Calcium phosphate/microgel composites for 3D powderbed printing of ceramic materials

Materials

N-Vinylcaprolactam (VCL, 98%), 2-(methacryloyloxy)ethyl acetoacetate (AAEM, 95%), 2,2′-azobis[2-methylpropionamidine] dihydrochloride (AMPA, granular, 97%), 1- vinylimidazole (VIm, ≥99%) and N,N′-methylene(bis)acrylamide (BIS, 99%) were purchased from Sigma Aldrich (Steinheim, Germany). Both VCL and VIm were purified by distillation. Moreover, EMPROVE^® exp calcium phosphate (TCP) and EMPROVE^® exp calcium hydrogen phosphate (CHP, CaHPO₄) were purchased from Merck (Darmstadt, Germany). 2-Propanol ROTIPURAN^® (≥99.8%, p.a.) were purchased from Carl Roth (Karlsruhe, Germany). CONTRASPUM CONC. was used as a defoamer (Zschimmer and Schwarz, Lahnstein, Germany). Water used in the experiments was purified using a Millipore water purification system with a minimum resistivity of 18 MΩcm.

Synthesis of microgels

PVCL/AAEM/VIm microgels (later VIm MG) were prepared according to a procedure reported elsewhere [1]. For the products 0 mol% VIm MG and 5 mol% VIm MG, appropriate amounts of VIm (0 g or 0.367 g), VCL (6.000 g), BIS (0.240 g), and AAEM (1.100 g) were mixed in 500 ml water and stirred at 200 rpm under a nitrogen flow. The solution was heated to 70°C and allowed to stir for 60 min. Afterwards, microgel synthesis was initiated by the addition of AMPA initiator (0.05 g). The reaction was carried out for 4 h at 70°C. The crude microgel products were purified via dialysis for 3 days with a composite regenerated cellulose membrane from Millipore (NMWCO 30,000).

Synthesis of β-TCP

Phase-pure β-TCP was prepared by calcination using a calcium-deficient hydroxyapatite with 3 wt% CHP at 1000°C for 2 h. CHP was added to adjust the Ca/P ratio to exactly 1.5 because small amounts of hydroxyapatite were determined via X-ray diffraction (XRD) of thermal-treated calcium-deficient HA, which indicates a higher Ca/P ratio than 1.5. Chemical and phase purity was confirmed by X-ray fluorescence (XRF) spectroscopy and XRD spectroscopy (see Supporting Information, Figure S1, Table S1).

Preparation of microgel β-TCP composite materials and granulates

The dialyzed microgel dispersions with concentrations ranging between 10 and 16 g/l were mixed with an appropriate amount of β-TCP. For example, a suspension of 500 ml (c=15.29 g/l) 5 mol% VIm MG dispersion and β-TCP (68.8 and 137.6 g, respectively) were either directly freeze-dried or spray-dried (Mini Spray Dryer B-290, Büchi, Flawil, Switzerland) at 220°C to yield granulates with a microgel content of 10 or 5 wt% (C10 – 5% VIm and C5 – 5% VIm, respectively). Before spray-drying, the suspensions were milled on a roller platform with zirconia milling balls (VTZ Grinding Media, Tosoh, Tokyo, Japan; in each case 1 kg of 1 cm and 2 cm in diameter) in 2-L polyethylene bottles for 1 day. After the addition of 1 to 2 drops of CONTRASPUM CONC., the suspensions were sieved with 125-μm sieves. According to the procedure, for the preparation of pure spray-dried β-TCP granulates, a suspension of 500 ml deionized water and 137.6 g β-TCP was used.

Characterization methods

Microgels with 5 mol% VIm were coated on a Si-wafer by spin-coating. The Si-wafers were cleaned with toluene and activated via air plasma treatment for 60 s at 2 mbar using a Plasma Cleaner Activate Flecto 10 USB-MFC. Directly after treatment, 50 μl of microgel solution was spin-coated onto the wafers for 60 s at 2000 rpm. After drying, the samples were analyzed using a Veeco Dimension ICON Atomic Force Microscope (AFM) with OTESPA tips at 284–338 kHz and spring constants of 12–103 N/m.

