Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite

G Model

ARTICLE IN PRESS

JECS-10375; No. of Pages 11

Journal of the European Ceramic Society xxx (2015) xxx–xxx

Contents lists available at www.sciencedirect.com

Journal of the European Ceramic Society journal homepage: www.elsevier.com/locate/jeurceramsoc

Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite Marie Lasgorceix, Eric Champion ∗ , Thierry Chartier Université de Limoges, CNRS, ENSCI, SPCTS, UMR 7315, F-87000 Limoges, France

a r t i c l e

i n f o

Article history: Received 24 July 2015 Received in revised form 16 November 2015 Accepted 18 November 2015 Available online xxx Keywords: Porous bioceramics Microstereolithography Sintering Silicon-substituted hydroxyapatite Bone graft substitutes

a b s t r a c t Additive manufacturing of silicon substituted hydroxyapatite (SiHA) ceramics with controlled macro–micro-porous architecture by microstereolithography and sintering is reported. Due to the role of silicon in bone calcification, the incorporation of silicate in hydroxyapatite has become of interest for applications in bone tissue engineering. But, the shaping and the sintering of SiHA remain few studied. For the shaping process, the formulation of a photopolymerizable suspension and microstereolithography parameters were optimized. Adjustment of the sintering parameters allowed the production of ceramics with controlled open microporosity in a wide range of variation, while preserving phase pure SiHA. A dimensioning model that takes into account the overcure due to light scattering during photopolymerization and the shrinkages during sintering was established. Using this method, macropores of various cross-sections, within the size range of interest for bone ingrowth, were shaped demonstrating the efficiency of microstereolithography for the direct manufacturing of bioceramic scaffolds with accurate architecture. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction Synthetic hydroxyapatite (HA) (Ca10 (PO4 )6 (OH)2 ) is commonly used as bone graft substitute because of its good biocompatibility and its faculty to conduct bone formation (i.e., cell adhesion, proliferation and differentiation) onto its surface [1,2]. According to the “diamond concept” described by Giannoudis et al. [3], in order to improve bone restoration, tissue engineering strategies involve porous scaffolds, growth factors, mesenchymal stem cells and an optimal mechanical environment. In this scheme, the architectural and microstructural properties of scaffolds are essential factors. An interconnected macroporosity, in the size range 300–600 ␮m, with interconnections of about 50–100 ␮m, is necessary for cell attachment, migration and proliferation within the implant, and for vascularization [4–6]. More, some studies showed increased osteoconduction, and even osteoinduction, in the presence of micropores in the macroporous architecture [7–12]. Cells are also sensitive to the geometry of the support on which they grow [13]. Very recently, Bidan et al. highlighted the effect of pore

∗ Corresponding author at: Université de Limoges, CNRS, SPCTS, UMR CNRS 7315, Centre Européen de la Céramique, 12 rue Atlantis, 87068 Limoges Cedex, France. Fax: +33 555 87 50 23 04. E-mail addresses: [email protected] (M. Lasgorceix), [email protected] (E. Champion), [email protected] (T. Chartier).

and surface curvatures on cell behavior [14,15]. But, the influence of macropore geometry and of open microporosity on cell behavior and vascularization remains misunderstood. In this aim, model ceramic supports with multiscale porosity have to be manufactured with good accuracy. Vector-by-vector stereolithography (SLA), an additive manufacturing (AM) technology, was used by several researchers to produce 3D scaffolds with complex shapes for tissue engineering [16,17]. This technique consists of shaping parts layer-by-layer by scanning a cross-sectional pattern of the part on each layer using a UV laser beam. The principle is based on the photopolymerization of a photosensitive system. This method was mainly applied for the manufacturing of polymeric scaffolds. Indeed, as pointed out by Skoog et al. [17], “direct stereolithography of bioceramic composite scaffolds (or bioceramic-polymer) is difficult since the addition of inorganic components increases the resin viscosity; in addition, bioceramic particles may interfere with irradiation of the photo-polymer. Due to limitations on the bioceramic concentration via the direct stereolithography approach, a number of studies have utilized indirect stereolithography (e.g., fabrication of a negative mold using stereolithography)”. Despite these limitations, this technique can be used for the direct fabrication of ceramic parts [18]. It has been successfully used to produce complex on demand cranial implants made of hydroxyapatite (HA) [19]. Projection microstereolithography (P␮SLA) derives from SLA. In this case, the whole surface of each layer that must be polymerized

http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020 0955-2219/© 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model JECS-10375; No. of Pages 11

ARTICLE IN PRESS M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

2

Fig. 1. Masks used for the shaping of model parts. The black portion corresponds to pixels “on”, i.e., light oriented by the DMD toward the surface of slurry to be polymerized. (a) Configuration 1; (b) configuration 2.

Fig. 2. Principle of overcure detection by image processing: (a) image of the mask, (b) photograph top view of a monolayer part shaped according to the mask, (c) superposition of the mask on the part photograph (d) red coloration of the overcured areas of the part (e) overcured areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

is fully illuminated by a UV light which is selectively reflected by a dynamic mask using an optical system [16,20]. Thus, P␮SLA permits to reach a high geometric accuracy with a low building time [21]. Nevertheless, the same difficulties as those mentioned above arise. They come from the slurry formulation that must be optimized and from the interference between the UV light and the ceramic particles during the photopolymerization process of the resin. For ceramic particles loaded systems, the interaction between the UV beam and the photosensitive resin can be described by Jacobs equation (Eq. (1)). In this equation, the cure depth (Cd ) relates to the incident energy (Ei ), the critical energy requied for photopolymerization (Ec ) and the penetration depth (Dp ) as follows: Cd = Dp × ln(

Ei ) EC

(1)

