Highly Porous Polymer-Derived Bioceramics Based on a Complex Hardystonite Solid Solution

Highly porous bioceramics, based on a complex hardystonite solid solution, were developed from silicone resins and micro-sized oxide fillers fired in air at 950 °C. Besides CaO, SrO, MgO, and ZnO precursors, and the commercial embedded silicone resins, calcium borate was essential in providing the liquid phase upon firing and favouring the formation of an unprecedented hardystonite solid solution, corresponding to the formula (Ca0.70Sr0.30)2(Zn0.72Mg0.15Si0.13) (Si0.85B0.15)2O7. Silicone-filler mixtures could be used in the form of thick pastes for direct ink writing of reticulated scaffolds or for direct foaming. The latter shaping option benefited from the use of hydrated calcium borate, which underwent dehydration, with water vapour release, at a low temperature (420 °C). Both scaffolds and foams confirmed the already-obtained phase assemblage, after firing, and exhibited remarkable strength-to-density ratios. Finally, preliminary cell tests excluded any cytotoxicity that could be derived from the formation of a boro-silicate glassy phase.


Introduction
Silicone polymers, homogeneously mixed with oxide fillers, have been extensively investigated as precursors for silicate bioceramics for the last ten years [1]. A fundamental advantage in all polymer-derived ceramics concerns the possible application of polymer forming techniques, both conventional (e.g., foaming) and advanced (e.g., additive manufacturing technologies), so that a component is first shaped at low temperature and then 'ceramised' [2]. Ideally, the ceramic conversion implies just a homogeneous shrinkage, with no microcracking; in actual polymer-derived components, if not thin-walled (e.g., films, fibres, foams) [2], the structural integrity is favoured by the adoption of fillers (either 'passive', i.e., simply 'diluting' the mass of polymer undergoing transformation, or 'active', when fillers react with the matrix and/or with the firing atmosphere) [3][4][5].
MK was first dissolved in isopropyl alcohol under magnetic stirring. Clear solutions were then added with the fillers (powders were slowly cast in the MK solutions), again under magnetic stirring, for 15 min. After sonication for 15 min, the mixtures were left to dry at 80 • C overnight in Teflon containers. Finally, dried powders were dry ball milled (Pulverisette 7 planetary ball mill, Fritsch, Idar-Oberstein, Germany), sieved below 90 µm, and pressed in a cylindrical die (diameter of 16.5 mm) at 40 MPa.
All pressed tablets were fired in air at 950 • C, with a 1 h holding time and a heating rate of 2 • C/min, followed by natural cooling.

Direct Ink Writing of Hardystonite Scaffolds
Pastes for direct ink writing experiments were based on MK. As previously done [21], the viscosity of MK-based pastes was adjusted by using colloidal silica, which replaced MK in an amount corresponding to 10 wt% of the total silica content. Colloidal silica as well as the other fillers (CaCO 3 , ZnO, anhydrous colemanite (Ca 2 B 6 O 11 ), MgO and SrCO 3 ) were all added in an MK paste formed by mixing the silicone with an appropriate solvent (30 vol%). The oxide balance corresponded to X30Y15 and X30Y15B formulations ( Table 1). The mixtures were homogenised by ball milling for 5 h at 300 rpm before printing. The printing process was carried out, in air, through a conical nozzle (with a diameter of 0.81 mm-Nordson EFD, Westlake, OH, USA), using a Delta printer (Wasp, Massa Lombarda, Italy). Reticulated structures resulted from the overlapping of filaments of about 0.8 mm diameter, with two different spacing distances (0.8 and 1.6 mm). After printing, the scaffolds were dried in air overnight. The ceramisation process was carried out in air at 950 • C for 1 h, with a heating rate of 0.3 • C/min, followed by natural cooling. During the heating phase, selected samples underwent an intermediate holding step at 590 • C for 3 h.

