Ti64/20Ag Porous Composites Fabricated by Powder Metallurgy for Biomedical Applications

We present a novel Ti64/20Ag highly porous composite fabricated by powder metallurgy for biomedical applications and provide an insight into its microstructure and mechanical proprieties. In this work, the Ti64/20Ag highly porous composites were successfully fabricated by the space holder technique and consolidated by liquid phase sintering, at lower temperatures than the ones used for Ti64 materials. The sintering densification was evaluated by dilatometry tests and the microstructural characterization and porosity features were determined by scanning electron microscopy and computed microtomography. Permeability was estimated by numerical simulations on the 3D real microstructure. Mechanical properties were evaluated by simple compression tests. Densification was achieved by interparticle pore filling with liquid Ag that does not drain to the large pores, with additional densification due to the macroscopical deformation of large pores. Pore characteristics are closely linked to the pore formers and the permeability was highly increased by increasing the pore volume fraction, mainly because the connectivity was improved. As expected, with the increase in porosity, the mechanical properties decreased. These results permitted us to gain a greater understanding of the microstructure and to confirm that we developed a promising Ti64/20Ag composite, showing E of 7.4 GPa, σy of 123 MPa and permeability of 3.93 × 10−11 m2. Enhanced adaptability and antibacterial proprieties due to Ag were obtained for bone implant applications.


Introduction
Ti6Al4V (Ti64) alloy is one of the most used materials to fabricate and design devices and bone implants for medical applications. This is because of its excellent mechanical properties, low weight, biocompatibility and good corrosion resistance [1][2][3]. Nonetheless, Ti64 alloy has two main disadvantages for bone implant applications: high stiffness (110 GPa) compared to the human bones (7-30 GPa) and a low antibacterial activity, which means that post-operative infections can appear by the development of bacterial colonies, generating serious complications that can lead to failure of the implants. In order to reduce the stiffness of the Ti64 alloy, porous materials called "scaffolds" have been widely studied by different techniques such as additive manufacturing (AM) [4,5], partial powder sintering [6,7], and the space holder technique [8][9][10]. Although AM is the most outstanding, it is

Sample Preparation
To prepare the Ti6420Ag porous composites, Ti64, Ag and ammonium bicarbonate powders were used. The first ones were spherical with a particle size distribution lower than 45 µm. These powders were furnished by Raymor Company, located in Quebec, Canada. The Ag powders had an irregular shape with a particle size distribution between 1 and 5 µm, and they were furnished by the Sigma Aldrich company. These two powders were used to obtain a composite of Ti64 with 20 vol.% of Ag by mixing them in a TURBULA ® shaker mixer during 30 min under dry conditions. The ammonium bicarbonate powders had an irregular shape, and they were sieved to have a particle size distribution between 100 and 200 µm. These powders were furnished by Alfa Aesar and used as pore formers, thus, they were mixed with the Ti6420Ag mixture of powders for 30 min. The quantity of pore formers was varied from 30 to 50 vol.% with respect to the solid of the Ti6420Ag composite. After that, 1 wt.% of PVA was added to the mixture of the three powders as a binder to bring a higher mechanical strength to the green samples. Then, the mixture of powders was poured into an 8 mm cylindrical stainless steel die to uniaxially press the mixture of powders at 450 MPa. This was carried out on an 1150 Instron Universal machine. In order to obtain the porous composites, the pore formers were thermally evaporated by heating the samples at 5 • C/min up to 180 • C at a holding time of 6 h under high purity argon. Next, the samples were introduced into a vertical dilatometer Linseis L75, and the PVA was eliminated by heating the samples at 10 • C/min up to 500 • C with a plateau of 45 min. Finally, the samples were sintered at 1100 • C with a heating rate of 10 • C/min and a soaking time of 5 min in the vertical dilatometer. The sintering thermal cycle was determined in a previous work [34].
The density of the sintered samples was estimated by weighting and measuring their volume and the theoretical density of the Ti6420Ag composite was calculated by the rule of mixtures by using the theoretical density of Ti64 (4.45 g/cm 3 ) and Ag (10.49 g/cm 3 ), which led to a value of 5.658 g/cm 3 for the composite with 20 vol.% of Ag.