Dp illustrates the divergence from unloaded systems and takes into account the effects of interaction phenomena between the UV beam and the solid particles. Chartier et al. [22] showed that the interaction of a UV light with ceramic particles loaded slurries was mainly governed by scattering phenomenon rather than absorption. This study investigated the influence of particle size and powder loading on the reflectance and transmittance values and established a relationship between the penetration depth (Dp ) and the absorption and scattering coefficients. A quantitative model was proposed to predict the cure depth (Cd ) knowing these coefficients. Other physics-based or experimental models for ceramic particles loaded slurries have been developed to describe light attenuation in the depth direction and to predict solidified layer thickness [23–25]. However, there is no physical model for light attenuation in the direction perpendicular to the incident beam, i.e., in the horizontal direction. Consequently, the lateral overcure phenomenon, in the horizontal direction and its consequences on the geometry of a ceramic part, cannot be anticipated. It is of prime importance to determine this effect in the case of complex geometries including angular zones. The monitoring and the quantification of roundness errors were proposed by Bail et al. [26] for the building of cylindrical features using unloaded resins. Kang et al. [27] proposed a model to predict solidified 2D horizontal pattern profiles for projection

stereolithography technologies. This model appeared successful in the case of unloaded resins, but it cannot be applied to the curing of ceramic loaded slurries because it does not take into account the scattering effect due to ceramic particles. Only Gentry and Holloran investigated recently the cure width after curing straight lines of ceramic loaded suspensions [24,28]. But, to our knowledge, there is not any study reporting the monitoring and the quantification of the lateral overcure phenomenon during the shaping of complex and angular geometries such as those of interest for the pores structure in bioceramic scaffolds. Finally, little attention has been paid to the design of complex ceramic parts with accurate dimensions and geometries such as scaffolds. This requires a study of overcure phenomenon, optimization of the slurry formulation and precise adjustment of the shaping parameters. Silicon substituted hydroxyapatite (SiHA) (Ca10 (PO4 )6−x (SiO4 )x (OH)2−x ) was chosen. Its biocompatibility was shown in a previous work [29]. Additionally, as stated by several authors, its bioactivity would be higher than that of pure HA [30–34]. Some of these results are discussed [35]. The doubts about the bioactivity of SiHA result from the incomplete material characterization (possible presence of secondary or amorphous phases containing silicon) and the lack of quantitative information concerning the in vivo Si release and its effect on bone growth. The incorporation of silicate groups in the apatite lattice causes formation of Si OH groups at the surface of the material, which may constitute favorable sites to organo-alkoxysilanes used in the silanization step for the grafting of bioactive molecules [36,37]. This aspect constitutes an additional criterion for the incorporation of silicon in HA. Doubts and lack of knowledge about this material, promising for bone tissue engineering, lead researchers to better investigate it. But, up to now, the sintering of SiHA material has been the subject of very few studies [38–41] in contrast to the sintering of pure HA which is well known. In this context, this work aimed at investigating the processability by P␮SLA and further sintering of SiHA bioceramic parts with various geometries of macropores in the desired size range (300–600 ␮m) and with a controlled amount of open microporosity in a wide range of variation. It focusses on the analysis of overcure phenomenon during the photopolymerization, the sintering

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model JECS-10375; No. of Pages 11

ARTICLE IN PRESS M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

3

Fig. 3. Determination of overcure inside angles: photograph of the part (a); staining the overcured areas in the pores (b); extraction of the angles of same total area (c); calculation of the overcure inside angles (d).

of SiHA ceramics and the establishment of an empirical dimensioning model for the reliable production of these macro–micro-porous parts. 2. Materials and methods 2.1. Material preparation 2.1.1. SiHA powder A powder of chemical composition Ca10 (PO4 )5.6 (SiO4 )0.4 (OH)1.6 was used. The raw powder was synthesised by an aqueous precipitation method as described previously [42]. In order to incorporate the silicon in the apatite lattice, the raw powders were first heat treated at 1200 ◦ C for 30 min in a Nabertherm furnace in air atmosphere, with heating and cooling rates of 10 ◦ C min−1 . Then, it was milled in ethanol with the addition of phosphate ester, by attrition at 800 rpm during 8 h (attritor Netzsch R41-25/4), to break the powder aggregates formed during the calcination. 2.1.2. SiHA slurry preparation A UV curable slurry was prepared by dispersing SiHA milled powder in an organic mixture containing an amine modified polyester acrylate resin, reactive at 365 nm (wavelength of the UV source of the P␮SLA equipment), a reactive diluent, a photoinitiator (ethanone, 2,2-dimethoxy-1,2-diphenyl-; EDMD) also absorbing in the range of wavelength of the UV source and a phosphate ester used as electrosteric dispersing agent. The slurry formulation results from a good compromise between the reactivity of the slurry, a high ceramic powder loading and a high definition of the shaped parts. The slurry needs to be sensitive to the working wavelength (365 nm), loaded in ceramic powder over 50 vol%, to confer a good cohesion to the green parts and to prevent deformation/degradation during handling and further debinding step (elimination of the organic phase). The slurry must have a shear thinning behavior with a yield stress  0 preventing its spontaneous

flow and it must have a viscosity lower than 5 Pa s, when submitted to spreading shear rate so as to deposit thin and homogeneous layers. Indeed, for higher values of viscosity, the slurry could not be spread homogeneously. On these bases, the formulation selected for the study corresponded to a ratio resin/diluent equal to 70/30 and a ceramic content of 55 vol%. It was chosen from a preliminary study which will not be detailed here [43]. 2.1.3. Shaping of green parts The P␮SLA equipment is an integral projection dynamic mask process [16]. This technique is based on the full illumination of a 2D pattern on the surface of a UV photosensitive slurry spread on the support by a blade. The sample is produced layer by layer by polymerization of successive patterns generated by the projection of a bidimensional (2D) image on the slurry surface. The UV light, with an emission peak at 365 nm, is partially reflected to the slurry surface according to the tailored pattern thanks to a dynamic Digital Micro-mirror Device (DMD, Texas Instrument). The DMD is constituted by a network of 1024 × 768 square micro-mirrors of 14 × 14 ␮m2 being individually able to tilt at ±12◦ along their diagonal thanks to electrostatic forces [22]. The theoretical definition allowed by the system is 14 ␮m, which corresponds to the micromirror dimensions. The configuration of the DMD is governed by CAD files of patterns called “masks” (Fig. 1). The exposure time (texp ) is monitored by an electronic shutter. Green parts were shaped according to two configurations (Fig. 1). The configuration 1 without macropore was used for the reactivity tests and the sintering study. The configuration 2, containing five pore geometries and three pore sizes for each geometry, was considered for the quantification of the overcure and for the establishment of the predictive dimensioning model. The five geometries were chosen so as to assess the effect of corners on the lateral overcure. The circular shape was used as a reference and the other four geometries gave a wide range of angles values (Table 1). These geometries were selected in view of further biological

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model JECS-10375; No. of Pages 11 4

ARTICLE IN PRESS M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

Fig. 4. Granulometric distribution of SiHA powder versus attrition milling time from 0 min (as thermally treated at 1200 ◦ C for 30 min) up to 8 h.

