Bioactive Bilayer Glass Coating on Porous Titanium Substrates with Enhanced Biofunctional and Tribomechanical Behavior

: The use of porous titanium samples fabricated by space-holder powder metallurgy with bioactive coatings has already been reported to prevent resorption of the bone surrounding the implant and improve osseointegration, respectively. However, the presence of pores as well as the poor adherence and the brittle behavior inherent to glassy coatings affect the service behavior of implants fabricated from these samples. Therefore, they need to be optimized. In this work, 50 vol.% of porosity titanium substrates were manufactured with different pore range size (100–200 and 355–500 µ m) spacer particles and coated with a bilayer of bioactive glasses (45S5/1393). The effect of the pores on the tribomechanical properties and inﬁltration of the bioactive glass 1393 along with the bioactivity of the bioactive glass 45S5 were evaluated by instrumented micro-indentation and scratch tests and the formation of hydroxyapatite in simulated body ﬂuid. The results obtained were very promising as potential implants for the replacement of small tumors in cortical bone tissues, mainly due to the smaller pores that present an improved biomechanical and biofunctional balance.


Introduction
The area of implant manufacture is attracting growing attention due to the result of a longer life expectancy (i.e., Spain and Italy had the highest life expectancy among EU countries, reaching over 83 years in 2018). However, living longer has associated the need for tissue or prosthesis replacements if they fail, for reasons such as trauma, unhealthy lifestyle, or weakening due to calcium deficiency. Furthermore, it is well known that implants for young people are subject to a higher mechanical load for longer periods of time, so the development of new strategies and methodologies to improve knowledge about these devices plays a crucial role in the field. In this context, more specialized and multidisciplinary research is needed to implement prototypes and in vivo performance of small implantable devices to replace damaged bone tissues in small bone-localized tumors [1,2].
Pure titanium is one of the most appropriate materials for cortical bone replacement due to its excellent properties regarding mechanical, chemical, and physical properties, such as high specific strength, corrosion resistance, low density, and biocompatibility/biofunctional behavior. Unfortunately, the widely described stress-shielding phenomenon associated with bone resorption, caused by the different Young's modulus between bone and implant, compromises the reliability of implants [3,4]. The use of porous substrates could implement stress-shielding behavior. At this point, the space-holder manufacturing technique described it will increase the area of contact with the physiological fluid and/or the bioactivity of the implant. Therefore, this is the enhanced combination of BGs, as it has been corroborated by previous studies of the authors [21,41].
In this scenario, it seems to be a critical point to achieve a better approximation to overcome prosthesis problems such as stiffness or stress-shielding, via a balanced introduction of pore content with adequate size, to avoid compromising mechanical behavior. In addition, it is also crucial to improve poor titanium osseointegration, adherence, and homogeneous infiltration, employing bioactive coatings with good adhesion and without fragility problems. In this sense, we have reported previous studies focused on a singular approach that includes the use of controlled porous implants covered with a bilayer coating; the inner layer that is in contact with the titanium substrate that has a greater infiltration capacity, a closer dilatation coefficient compared to Ti, and significant adherence to the pores (BG 1393); and the outer layer (BG 45S5), which exhibits better bioactivity with physiological fluids. However, despite the novelty of these findings and the viability of the proposal, it is still necessary to optimize the porosity of the titanium substrate, improve the tribomechanical balance (rigidity, microhardness, mechanical, and scratch resistance), and at the same time, the biofunctionality (infiltration, structural integrity, and adhesion of the coating).
Thus, the main objective of this work is to evaluate the tribomechanical behavior and in vitro osteointegration capacity of porous titanium substrates coated with this specific bilayer coating. In this case, porous titanium discs were fabricated using a space-holder technique with 50 vol.% porosity and two pores size ranges: 100-200 and 355-500 µm spacer particles [42][43][44][45], as they have already been tested for adequate characteristics to promote infiltration, adhesion, structural integrity, and mechanical balance (stiffness and yield stress). To promote adherence, BG 1393 was used with a CTE such as Ti; however, BG 45S5 was placed in the outer layer to promote osseointegration. In vitro experiments were performed in SBF to evaluate bioactivity.

Commercially Pure Titanium Substrates
Porous substrates were manufactured using the space-holder technique (SH) following the protocol already described by the authors [7,[46][47][48]. For this purpose, commercially pure titanium (c.p. Ti) grade IV supplied by SEJOING Materials Co. Ltd. (Seoul, Korea) was used (mean particle size d [50] = 23.3 µm) [49]. Ammonium bicarbonate (NH 4 HCO 4 ) (BA), 99% purity, supplied by Cymit Química S.L. (Barcelona, Spain) of two different range sizes (100-200 and 355-500 µm) was used as spacer particles. They were mixed with the c.p. Ti (50 vol.%), pressed at 800 MPa using an Instron 5505 machine and then, the spacer particles were removed in a furnace under low vacuum conditions (~10 −2 mbar) in two steps, first at 333 K and then at 380 K, for 12 h each cycle. Subsequently, the samples were sintered in a molybdenum chamber furnace (Termolab, Agueda, Portugal) under high vacuum conditions (~10 −5 mbar), 1523 K for 2 h. Cylindrical discs of~12 mm diameter and 5 mm height were obtained. To perform the different studies, the samples were prepared following a standard metallography procedure on the one side (grinding and mechanicalchemical polishing). As a reference, fully-dense samples (without spacer particles) were fabricated to be compared to those of the porous substrates.