Microstructural Characterization
The microstructure of the sintered porous composites was observed by scanning electron microscopy (SEM). For that, the samples were half-cut and the transversal section was polished with SiC papers, whereas 1 µm alumina powder was used to obtain a mirrorlike surface. Surface analysis was performed using a Tescan, Mira 3 LMU Field Emission Scanning Electron Microscope (FE-SEM). As the SEM images bring only 2D information, 3D images with a voxel size resolution of 1 µm were acquired with the aim of determining the Ag distribution in 3D. To achieve such voxel resolution, samples were cut in prisms of 0.7 mm × 0.7 mm × 4 mm. The 3D image acquisition was performed using a Zeiss Xradia 510 Versa 3D X-ray microscope with a beam intensity of 140 kV, which was high enough to pass through the Ti64/20Ag porous samples of around 0.7 mm thick. An X-ray detector of 20× was used to collect 1600 projections with a CCD camera of 1024 × 1024 pixels by rotating the sample a total of 360 • . In addition, 3D images of the whole sample with a 6.5 µm resolution were also acquired to determine the porosity features, such as pore size distribution, connectivity, pore volume, etc. For such images, the beam intensity used was 160 kV with an X-ray detector of 0.4x, used to collect 1600 projections all around the sample.

Permeability Evaluation
The permeability of porous samples was estimated by numerical simulations on the real microstructure issued from the 3D images, acquired with a voxel size of 6.5 µm, and in this way, we obtained the whole image of the porosity. For that, the module of "Absolute Permeability Experiment Simulation" in the Avizo ® software was used. This module works with binary images containing the pores with an intensity of 1 and the solid with the intensity of 0. Simulations were carried out by solving the Navier-Stokes equations with an iterative method that is stopped with an error criterion of 10 −5 . The simulations take into account the Darcy law, a Newtonian fluid with a steady state laminar flow where it is possible to introduce the fluid viscosity and the boundary conditions of inlet and outlet pressure. In this work, the inlet and outlet pressures were 130 and 100 kPa, respectively, and the fluid viscosity was 0.045 Pa, which mimics that of human blood. As the simulation module demands a lot of computational requirements, it was necessary to determine a minimum representative volume (MRV) to reduce the computational time of the permeability simulations. For that, the methodology proposed by Okuma et al. [36] was followed, finding a MRV of 250 3 voxels 3 . Therefore, the permeability simulations were run in a MRV of 300 3 voxels 3 that represents a 7.4 mm 3 of the sample. The MRV was virtually cropped from the center of the sample to avoid the border effects.

Compression Tests
The mechanical strength of the Ti6420Ag porous composites was evaluated by simple compression tests. For that, samples with a 6 mm diameter and 8 mm height were sintered.
The tests were carried out with a 1150 Instron Universal machine, following the ASTM D695-02 requirements. Tests were performed with a strain rate of 0.5 mm·min −1 , as reported elsewhere [37]. The data collected by the software of the Instron during the tests were the displacement and load. These data obtained from the compression tests were used to calculate the strain and the compression stress, using the initial dimensions of the samples. The Young's modulus (E) was obtained by fitting a straight line in the elastic zone of the curves. The Yield stress was also determined from the stress-strain curves by taking into account the values used for calculating the E values.