Fig. 5. Experimental data of shear stress (rheogramme; red points) and viscosity (flow curve; blue crosses) for an increasing shear rate, and simulated Herschel–Bulkley model (Eq. (4), black curve). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

purposes on the basis of previous works performed by Bidan et al. [14,15] highlighting the influence of pore geometries and surface curvatures on cell behavior. The geometries studied in the present work were also chosen in order to extend the range of macropore angles values. The P␮SLA processing parameters, i.e., exposure time (texp ) and spread slurry thickness (Ep), were determined from the working curve established as mentioned in Section 2.2.2. All the samples were produced using the same batch of slurry and were cleaned post building in an ultrasonic solvent bath for 5 min, several times, until the getting of a complete clear solution. Then, they were air-dried overnight at room temperature.

rate of 20 ◦ C min−1 , at different temperatures for several times. The sintering parameters were established, according to dilatometric assays performed in a horizontal dilatometer (Netzsch DIL 402C). The thickness variation of green samples was measured by the displacement of an alumina probe applying a load of 0.3 N on the sample. An air flow of 100 mL min−1 was maintained during the whole thermal cycle. A correction was made to remove the contribution of the alumina thermal expansion in order to obtain the sole dimensional changes of the sample.

2.1.4. Binder removal and sintering of ceramic parts The composite green samples were debinded with a slow heating rate of 1 ◦ C min−1 up to 400 ◦ C in air atmosphere to remove the polymeric phase. The debinding temperature was determined from thermogravimetric analysis (Netzsch, STA 449 F3) performed on a green sample (data not shown). Then, the parts were sintered in a Super Kanthal furnace in air atmosphere with a heating

2.2.1. Powder characterization The purity of the powder was verified according to the ISO 13779-3:2008(F) [44] (data not shown). The silicon content (i.e., 0.42 ± 0.05 mol) was measured by Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP/AES), after dissolution of 0.1 g of calcined powder in acidic solution prepared with 4 mL HNO3 and 10 mL HF. Powder XRD data were collected using CuK˛

2.2. Material characterization

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model

ARTICLE IN PRESS

JECS-10375; No. of Pages 11

M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

5

Table 1 Macropore geometries with corresponding angles values; overcure values (S) measured in the salient angles, on green part; limit of surface area values (Sth ) of pores on the mask below which the pores are completely obstructed during the shaping; correlation coefficient (R2 ) of the linear model established between the real surface value (S0 ) of macropores on the green part and the theoretical surface (Sth ) of macropores on the mask. Macropore geometry

Angle value (◦ )

S (%) measured in salient angles

Limit Sth (mm2 )

R2 (model)

Star Triangle Rhombus

53 60 80 100 / 90

75 71 66 47 / 50

0.150 0.151 0.141

0.996 0.995 0.998

0.118 0.124

1.000 1.000

Circle Square

Fig. 6. (a) Top view of a 5-layers green part, shaped by P␮SLA with an exposure time of 0.83 s (Ei = 71.5 mJ cm−2 ) and a spread layer thickness of 255 ␮m (non-adherent layers); (b) top view of a 5-layers green part with an exposure time of 0.83 s and an insufficient spread layer thickness leading to partial macropores obstruation; (c) view of a 10-layers green part shaped with an exposure time of 0.83 s and a spread slurry thickness of 250 ␮m (adherent layers).

radiation on a /2 diffractometer (Siemens, model D5000) over the 25–40◦ range with a step size of 0.02◦ and a count time of 4 s, in order to identify the crystalline phases. Phase identification was achieved by comparing the diffraction patterns with International Centre for Diffraction Data (ICDD) standards (PDF cards) processed on EVA software (Brukers AXS). Complementary, Fourier transformed infra-red (FTIR) absorption spectra were collected using a Perkin Elmer Spectrum One FTIR spectrometer, to determine the ionic groups present in the powders. They were recorded over the 400–4000 cm−1 region with a resolution of 2 cm−1 . All the spectra were normalized in respect to the 4 band of phosphate (602 cm−1 ) [45]. The particle size distribution was determined with a laser granulometer provided with a LD source of 655 nm and a LED source of 405 nm (Horiba LA950V2). The specific surface area of powders was determined by the BET method 8 points using N2 gas (Analyser Micromeritics ASAP 2010) after degassing under vacuum at 250 ◦ C. 2.2.2. Slurry characterization Rheological characterization of the slurry was performed at 25 ◦ C using a cone-plane configuration on a controlled stress rheometer (AR-G2, TA Instruments) equipped with a stainless steel cone of 40 mm diameter and 2◦ angle. The gap between the cone and the plane was set to 800 ␮m. Shear stress and viscosity values were recorded during an increase of shear rate from 0 to 160 s−1 . The rheological parameters were adjusted according to the Herschel–Bulkley law (Eq. (2)) in the domain of interest (0–50 s−1 ) using TRIOS® software (TA Instrument).  = 0 + k ×  n

(2)

where  is the shear stress (Pa),  0 the yield stress (Pa), k the consistency index,  the shear rate (s−1 ) and n the power law order (n < 1 for shear thinning behavior). The viscosity value was considered at 150 s−1 , which corresponds to the shear rate applied by the blade on the slurry during its spreading. The reactivity of the slurry was evaluated determining the minimal energy necessary to induce polymerization. This critical energy (Ec ) was assessed measuring the thickness (Cd ) of polymerized monolayers after several exposure times (texp ). Experimentally, the slurry was kept in a glass container of 5 mm depth, so that the thickness of slurry was not limiting for the beam penetration depth. The

