a very slow cooling regime the layer (or pore) thickness can be noticeably increased.

JJ. Klawitter and S.F. Hulbert23 established a minimum pore size of-100 [mu]m for bone growth into ceramic structures and a similar conclusion was reached by SJ. Simske et al.24 More recently, AJ. Itala claimed that bone ingrowth occurred in pores as small as 50 [mu]m25 and similar results were obtained for porosity engineered in the range 15-40 [mu]m.26 The porosity must be interconnected to allow the penetration of cells, vascularization, and diffusion of nutrients. Typical porosities for HA scaffolds described in the literature range from 35%27,28 to 75%28 with pore sizes of 50 [mu]m25 to 400 [mu]m.28 A preponderance of pores this large can weaken the implant material. Moreover, even small movements of the implant can cause complications by cutting off blood supply to tissue in the pores, which can lead to inflammation. Also, bone response reflects mesoscale structure and direct bone attachment and conduction along the struts of porous scaffolds has been reported,29 even for small porosity,26 revealing that pore connectivity and orientation are at least as important as pore size. Hence, the porosity range of freeze-casted materials might be suitable for osseous regeneration.


The techniques described show promise for applications that require tailored scaffolds. Because the processes rely on physical rather than chemical interactions, they can be easily extended to any calcium phosphate or suitable ceramic for biomedical applications, opening the way to systematically manipulate the degradation rates and prepare materials with an optimum combination of mechanical properties and bioresorption. Control of the microporosity and roughness of the ceramic walls and struts will also help to manipulate cell attachment and osseointegration.


This work was supported by the National Institutes of Health under Grant No. 5R01 DE015633.

How would you...

...describe the overall significance of this paper?

Every year, millions of patients require bone graft procedures or orthopedic implants. These procedures have profound economic and humane implications. The advance of bone tissue engineering requires the development of new processing technologies for the fabrication of scaffolds with complex architectures. These technologies are also of interest in many different areas from microelectronics to structural materials.

...describe this work to a materials science and engineering professional with no experience in your technical specialty?

This paper describes new techniques for the fabrication of ceramic scaffolds for tissue engineering. The main goal is the development of techniques flexible enough to produce materials with a wide spectrum of solubilities fbioresorption rates) and mechanical properties matching those of calcified tissues: low density, low stiffness, and high strength. These materials could then be tailored for specific applications.

...describe this work to a layperson?

The demand for new therapies to repair bone defects is rapidly increasing. An alternative that has attracted widespread attention is the engineering of new hone to replace the damaged or diseased tissue. A critical component of this approach is the development of porous scaffolds that will provide cell support and guide bone formation. This paper describes new technologies for the production of ceramic scaffolds with complex architectures for bone tissue engineering.


Robocasting inks were prepared by the following two different approaches. The first starts with the preparation of a stable hydroxyapatite suspension with 40-45 vol.% of powder in distilled water. The stability of the suspension was achieved by dissolving the appropriate amount of Darvan C-N(R) dispersant (R.T. Vanderbilt Company, Norwalk, Connecticut) in water and then gradually adding the hydroxy apatite powder (Hydroxyapatite#30, Trans-Tech, Adamstown, Maryland) while shaking vigorously after each addition to improve its homogeneity and stability using a shaker and zirconia balls. An appropriate amount (-7 mg per mL of solution) of previously dissolved hydroxypropyl methylcellulose (Methocel F4M. Dow Chemical Company, Midland, Michigan) was then added to the mixture to increase viscosity. Subsequently, the ink was gellified by adding the proper amount of polyethylenimine as ftocculanl. The viscosity of the final ink was modified to desired consistency by adjusting its pH with HNO^sub 3^ or NH^sub 4^OH as needed. Each addition to the mixture was followed by mixing for ~1 h in the shaker.

The second approach is based on the use of a commercial hydrogel, Pluronic(R) F-127 (BASF Corporation) as a printing vehicle. Pluronic consists of approximately 70 wt.% ethylene oxide and 30 wt.% propylene oxide. It is soluble in cold water (T~5[degrees]C) and it forms a hydrogel at room temperature. The HA powders are dispersed in a mixture of water and Pluronic (10-30 wt.% of Pluronic) with the appropriate amount of Darvan C-N as a dispersant. After adding the hydroxyapatite, the suspension is placed for 1 h in the shaker using zirconia balls as mixing media in order to achieve a homogeneous dispersion and break large aggregates.

In both cases, the final quality of the inks was assessed in terms of printability, measured as the minimum tip diameter suitable to extrude the ink without clogging, and stability (i.e., shape retention capacity during drying and sintering) of the assembled structures.

