*3.1. Microstructural Characterization*

Computer tomographic images of an exemplary hybrid structure confirmed that the form and material fit between the columns and the foam (Figure 4). From left to right, first the VPP column and then the hybrid, both in the green state, are displayed, followed by the sintered hybrid.

**Figure 4.** CT images of an exemplary hybrid structure; from left to right: left: VPP support structure (green state); middle: hybrid (green state); right: sintered hybrid (**A**: plan view, **B**: side view).

SEM images clearly showed the denser VPP-manufactured round column and the porous Freeze Foam (Figure 5). At the junction between them, several gaps appear. Foam and the VPP part fused together, but only partially and only in a few spots. However, this SEM analysis only shows one specific location within one hybrid structure. More hybrids need to be manufactured and examined to come to a general conclusion.

**Figure 5.** FESEM cross-cutting images displaying interface between VPP-manufactured column and Freeze Foam; magnification/HFW 35×/3334 μm (**left**); 100×/1160 μm (**middle**); 500×/242 μm (**right**); 8 kV acceleration voltage.

The gap between column and foam was measured at one location (Figure 6), which varied between 1 and 13 μm. In general, the column was much denser than the foam, with macropores of around 100 to <600 μm. A closer look at a higher magnification showed mesopores of around 1–2 μm in the foam and in the struts (Figures 5 and 6).

One hybrid foam was analyzed in the fractured view (Figure 7). The gap at the interface between the foam and column is obvious as well as the interconnected pores in the Freeze Foam. At this location, a higher magnification showed that the material fit between the foam and the column and formed the TCP microstructure, with mesopores of around 1–2 μm.

**Figure 6.** FESEM image: measured distance in the gap interface of VPP column and Freeze Foam.

**Figure 7.** SEM fractured view of gap interface of VPP column and Freeze Foam: magnification/HFW 26×/4446 μm (**left**); 100×/1109 μm (**middle**); 500×/227 μm (**right**); 6 kV acceleration voltage.

Fractured and cross-sectioned images (Figure 8) once again show a good material fit between the column and the foam. As stated before, the interface connection was not thoroughly complete, and its state/appearance depended on the location in the hybrid (referring to each column and each column length).

**Figure 8.** SEM fractured and cross-section view of gap interface of VPP column and Freeze Foam: magnification/HFW 35×/3550 μm (**left**); 100×/1109 μm (**right**); 8 kV acceleration voltage.

The surface roughness Sa was determined to be 5.99 ± 1.43 μm for the Curasan control, 3.73 ± 1.94 μm for the outer ceramics (column's) ring of the hybrid foam and 6.54 ± 2.93 μm for the inner foam structure of the hybrid foams (Figure 9).

**Figure 9.** 3D Laser scanning image (top) and 3D reconstruction (bottom) of: (**a**) Curasan control; (**b**) Hybrid Foam, outer ceramic ring; (**c**) Hybrid Foam, inner foam. Images were taken with the KEYENCE VK-X210 3D Laser scanning microscope, 400× magnification.

#### *3.2. Mechanical Characterization and Comparison to FE Simulation*

One of the aims of this study was to enhance the mechanical stability of the TCP. Therefore, Freeze Foams and hybrid foams were tested for compressive strength. The following table summarizes the compressive strength (relating to the cylindrical crosssection) and porosity (geometrical and foam cells determined via foam structure analysis based on CT images) of tested samples (five each, mean values shown in Table 2).


**Table 2.** Compressive strength of manufactured components (foams, hybrids, and columns).

\* Archimedes method.

Freeze Foams and hybrid foams exhibited similar porosity. However, the hybrid's compressive strength was 25 times higher (23 MPa) than the Freeze Foam alone (0.9 MPa). Surprisingly, the VPP columns alone showed an even higher compressive strength. Those values lie, however, within the standard deviation. It must be noted that the standard deviation was quite large. There were microdefects leading to failures in the macrostructure and/or the loaded surface was not plane, leading to varying forces upon contact with the compression stamp. The Curasan component provided the lowest porosity of all CaP scaffolds and showed much lower compressive strength than the hybrid foam (roughly one-seventh).