Dilatometry Analysis
The sintering behavior of the Ti64/20Ag with and without pore formers is shown in Figure 1. The axial strain of the samples is very similar to a thermal expansion up to 750 • C, where a small swelling is observed (Figure 1a). The maximum dilation was obtained at 860 • C, and then a sharp shrinkage was observed in all the samples until the cooling stage was reached. Finally, a linear behavior was observed during the cooling down period to room temperature. From the strain rate as a function of temperature, it is possible to notice that the axial deformation is mainly due to the liquid Ag formation. There are two sharp increments in the strain rate, at 940 • C and 962 • C, that accomplish most of the densification of the samples. At 940 • C, the softening of Ag particles can generate a densification by rearrangement since the melting point is close. At 962 • C, the liquid Ag can fully spread among the interparticle pores left by the Ti64 particles, as it was reported in [34]. The strain rate could also confirm that the addition of large pores does not change the behavior of the densification by liquid phase. At the end, the shrinkage (negative axial strain) increased along with the pore former volume. This could suggest that the liquid Ag drains into the large pores; however, this is not true, as is shown in Figure 2. Therefore, the additional shrinkage obtained from the samples is due to the pore deformation during sintering. This is possible because the densification generates compression stresses that can reduce the pore size initially created by the pore formers. sintering. This is possible because the densification generates compression stresses that can reduce the pore size initially created by the pore formers.  The densities of all samples are listed in Table 1, and it can be noticed that a reduction in the values is obtained. The sample with 50 vol.% of pore formers shows a sintered density of 3.2 g/cm 3 , which is a lower value of the Ti64 alloy and close to the ideal value reported for metallic implants elsewhere of 1.8 g/cm 3 [38]. The densification of the porous samples is larger than the one without pore formers, which increases as the pore former's sintering. This is possible because the densification generates compression stresses that can reduce the pore size initially created by the pore formers.  The densities of all samples are listed in Table 1, and it can be noticed that a reduction in the values is obtained. The sample with 50 vol.% of pore formers shows a sintered density of 3.2 g/cm 3 , which is a lower value of the Ti64 alloy and close to the ideal value reported for metallic implants elsewhere of 1.8 g/cm 3 [38]. The densification of the porous samples is larger than the one without pore formers, which increases as the pore former's The densities of all samples are listed in Table 1, and it can be noticed that a reduction in the values is obtained. The sample with 50 vol.% of pore formers shows a sintered density of 3.2 g/cm 3 , which is a lower value of the Ti64 alloy and close to the ideal value reported for metallic implants elsewhere of 1.8 g/cm 3 [38]. The densification of the porous samples is larger than the one without pore formers, which increases as the pore former's volume does, too. However, the difference in densification is lower in comparison to the Materials 2022, 15, 5956 6 of 14 one measured by the axial strain. This suggests that deformation of large pores during sintering is anisotropic, which can be associated with the pore shape and orientation, as was pointed out by Cabezas et al. [8].