Preparation of Hardystonite Foams
H62C was used, instead of the MK polymer, as a silica source for foams. The formulations followed those previously shown in Table 1, except for the replacement of MgO with Mg(OH) 2 (<10 µm, Industrie Bitossi, Vinci, Italy) and the use of colemanite in the hydrated form (Ca 2 B 6 O 11 ·5H 2 O). After drying at 60 • C overnight, any H62C-based mixture turned into a viscous paste and was later cast in an aluminium cylindrical moulds (diameter of~16 mm). The paste was foamed by direct insertion in a furnace set at 420 • C for 10 min. Once extracted from the furnace, hardened foams were separated from the aluminium moulds and subjected to a final thermal treatment. Finally, the ceramisation was performed with the same schedule adopted in the case of the scaffolds.

Characterisations
The bulk density of foams and scaffolds was determined from the weight-to-volume ratio using a calliper and a digital balance. The apparent and true densities of these cellular parts were measured by a gas pycnometer (Helium gas, Micromeritics AccuPyc 1330, Norcross, GA, USA). Morphological details and microstructural characterisations were achieved by optical stereomicroscopy (AxioCam ERc 5s Microscope Camera, Carl Zeiss Microscopy, New York, NY, USA) and SEM (FEI Quanta 200 ESEM, Eindhoven, The Netherlands and JSM JEOL 6490 SEM microscope, JEOL, Tokyo, Japan) equipped with Energy-dispersive X-ray spectroscopy (EDS).
The compressive strengths of hardystonite (HT)-based foams and scaffolds were measured at room temperature, using an Instron 1121 UTM (Instron, Danvers, MA, USA) operating with a cross-head speed of 0.5 mm/min. Each data point represents the average value of at least 10 individual tests.
The identification of the crystalline phases was performed on finely ground powders by X-ray diffraction (XRD; Bruker AXS D8 Advance, Bruker, Karlsruhe, Germany) with the support of the databases of the crystallographic patterns PDF-2 (ICDD-International Center for Diffraction Data, Newtown Square, PA, USA), the Match! program (Crystal Impact GbR, Bonn, Germany), and the COD database (Crystallography Open Database, www.crystallography.net).
Furthermore, a detailed, multi-analytical approach, comprising microstructural, microchemical, and mineralogical investigations, was applied on pellets the X30Y15 formulation (derived from MK) to quantitatively verify the formation of a stable solid solution phase, including Sr 2+ , Mg 2+ , and B 3+ ions, in the hardystonite structure. The cross-section of a fired pellet was firstly characterised by SEM-EDS (CamScan MX2500 SEM, Waterbeach, United Kingdom; EDAX, Mahwah, NJ, USA). Subsequently, the same pellet was subjected to a quantitative micro-chemical characterisation by wavelength-dispersive electron microprobe (WDS-EMP; Cameca SX50, CAMECA, Gennevilliers, France). Forty point analyses were acquired from homogeneous areas of the samples, calculating the weight percentages of boron oxide by subtraction from the overall sum of recalculated oxides due to the insensitivity of the instrument to light elements. Then, the mean composition of oxides in the solid solution was determined and used as the input for the calculation of the experimental crystal chemical formula. Finally, quantitative phase analysis (QPA) and structural refinement, based on the Rietveld method [32], were performed on a highly detailed diffraction pattern (14 h data collection with PANalytical X´Pert PRO, PANalytical, The Netherlands), using HighScore Plus 4.7 program package (PANalytical, Almelo, The Netherlands).