For interpretation of the simulation results, the maximum principal stress was considered because of the known brittleness of the support structure's ceramic material. Tensile load cases are critical for ceramics. Results of the first load case (Figure 10a) showed the largest maximum principal stress at the ringed segments. Maximum tensile stress appeared at the bottom surface of the top ring at around 61 MPa. Its origin can be assumed by the expansion of the rings by given external loads. This leads to an increase of a tangential

component of normal stress in the ring. For the first load case, a reaction force of 780 N was considered. In experiments, a structural failure occurred at an average load of 3200 N for the support structure. This corresponds to 800 N for one-quarter of the structure. Therefore, reaction force was increased for the second load case, up to 806 N (results displayed in Figure 10b). The largest maximum principal stress reached 80 MPa. Maximum tensile stress appeared at the same location, similar to the first load case. However, in the middle region of the columns (front), a tensile stress suddenly evolved, which was likely due to the buckling sensitivity of the support structure. At a certain uniaxial load, the structure will collapse because of buckling if the tensile strength of the bulk material is larger compared to this. In the third load case, the reaction force was increased to 809 N. Results are shown in Figure 10c. In this load case, the maximum tensile stress appeared in the middle of the columns at 128 MPa. The maximum principal stress was also high in the ring segments but not at this level. This led to the conclusion that, in third load case, the structure started to collapse by buckling. In Figure 11, a comparison of the three load cases for the maximum principal stress is shown. This figure clearly demonstrates the buckling sensitivity of the support structure. Between the first and the second load cases, the axial force was increased by 26 N.

**Figure 10.** Results of analysis for the maximum principal stress: (**a**) @780 N load; (**b**) @806 N load; (**c**) @809 N load.

**Figure 11.** Comparison of uniaxial load sets (deformation shown in true scale).

This contributed to an expected increase of the maximum principal stress in the ringed segments but also led to an increase of tensile stress in the column of the structure. From the second to the third load case, a further small increase of axial force of 3 N significantly changed the load conditions. In the center region of the columns, maximum principal stress grew proportionally. The location of the maximum tensile stress changed from the ringed segments to the center region of the columns.

In Figure 12, experimental observed defects of the broken columns are shown. The red marked defects (primary defects) show those defects that initially occurred during experimental compression testing when the structure collapsed; the blue marked defects show the secondary defects that followed.

**Figure 12.** Experimental observed defects at the column structure.

The experimental failure pattern shown above leads to the assumption that the support structure collapsed by exceeding the tensile strength of the VPP-manufactured material, which corresponds to the FE analysis (first load case), and not by buckling, because cracks appeared at locations of maximum tensile stress in the ringed segments. However, some aspects must be considered. First, the simulation showed that the critical buckling loads appeared in the middle of the columns (third load case), which is very close to the experimental testing conditions (i.e., secondary defects shown in Figure 12). Therefore, complete structural failure by buckling should not be excluded. Second, the distribution of micropores in the VPP-manufactured column structure was considered homogenous for the simulated model. However, manufactured structures probably also include microdefects leading to the observed failure pattern.

Third, technological tolerances (by 3D printing, sintering, etc.) can lead to geometric imperfections (e.g., flatness of the ringed cross-section), which induce critical tensile stresses. Such effects were not included in the described FE model. At this stage of the presented research, there is still opportunity for adjustment if indeed microdefects appear, e.g., adjusting CerAM VPP exposure parameter, suspension parameters, debinding/sintering regime or the VPP design. By neglecting these aspects, a tensile strength of our manufactured material with 5% porosity could be obtained by further optimizing the VPP process. The tensile strength of a *β*-TCP 5% porous material should lie between 61 MPa and 80 MPa and take the mentioned restrictions into account.