Fourier transform infrared (FT-IR) spectra were recorded on a Thermo Nicolet 470 FT-IR spectrometer in transmission mode. The freeze-dried microgel sample was mixed with KBr powder and then pressed to form a transparent KBr pellet.

The electrophoretic mobility at varying pH of the microgel particles and β-TCP were measured using a Zetasizer NanoZS (Malvern, UK). The acidity and basicity were measured from pH=3 to 10 in 0.5 steps using 0.1 M HCl and NaOH, respectively. At each pH, 10 measurements were taken at 25°C after equilibrating the samples for at least 10 min.

X-ray photoelectron spectroscopy (XPS) measurements were done using a spectrometer (Axis HSi 165 Ultra, Kratos Analytical, Kyoto, Japan) with a focused monochromatic Al K_α source (1486.6 eV) for excitation. The electron take-off angle was 45°, and the analyzer was operated in the constant energy mode.

Scanning electron microscopy (SEM) combined with energy dispersive X-ray spectroscopy (EDX) analyses were done using a Hitachi SU9000 electron microscope (Philips, Eindhoven, The Netherlands) at a beam voltage of 30 kV. The EDX maps were recorded with acquisition times ≥30 min. For SEM/EDX observations, particles were dispersed in distilled water. A drop of this dispersion was placed on a silicon-oxide-coated gold grid and dried in air.

Moisture sorption-desorption isotherms were measured using an IGAsorp moisture sorption analyzer (Hiden Isochema, Kochem am See, Germany) at 37°C. The relative humidity (RH) was increased from 2% to 94% in increments of 10%.

Thermogravimetric analysis (TGA) measurements were performed using a Netzsch TG 209c unit (Netzsch Gerätebau, Selb, Germany) operating under a nitrogen atmosphere with a flow rate of 10 ml/min. Samples of 20–30 were placed in standard Netzsch alumina 85 μl crucibles and heated at 20 K/min to 800°C.

The dry and wet granulometric analysis of the spray-dried granulates were measured by a laser granulometer (Mastersizer 2000, Malvern, Worcestershire, Great Britain). The morphology of the samples was determined using an XL30 ESEM FEG device (FEI, Philips, Eindhoven, The Netherlands). The non-sputtered samples were measured under different conditions (0.8 Torr or high vacuum, 3–20 kV) depending on the detector used (SE, BSE, GSE).

TEM images were obtained with a Philips EM 400 T/ST (Eindhoven, The Netherlands) equipped with an 11-megapixel TEM CCD camera (Morada, SIS/Olympus, Hamburg, Germany) operating at 60 kV. The particles were diluted with deionized water, placed on a Formvar/carbon-coated Ni grid (200 mesh) and dried at room temperature (method A). The particles were embedded using an epoxy embedding medium kit (Sigma-Aldrich, Taufkirchen, Germany), cut into ultrathin layers (90 nm) with a diamond ultra-knife (DiATOME) using an ultracut (Leica EM UC 6) and analyzed before and after staining in an LKB 2168 Ultrostainer (Bromma, Langenhagen, Germany) using uranyl acetate and lead citrate (method B).

Ball-on-three-balls tests [4] were performed on unpolished uniaxially pressed discs (13–15 mm diameter, 1.5–1.6 mm thickness) by using an universal testing machine (Z030, Zwick, Ulm, Germany) with a 10 kN load cell (value ±0.02%) without preload, and balls with a diameter of 7.96 mm. Before it, the discs were fabricated using the same universal testing machine under the following conditions: 400 mg pestled powder, 3000 N maximum load (top dead center), 50 N preload, 5 mm/min testing speed and 120 s hold time at maximum load. The tested 3×3-matrix was composed as follows: pressed discs of pure β-TCP, C10-0%VIm, and C10-5%VIm with or without 2 h heat-treatment at 200°C and 1100°C (L9/13, Nabertherm, Lilienthal, Germany), respectively. Before the compacts were sintered in air with a heating/cooling rate of 5°C/min, the samples were held at 600°C for 2 h in order to remove the microgels. Thirty specimens were tested for every test series with the exception of too fragile non-treated β-TCP samples. The biaxial strength of brittle ceramic discs was calculated using the equation of Börger et al [4]. The Poisson’s ratio ν=0.22 of pure β-TCP was used for all calculations [20]. WeibPar 4.3 (WeibPar, Connecticut Reserve Technology, Gates Mills, OH, USA) was used for the calculation of the Weibull modulus m and the characteristic strength σ₀ via MLE2-U.