Microstructure Analysis
The microstructure observed by SEM is shown in Figure 2. Three different characteristics can be observed from Figure 2a: the large pores, the Ti64 matrix in dark grey and the Ag in light grey. The large pores are elongated and randomly distributed in the sample. The distribution of Ag around the Ti64 particles is better observed in Figure 2b. Figure 2c shows the microstructure inside of a large pore, in which it is possible to confirm that the Ag filled the interparticle pores but not the large pores. Figure 2d shows the surface of the particles exposed at the surface of large pores, in which the Ag is confined by capillarity between the smaller pores. It can also be noticed that the surface of the particles is not covered by Ag, which confirms that the liquid only densified the porosity left by the particle packing. In addition, it was confirmed that some Ag diffuses in a solid state into the Ti64 net to form thin lamellae of the Ti 2 Ag intermetallic (Figure 2e,f), as was reported for alloys processed by casting [20,22] or composites fabricated by powder metallurgy [34].
In order to observe the distribution of the Ag in the porous composite and quantify the quantity of the volume occupied by the liquid, 3D images with a 0.7 µm pixel resolution were acquired (Figure 3). Figure 3a shows a 2D virtual slice of the sample with 30 vol.% of pore formers, and three different gray levels can be distinguished in the image. Very dark grey is associated with the pores, the dark grey corresponds to the Ti64 and the light grey corresponds to the Ag. In order to segment the different phases, a thresholding procedure is performed by selecting the gray level value that corresponds to each phase in the image, as indicated in Figure 3e. Thanks to the good contrast generated by the difference in densities of the Ti64 and Ag, it is possible to obtain binary images that represent each phase, Ti64- Figure  A 3D rendering of each phase is shown in Figure 4, which allows us to qualitatively determine the interaction between Ti64, Ag and pores in 3D. However, the quantitative data are obtained from binary images, for example, the volume fraction of each phase is determined by counting the voxels corresponding to the Ti64, Ag and pores, respectively, and dividing each one by the total volume. However, these kinds of images were acquired mainly to quantitatively determine the distribution of Ag in the composite and not the pores because only a small volume is observed. From these images, the average volume fraction of Ag measured from 3D images of samples with different quantities of pore formers was 22%, which is close to the 20 vol.% added. The small difference can be attributed to the image treatment that could slightly overestimate the Ag. The interconnectivity of the Ti64 and Ag was determined by following the procedure described elsewhere [8]. It was determined that Ti64 is fully interconnected in 3D, and the Ag shows a 96% interconnectivity with the small, isolated volumes. From the 3D image analysis, it was also demonstrated that liquid Ag remains into the interparticle pores formed by the Ti64 packing and that liquid does not drain into the large pores.
vol.% of pore formers, and three different gray levels can be distinguished in the image. Very dark grey is associated with the pores, the dark grey corresponds to the Ti64 and the light grey corresponds to the Ag. In order to segment the different phases, a thresholding procedure is performed by selecting the gray level value that corresponds to each phase in the image, as indicated in Figure 3e. Thanks to the good contrast generated by the difference in densities of the Ti64 and Ag, it is possible to obtain binary images that represent each phase, Ti64- Figure  A 3D rendering of each phase is shown in Figure 4, which allows us to qua determine the interaction between Ti64, Ag and pores in 3D. However, the qua data are obtained from binary images, for example, the volume fraction of each determined by counting the voxels corresponding to the Ti64, Ag and pores, resp and dividing each one by the total volume. However, these kinds of images were mainly to quantitatively determine the distribution of Ag in the composite and pores because only a small volume is observed. From these images, the average fraction of Ag measured from 3D images of samples with different quantities of mers was 22%, which is close to the 20 vol.% added. The small difference can be a to the image treatment that could slightly overestimate the Ag. The interconnec the Ti64 and Ag was determined by following the procedure described elsewhe was determined that Ti64 is fully interconnected in 3D, and the Ag shows a 96% i nectivity with the small, isolated volumes. From the 3D image analysis, it was also strated that liquid Ag remains into the interparticle pores formed by the Ti64 pac that liquid does not drain into the large pores.  A 3D rendering of each phase is shown in Figure 4, which allows us to qualitatively determine the interaction between Ti64, Ag and pores in 3D. However, the quantitative data are obtained from binary images, for example, the volume fraction of each phase is determined by counting the voxels corresponding to the Ti64, Ag and pores, respectively, and dividing each one by the total volume. However, these kinds of images were acquired mainly to quantitatively determine the distribution of Ag in the composite and not the pores because only a small volume is observed. From these images, the average volume fraction of Ag measured from 3D images of samples with different quantities of pore formers was 22%, which is close to the 20 vol.% added. The small difference can be attributed to the image treatment that could slightly overestimate the Ag. The interconnectivity of the Ti64 and Ag was determined by following the procedure described elsewhere [8]. It was determined that Ti64 is fully interconnected in 3D, and the Ag shows a 96% interconnectivity with the small, isolated volumes. From the 3D image analysis, it was also demonstrated that liquid Ag remains into the interparticle pores formed by the Ti64 packing and that liquid does not drain into the large pores.

Porosity Analysis and Permeability
The 2D virtual slices of Ti64/20Ag samples with different volume fractions of pore formers acquired with a pixel resolution of 6.5 µm allow us to observe the homogeneous distribution of large pores in the whole sample ( Figure 5). It can also qualitatively be observed that no pore agglomeration or segregation is found, even in the sample containing 50 vol. % of pore formers. In order to obtain the quantitative data, the grey level images were transformed into binary ones following a similar process as described above, with the difference being that the Ag cannot be detected with this pixel size resolution. In this case, only the pores and the solid that includes Ti64 and Ag were segmented, but the main