Eluates of hardystonite scaffolds were subjected to cytotoxicity assessment tests. Therefore, a suitable bone marrow stromal cell line (ST-2, Deutsche Sammlung für Mikroorganismen und Zellkultur, Germany), isolated from bone marrow of BC8 mice, was used for this purpose. The cells were cultured in CMM (RPMI 1640 medium (Gibco, Germany)) containing 10 vol% fetal bovine serum (FBS) (Sigma-Aldrich, Germany), 1 vol% penicillin/streptomycin (Sigma-Aldrich), and 1 vol% Glutamax (Gibco)). For these tests, 100,000 ST2 cells were seeded in 1 mL CCM in 24-well plates for 24 h (incubated at 37 • C in a humidified atmosphere of 95% air and 5% CO 2 ). Simultaneously, scaffold samples were incubated in CCM for 24 h in a ratio 5 g:5 mL. CCM was removed from the samples and samples were then diluted with CCM to form 1%, 0.1%, and 0.01% dilutions. Finally, these dilutions were used to cultivate the cells for 48 h (with pure CCM as control). For assessing the influence of different material supernatant concentrations on the viability of the cultivated cells, a WST-8 assay (Sigma-Aldrich) was used. The amount of released VEGF from ST-2 cells into the cell culture medium was measured by using a RayBio Human VEGF ELISA (Enzyme-Linked Immunosorbent Assay) kit. In order to observe the morphology of the bone marrow stromal cells cultivated with different dilutions of the of hardystonite scaffolds, H&E (Hematoxylin & Eosin) staining was performed.

Phase Evolution
The starting point for the present investigation consisted of the theoretical solid solution between Zn-and B-based melilites, such as hardystonite (HSt, Ca 2 ZnSi 2 O 7 ) and okayamalite (Ok, Ca 2 B 2 SiO 7 ), as previously studied [22]. Silicone-filler mixtures designed for a stoichiometry 75 mol% HSt-25 mol% Ok (Ca 2 Zn 0.75 B 0.5 Si 1.75 O 7 ) did not yield a single phase, forming a solid solution with approximate stoichiometry of Ca 2 Zn 0.83 B 0.33 Si 1.83 O 7 , accompanied by traces of wollastonite (CaSiO 3 ) and calcium borate [22]. However, the formulation had significant advantages in producing hardystonite in a single process (no treatment of a previously synthesised powder) at a particularly low temperature (<1000 • C).
The ceramic from formulation X15 confirmed the feasibility of a single process production at low temperature, but it evidently did not follow the ideal structure (single phase, consisting of Ca 1.7 Sr 0.3 Zn 0.75 B 0.5 Si 1.75 O 7 ), as seen in the diffraction patterns in Figure 1. Most diffraction peaks matched the ones of pure hardystonite (PDF#35-0745), except for a small downshift in the 2θ positions. This downshift is reasonable, owing to the Sr doping; Sr hardystonite (Sr 2 ZrSi 2 O 7 ) is known to feature a diffraction pattern similar to that of hardystonite, but with all peaks shifted at lower 2θ positions [26]). The peak at 2θ~30 • could be attributed, as in previous studies, to wollastonite (CaSiO 3 , PDF#10-0489). Materials 2019, 12, x FOR PEER REVIEW 6 of 14 Secondary phases were likely reduced in the X30 ceramic (stoichiometry Ca1.4Sr0.6Zn0.75B0.5Si1.75O7) since all diffraction peaks could be ascribed to hardystonite solid solution; the increased downshift in the 2θ positions was consistent with the enhanced incorporation of extra ions.
For the formulations Y15 and Y30 (corresponding to the replacement of 15% and 30% of Zn 2+ ions, respectively), the assessment of the incorporation effect of Mg 2+ from shifts in the 2θ positions was difficult since the pattern of magnesium-based melilite, akermanite (Ca2MgSi2O7, PDF#79-2425), is nearly indistinguishable from that of hardystonite. Again, a secondary phase was evident.
X30Y15 and X30Y30 ceramics, corresponding to the simultaneous inclusion of Sr 2+ , Mg 2+ , and B 3+ ions in the hardystonite structure, confirmed the reduction of secondary phases found with X30. Strontium ions likely had a 'triggering' action. That is, the distortion in the spacing between zinc silicate layers, caused by the larger Sr 2+ ions compared to the Ca 2+ ions, evidently balanced the distortions in the same layers occurring upon replacement of Zn 2+ and Si 4+ ions with Mg 2+ and B 3+ ions.
