Statistical analysis

The one-way analysis of variance (ANOVA) was used to evaluate significant differences between means in the measured data. Scheffe’s multiple comparison testing was used to determine the significance of standard deviations in the measured data from each specimen under different experimental conditions. In all cases the results were considered statistically significant with a p-value of less than 0.05.

5. Results and Discussion

5.1. Gelatin distribution and porosity

The burnout test was used to confirm the distribution extent of gelatin in the composites because of the gelatin combustion at 600ºC. The actual gelatin content agrees well with the theoretical values (Fig. 1). Additionally, different parts of the same composite had almost identical gelatin content, indicating that the gelatin dispersed uniformly in the calcium silicate phase and that there was no phase separation between the two phases.

The porosity was 16, 12, 10, and 11%, respectively, for the specimens containing 0, 5, 10, and 15 wt% gelatin. The control without gelatin was significantly (p < 0.05) higher than the gelatin-containing composites (Table 1). The porosity of the three bio-inspired hybrid composites was close to that of cortical bone (312%). No significant difference (p > 0.05) was found in the density (approximately 1.8 g cm^-3) of the specimens, as listed in Table 1.

5.2. Phase composition

The XRD patterns of the as-prepared CaSi powders sintered at 800ºC, indicating that the majordiffraction peaksat2θ between 32and 34 areattributed to theβ-Ca2SiO4(β-dicalcium silicate) phase (Fig. 2a). For the bulk composites with and without gelatin, an obvious

diffraction peak near2θ = 29.4,corresponding to the calcium silicate hydrate (C–S–H) gel overlapped with calcite (CaCO3),and incompletely reacted inorganiccomponentphasesof β-Ca₂SiO₄were found. The hydration reaction of the β-Ca₂SiO₄powder is initiated upon contact with water, which was featured the present study. The hydration reaction connected the originally hydrophilic particles together, resulting in a C–S–H gel that developed bonding properties and calcium hydroxide (Ca(OH)2) that may transform into the calcite phase. It is clear that the peak intensity of C–S–H formed in the gelatin-containing specimen was lower, in particular for the 15 wt% gelatin, when compared to the control. The diffraction peaks of the gelatin- containing composites resulted mainly from poorly crystalline C––S–H, CaCO3, and β-Ca2SiO4, confirming that the amorphous gelatin filler did not affect the ceramic phase, although the gelatin did contribute to the lower peak intensities of the C–S–H and β-Ca₂SiO₄ phases.

In the Raman spectra the broad peak between 590 and 735 cm^-1 is assigned to the SiOSi symmetric bending mode of the silicate species [8], corresponding to CSH (Fig.

2b). The observed Raman activity at 860 cm^-1 for the symmetric vibration of silicate was ascribed to β-Ca₂SiO₄.³¹ Furthermore, the broad band at 984 cm^-1 was attributed to the antisymmetric vibration of the dimeric silicate species [9]. A wavenumber of 1086 cm^-1 was assigned to the symmetric stretching of the carbonate ion [10], whose appearance is due to the rapid uptake of CO₂ from air. Carbonation led to the formation of calcium carbonate at the expense of Ca(OH)₂, whose characteristic Raman peak disappeared at 356 cm^-1 [8], consistently with the XRD results. Additionally, the peaks at 1392 and 822 cm^-1 were suggested to be the functional group of gelatin in the Raman spectrum [11], which indicated the existence of gelatin in the CaSi-gelatin composites.

5.3. Morphology

Fig. 3 shows the surfaces of specimens with different ratios of gelatin. The control without gelatin has an appearance of entangled particles and exhibited several pores (Fig. 3a).

When adding gelatin, the hybrid composite becomes a heterogeneous structure with a homogenous distribution of gelatin within the CaSi matrix, and there are many filament crystals surrounding the CaSi body, which become more obvious with increasing gelatin content (Figs. 3bd). In sharp contrast, it seems that the morphology of the hybrid composites was more compact than the control, consistent with the porosity.

The analysis of the fractured surface using fractographic principles is a well-established analytical tool to determine the failure behavior of brittle materials. The control presents a sandy, smooth, and loose structure, while the gelatin-containing specimens are of solid and dense features (Fig. 4); this finding is similar to another study [12]. The added gelatin, which was examined in this study, may play a crucial role in the properties of the bio-inspired composites. A colloidal gel facilitates the formation of a dense structure of hydrated composites that entangle the original C–S–H structure. Such dense structures of gelatin-containing composites might be due to the existence of negatively charged gelatin. Type B gelatin with an isoselectric point of 5 has a high density of carboxyl groups, which makes the gelatin negatively charged [13,14]. The carboxyl groups might bind calcium ions on the surface of the CaSi particles. Hence, the gelatin structure not only serves as the filler incorporating to the CaSi matrix but also provides an anchoring site for the CaSi particles in the structure, binding them together to form a composite.