Porosity Analysis and Permeability
The 2D virtual slices of Ti64/20Ag samples with different volume fractions of pore formers acquired with a pixel resolution of 6.5 µm allow us to observe the homogeneous distribution of large pores in the whole sample ( Figure 5). It can also qualitatively be observed that no pore agglomeration or segregation is found, even in the sample containing 50 vol. % of pore formers. In order to obtain the quantitative data, the grey level images were transformed into binary ones following a similar process as described above, with the difference being that the Ag cannot be detected with this pixel size resolution. In this case, only the pores and the solid that includes Ti64 and Ag were segmented, but the main purpose of such images was to evaluate the porosity. The volume fraction of pores was determined as mentioned above, and the values are listed in Table 2. For all the samples, a higher value of porosity than the volume of pore formers added is found. The difference between the added value and the found one might be due to the packing arrangement during powder compaction. This is because the coordination number of the salt particles increases, which leads to larger interparticle pores among the salt particles that could not be filled by the Ti64 and Ag particles.
purpose of such images was to evaluate the porosity. The volume fraction of pores was determined as mentioned above, and the values are listed in Table 2. For all the samples, a higher value of porosity than the volume of pore formers added is found. The difference between the added value and the found one might be due to the packing arrangement during powder compaction. This is because the coordination number of the salt particles increases, which leads to larger interparticle pores among the salt particles that could not be filled by the Ti64 and Ag particles.  The pore connectivity is determined by analyzing the continuity of pores inside the measured volume; this procedure is detailed in [39]. The values obtained were higher than 95% for all the samples, which can also be qualitatively observed in Figure 6b,e,h. A 3D rendering of the porosity in the analyzed volumes is presented, in which pseudo-colors indicate the interconnected pores. Thus, each color in the image indicates a group of pores interconnected among them. It can be noticed that one color is predominant in the image, but also that more isolated pores are found in the sample with 30 vol.% of pore formers (Figure 6b).  The pore connectivity is determined by analyzing the continuity of pores inside the measured volume; this procedure is detailed in [39]. The values obtained were higher than 95% for all the samples, which can also be qualitatively observed in Figure 6b,e,h. A 3D rendering of the porosity in the analyzed volumes is presented, in which pseudo-colors indicate the interconnected pores. Thus, each color in the image indicates a group of pores interconnected among them. It can be noticed that one color is predominant in the image, but also that more isolated pores are found in the sample with 30 vol.% of pore formers (Figure 6b).
The pore size distribution of the samples with different quantities of pore formers is very similar, which confirms a good distribution of the pore formers as well as the repeatability of the process (see Figure 7). The pore sizes go from 30 to 200 µm for every sample. Moreover, the median pore size (d 50 ) presents similar values ranging from 75 to 89 µm (see Table 2), that suggests a good distribution of the pore formers during the whole fabrication process. Furthermore, the median pore size also indicates a reduction of 47% with respect to the median size of the initial pore sizes (140 µm). This can support the discussion about the pore deformation undergone by the samples during sintering, as mentioned above. The 3D rendering of the samples with different quantities of pore volumes illustrates the increment in the void spaces but with similar size (Figure 6a,d,g). The pore size distribution finds itself in the range reported as optimal to favor the cell ingrowth by Itälä et al. (100-200 µm) [40]. However, the median pore size we obtained is low in comparison to the one used by Pérez et al. to regenerate the bone defects, with macropores of 210 µm on average [12]. Materials 2022, 15, x FOR PEER REVIEW 9 of 14 The pore size distribution of the samples with different quantities of pore formers is very similar, which confirms a good distribution of the pore formers as well as the repeatability of the process (see Figure 7). The pore sizes go from 30 to 200 µm for every sample. Moreover, the median pore size (d50) presents similar values ranging from 75 to 89 µm (see Table 2), that suggests a good distribution of the pore formers during the whole fabrication process. Furthermore, the median pore size also indicates a reduction of 47% with respect to the median size of the initial pore sizes (140 µm). This can support the discus-   Table 2 for the Ti64/20Ag samples with different quantities of pore formers. An increment of eight times was found when the pore volume fraction of samples increased from 33 to 57% and for a small difference in the pore size. This can be associated to the reduction in the tortuosity from a value of 1.83 to 1.32 (see Table 2). However, this increment in the permeability was also obtained because an increment in the porosity generates more paths for the fluid to pass throughout the porous sample, as can be qual-itatively observed from Figure 6c,f,i, in which there are more paths for the sample with 50 vol.% of pore formers, which are also low-tortuous, allowing higher velocities of the fluid, although the pore connectivity is the same with respect to the sample with 40 vol.% of pore formers. The permeability values of the samples are between 0.47 and 3.93 × 10 −11 m 2 , which are in the range of the human trabecular bones in the proximal femur (3 × 10 −11 to 5 × 10 −10 m 2 ), but low compared to the values for the permeability of the human vertebral trabecular bone, 10 −8 to 10 −9 m 2 [13]. Permeability can be increased by the addition of higher quantities of pore formers, but this can compromise the structure of the Ti64/20Ag, as the liquid Ag may collapse the thinner struts obtained. The second method to improve permeability and obtain values closer to the human trabecular bones is to increase the pore former's size. Additional tests should be carried out to demonstrate whether one of these two methods could be viable to increase the permeability. Nevertheless, the obtained values of permeability are in the range of different scaffolds fabricated by space holder [41] or AM [42] with similar volume fraction of pores.    Table 2 for the Ti64/20Ag samples with differen formers. An increment of eight times was found when the pore volume increased from 33 to 57% and for a small difference in the pore size. Th to the reduction in the tortuosity from a value of 1.83 to 1.32 (see Tab increment in the permeability was also obtained because an increment erates more paths for the fluid to pass throughout the porous sampl tively observed from Figure 6c,f,I, in which there are more paths for vol.% of pore formers, which are also low-tortuous, allowing higher v although the pore connectivity is the same with respect to the samp pore formers. The permeability values of the samples are between 0.4