The successful formation of a solid solution with such a high degree of complexity (six oxides simultaneously present) is promising since it could be the basis for the incorporation of a number of dopants. In other words, hardystonite solid solutions could represent 'hosts' for many ions, each with specific functions in tissue engineering, thereby offering an alternative to 45S5 bioactive glass (the doping of 45S5 is one of the most promising topics in the research on biomaterials) [29,33,34].
Considering the Sr and Mg amounts in the already recognised hardystonite-based biomaterials [26,27], we selected the X30Y15 formulation for additional experiments, aimed at evidencing (for the first time) the specific role of preceramic polymers. In fact, silicones, as a silica source, had a synergistic effect with oxide fillers on phase evolution. Figure 2 shows the diffraction patterns of ceramics corresponding to the X30Y15 formulation obtained (according to the same firing schedule) For the formulations Y15 and Y30 (corresponding to the replacement of 15% and 30% of Zn 2+ ions, respectively), the assessment of the incorporation effect of Mg 2+ from shifts in the 2θ positions was difficult since the pattern of magnesium-based melilite, akermanite (Ca 2 MgSi 2 O 7 , PDF#79-2425), is nearly indistinguishable from that of hardystonite. Again, a secondary phase was evident.
X30Y15 and X30Y30 ceramics, corresponding to the simultaneous inclusion of Sr 2+ , Mg 2+ , and B 3+ ions in the hardystonite structure, confirmed the reduction of secondary phases found with X30. Strontium ions likely had a 'triggering' action. That is, the distortion in the spacing between zinc silicate layers, caused by the larger Sr 2+ ions compared to the Ca 2+ ions, evidently balanced the distortions in the same layers occurring upon replacement of Zn 2+ and Si 4+ ions with Mg 2+ and B 3+ ions.
The successful formation of a solid solution with such a high degree of complexity (six oxides simultaneously present) is promising since it could be the basis for the incorporation of a number of dopants. In other words, hardystonite solid solutions could represent 'hosts' for many ions, each with specific functions in tissue engineering, thereby offering an alternative to 45S5 bioactive glass (the doping of 45S5 is one of the most promising topics in the research on biomaterials) [29,33,34].
Considering the Sr and Mg amounts in the already recognised hardystonite-based biomaterials [26,27], we selected the X30Y15 formulation for additional experiments, aimed at evidencing (for the first time) the specific role of preceramic polymers. In fact, silicones, as a silica source, had a synergistic effect with oxide fillers on phase evolution. Figure 2 shows the diffraction patterns of ceramics corresponding to the X30Y15 formulation obtained (according to the same firing schedule) from two different silicones and two different forms of silica. One could suppose that using calcium borate which leads to a liquid phase, could provide a strong 'fluxing' action and dissolve most parts of the other components at the early stages of firing, forming a viscous mass and later crystallising into a hardystonite solid solution. It was evident that neither quartz sand nor the more reactive colloidal silica led to the expected phase. Calcium and calcium-magnesium silicates (Ca 2 SiO 4 , PDF#31-0298, and CaMgSi 2 O 6 , PDF#89-1484) could be found along with quartz (PDF#86-1560) and ZnO (PDF#89-0511).