3D powderbed printing

Cylinders with different sizes were prepared by 3D powderbed printing (Spectrum Z510, ZCorporation, Burlington, VT, USA) using Zprint 7.6 as well as synthesized spray-dried granulates as powder and deionised water/2-propanol (80/20) as a binding system. Printing was performed with a layer thickness of 50 or 100 μm/step and a binder/volume ratio of 0.45. The printed structures were taken out of the powder bed after 1 day and cleaned using oil-free compressed air. The microgels of the printed ceramic green bodies were pyrolyzed at 600°C for 2 h. The samples were then sintered at 1100°C for 2 h to increase mechanical strength.

Characterization of freeze-dried β-TCP/microgel composites

Composites of PVCL/AAEM/VIm microgels (with 0 mol% and 5 mol% VIm content) and β-TCP with a microgel content of 5 wt%, 10 wt% and 15 wt% have been successfully manufactured via freeze-drying and spray-drying of suspensions, respectively.

The aqueous PVCL/AAEM/VIm microgel particles, having heterogeneous core-shell structures, were prepared by precipitation polymerization (Figure 1). AAEM has a higher reactivity ratio as compared to VCL. Therefore, AAEM is consumed faster during polymerization, resulting in a hardly cross-linked AAEM-rich core and a weakly cross-linked VCL-rich shell with selectively incorporated VIm groups [1, 34]. The chemical composition of microgels used in the present study was confirmed by FT-IR spectroscopy, and the number of VIm groups was determined by potentiometric titration.

Figure 1:

Schematic structure of Poly(N-vinylcaprolactam (PVCL)/AAEM/VIm microgels and AFM image of 5 mol% VIm MG.

To investigate the interactions of the microgels and the β-TCP in composites, we performed FT-IR analysis of selected samples. FT-IR spectra of pure β-TCP, PVCL/AAEM/VIm MGs with 5 mol% VIm content, and composite C10-5% VIm are summarized in Figure 2. The peak at 1628 cm^-1 corresponds to the -C=O group of PVCL (Figure 2A). A shift to 1616 cm^-1 was observed for this peak in the FT-IR spectrum of the composite material. This can be explained based on the interactions between the -C=O group of PVCL and the water molecules adsorbed to the surface of β-TCP (Figure 2B). As AAEM is present in the core, the peak for its -C=O group remains unaffected. Further possible electrostatic interactions between VIm and β-TCP are not detectable, because of the low signal intensity of VIm due to its low content within the microgel. Other peak shifts could be neither observed for microgels nor for β-TCP. In addition, a shift of the -C=O group of PVCL from 1624 cm^-1 to 1634 cm^-1 was observed in the overlay Raman spectra (see Supporting Information, Figure S2), which additionally confirms the interactions of the side groups of polymer chains with the surface of β-TCP. To confirm the possible electrostatic interactions in the composite, the electrophoretic mobility of pure β-TCP and 5% VIm MGs was measured as a function of pH (Figure 2C). As expected, 5% VIm MGs are positively charged in an acidic pH range due to the protonation of VIm groups, and this charge decreases with an increasing pH value. However, pure β-TCP is negatively charged through the entire pH range from 5 to 10. A strong aggregation was observed for pure β-TCP between pH 3 and 5. The measured pH value of a suspension for the preparation of composite C5-5% VIm was 7.04. These experimental data allow us to conclude that electrostatic interactions exist between pure β-TCP and 5% VIm MGs, leading to successful composite formation, and support the FT-IR data.

Figure 2:

(A) Overlay FT-IR spectra of pure β-TCP, 5% VIm MG and composite C10-5% Vim. (B) Schematic illustration of possible interactions between β-TCP and microgels adapted from Schachschal et al [40]. (C) Electrophoretic mobility of 5% VIm MG and pure β-TCP as a function of pH.