surface of the slurry was smoothed with a glass slide. The UV light was projected according to the mask Fig. 1a. After illumination, the square solid part was removed, rinsed three times with a solvent in an ultrasonic bath, and then air-dried overnight at room temperature. Five thickness measurements were performed on each monolayer part using a caliper (OTMT microextDigit, 0–25 mm) and averaged. The working curve, giving the values of the polymerized thickness (Cd ) versus the logarithm of the exposure time (texp ), was then plotted. A calibration curve was obtained by measuring, for several exposure times (texp ), the energy (Ei ) received by a UV radiometer (VLX-3W). The radiometer was provided with a photoelectric cell type probe calibrated for the spectral region of UV A (355–375 nm), therefore sensitive to the wavelength of the irradiating light (365 nm). The incident energy values (Ei ) were then plotted versus the exposure time (texp ) to establish the calibration curve. Finally, the thickness values (Cd ) were plotted versus the incident energy (Ei ) in order to extract the critical energy (Ec ) characteristic of the slurry reactivity, in agreement with Eq. (1). 2.2.3. Characterization of green and sintered parts External dimensions (length, width and height) of the green parts (configuration 1) were measured with a digital caliper. Macroporous green parts (configuration 2) were observed by optical microscopy (Zeiss Stemi SV6). The mask picutre was overlaid on the part picture in order to highlight the overcure portions (Fig. 2). In each macropore, the overcure amount S was defined as the ratio of the overcured surface to the macropores surface on the mask. These two surface values were assessed with Photoshop-CS6 Software, in px2 . In order to evaluate the effect of the angle value, the overcure amount was also determined in the angles of the different geometries by closing them with a fixed total surface (Fig. 3). The open microporosity amount of the sintered samples (according to configuration 1) was measured by the Archimedes’ method in water with a microbalance of 10−4 g accuracy. The results were averaged on three samples. The specific surface area was determined by the five points BET method using Kr gas (Analyser Micromeritics ASAP 2020 V4.01) after degassing under vacuum at 200 ◦ C during 24 h. The results were averaged on 2 samples. The microstructure was observed by scanning electron microscopy (SEM), using a Philips XL30 microscope with a 10 kV accelerating voltage, after a platinum metallization of 15 nm thickness.

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model

ARTICLE IN PRESS

JECS-10375; No. of Pages 11

M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

6

Fig. 7. (a) Evolution of the overcure S in macropores of a monolayer green part versus the theoretical surface Sth of the pore according to the mask, for five geometries of macropores (same shape as the experimental dots on figs), and fitted mathematical laws; (b) logarithmic transformation with linear extrapolation; (c) theoretical macropore surface values Sth (in px2 ) on the mask of 1024 × 768 px2 versus macropore real surface values on the green parts S0 (in mm2 ); (d) logarithmic transformation with linear extrapolation, zoom on the convergence zone.

Table 2 Relative changes of sample thickness, measured by horizontal dilatometry for three sintering temperatures. Sintering Temperature

Shrinkage between 0 h and 2 h

Shrinkage between 1 h 55 min and 2 h 05 min

Proportion of shrinkage between 1h55 min at 2h05 min versus the shrinkage at 2 h 00 min

Global Shrinkage

1160 ◦ C 1180 ◦ C 1200 ◦ C

7.0% 8.3% 9.4%

0.12% 0.08% 0.03%

1.7% 1.0% 0.3%

19.8% 20.8% 21.2%

The purity of the sintered parts was checked by XRD and FTIR as described above. The horizontal and vertical shrinkages were determined with a digital caliper measuring the external dimensions of the samples after sintering.

3. Results and discussion 3.1. SiHA Powder and photosensitive slurry SiHA powder heat treated at 1200 ◦ C for 30 min is pure phase apatite, indexed according to PDF card 9-432, without any detectable secondary crystalline phase (XRD data not shown). The FTIR spectrum (data not shown) exhibits the typical vibration bands of phosphate and hydroxide groups of HA [46], the specific bands of silicate groups and some vibration bands of Si OH bonds present on SiHA surface [47]. No band corresponding to

Si O bond of amorphous silica (at about 873, 798, 683, 535 and 515 cm−1 [47]) appears. The calcination treatment results in a low specific surface area (2.4 ± 0.4 m2 g−1 ) with coarse particle size distribution due to the grain coalescence (Fig. 4). It evolves to a narrower distribution that becomes almost monomodal after 8 h of attrition milling with an average particle size Dv50 = 0.30 ␮m (Dv10 = 0.15 ␮m, Dv90 = 0.80 ␮m). The final specific surface area is 10.9 ± 1.2 m2 g−1 . The rheogramme and flow curve indicate that the SiHAorganic slurry is a plastic shear-thinning fluid (Fig. 5). A very good correlation is found between the experimental data and the Herschel–Bulkley law (Eq. (2)) with a power order n = 0.84 and a yield stress value  0 = 140 Pa. The slurry possesses a viscosity of  = 1.7 Pa s at 150 s−1 . Therefore, it exhibits adequate flow properties for P␮SLA process. Moreover, it is sensitive to the UV wavelength, with a low critical energy (Ec ) of 14 mJ cm−2 accord-

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model JECS-10375; No. of Pages 11

ARTICLE IN PRESS M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

7

Fig. 8. (a) X-ray diffraction patterns of samples sintered for 2 h at 1160 ◦ C, 1180 ◦ C, 1200 ◦ C, 1220 ◦ C and 1240 ◦ C; (b) zoom of X-ray diffraction patterns of samples sintered for 2 h at 1220 ◦ C and 1240 ◦ C; phases identified with PDF cards of ICCD numbers 9–432 and 9–348 for apatite and ␣-TCP respectively.

ing to the linear regression, established from the experimental data of the thickness values (Cd ) versus the incident energy (Ei ), which was in very good correlation with Eq. (1). The experimental values of the working curve, giving the polymerized thickness Cd (␮m) versus the exposure time texp (s) fitted with very good correlation (R2 = 0.983) to the following empirical linear regression (Eq. (3)): C d = 129.6Ln(t exp ) + 279.5

(3)

3.2. Shaping of green parts by PSLA

Fig. 9. FTIR spectra of samples sintered at 1160 ◦ C, 1180 ◦ C, 1200 ◦ C, 1220 ◦ C and 1240 ◦ C, for 2 h.