Materials 2019, 12, x FOR PEER REVIEW 7 of 14 from two different silicones and two different forms of silica. One could suppose that using calcium borate which leads to a liquid phase, could provide a strong 'fluxing' action and dissolve most parts of the other components at the early stages of firing, forming a viscous mass and later crystallising into a hardystonite solid solution. It was evident that neither quartz sand nor the more reactive colloidal silica led to the expected phase. Calcium and calcium-magnesium silicates (Ca2SiO4, PDF#31-0298, and CaMgSi2O6, PDF#89-1484) could be found along with quartz (PDF#86-1560) and ZnO (PDF#89-0511). Amorphous silica from the oxidation of silicone had a confirmed high reactivity towards oxide fillers. Switching from MK to H62C did not modify the developed phase given the position and the intensity of diffraction peaks; H62C likely resulted in just a reduction in crystal size, considering the broadening of diffraction peaks, as seen in Figure 2. The pellets from MK were subjected to a refined microstructural and mineralogical analysis. SEM-EDS analyses ( Figure 3) showed that the X30Y15 sample constituted a fairly homogeneous porous matrix with a dominating content of Ca and Si, but also included Zn, Sr, and Mg (spectrum 1 in Figure 3). Such experimental evidence indicates a successful incorporation of the elements in the solid solution (B was not detectable by EDS analysis). However, some portions with different chemical composition, i.e., with a darker colour in the backscattered electron image, were detected. The Ca/Si ratio in these portions presented significant variations, passing from high levels (spectra 2-3), typically presented by wollastonite (CaSiO3, i.e., Ca/Si = 1), to particularly low levels, corresponding to almost pure silica (spectrum 4). Such experimental data indicate that the reaction process still did not lead to a full incorporation of all the chemical constituents into a single crystalline phase; however, hardystonite solid solution was largely dominant. Amorphous silica from the oxidation of silicone had a confirmed high reactivity towards oxide fillers. Switching from MK to H62C did not modify the developed phase given the position and the intensity of diffraction peaks; H62C likely resulted in just a reduction in crystal size, considering the broadening of diffraction peaks, as seen in Figure 2.
The pellets from MK were subjected to a refined microstructural and mineralogical analysis. SEM-EDS analyses (Figure 3) showed that the X30Y15 sample constituted a fairly homogeneous porous matrix with a dominating content of Ca and Si, but also included Zn, Sr, and Mg (spectrum 1 in Figure 3). Such experimental evidence indicates a successful incorporation of the elements in the solid solution (B was not detectable by EDS analysis). However, some portions with different chemical composition, i.e., with a darker colour in the backscattered electron image, were detected. The Ca/Si ratio in these portions presented significant variations, passing from high levels (spectra 2-3), typically presented by wollastonite (CaSiO 3 , i.e., Ca/Si = 1), to particularly low levels, corresponding to almost pure silica (spectrum 4). Such experimental data indicate that the reaction process still did not lead to a full incorporation of all the chemical constituents into a single crystalline phase; however, hardystonite solid solution was largely dominant.  The mean oxide composition of the hardystonite solid solution, reported in Table 3, may be summarised by the crystal chemical formula of (Ca0.70Sr0.30)2(Zn0.72Mg0.15Si0.13) (Si0.85B0.15)2O7. This formula was confirmed by the Rietveld refinements, illustrated in Figure 4, performed on a high-resolution diffraction pattern (14 h data collection, Co anode instead of Cu anode used for preliminary phase identification studies). The chemical formula of hardystonite from EMP-WSD analysis was implemented in a melilite structural model Panalytical ICSD database (akermanite 280405). Atomic positions and occupancies were refined (see Table 4) with the exception of the 2a Wyckoff position, including 3 atomic species (Si, Zn, Mg). The Rietveld refinement led to an excellent fitting of experimental data and evidenced secondary phases, in the forms of pseudo-wollastonite, wollastonite 2M, and quartz, in limited amounts (not exceeding 7 wt%). The presence of pseudo-wollastonite was also confirmed by the characteristic cathodo-luminescence [35] observed during EMP-WDS analysis.  The mean oxide composition of the hardystonite solid solution, reported in Table 3, may be summarised by the crystal chemical formula of (Ca 0.70 Sr 0.30 ) 2 (Zn 0.72 Mg 0.15 Si 0.13 ) (Si 0.85 B 0.15 ) 2 O 7 . This formula was confirmed by the Rietveld refinements, illustrated in Figure 4, performed on a high-resolution diffraction pattern (14 h data collection, Co anode instead of Cu anode used for preliminary phase identification studies).  