To investigate the chemical composition of the composites, XPS measurements were performed. Figure 3 shows the overlay XPS spectra of β-TCP, 5% VIm MG and composite C10-5% VIm. Pure β-TCP showed the characteristic signals of Ca 2p and P 2p at 344 eV and 130 eV, respectively. A smaller signal also appeared for C 1s, which could be caused by environmental impurity. However, a 5% VIm MG sample showed three main signals that could be assigned to C 1s at 282 eV, N 1s at 396.5 eV and O 1s at 528.5 eV. As expected, no signal related to Ca 2p and P 2p was observed. The combination of both microgels and β-TCP peaks was found for the C10-5% VIm composite. The presence of a prominent peak for carbon in combination with nitrogen supported the successful incorporation of the microgel in the composite.

Figure 3:

Overlay XPS spectra of β-TCP, PVCL/AAEM/VIm MGs (5 mol% VIm) and composite C10-5%VIm.

Water sorption/desorption and thermal properties

As potential material for powderbed printing, the composite powder should absorb the binding liquid fast and localize it to ensure a high printing resolution. Therefore, moisture sorption/desorption was systematically investigated. This provided useful information about water-uptake-induced changes in the material and its behavior as a function of environmental humidity. Figure 4A represents the sorption isotherms for pure β-TCP, PVCL/AAEM/VIm MGss without VIm, and C10-0% VIm measured at 37°C. By comparing these isotherms, it can be seen that moisture sorption of the microgels was greatly reduced from 34% to 4% for the pure microgels and microgels in composite, respectively. Pure β-TCP showed almost no changes with varying humidity. Similar results were obtained for pure 5% VIm MG and its respective composite C10-5% VIm (Figure 4B).

Figure 4:

(A) Sorption isotherms measured at 37°C for pure β-TCP, pure 0% VIm MG and the respective composite C10-0% VIm; (B) pure β-TCP, pure 5 % VIm MG and C10-5% VIm.

Figure 5 depicts the adsorption and desorption rates for all the samples derived from kinetic measurements. In general, adsorption rates decreased for all the investigated samples with increasing humidity, which was based on decreasing the adsorption energy of the sample with increasing surface coverage in accordance to the Temkin adsorption model (Figure 5A) [45]. The adsorption rate constants were comparable for all the samples, but microgels took up more water over a longer period of time, resulting in a higher equilibration time (Figure 5B). However, the desorption rates were independent of surface coverage as shown in Figure 5C.

Figure 5:

Adsorption (A) and desorption (C) rates of moisture at 37°C of 0% VIm MG, 5% VIm MG, C10-0% VIm and C10-5%VIm; (B) moisture adsorption and desorption as a function of time at 37°C for pure β-TCP, 0% VIm MG, 5% VIm MG, C10-0% VIm and C10-5% VIm.

The composite materials were freeze-dried and subsequently investigated by TGA. Figure 6 demonstrates the TGA curves of various composites prepared with different microgel contents. The pure microgel curve showed an initial weight loss at T≈100°C, which appeared due to the evaporation of water. In the temperature range from 150° to 350°C, weight loss occurred due to the decomposition of the acrylic groups of the microgel network. Thermal decomposition of the C–C backbone then arose, and complete polymer degradation took place. The inflection point for pure microgels was at 441°C, which shifted to 450°C in case of composite, indicating improved thermal stability. The residual mass remaining after heating of the samples to 500°C could be related to β-TCP, which is not degradable under these conditions. Experimental residual amounts from TGA for C5-5% VIm, C10-5% VIm and C15-5% VIm were 95%, 90% and 87%, respectively. These TGA results indicated that a hardening step of composite specimens via a sintering process at 1100°C would yield pure β-TCP due to a complete pyrolysis of the microgel that occurred at a temperature of T <500°C.

Figure 6:

TGA thermographs of pure β-TCP, pure 5% VIm MG and composites prepared by using 5 wt%, 10 wt% and 15 wt% of 5% VIm MG.