By setting the spread layer thickness (Ep ) at the value of cured thickness (Cd ) corresponding to a given exposure time (texp ) (for example Ep = Cd = 255 ␮m for texp = 0.83 s, Eq. (3)), the final part consists of a stack of layers non-adherent to each other (Fig. 6a). It is necessary to reduce the thickness of spread layers (Ep ) to compensate the slight shrinkage due to the polymerization process and to generate a vertical overcure at the interface between the layers, which will induce their adhesion. However, this can also cause a slight lateral overcure that affects the definition of the porous geometries [25]. The more the spread layer thickness is reduced the higher is the lateral overcure. Too high lateral overcure can lead to partial or complete obstruation of the porous geometries (Fig. 6b). A minimum reduction of 5 ␮m of the spread slurry thickness (Ep ), compared to the nominal cured thickness (Cd ), is necessary to allow the layers to adhere perfectly together (Fig. 6c). The lateral overcure, which directly affects the definition of porous geometries, must be quantified. This quantification is intended to determine the achievable geometry and dimensions in the shaping conditions (i.e., for a given slurry and given process parameters) and to provide data in order to adapt the masks and/or the process parameters to the desired final geometry. The overcure due to the lateral scattering of the UV beam on ceramic particles and to the pixel size (micromirror size) which both limit the lateral definition (i.e., the total lateral polymerization that exceeds the masks limits), was determined on a monolayer part. The lateral overcure values measured in the macropores are plotted versus the respective theoretical surface values of macropores on the mask, for several macropore geometries (Fig. 7a). Each point is the average of values obtained on two identical macropores. Regardless the pore geometry, the overcure increases with the decrease of the pore size, in very good correlation with a power law. These laws can be linearized by logarithmic transformation (Fig. 7b). The linear relationships established between the real surface S0 of a macropore on the green part and its theoretical surface Sth on the mask, define a predictive linear model for the design of pores on the masks, i.e., the pore size in px2 on the mask versus its real size S0 (mm2 ) on the shaped part (Fig. 7c). The correlation coefficients R2 (Tab. 1) indicate that the model is accurate for simple geometries with few sharp angles (i.e., circular or square geometries). But, the correlation coefficients testify a slight distance to the linear model for the other geometries and particularly in the case

of acute angles. Fig. 7d, focused on the area of theoretical obstruation of pores, highlights the effect of the macropore geometry. The theoretical limits of pore size on the mask, from which the pores are completely filled (overcure S = 100%) are determined by linear extrapolation (Fig. 7b,d). The logarithms of the theoretical minimum areas correspond to the abscissa of the intersection points of the lines representative of these linear relationships with the straight line of equation y = ln(100). Though the intersection points overall converge toward a narrow area (Fig. 7d), the limit value of Sth (Tab. 1) is more rapidly reached for the geometries having acute angles (triangle, rhombus and star) (about 0.15 mm2 ) than for circular or square geometries (about 0.12 mm2 ) (Table 1). The overcure is favored by the presence of sharp corners and it increases with the decrease of the angles value. To verify this hypothesis, the overcure was measured inside the salient angles of each geometry after cutting the angles at the same surface area on the mask (Fig. 3). The results, summarized in Table 1, confirm the increase of the overcure when decreasing the value of the salient angle. The overcure causes rounding of pore edges and the magnitude of this phenomenon increases with decreasing the pore size (see example of star geometry in Fig. 14). All the geometries tend towards a substantially circular shape with decreasing pore size, which explains the convergence of the limit values noted in Fig. 7b,d. However, the geometries are well defined for pore having a minimum size of about 300 ␮m, which corresponds to the interesting size range of macropores for the field of applications considered in this study. In addition to the diffusion effect by the ceramic particles [22], corners rounding can be attributed to a superposition effect assuming that the distribution of the beam reflected by each micromirror is Gaussian [48]. Indeed, Sun et al. [49] showed that the intensity of a pixel on the image plane may be considered as the first order approximation of Gaussian distribution. This distribution was observed experimentally [24,28]. This concept was considered by some authors for modeling the exposure in P␮SLA, i.e., the energy distribution received by the slurry surface [50]. So, because of these technical limitations, there is a pore size range for which the geometries are poorly defined, especially for those having small angles.

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model JECS-10375; No. of Pages 11 8

ARTICLE IN PRESS M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

Fig. 10. Specific surface area (measured by BET method, averaged over two samples) and open microporosity amount (measured by Archimedes method, averaged over three samples) versus the sintering time for a sintering temperature of 1160 ◦ C (a) and versus the sintering temperature for a sintering time of 2 h (b).

Fig. 11. SEM micrographs of SiHA P␮SLA samples sintered at 1160 ◦ C for different sintering times or for 2 h at different temperatures (scale bars = 5 ␮m).

3.3. Sintering and microstructure of SiHA ceramics For the sintering cycles, an initial slow rate of temperature increase, set at 1 ◦ C min−1 up to 400 ◦ C, was applied to eliminate slowly the organic phase while maintaining the organization of the ceramic grains according to the architecture initially defined by the polymer network. Then, the heating rate was set at 20 ◦ C min−1 up to the sintering temperature. In order to control the open microporosity of the ceramic parts, the strategy was to define sintering temperature and time so as to obtain reproducible porosity amounts in a wide range. Since the SiHA phase can decompose during the thermal treatment, sintering temperatures up to 1240 ◦ C were tested. Only the apatite phase is detected on the XRD patterns of the samples sintered at temperatures up to 1220 ◦ C (Fig. 8a). The absence of secondary phase in these samples is confirmed by FTIR spectroscopy (Fig. 9). The spectra also show the disappearance of OH libration band of hydroxyapatite at about 630 cm−1 after sintering. Dehydroxylation occurs during heating above 1000 ◦ C. The reverse reaction of hydroxylation during cooling can only proceed at the grains surface in direct contact with the surrounding atmosphere (i.e., ambient air

containing moisture for natural sintering) and the bulk material is mainly oxyapatite as it can be the case for pure hydroxyapatite [41]. ␣-TCP secondary phase resulting from the thermal decomposition of SiHA is detected in the samples sintered at 1240 ◦ C. In addition, the relative intensity of the peak at 33◦ , relative to the other peaks of the same pattern, is much higher than in the case of the samples sintered at lower temperatures (Fig. 8b). This increase of intensity is due to the contribution of the main peak of Ca2 SiO4 , another secondary phase resulting from the thermal decomposition of the SiHA. The presence of secondary phases in the sample sintered at 1240 ◦ C also results in additional shoulders on the FTIR spectra (Fig. 9). In order to obtain pure SiHA without phase decomposition and residual secondary phases, a sintering temperature of 1220 ◦ C for 2 h, cannot be exceeded. These results are in agreement with previous works [38–41]. The linear shrinkage values for the three studied temperatures (1160, 1180 and 1200 ◦ C), between the beginning of the sintering and a sintering time of 2 h are reported in Table 2. The differences of linear shrinkage recorded are significant. Moreover, the rate of linear shrinkage registered for two hours of sintering, measured over an interval of 10 min (i.e., between 1 h55 min and 2 h05 min),

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model JECS-10375; No. of Pages 11

ARTICLE IN PRESS M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

9

Fig. 12. Specific surface area values (measured by BET method, averaged over two samples) versus the open microporosity amount (measured by Archimedes method, averaged over three samples); Standard variation of specific surface area values is less than 0.01 m2 /g (not noticeable on the graph) ensuring significant differences.