The mean oxide composition of the hardystonite solid solution, reported in Table 3, may be summarised by the crystal chemical formula of (Ca0.70Sr0.30)2(Zn0.72Mg0.15Si0.13) (Si0.85B0.15)2O7. This formula was confirmed by the Rietveld refinements, illustrated in Figure 4, performed on a high-resolution diffraction pattern (14 h data collection, Co anode instead of Cu anode used for preliminary phase identification studies). The chemical formula of hardystonite from EMP-WSD analysis was implemented in a melilite structural model Panalytical ICSD database (akermanite 280405). Atomic positions and occupancies were refined (see Table 4) with the exception of the 2a Wyckoff position, including 3 atomic species (Si, Zn, Mg). The Rietveld refinement led to an excellent fitting of experimental data and evidenced secondary phases, in the forms of pseudo-wollastonite, wollastonite 2M, and quartz, in limited amounts (not exceeding 7 wt%). The presence of pseudo-wollastonite was also confirmed by the characteristic cathodo-luminescence [35] observed during EMP-WDS analysis. The chemical formula of hardystonite from EMP-WSD analysis was implemented in a melilite structural model Panalytical ICSD database (akermanite 280405). Atomic positions and occupancies were refined (see Table 4) with the exception of the 2a Wyckoff position, including 3 atomic species (Si, Zn, Mg). The Rietveld refinement led to an excellent fitting of experimental data and evidenced secondary phases, in the forms of pseudo-wollastonite, wollastonite 2M, and quartz, in limited amounts (not exceeding 7 wt%). The presence of pseudo-wollastonite was also confirmed by the characteristic cathodo-luminescence [35] observed during EMP-WDS analysis.

Obtainment of Scaffolds and Foams
Scaffolds and foams were prepared starting from MK-and H62C-based mixtures, respectively. Figure 5 shows selected images from scaffolds based on the X30Y15 formulation from direct ink writing of silicone pastes. Owing to the use of anhydrous borate, the residual porosity was quite limited; however, some microcracks were still visible (Figure 5a,b). These cracks are ascribable to the gas released from fillers and from the ceramic transformation of the MK silicone (still representing the dominant silica source). The filaments, in any case, showed a good interpenetration ( Figure 5b); high magnification details (Figure 5c) showed the formation of a multitude of tiny crystals. The scaffolds, in terms of strength (Table 3) for both settings of filament spacing (800 and 1600 µm), compared favourably with other hardystonite scaffolds with similar porosity [36].

Obtainment of Scaffolds and Foams
Scaffolds and foams were prepared starting from MK-and H62C-based mixtures, respectively. Figure 5 shows selected images from scaffolds based on the X30Y15 formulation from direct ink writing of silicone pastes. Owing to the use of anhydrous borate, the residual porosity was quite limited; however, some microcracks were still visible (Figure 5a,b). These cracks are ascribable to the gas released from fillers and from the ceramic transformation of the MK silicone (still representing the dominant silica source). The filaments, in any case, showed a good interpenetration ( Figure 5b); high magnification details (Figure 5c) showed the formation of a multitude of tiny crystals. The scaffolds, in terms of strength (Table 3) for both settings of filament spacing (800 and 1600 μm), compared favourably with other hardystonite scaffolds with similar porosity [36]. The water vapour released from within H62C still in its polymer state, from the use of hydrated fillers (hydrated Ca borate and Mg(OH)2), led to a very remarkable foaming at 420 °C. The ceramic transformation confirmed an abundant porosity (>80 vol%), completely open and with wide interconnections (>100 μm), reflecting the main requirements of a scaffold for tissue engineering applications [37], as shown in Figure 5d. The high magnification details illustrate the presence of a The water vapour released from within H62C still in its polymer state, from the use of hydrated fillers (hydrated Ca borate and Mg(OH) 2 ), led to a very remarkable foaming at 420 • C. The ceramic transformation confirmed an abundant porosity (>80 vol%), completely open and with wide interconnections (>100 µm), reflecting the main requirements of a scaffold for tissue engineering applications [37], as shown in Figure 5d. The high magnification details illustrate the presence of a substantial porosity in the cell walls (Figure 5e,f), resulting from the gas released upon ceramic conversion, as well as the macro-porosity determined by the low temperature foaming, caused by dehydration of some fillers. Porous walls are known to promote cell adhesion, absorption of metabolites, and faster controlled rates for the release of ionic dissolution products [38]. The observed crushing strength (Table 5) is in the order of that of silicate foams with the same porosity [14,39].