Mechanical testing using ball-on-3-balls test

The strength of the specimens was tested using the ball-on-three-balls test [4]. The tested samples consisted of pure β-TCP, β-TCP with 10 wt% of 0 wt% VIm microgel (C10-0% VIm), and β-TCP with 10 wt% of 5 wt% VIm microgel (C10-5% VIm), respectively. The mechanical data set of all non-sintered test series is presented in Supporting Information, Table S2. These data indicate that the strength for all non-sintered test series is still too low (<2.6 MPa) for a potential application as bone substitute.

The strength of the samples was improved considerably by sintering (Table 1). The microgels pyrolized, yielding pure β-TCP samples that are referred to in the following as sintered composites. It was found that the sintered β-TCP test series showed a two-fold increase in strength compared to those of the sintered composites (28.4±2.8 MPa vs. 14.1±1.1 MPa and 15.9±0.9 MPa, respectively).

Reference values for the flexural strength of pressed sintered β-TCP samples, without consideration of reference values for hot isostatic pressed β-TCP samples, vary between 6 and 91 MPa [17, 25, 26, 32]. Compared to those values, the biaxial strength of β-TCP determined in our study was in the lower range. In addition to the sintering temperature and time, the reason for the low biaxial strength could be that the pressed samples were fabricated using pestled powder only. The absence of organic pressing aids/lubricating additives caused moulding defects and jamming of the pressing die, resulting in additional sample defects. A requirement for ball-on-three-balls test is that the samples should start breaking at their center because of center loading during the test. All sintered samples fulfilled this requirement and broke into two or three parts. However, in the case of the β-TCP test series, some samples showed an irregularity close to the sample edge (see Supporting Information, Picture S3), indicating moulding defects. In contrast, none of the sintered composite samples showed such defects. This means that the sample fabrication was facilitated by the microgels, which acted as lubricant.

The bending strengths of quasi-brittle materials like ceramics are Weibull-distributed. Therefore, the characteristic strength σ₀ should be calculated using Weibull statistics and the maximum-likelihood estimation [37, 49]. The Weibull modulus m specifies the precision of the determined characteristic strength. The higher the m value, the more probable is the determined characteristic strength. In case of ceramics, m values range from 10 to 20. The sample defects discussed are reflected in the lower modulus for the β-TCP test series compared to those of the sintered composites (m=11.8 vs. 14.3 and 18.3, respectively). This means that the determined characteristic strengths of the sintered composite test series are more reliable than those of the β-TCP test series. The characteristic strength specifies the strength at which 63.21% of the samples fail. For all sintered test series, the characteristic strength is slightly higher than the respective mean failure stress.

The strength depends, among other factors, on the method of β-TCP synthesis (particle size), the method of sample fabrication (use of lubricant), included defects, surface characteristics of the samples (polished or unpolished), the testing method and Poisson’s ratio (ν=0.22-0.30 for β-TCP) [7, 20] used for the strength calculations. Furthermore, the sintering parameters temperature and time have an influence on porosity as well as on the strength of sintered samples [32]. This makes a comparison between measured and reported values difficult. A higher sintering temperature of up to 1200°C and a higher sintering time of 3 h could enhance the strength. However, an increase in strength correlates with a decrease in porosity, which is also important for bone substitutes [10].

The density and, consequently, the total porosity of each test series were determined as an average of five samples by the weight to volume ratio using ρ=3.14 g/cm³ for the specific density of β-TCP. The calculated total porosity of sintered β-TCP samples was between 42%–45%, whereas the sintered composites showed a total porosity of 52%–53% (sintered C10-0% VIm) and 56%–57% (sintered C10-5% VIm), respectively. The difference in porosity is the reason for the distinction in the biaxial strength of sintered samples. A 10% to 15% higher porosity caused a 50% reduction of biaxial strength.

Characterization of spray-dried granulates

The following measurements were performed to study the particle-size distribution and morphology of the β-TCP/microgel granulates, and to test, whether this mixture is homogeneous. These results are important for a potential application as powder for powderbed printing, i.e., the microgels should act as glue, and should therefore be homogeneously distributed.