Fig. 13. Linear shrinkage versus the sintering time for a temperature of 1160 ◦ C (a) and versus the temperature for a sintering time of 2 h (b) (mean values obtained from three samples for each thermal cycle).

is less than 2% of the shrinkage value at 2 h (Table 2). These low dimensional changes ensure a good reproducibility of the results for this sintering time of 2 h. Open microporosity and specific surface area values are plotted versus the sintering temperature in Fig. 10. An open microporosity is kept for sintering temperatures up to 1220 ◦ C and the amount increases from 15 vol% at 1220 ◦ C up to about 32 vol% at 1160 ◦ C. A decrease of the sintering time at 1160 ◦ C allows similarly the controlled expansion of open microporosity amount up to approximately 37 vol%, value obtained for 0.2 h at 1160 ◦ C (Table 2). The fast fall in surface area (from 2.2 down to 1.6 m2 g−1 ) while increasing the sintering duration from 0.2 h to 0.5 h at 1160 ◦ C is due to the formation of necks between the grains, notable on SEM micrographs (Fig. 11). This phenomenon matches the primary stage of sintering. A decrease of the sintering temperature or time is not possible because it leads to a lack of grains welding, resulting in a low cohe-

sion of the ceramic. Finally, the range of interest to produce SiHA ceramics without phase decomposition and having a controlled and reproducible open porosity is limited to 1160 ◦ C–0.5 h up to 1220 ◦ C–2 h. The values of specific surface area versus the values of open microporosity amount are plotted in (Fig. 12). The evolution is almost linear, from 0.5 m2 g−1 for an open microporosity of 15 vol% (samples sintered at 1220 ◦ C for 2 h) up to a maximum value of 1.6 m2 g−1 for 35 vol% of porosity (samples sintered at 1160 ◦ C for 0.5 h). 3.4. Dimensioning predictive model Shrinkage values, averaged over three samples for each sintering cycle, are shown in Fig. 13. They vary in a range of 5–20% depending on the sintering parameters in agreement with the experimen-

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model JECS-10375; No. of Pages 11 10

ARTICLE IN PRESS M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

Fig. 14. Evolution of the overcure with macropore dimensions for five geometries and illustration of corners rounding in the case of the star.

tal values of residual porosity volume. The horizontal shrinkage was measured in two perpendicular directions and could be considered homogeneous in the horizontal plane assuming that there was no local deformation of the part during heat treatment, which was verified by image analysis. The vertical shrinkage is always larger than the horizontal one. This can be attributed to the nonuniform arrangement of particles during the spreading of the layers, similarly to tape casting. These results of shrinkage enable the dimensioning of the green parts. The thickness of the green parts is calculated according to the desired final thickness, knowing the vertical linear shrinkage. The thickness of the green part can be reached choosing the layer thickness and the number of layers, and adapting the exposure time according to the experimental working curve (Eq. (3)). The 2D horizontal geometries of the green part can be calculated similarly from the horizontal shrinkage according to the desired open microporosity or to the target specific surface area values. Then, the 2D dimensions of the geometries on the mask, that are necessary to produce the green part geometries, are fixed according the predictive dimensioning model resulting from the overcure quantification (Fig. 7c). For example, for the fabrication of a macropore with an horizontal square section of 600 ␮m length (area of 0.360 mm2 ) in a sintered part having an open microporosity amount of 23 vol%, there must be a square macropore of 669 ␮m length (area of 0.448 mm2 ) on the green part according to the horizontal linear shrinkage and thus a length on the mask of 58 px according to the predictive dimensioning model of overcure. 4. Conclusion Pure SiHA ceramic without secondary crystalline phase and having a wide and controlled range of open microporosity could be sintered using adjusted sintering parameters (time and temperature). A predictive model for the design of the P␮SLA masks was established from image analysis of the overcure with pore dimensions and geometry. As summarized in Fig. 14, this model demonstrates the feasibility of complex macropores design with various angular geometries in the desired size range (i.e., 300–600 ␮m) for further applications in bone tissue engineering. Additionally, the knowledge of the horizontal and vertical shrinkages during sintering allows dimensioning accurately these macropores on the masks patterns for microstereolithography. Finally, microstereolithography was found efficient for shaping macro–micro-porous parts and may be considered in particular for the manufacture of tailor ceramic scaffolds with complex architectures and controlled multiscale porosity. The main limitation of P␮SLA lies in the difficulty of producing large parts. To overcome this difficulty hybrid scanning-projection methods using both