Compression Analysis
The behavior under simple compression of the Ti64/20Ag samples with and without pore formers was plotted in Figure 8. It was found that the sample without pore formers shows a higher strength in comparison to the porous samples. Further, all the samples show an elastic zone, then a plastic deformation with a small increment in the stress and, then, a plastic deformation at more or less constant stress until the failure. As was expected, the addition of large pores generates a high reduction in the maximal compression strength of samples from 916 MPa to a maximum of 117 MPa. The Young's modulus (E) was obtained from the elastic zone of the curves, and it is listed in Table 3, as well as the yield stress (σ y ). It was found that E is reduced seven times by adding 50 vol.% of pore formers, meanwhile, the σ y is also reduced, but only five times. The E values of the Ti64/20Ag porous samples are in the range of the human bones, under 20 GPa [4]. The σ y values are also in the range reported for human bones (50-200 MPa), although the sample with 30 vol.% of pore formers is a little higher, 220.9 MPa. Thus, the compression values are close to the ones reported for Ti64 porous materials fabricated by space holder [43] or AM [44] with similar volume fraction of pores, however, there are no reports of Ti64Ag porous materials to directly compare with our results. It was also pointed out by Liu et al. [45] that the σ y /E ratio, called "admissible strain", is a good parameter to improve the performance of the metallic implants, and it is recommended that this value should be the highest possible. The σ y /E was calculated and listed in Table 3 for all the samples. The values of σ y /E range from 12 to 16 × 10 −3 and the highest value corresponds to the Ti64/20Ag sample with 50 vol.% of pore formers. In conclusion, these values are higher than the ones reported for the solid Ti64 [46] but lower than the ones obtained for porous composites of Ti64/xTa [47]. More, the σ y /E values are in the range of the compact bones (11 × 10 −3 ) but lower than the ones reported for the human vertebral trabecular bone (35 × 10 −3 ) [3].

Conclusions
An innovative strategy to fabricate Ti64/20Ag highly porous composites by the space holder technique and consolidated by liquid phase sintering, at lower temperatures than the ones used for Ti64 materials, was successfully studied. It was demonstrated that the liquid Ag spreads over the interparticle pores, densifying the composite without filling the large pores. Full pore connectivity was obtained, with a pore size distribution controlled by the initial size of the pore formers and with the help of the sintering densification. The porous samples presented the characteristics of the pores, the permeability and the mechanical properties that make them suitable for bone implants. However, the best results for such applications were obtained for the Ti64/20Ag sample with 50 vol.% of pore formers, as its structural properties come closest to those of human bones. Additional corrosion, antibacterial and biological tests are in progress to establish the benefits of the Ag addition in Ti64 porous materials.

Data Availability Statement:
The raw/processed data required to reproduce these findings cannot be shared at this time as the data also form part of an ongoing study.