The particle size distribution of granulates is crucial for an application as powder in a powderbed printer. Table 2 gives an overview of the evolution of the average spherical diameter of the particles. The specific values D10, D50 and D90 represent 10 vol%, 50 vol% and 90 vol% of the total volume of the particles, respectively. The specific values are referred to the mean particle size (D50), to the smallest particles (D10) and to the bigger particles (D90). According to our theory, an increase in the solid content of the suspensions caused an increase in particle size. For example, a doubling of the β-TCP content of the suspension changed the mean particle size (D50) from 7.8 to 12.4 μm. The D50 of β-TCP granulates was 9.9 μm. The size of the primary grains gives information about the sintering process: the smaller the primary grains, the more energy is necessary for the sintering process. The size of the primary grains was determined by wet granulometric analysis and was <0.8 μm (D10, wet after ultrasonic). The decreased particle sizes of C10-5% VIm and C5-5% VIm (wet before US vs. dry) indicates that granulates are rather unstable in water.

The morphology of granulates is important for its flowability, and finally for a good layer formation during 3D powderbed printing. The particles should be neither too rough nor too round. In Figure 7, the ESEM micrographs of spray-dried granulates clearly indicate higher particle sizes for C5-5% VIm compared to C10-5% VIm, which agrees with the granulometric analysis (Figure 7A and B). Further, it was not possible to discriminate between β-TCP and microgels because the primary grain size (<0.8 μm) and the microgel size (<0.5) were close together. In addition, despite a change of detector (BSE or GSE), no difference between the components was visible. Furthermore, we did not observe any doughnut-formed granulate, which would be an indicator for a hollow granulate. In contrast, the formation of β-TCP granulates failed (Figure 7C). Some granules are visible, but the main part consists of angular grains of different sizes. This can be explained by the lack of any organic additive, i.e., microgels. To conclude, a microgel content of 5 or 10 wt% of total solid content of β-TCP-suspension facilitated the spray-drying process because flowable granulates were formed, and, furthermore, no filter-blocking occurred during spray-drying. In addition, the microgels could also be beneficial as glue during a 3D powderbed-printing process due to its demonstrated ability to glue together the primary grains within a granulate.

Figure 7:

ESEM micrographs of the three spray-dried granulates C10-5% VIm (A), C5-5% VIm (B) and pure β-TCP (C) using the BSE detector.

The presence of microgels in the composite was further confirmed by EDX mapping. Figure 8 shows SEM images and elemental mapping images for C10-5% VIm composite. From the SEM image, it was difficult to distinguish between microgel and β-TCP due to their similarity in morphology. Therefore, elemental mapping corresponding to C present in microgels and Ca and P present in β-TCP was performed. These elemental mapping images, which correlated with the SEM image, showed that the elemental distribution was relatively homogeneous. As a consequence, pyrolysis of composites would lead to relatively homogeneously distributed pores.

Figure 8:

(A) SEM image of C10-5% VIm composite. (B–D) EDX elemental mapping for carbon (red), calcium (green) and phosphor (blue) for the region shown in (A).

Further investigations concerning the morphology of β-TCP/microgel composites were performed using TEM to study the condition and behavior of microgels as well as their distribution within the composites. The wet preparation of the spray-dried C10-5% VIm on a Formvar/ Carbon-coated Ni grid (method A) caused a leaching of the water-soluble microgels (grey and round) out of the granulates (black), which allowed a good view of the microgels (Figure 9A and B). Despite having undergone milling and spray-drying processes, the microgels appeared intact. The TEM images of Figure 9C and D, which were observed on a stained ultrathin layer, show primary grains of β-TCP (black), crystallized β-TCP from the suspension (crystals) and microgels (gray and round). The white regions are referred to as holes within the ultrathin sections due to the natural removal of bigger β-TCP particles during the cutting of the embedded sample. According to the images, the granulates are not compact inside, the microgel content looked relatively low within the 90-nm-thin section and the microgel distribution seemed to be inhomogeneous. Furthermore, the microgels are connected to each other as well as to the β-TCP particles via entanglements formed by interpenetrating polymer chains. It could be clearly seen that the microgels glued the β-TCP particles together.