advantages of P␮SLA and scanning methods could be developed [48,50]. This new approach seems to provide good accuracy for the shaping of polymer parts [48,50]. But, it has not been yet operated for the manufacture of ceramic parts, for which the difficulty is not limited to the modelling of the exposure but also linked to the light diffusion on the ceramic particles. Currently, these hybrid methods present more complexity of implementation than “conventional” P␮SLA. Unidirectional crossing macropores have been considered in this study in order to produce simple model parts for investigating cells colonization and vascularization in vitro [43]. Nevertheless, all the chosen macropore structures have an axial symmetry as cylinders, only the cross-section geometry changed. So, 3D porous architectures using these pore geometries can be shaped by microstereolithography as it is already the case for scaffolds with pores of “classical” circular cross-section shaped by stereolithography. The main difference comes from the fact that shaping accurate geometries requires the reduction of the layer thickness, which may increase noticeably the processing time. 3D scaffolds with pores of triangular cross-section have been produced and implanted in rats for advanced in vivo investigation. The results will be considered for publication. The simple and fast imaging method set up in this study to monitor and quantify the overcure presents applicability to other geometries, slurry compositions, P␮SLA equipment, hybrid scanning-projection ␮SLA and even other layer-by-layer manufacturing processes. More generally, the implementation of this additive manufacturing technology, innovative for the shaping of ceramics, provides technological advance of interest that can find applications in several other fields than the medical one. Acknowledgments The authors are grateful to Dr David Marchat (CIS EMSE, CNRS UMR 5307, Saint-Etienne, France) for BET analysis of the sintered samples. This work is supported by institutional grants from the LabEX SigmaLim (ANR-10-LABX-0074-01). References [1] M. Bohner, L. Galea, N. Doebelin, Calcium phosphate bone graft substitutes: failures and hopes, J. Eur. Ceram. Soc. 32 (2012) 2663–2671, http://dx.doi.org/ 10.1016/j.jeurceramsoc.2012.02.028. [2] F. Barrère, C.A. Van Blitterswijk, K. De Groot, Bone regeneration: molecular and cellular interactions with calcium phosphate ceramics, Int. J. Nanomed. 1 (2006) 317–332. [3] P.V. Giannoudis, T.A. Einhorn, G. Schmidmaier, D. Marsh, The diamond concept-open questions, Injury 39 (Suppl. 2) (2008) S5–S8, http://dx.doi.org/ 10.1016/s0020-1383(08)70,010-x. [4] O. Gauthier, J. Bouler, E. Aguado, P. Pilet, G. Daculsi, Macroporous biphasic calcium phosphate ceramics: influence of macropore diameter and macroporosity percentage on bone ingrowth, Biomaterials 19 (1998) 133–139. [5] C.M. Murphy, M.G. Haugh, F.J. O’Brien, The effect of mean pore size on cell attachment, proliferation and migration in collagen—glycosaminoglycan scaffolds for bone tissue engineering, Biomaterials 31 (2010) 461–466. [6] V. Karageorgiou, D.L. Kaplan, Porosity of 3D biomaterial scaffolds and osteogenesis, Biomaterials 26 (2005) 5474–5491. [7] H. Yuan, K. Kurashina, J.D. de Bruijn, Y. Li, K. de Groot, X. Zhang, A preliminary study on osteoinduction of two kinds of calcium phosphate ceramics, Biomaterials 20 (1999) 1799–1806. [8] L. Cheng, F. Ye, R. Yang, X. Lu, Y. Shi, L. Li, et al., Osteoinduction of hydroxyapatite/␤-tricalcium phosphate bioceramics in mice with a fractured fibula, Acta Biomater. 6 (2010) 1569–1574. [9] J. Wei, J. Jia, F. Wu, S. Wei, H. Zhou, H. Zhang, et al., Hierarchically microporous/macroporous scaffold of magnesium–calcium phosphate for bone tissue regeneration, Biomaterials 31 (2010) 1260–1269. [10] K.A. Hing, B. Annaz, P. Saeed, P.A. Revell, T. Buckland, Microporosity enhances bioactivity of synthetic bone graft substitutes, J. Mater. Sci. Mater. Med. 16 (2005) 467–475. [11] J. Malmström, E. Adolfsson, A. Arvidsson, P. Thomsen, Bone response inside free-form fabricated macroporous hydroxyapatite scaffolds with and without an open microporosity, Clin. Implant Dent. Relat. Res. 9 (2007) 79–88.

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020

G Model JECS-10375; No. of Pages 11

ARTICLE IN PRESS M. Lasgorceix et al. / Journal of the European Ceramic Society xxx (2015) xxx–xxx

[12] S. Polak, S. Lan Levengood, M. Wheelen, A. Maki, S. Clark, A.J. Wagoner Johnson, Analysis of the roles of microporosity and BMP-2 on multiple measures of bone regeneration and healing in calcium phosphate scaffolds, Acta Biomater. 7 (2011) 1760–1771. [13] P. Kollmannsberger, C.M. Bidan, The physics of tissue patterning and extracellular matrix organisation: how cells join forces, Soft Matter 7 (2011) 9549–9560, http://dx.doi.org/10.1039/c1sm05588g. [14] C.M. Bidan, K.P. Kommareddy, M. Rumpler, P. Kollmannsberger, Y.J.M. Bréchet, P. Fratzl, et al., How linear tension converts to curvature: geometric control of bone tissue growth, PLoS One 7 (2012) e36336, http://dx.doi.org/ 10.1371/journal.pone.0036336. [15] C.M. Bidan, K.P. Kommareddy, M. Rumpler, P. Kollmannsberger, P. Fratzl, J.W.C. Dunlop, Geometry as a factor for tissue growth: towards shape optimization of tissue engineering scaffolds, Adv. Healthc. Mater. 2 (2013) 186–194, http://dx.doi.org/10.1002/adhm.201200159. [16] T. Chartier, C. Dupas, M. Lasgorceix, J. Brie, E. Champion, N. Delhote, et al., Additive manufacturing to produce complex 3d ceramic parts, J. Ceram. Sci. Technol. 6 (2015) 95–104, http://dx.doi.org/10.4416/JCST2014-00040. [17] S.A. Skoog, P.L. Goering, R.J. Narayan, Stereolithography in tissue engineering, J. Mater. Sci. Mater. Med. 25 (2014) 845–856, http://dx.doi.org/10.1007/ s10856-013-5107-y. [18] T. Chartier, C. Chaput, F. Doreau, M. Loiseau, Stereolithography of structural complex ceramic parts, J. Mater. Sci. 37 (2002) 3141–3147. [19] J. Brie, T. Chartier, C. Chaput, C. Delage, B. Pradeau, F. Caire, et al., A new custom made bioceramic implant for the repair of large and complex craniofacial bone defects, J. Craniomaxillofac Surg. 41 (2013) 403–407, http:// dx.doi.org/10.1016/j.jcms.2012.11.005. [20] A. Bertsch, S. Jiguet, P. Bernhard, P. Renaud, Microstereolithography: a review, Mat. Res. Soc. Symp. Proc. 758 (2003) 3–15. [21] J. Choi, R. Wicker, S. Lee, K. Choi, C. Ha, I. Chung, Fabrication of 3D biocompatible/biodegradable micro-scaffolds using dynamic mask projection microstereolithography, J. Mater. Process. Technol. 209 (2009) 5494–5503. [22] T. Chartier, C. Duterte, P. Abelard, Y. Abouliatim, C. Delage, C. Chaput, Net-shape forming of complex ceramic parts with a high dimensional resolution stereo- and microstereo- lithography, in: J.G. Heinrich (Ed.), ¨ Proceedings of 10th ECeRS Int. Conference, Baden–Baden Goller, Berlin, 2007. [23] V. Tomeckova, J.W. Halloran, Predictive models for the photopolymerization of ceramic suspensions, J. Eur. Ceram. Soc. 30 (2010) 2833–2840. [24] S.P. Gentry, J.W. Halloran, Light scattering in absorbing ceramic suspensions: effect on the width and depth of photopolymerized features, J. Eur. Ceram. Soc. (2015) 1895–1904. [25] A. Limaye, D. Rosen, Process planning to build Mask Projection Stereolithography parts with accurate vertical dimensions, Proc. 18th Solid Free Fabr. Symp. (2007) 159–173. [26] R. Bail, J.I. Segal, S.M. Ratchev, Monitoring diameter and roundness errors in projection-microstereolithography of biomedical tooling and scaffolds, Procedia CIRP 5 (2013) 57–60, http://dx.doi.org/10.1016/j.procir.2013.01.011. [27] H.-W. Kang, J.H. Park, D.-W. Cho, A pixel based solidification model for projection based stereolithography technology, Sens. Actuators Phys. 178 (2012) 223–229, http://dx.doi.org/10.1016/j.sna.2012.01.016. [28] S.P. Gentry, J.W. Halloran, Depth and width of cured lines in photopolymerizable ceramic suspensions, J. Eur. Ceram. Soc. 33 (2013) 1981–1988, http://dx.doi.org/10.1016/j.jeurceramsoc.2013.02.033. [29] M. Lasgorceix, A. Machado Costa, E. Mavropoulos, M. Sader, M. Calasans, M. Tanaka, et al., In vitro and in vivo evaluation of silicated hydroxyapatite and impact of insulin adsorption, J. Mater. Sci. Mater. Med. 25 (2014) 2383–2393, http://dx.doi.org/10.1007/s10856-014-5237-x. [30] M. Vallet-Regi, D. Arcos, Silicon substituted hydroxyapatites. A method to upgrade calcium phosphate based implants, J. Mater. Chem. 15 (2005) 1509–1516. [31] K.A. Hing, P.A. Revell, N. Smith, T. Buckland, Effect of silicon level on rate, quality and progression of bone healing within silicate-substituted porous hydroxyapatite scaffolds, Biomaterials 27 (2006) 5014–5026, http://dx.doi. org/10.1016/j.biomaterials.2006.05.039.