5.4. Compressive properties

Table 1 lists the relationships between the compressive strength, modulus, work-of-fracture, or hardness of the composites and the gelatin content. The specimens showed non-monotonous changes in strength and a significant difference (p < 0.05) among the specimens was found. Composites containing 5 wt% gelatin (CSG5) had a compressive strength value of 105.0 MPa on average, significantly (p < 0.05) higher than the CSG0 control (86.1 MPa). The addition of gelatin up to 10 wt% (CSG10) achieved a significantly (p < 0.05) increased compressive strength of up to 141.7 MPa (p < 0.05), which is within the reported compressive strength for cortical bone [15]. However, higher gelatin content at 15 wt% (CSG15) adversely affected the mechanical strength with a reduction of up to 31% of the highest value.

Although a high isostatic pressure of 500 MPa was applied to the specimens, which partly contributed to the strength measured, the obtained high strength could result from the hydrothermal factor. To confirm the speculation, the CaSi powders containing 0, 5, 10, and 15 wt% gelatin were directly pressed by cold isostatic pressing at 500 MPa without the further hydrothermal treatment in water at 60ºC. The compressive strength values of the directly pressed compact specimens only were 6.3, 4.2, 11.4, and 14.3 MPa, respectively, indicating significantly (p < 0.05) lower values than the corresponding specimens prepared by the dual pressing-hydrothermal method. The hydrothermal treatment-induced increase in strength is attributable to the more complete hardening during soaking in solution, which has also been observed in other studies [5,13,16]. The development of mechanical properties of the present materials is mainly the result of the reaction of liquid phases, such as water, with the components. Nevertheless, high applied pressure is beneficial via the densification mechanisms of nanoparticle rearrangement and sliding, plastic deformation, and pore shrinkage. More importantly, such densification procedures using a simple pressing-hydrothermal route can provide the possibility of incorporating proteins and drugs without damaging their biological activity, featuring a drug delivery system.

One concern is that the incorporation of gelatin into CaSi ceramics might result in degradation of the mechanical strength of the composites. However, in this study, this concern did not occur. When increasing the gelatin, to the result is an increase in the compressive strength; however, once it reaches the maximum value (i.e., 10% of gelatin), the compressive strength decreases drastically. The resulting high strength of the hybrid composites are due to a combination between the progressive hardening originating from the main CaSi reactant and reinforcementeffectfilling thedefectsby thegelatin phase,which servesasa‘glue’to fusethe particles together, as confirmed by SEM. Some surface chemistry properties unique to the gelatin enhance the interactions between the filler and the matrix [17]. It is also possible that the gelatin in the composites reduces the effect of the flaws and the porosity on the mechanical behavior of these hybrid composites, as indicated by the results of the porosity testing.

Porosity has a deleterious effect on the properties of ceramic materials, acting as a stress raiser and reducing the mechanical strength [18]. More recently, Martínez-Vázquez et al. suggested that the presence of polymer within the macropores of the ceramic modified the stress acting on the ceramic during the uniaxial compression tests for strengthening [19]. The proper stress transfer occurring between the reinforcement and the matrix may govern the mechanical characteristics of the hybrid composites [20]. Concerning the present study, chemical and mechanical interlocking between the CaSi and gelatin accounts for the efficient stress transfer in the composite system. It could be expected that excessive gelatin might disrupt the entanglement structure, resulting in weaker interfacial bonding between the CaSi matrix and gelatin. Moreover, the loose connection throughout the excessive gelatin might be another factor that decreases the strength. Additionally, the decrease in the strength of the CSG15 group can be attributed to the poor mechanical strength of the excessive amorphous phase that

lacks a crystalline configuration.

The values of the mean elastic moduli obtained in the range of 2.7 to 3.1 GPa during the measurement of compressive strength are also listed in Table 1 with the respective standard deviations. The amount of added gelatin influenced the elastic modulus of the composites;

however, the trend presented was not similar to the changes in compressive strength. The modulus decreased somewhat after the incorporation of either 5 wt% or 10 wt% gelatin to the control, but there was no significant difference (p > 0.05) after Scheffe’smultiplecomparison testing. A reason for the observed decrease in the compressive modulus of the composites is that gelatin is not stiffer than the surrounding ceramic matrix. Hence, it can be expected that the increased contents of the gelatin lead to a lower modulus, as confirmed by the result of CSG15. It is noted that the modulus of all of the bio-inspired composites is slightly lower than thelowerbound forthoseofthehuman corticalbone.When thematerialhasahigherYoung’s modulus than bone, they can cause stress shielding and lead to bone resorption. Regarding the work-of-fracture, various specimens increased with increasing gelatin content up to 10 wt%, indicating there was a significant difference (p < 0.05). After that, the work-of-fracture began to decrease, eliciting a trend similar to the variations in the strength. Although the detailed mechanism of this change is not fully understood, enhanced plastic deformation is achieved when the small organic molecules are uniformly dispersed throughout the ceramics and interacted strongly with the ceramic matrix.