Figure 9:

TEM images of spray-dried C10-5% VIm using method A (A, B) and method B (C, D).

Application of β-TCP/microgel composites for 3D powderbed printing

Using the rapid prototyping technique of 3D powderbed printing, β-TCP granulates are usually processed into scaffolds by hydraulic bone-cement reactions using diluted phosphoric acid as a binder [6, 13] or by using a polymer/water-based binding system [23, 2, 41]. To date, a microgel/water-based binding system for 3D powderbed printing has not been published. We became interested in microgels as polymeric components due to their nearly instantaneous swelling properties. Our assumption was that during the printing process, the binding solution would be absorbed fast and only in the desired region, which is important for the resolution of the printing process.

The first 3D printing tests using β-TCP granulates as powder and microgel dispersions (0 or 5% VIm MG) directly or diluted (deionized water or deionized water/2-Propanol) as a binding solution failed because 2.5 wt% of microgel dispersions clogged the injector of the printing head. Further 3D printing tests were done using the spray-dried β-TCP microgel composites (C5-5% VIm and C10-5% VIm) as powder and diluted 2-propanol (20 vol%) as a binder. Prior to this, XRD measurement of a sintered mixture of C5-5% VIm granulates and diluted 2-propanol (20 vol%) confirmed the desired preservation of phase purity of β-TCP (see Supporting Information, Figure S4). This is a main advantage compared to a phosphoric acid binder, which reacts with β-TCP. The first printed samples are shown in Figure 10. Loose powder was removable using compressed air. The typical size of the curved channels, where loose, spray-dried powder was removable without the use of compressed air was 2 mm. Typically, the subsequent hardening step via a sintering process at 1100°C caused shrinkage of the printed specimens (Figure 10). The images of printed specimens show that a microgel content of 5 or 10 wt% was sufficient to glue the granulates together.

Figure 10:

Printed samples of C5-5% VIm before and after sintering (A); printed samples of C10-5% VIm before and after sintering (B).

Characterization of printed specimens

The microstructures of different C10-5% VIm samples were studied by ESEM. Figure 11 shows ESEM micrographs of cross-sections of pressed and printed C10-5% VIm granulate (freeze- and spray-dried, respectively) before and after sintering at 1100°C. Accordingly, the printed samples have a higher porosity than the pressed samples, but this difference in porosity is smaller in the sintered samples. The micrographs of the cross-sections of printed green bodies show that, in both cases, disruptions occurred between preserved granulates (Figure 11b, c). The cohesion between granulates seemed to be weaker than the cohesion within granulates. Printed C10-5% VIm was packed closer than printed C5-5% VIm. As can be seen in Figure 11 (E and F), the heat-treatment at 1100°C caused a decrease in pore size because the granulate sintered together. Only some sintered granulate particles were visible as a whole. Visual inspection indicated a similar porosity for both sintered printed samples (Figure 11E and F), which was increased in comparison to the porosity of the sintered pressed sample (Figure 11D). According to the visual evaluation of ESEM micrographs, the sintered pressed sample only contains micropores <5 μm, whereas the sintered printed samples contain micropores <5 μm as well as macropores >100 μm (not shown).

Figure 11:

ESEM micrographs of the cross-sections before (A–C) and after (D–F) sintering at 1100°C: pressed freeze-dried C10-5% VIm (A, D); printed spray-dried C10-5% VIm (B, E) and printed spray-dried C5-5% VIm (C, F).

Supporting information

Overlay XRD graphs of calcinated TCP and pure β-TCP (Figure S1), XRF spectroscopy data of synthesized pure β-TCP (Table S1), Overlay Raman spectra of pure β-TCP, 5% VIm MG, and composite C10-5% VIm (Figure S2), results of all test series of the ball-on-three-balls tests (Table S2), pictures of all sintered samples tested by ball-on-three-balls tests (Figure S3), ESEM micrographs (BSE and GSE) of the surface of printed ceramic green bodies of C10-5% VIm and C5-5% VIm (Figure S4) and XRD graph of a sintered mixture of C5-5% VIm and diluted 2-propanol (Figure S5).