Preliminary Cell Tests
The phase assemblage, resulting from the mineralogical analysis, supports the hypothesis of a crystallinity degree not exceeding 80 wt%, as seen in the relatively high content of MgO detected in the hardystonite solid solution (corresponding to 93% of the crystals), compared to the reference stoichiometry (X30Y15). This estimation is consistent with the assumption of MgO completely embedded in the hardystonite solution and the other oxides being distributed, except for minor crystal phases, in a glass phase surrounding the silicate crystals. This implies, above all, that the concentration of B 2 O 3 (in an amount of about 30 wt% of the glassy phase) could be critical for the application of the developed porous materials in tissue engineering. In fact, although present in many bioceramics [29,30], B 2 O 3 is quite controversial because of the potential toxicity of boron released in solution as borate ions (BO 3 ) 3− [40,41].
Preliminary cell tests were performed on X30Y15 scaffolds with the specific purpose of elucidating any cytotoxic effect. The cell viability of ST2 cells in the presence of the scaffold eluates is shown in Figure 6. The reference, which relates to cells cultured only with CCM, was normalised to 100%. Quantitative assessment after 48 h of culture shows that there was an increase in cell viability with a decrease in the concentration of the scaffold eluates in the cell culture medium. In the case of 0.01% dilution, the highest cell viability was achieved by the 0.1% concentration. The values are statistically relevant as the difference was greater than p = 0.001. The scaffold eluates were found to not be cytotoxic after 48 h of incubation in contact with bone marrow stromal cells. substantial porosity in the cell walls (Figure 5e,f), resulting from the gas released upon ceramic conversion, as well as the macro-porosity determined by the low temperature foaming, caused by dehydration of some fillers. Porous walls are known to promote cell adhesion, absorption of metabolites, and faster controlled rates for the release of ionic dissolution products [38]. The observed crushing strength (Table 5) is in the order of that of silicate foams with the same porosity [14,39].

Preliminary Cell Tests
The phase assemblage, resulting from the mineralogical analysis, supports the hypothesis of a crystallinity degree not exceeding 80 wt%, as seen in the relatively high content of MgO detected in the hardystonite solid solution (corresponding to 93% of the crystals), compared to the reference stoichiometry (X30Y15). This estimation is consistent with the assumption of MgO completely embedded in the hardystonite solution and the other oxides being distributed, except for minor crystal phases, in a glass phase surrounding the silicate crystals. This implies, above all, that the concentration of B2O3 (in an amount of about 30 wt% of the glassy phase) could be critical for the application of the developed porous materials in tissue engineering. In fact, although present in many bioceramics [29,30], B2O3 is quite controversial because of the potential toxicity of boron released in solution as borate ions (BO3) 3− [40,41].
Preliminary cell tests were performed on X30Y15 scaffolds with the specific purpose of elucidating any cytotoxic effect. The cell viability of ST2 cells in the presence of the scaffold eluates is shown in Figure 6. The reference, which relates to cells cultured only with CCM, was normalised to 100%. Quantitative assessment after 48 h of culture shows that there was an increase in cell viability with a decrease in the concentration of the scaffold eluates in the cell culture medium. In the case of 0.01% dilution, the highest cell viability was achieved by the 0.1% concentration. The values are statistically relevant as the difference was greater than p = 0.001. The scaffold eluates were found to not be cytotoxic after 48 h of incubation in contact with bone marrow stromal cells. In Figure 7, the VEGF released from the ST2 cells cultured in CCM with different dilutions (1%, 0.1%, and 0.01%) of scaffold is shown. Measuring the increase in VEGF concentration when cells are exposed to dissolution products of biomaterials is a suitable in vitro methodology [42] to assess the potential angiogenic effect of the material being investigated, given that increased VEGF will attract In Figure 7, the VEGF released from the ST2 cells cultured in CCM with different dilutions (1%, 0.1%, and 0.01%) of scaffold is shown. Measuring the increase in VEGF concentration when cells are exposed to dissolution products of biomaterials is a suitable in vitro methodology [42] to assess the potential angiogenic effect of the material being investigated, given that increased VEGF will attract endothelial cells, being thus a marker for the vascularisation potential of the biomaterial. The eluates of this scaffold increased VEGF secretion with increasing supernatant concentration. The 1% supernatant sample showed the highest release of VEGF for all three dilutions. These results were also in accordance with the data obtained from the cell viability study. However, the results also indicated no significant variation as a function of the reference sample.