5.5. Weibull analysis

To understand the level of the structural reliability of the materials, a Weibull analysis is commonly used to analyze historical failure data and produce failure distributions. The twenty measurements of failure strength from each group were ranked in order from the weakest to the strongest. Fig. 5 shows a double logarithmic plot of the experimental compressive strength data. It can be seen that the data points fell approximately along a straight line, which indicated that the two-parameter Weibull distribution is a reasonable assumption for the compressive strength distribution of the four specimens. The description of the Weibull distribution is given by the formula [21]:

F(σ)= 1-exp [-(σ/σ0)^m] (2)

whereF(σ)isthefailureprobability (defined by therelation ofF(σ)= i/(N+1),in which iisthe rank orderofthecompressivestrength and N isthenumberofspecimens),σ isthestrength at a given F(σ),σ0 is the characteristic strength (scale parameter) at the fracture probability of 63.2%,and m istheWeibullmodulus(shapeparameter).Largervaluesofσ0 imply greater strength, and larger values of m imply more reliably reproducible specimens in a narrower distribution of values. The characteristic strengths of the composites were 90.1, 106.4, 144.7, and 102.4 MPa for the 0, 5, 10, and 15 wt% gelatin, respectively, and the corresponding dependent behaviors. The 10 wt% gelatin-containing composite had a higher Weibull modulus (17.5) than the other three specimens, indicating more uniform strength distribution and Weibull moduli were 12.2, 8.9, 17.5, and 8.1, respectively, which were calculated from the slope of the fitting curves. The Weibull characteristic strength and Weibull modulus did display gelatin-higher structural reliability. In general, ceramics have wide variability in failure strength because of the flaws incorporated during processing. Most ceramics are reported to have m values in the range of 5 to 15 [22].

5.6. Hardness

Hardness is intended to be a measure of the resistance to plastic deformation, which may include effects such as material displacement and fracture. Table 1 shows that CSG5 has a

significantly (p < 0.05) higher Vickers’microhardnessvalue(120.8)than thatoftheCSG0 control (99.1). The microhardness value of CSG10 (103.5) was not much higher than that of the CSG0 control, indicating no significant difference (p > 0.05). Different from the trend in the compressive strength, when the composite contained 15 wt% gelatin (CSG15), the specimen hardness (82.0) was significantly (p < 0.05) lower than that of the control. Such a large decrease in hardness might be related to the soft feature of the gelatin.

Fig. 6 shows the typical indentations produced by Vickers’ microhardness tester, indicating the appearance of well-defined features. A number of surface cracks of varying lengths were associated with the impression of the indentation on the CSG0 control and CSG15 surfaces. The observed surface cracks are believed to be from the mismatch between the ceramic matrix and polymer additive. In contrast to the findings, the CSG5 and CSG10 composites had no signs of crack motion at the impression tip. The amorphous gelatin phase of the composites during the compression could be reorienting up to breakage. The high resistance to crack-induction makes the present CSG10 composite highly interesting in both fundamental and practical aspects.

5.7. In vitro fatique

The load-bearing implant materials must be evaluated not only by their strength as described previously but also by their resistance to fatigue in solution because of the cyclic nature of in vivo loading. However, to the knowledge of the authors, relatively little information is available in the published literature concerning fatigue behavior in a physiological solution for bio-inspired ceramic/polymer composites. The results of fatigue experiments are presented as S-N diagrams (the so-called fatigue life diagram), where S is the maximum stress in a cyclic loading and N is the number of cycles until fracture. Fig. 7 is a plot of the maximum stress applied in the compressive cyclic fatigue against the number of cycles to failure at 37ºC in SBF. The graph showed that the stability of the composites was apparently affected by the cyclic loading with a remarkable decrease in the strength as the number of cycles increased. For example, the control fatigued in SBF for 2 ×10³cycles had a significant degradation down to 35% of the original strength. When a loading stress of 37 MPa was applied, the specimen containing 10 wt% gelatin (CSG10) lasted approximately 40 min in the in vitro fatigue test until failure occurred. As for the 0, 5, and 15 wt% gelatin-containing specimens, 40-min fatigue-induced failure in SBF solution required only 27, 32, and 26 MPa, respectively.