11

[32] N. Patel, R.A. Brooks, M.T. Clarke, P.M.T. Lee, N. Rushton, I.R. Gibson, et al., In vivo assessment of hydroxyapatite and silicate-substituted hydroxyapatite granules using an ovine defect model, J. Mater. Sci. Mater. Med. 16 (2005) 429–440. [33] A.E. Porter, Nanoscale characterization of the interface between bone and hydroxyapatite implants and the effect of silicon on bone apposition, Micron 37 (2006) 681–688, http://dx.doi.org/10.1016/j.micron.2006.03.006. [34] A.M. Pietak, J.W. Reid, M.J. Stott, M. Sayer, Silicon substitution in the calcium phosphate bioceramics, Biomaterials 28 (2007) 4023–4032, http://dx.doi.org/ 10.1016/j.biomaterials.2007.05.003. [35] M. Bohner, Silicon-substituted calcium phosphates—a critical view, Biomaterials 30 (2009) 6403–6406, http://dx.doi.org/10.1016/j.biomaterials. 2009.08.007. [36] M.C. Durrieu, S. Pallu, F. Guillemot, R. Bareille, J. Amédée, C.H. Baquey, et al., Grafting RGD containing peptides onto hydroxyapatite to promote osteoblastic cells adhesion, J. Mater. Sci. Mater. Med 15 (2004) 779–786. [37] N. Douard, Élaboration et fonctionnalisation de biocéramiques phosphocalciques (Elaboration and functionalization of calcium phosphate bioceramics), in: PhD Thesis, University of Limoges, France, 2010. [38] I.R. Gibson, S.M. Best, W. Bonfield, Effect of silicon substitution on the sintering and microstructure of hydroxyapatite, J. Am. Ceram. Soc. 85 (2002) 2771–2777. [39] V. Putlayev, A. Veresov, M. Pulkin, A. Soin, V. Kuznetsov, Silicon-substituted hydroxyapatite ceramics (Si-HAp): densification and grain growth through the prism of sintering theories, Mater. Werkst. 37 (2006) 416–421, http://dx. doi.org/10.1002/mawe.200600007. [40] M. Palard, J. Combes, E. Champion, S. Foucaud, A. Rattner, D. Bernache-Assollant, Effect of silicon content on the sintering and biological behaviour of Ca10 (PO4 )6−x (SiO4 )x (OH)2−x ceramics, Acta Biomater. 5 (2009) 1223–1232. [41] E. Champion, Sintering of calcium phosphate bioceramics, Acta Biomater. 9 (2013) 5855–5875. [42] M. Palard, E. Champion, S. Foucaud, Synthesis of silicated hydroxyapatite Ca10 (PO4 )6−x (SiO4 )x (OH)2−x , J. Solid State Chem. 181 (2008) 1950–1960. [43] M. Lasgorceix, Mise en forme par microstéréolithographie et frittage de céramiques macro-micro-poreuses en hydroxyapatite silicatée et évaluation biologique (Shaping by microstereolithography and sintering of macro–micro-porous ceramics in silicated hydroxyapatite and biological evaluation), in: PhD Thesis, University of Limoges, France, 2014. [44] ISO 13779-3:2008 – Implants for surgery – Hydroxyapatite – Part 3: chemical analysis and characterization of crystallinity and phase purity. [45] C. Rey, B. Collins, T. Goehl, I.R. Dickson, M.J. Glimcher, The carbonate environment in bone mineral: a resolution-enhanced fourier transform infrared spectroscopy study, Calcif. Tissue Int. 45 (1989) 157–164. [46] J.C. Elliott, Structure and chemistry of the apatites and other calcium orthophosphates, Stud. Inorg. Chem. 18 (1994) 11–89, Elsevier. [47] D. Marchat, M. Zymelka, C. Coelho, L. Gremillard, L. Joly-pottuz, F. Babonneau, et al., Accurate characterization of pure silicon-substituted hydroxyapatite powders synthesized by a new precipitation route, Acta Biomater. 9 (2013) 6992–7004, http://dx.doi.org/10.1016/j.actbio.2013.03.011. [48] I.-B. Park, Y.-M. Ha, S.-H. Lee, Still motion process for improving the accuracy of latticed microstructures in projection microstereolithography, Sens. Actuators Phys. 167 (2011) 117–129, http://dx.doi.org/10.1016/j.sna.2010.12. 023. [49] C. Sun, N. Fang, D.M. Wu, X. Zhang, Projection micro-stereolithography using digital micro-mirror dynamic mask, Sens. Actuators A 121 (2005) 113–120. [50] M.M. Emami, F. Barazandeh, F. Yaghmaie, An analytical model for scanning-projection based stereolithography, J. Mater. Process. Technol. 219 (2015) 17–27, http://dx.doi.org/10.1016/j.jmatprotec.2014.12.001.

Please cite this article in press as: M. Lasgorceix, et al., Shaping by microstereolithography and sintering of macro–micro-porous silicon substituted hydroxyapatite, J Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2015.11.020