Materials 2019, 12, x FOR PEER REVIEW 11 of 14 endothelial cells, being thus a marker for the vascularisation potential of the biomaterial. The eluates of this scaffold increased VEGF secretion with increasing supernatant concentration. The 1% supernatant sample showed the highest release of VEGF for all three dilutions. These results were also in accordance with the data obtained from the cell viability study. However, the results also indicated no significant variation as a function of the reference sample.

Conclusions
Although still not phase pure, porous ceramics based on a complex hardystonite solid solution were successfully manufactured by the direct firing of silicone-based mixtures at only 950 °C in air. The reactivity of amorphous silica provided by silicones was much higher than that of more conventional silica sources when combined with the same additives (precursors of the other oxides). The production of the hardystonite solid solution involved the unprecedented combination of six oxides, with Mg 2+ and B 3+ ions modifying the ZnSi2O7 4− sheets, and Sr 2+ ions partially replacing the Ca 2+ ions sandwiched between the same sheets.
The use of silicones mixed with fillers, besides favouring synthesis, enabled the application of simple shaping technologies, such as direct ink writing and direct foaming, at low temperature, with silicones still in the polymer state. All developed porous hardystonite ceramics compared well, in endothelial cells, being thus a marker for the vascularisation potential of the biomaterial. The eluates of this scaffold increased VEGF secretion with increasing supernatant concentration. The 1% supernatant sample showed the highest release of VEGF for all three dilutions. These results were also in accordance with the data obtained from the cell viability study. However, the results also indicated no significant variation as a function of the reference sample.

Conclusions
Although still not phase pure, porous ceramics based on a complex hardystonite solid solution were successfully manufactured by the direct firing of silicone-based mixtures at only 950 °C in air. The reactivity of amorphous silica provided by silicones was much higher than that of more conventional silica sources when combined with the same additives (precursors of the other oxides). The production of the hardystonite solid solution involved the unprecedented combination of six oxides, with Mg 2+ and B 3+ ions modifying the ZnSi2O7 4− sheets, and Sr 2+ ions partially replacing the Ca 2+ ions sandwiched between the same sheets.
The use of silicones mixed with fillers, besides favouring synthesis, enabled the application of simple shaping technologies, such as direct ink writing and direct foaming, at low temperature, with silicones still in the polymer state. All developed porous hardystonite ceramics compared well, in

Conclusions
Although still not phase pure, porous ceramics based on a complex hardystonite solid solution were successfully manufactured by the direct firing of silicone-based mixtures at only 950 • C in air. The reactivity of amorphous silica provided by silicones was much higher than that of more conventional silica sources when combined with the same additives (precursors of the other oxides). The production of the hardystonite solid solution involved the unprecedented combination of six oxides, with Mg 2+ and B 3+ ions modifying the ZnSi 2 O 7 4− sheets, and Sr 2+ ions partially replacing the Ca 2+ ions sandwiched between the same sheets. The use of silicones mixed with fillers, besides favouring synthesis, enabled the application of simple shaping technologies, such as direct ink writing and direct foaming, at low temperature, with silicones still in the polymer state. All developed porous hardystonite ceramics compared well, in terms of strength-to-density ratio, with the analogous materials presented in the literature. The direct foaming, in particular, is promising for the possibility of obtaining cellular bodies with both well-interconnected macropores and porous cell walls. According to the results of preliminary cell tests, the developed ceramics were not cytotoxic.