In order to formulate stable hydroxyapatite slurries with the right composition for the freeze-drying process, the sedimentation of slurries in narrow glass test tubes (~1 cm in diameter) was analyzed as a function of the dispersant (typically ammonium polymethacrylate anionic dispersant Darvan C or Darvan 811 (R.T. Vanderbilt Company, Norwalk, Connecticut) and binder (polyvinyl alcohol) concentrations. Once the optimum compositions were identified, the hydroxyapatite slurries were prepared by mixing distilled water with the dispersant, the binder, and the hydroxyapatite powder in various contents ranging from 15 vol.% to 40 vol.%. Slurries were ball-milled for 20 h with alumina balls and de-aired by stirring in a vacuum dessicator. Controlled unidirectional freezing of the slurries was done by pouring them into a Teflon mold, with two copper rods on each side, which were cooled using liquid nitrogen. Freezing kinetics were controlled by heaters placed on the metallic rods and thermocouples placed on each side of the mold.15-17 Frozen samples were freeze dried for 24 h in order to sublimate the water and leave the green ceramic scaffold.

Sintering of the green bodies produced using both techniques was done in an air furnace with dwell temperatures ranging between 1,25O0C and 1,350"C (heating rate 30C/ min.) and sintering times from 2 h to 6 h. In the case of materials prepared by robocasting, the samples were dried in air at room temperature for 24 h and then at 40O0C (I0C/ min. heating rate) for 1 h to evaporate organics prior to sintering.

The starting powders and sintered samples were characterized by x- ray diffraction and scanning-electron microscopy. In addition, the sintered density and open porosity of the samples were characterized using Archimede's method. Preliminary uniaxial compression tests were performed using crosshead speeds ranging from 0.01 mm/s to 0.015 mm/s in ~4 x 4 x 5 mm^sup 3^ samples cut from the scaffolds using a diamond saw.


1. R. Langer and J.R Vacanti, Science, 260 (1993), pp, 920-926.

2. A. Almirall et al., Biomat., 25 (2004), pp. 3671-3680.

3. R.P. del Real et al., Biomat., 23 (2002), pp. 3673-3680.

4. H.R. Ramay and M. Zhang, Biomat., 24 (2003), pp. 3293-3302.

5. M. Sous et al., Biomat, 19 (1998), pp. 2147-2153.

6. M. Kawata et al., J. Mater. Sci. Mater. Med., 15 (2004), pp. 817-823.

7. J.C. Le Huec et al., Biomat., 16 (1995), pp. 113-120.

8. D.-M. Liu, Ceramics International, 23 (1997), pp, 135-139.

9. M. Milosevski et al., Ceramics International, 25 (1999), pp. 693-696.

10. A. Bignon (Ph.D. Thesis, National Institute of Applied Science, Lyon, France, 2002).

11. S. Michna, W. Wu, and J.A. Lewis, Biomat., 26 (2005), pp. 5632-5639.

12. J. Russias et al., Journal of Biomedical Materials Research Part A, 83A (2007), pp. 434-445. 13. J.E. Smay, J. Cesarano, and J.A. Lewis, Langmuir, 18 (2002), pp. 5429-5437.

14. P Miranda et al., Acta Biomaterialia, 2 (2006), pp. 457-466.

15. S. Deville et al., Science, 311 (2006), pp. 515-518.

16. S. Deville, E. Saiz, and A.P.Tomsia, Acte Materialia, 55 (2007), pp. 1965-1974.

17. S. Deville, E. Saiz, and A.P. Tomsia, Biomat, 27 (2006), pp. 5480-5489.

18. H.I. Zhang et al., Nature Materials, (2005), pp. 787-793.

19. M.C. Gutierrez et al., Advanced Materials, 18 (2006), pp. 1137-1140.

20. R. Murugan and S. Ramakrishna, Composites Science and Technology, 65 (2005), pp. 2385-2390.

21. V. Karageorgiou and D. Kaplan, Biomat, 26 (2005), pp. 5474- 5491.

22. J.D. Hunt, Materials Science & Technology, 15 (1999), pp. 9- 14.

23. J.J. Klawitter and S.F. Hulbert, Journal of Biomedical Materials Research, 2 (1971), pp. 161-229.

24. S.J. Simske, R.A. Ayers, and T.A. Bateman, Porous Materials for Tissue Engineering (Enfield, NH: Transtech, 1997), pp. 151-182.

25. A.I, Itala et al., Journal of Biomedical Materials Research, 58 (2001), pp. 679-683.

26. Tithi Dutta Roy et al., Journal of Biomedical Materials Research, 66A (2003), pp. 283-291.

27. R.M. Pilliar et al., Biomat., 22 (2001), pp. 963-972.

28. N. Tamai et al., Journal of Biomedical Materials Research, 59 (2002), pp. 110-117.

29. S. Zmora, R. Glicklis, and S. Cohen, Biomat., 23 (2002), pp. 4087-4094.

E. Munch, J. Franco, S. Deville, P. Hunger, E. Salz, and A.P. Tomsia are with the Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley CA 94720. Current address for S. Deville: Laboratoire de Synthese et Fonctionnalisation des Ceramiques, FRE2770 CNRS/St Gobain CREE 550, Avenue Alphonse Jauffret, BP 224,84306 Cavaillon CEDEX. Dr. Tomsia can be reached at [email protected]

Copyright Minerals, Metals & Materials Society Jun 2008

(c) 2008 JOM. Provided by ProQuest Information and Learning. All rights Reserved.