When subjected to fatigue testing in SBF under a dynamic condition, the detrimental effect on the specimens is seen, consistent with a previous study [23]. In addition to the crack growth that occurs under stress, the decrease in strength of the bio-inspired materials is also caused by environmental factors, such as the penetration of water/ions [23, 24]. Water/ions can easily infiltrate the inner portion of the specimens through structural imperfections, particularly under applied stress, resulting in weakened bond strength due to particle dissolution. The addition of gelatin can also increase the ratio of water absorption of the composite, which reduced the retention of the mechanical properties of the CaSi-gelatin composite fatigued under moisture conditions. The conclusion may be drawn that the present specimens have a limited resistance to fatigue failure. However, it provides that the fatigue properties of composites should be a noticeable factor.

5.8. In vitro bioactivity and degradation

Broad face and cross-sectional SEM micrographs of the bone grafts after soaking in a SBF solution for 1 day are shown in Fig. 8. The surface morphology of CSG0 immersed for 1

day was similar to those of the other three specimens (Fig. 8 a-d). It is clear that precipitation took place on all specimen surfaces which were covered with clusters of precipitated spherulites. From the cross-sectional SEM micrograms (Fig. 8 e-h), it can be seen that such a precipitated layer with a distinct contrast was observed in all specimens, confirming the broad face examination. The precipitated apatite layer was smoother and denser than the original specimen structure. The average thickness of precipitated apatite layers was approximately 220, 200, 160, 100 nm, respectively, for CSG0, CSG5, CSG10, and CSG15. To further confirm that the observed layer was indeed ascribed to apatite precipitated from the SBF solution, SEM/EDS analyses were performed on the 1-day-immersed specimens, in addition to the XRD analysis. Ca/P ratios of the SBF-immersed specimens were 4.0, 4.5, 5.8, and 10.6 for CSG0, CSG5, CSG10, and CSG15, respectively, which significantly (p < 0.05) increased with increasing gelatin content. The reason for the much lower apatite precipitation rate on CSG15 was possibly due to the presence of gelatin in its as-prepared material. Nevertheless, all the bonegraftsshowed astrong tendency for“attracting”apatiteprecipitateonto theirsurfaces, as evidenced by the Ca/P molar ratio. The much higher Ca/P ratio (compared to the 1.67 stoichiometric Ca/P ratio of apatite) on 1-day-immersed surfaces was not surprising due to the fact that a large quantity of calcium originating from the underlying specimens was detected.

The lower Ca/P ratio on the surface of the CSG0 control without gelatin was possibly due to the faster apatite precipitation rate, consistent with the thickness of the precipitated layer. The in vitro bioactivity of the SiO₂–CaO-based materials indicates that the presence of PO43

ions in the composition is not an essential requirement for the development of an apatite layer, which consumes the calcium and phosphate ions. This is because PO43

ions originate from the in vitro assay solution. An increase in the pH of SBF at different time intervals was attributed to the release of Ca(OH)₂, which is conducive for the formation of apatite precipitation. The results of the higher pH value in the CSG0 control-immersed SBF paralleled the apatite precipitation rate.

It was of interest to immerse specimens in an SBF solution for extended time (up to 180 days) to investigate the variations in the activity and degradation of the composites. After soaking for 180 days, the surface morphology of the specimens was significantly altered in the presence of the etching-induced micropores on the apatite layer (Fig. 9). It appears that during the immersion test dissolution of the surface had taken place. To further understand the etching effects, porosity measurements were conducted using a liquid displacement technique.

Before soaking in SBF, the porosities were 16%, 12%, 10%, and 11% for the specimens containing 0, 5, 10, and 15 wt % gelatin, respectively. On day 180, the porosities became 17%, 22%, 23%, and 28%, respectively. Significant differences (p < 0.05) between the porosities before and after soaking were found in the gelatin-containing composites.

Four soaking regimes up to 180 days were selected for testing compressive strength of the specimens, as shown in Fig. 10. The results revealed that all the four different types of bone grafts gradually lost their strengths with the increase in soaking time. After immersion for 180 days, the strength values of immersed specimens were in the range of 77−39 MPa, lower than respective strength values on day 0. It is worth noting that the gelatin-containing composites had a significantly lower strength (p < 0.05) compared to corresponding as-prepared composites, but not for the CSG control. Additionally, CSG0, CSG5, and CSG10 had a similar strength at day 180. It is surprising that the CSG0 control has insignificantly bond

Four soaking regimes up to 180 days were selected for testing compressive strength of the specimens, as shown in Fig. 10. The results revealed that all the four different types of bone grafts gradually lost their strengths with the increase in soaking time. After immersion for 180 days, the strength values of immersed specimens were in the range of 77−39 MPa, lower than respective strength values on day 0. It is worth noting that the gelatin-containing composites had a significantly lower strength (p < 0.05) compared to corresponding as-prepared composites, but not for the CSG control. Additionally, CSG0, CSG5, and CSG10 had a similar strength at day 180. It is surprising that the CSG0 control has insignificantly bond