Characterization of the c.p. Ti Substrates
The porosity of the substrates was evaluated using the Archimedes' method [50], from which the percentage of total and interconnected porosity (P T and P i , respectively) was determined. Furthermore, image analysis (IA) was carried out on optical images acquired with a Nikon Eclipse MA100 N microscope to characterize the equivalent diameter of the pores, Deq, and the shape factor, F f = 4πA/(PE) 2 , where A is the area of the pores and PE is the experimental perimeter of pseudo-elliptic pores. For image analysis, Image-ProPlus 6.2 software (Mediacibernectic, Bethesda, MD, USA) was used.
In addition, scanning electron microscopy images were also acquired with a FEI Teneo microscope (FEI, Eindhoven, The Netherlands) operating at 15 keV for structural characterization. This microscope is equipped with an energy-dispersive X-ray spectroscopy system (EDS-SEM) for compositional analyses. Images were also acquired with a confocal laser microscope (CLM) to measure the roughness of the surface by evaluating parameters such as the arithmetical mean deviation (S a ) and the root-mean-square height (S q ).
Finally, mechanical properties of the porous c.p. Ti substrates, such as the yield strength and Young's modulus, were estimated from the porosity data, using equations already reported in the literature. In particular, dynamic Young's modulus, E d , [7] (Equation (1)); Nielsen equation for Young Modulus, E N ; [51] (Equation (2)) and yield strength, σ y , [52] (Equation (3)). For these equations, Young's modulus for bulk c.p. Ti grade IV was~110 GPa [53] and the yield strength of the bulk c.p. Ti grade IV~650 GPa [54].
The layers of the bioglasses were deposited onto the porous c.p. Ti substrates by dripping sedimentation with a suspension of 0.1 g of bioglass per 10 mL of ethanol that was stirred actively and subjected to an ultrasonic bath for 15 min to avoid sedimentation. Two layers of each bioglass were deposited on each substrate, starting with BG 1393, directly contacting the c.p. Ti substrates and then, BG 45S5 was deposited on the BG 1393. To allow the evaporation of the ethanol, the samples were dried for 24 h at room temperature after the deposition of each bioglass layer before coating with the next layer. Finally, a low vacuum (~10 −2 mbar) heat treatment was performed on the coated substrates in a Vita Vacumat 600 M furnace, with a heating rate of 20 • C/min until 600 • C and then 40 • C/min until 820 • C, a temperature that was held for 5 s, cooling the sample outside the furnace. This heat treatment was selected after trying different conditions, taking into account the chemical composition and critical temperatures of both BGs as well as the fluidity of BG 1393 to favor its infiltration into the macropores. Coating was performed for the two pores range size spacer particles and on the whole cylinder as well as D-shaped samples to determine the infiltration into the pores and thickness of the BG bilayer coating. In the last case, for the coating of the D-shaped, the c.p. Ti substrates were held together with a retractable polymer to achieve a homogeneous coating [21].
To evaluate the coating, SEM and CLM studies were also performed in the top-view to check the homogeneity and general aspect, and in the cross-sectional view to measure the thickness and infiltration of the coating into the pores. Furthermore, the tribomechanical properties of the coated substrates were measured on the surface of the samples using a Microtest machine (MTR3/50-50/NI, Microtest S.A., Madrid, Spain), which allowed loading and unloading tests (P-h curves) and scratch tests. On the one hand, P-h curves were performed by load control applying 1 N as maximum load and 10 s of holding time (1 N·min −1 as applied load rate) using a Vickers indenter. The micro-hardness and Young's modulus of the coated substrates were estimated from the P-h curves using the Oliver and Pharr method [57][58][59][60] (Equations (4)-(6)). These results were corrected for by considering changes due to the indenter geometry due to wear or damage due to its use. On the other hand, scratch tests were performed using a Rockwell diamond tip of 200 µm diameter and increasing the load from 0 to 3 N (0.5 N·min −1 as applied load rate) for 3 mm of groove scar following the standard [61]. Normal load was recorded continuously during scratching. For each test, a previous scan was performed to evaluate the roughness profile of the surface, from which different roughness parameters were measured: arithmetic average of the absolute values of all points of the profile (R a ), root mean square of the values of all points of the profile (R q ), maximum peak-to-valley height of the entire measurement trace (R y ), and arithmetic average of the maximum peak-to-valley height of the roughness values of five consecutive sampling sections over the filtered profile (R z ). Then, the scratch test was performed; penetration depth, applied load, and path were continuously recorded. From these data, permanent deformation and the estimated elastic recovery of the coating were determined from the real penetration of the depth of the groove scar. Additionally, the observation of the scar provided information on the damage generated on the surface by the applied load (cracking and chipping). All measures, P-h curves, and scratch tests, were performed at least three times.
Finally, in vitro osseointegration was evaluated by measuring bioactivity following the protocol established by Kokubo and Takadama [62] consisting of immersion of coated samples in simulated body fluid (SBF) for 21 days in a polystyrene container at 310 K, renewing the solution at 7, 14, and 21 days. SBF was prepared in the laboratory as in the reference [21]. After this time, the samples were rinsed with deionized water, dried, and characterized to evaluate the formation of hydroxyapatite at the surface as a measure of the bioactivity of the coated samples. For this purpose, inductively coupled plasma (ICP) atomic emission spectrometry (Horiba Jobin Yvon, Horiba LTD, Kyoto, Japan) was employed. Furthermore, SEM images were acquired to observe HA formation, which was also checked by thin film X-ray diffraction (XRD, Bruker, Rivas-VaciaMadrid, Spain) performed on a Bruker D8 Advance A25 diffractometer with a Cu Kα radiation of 0.154 nm.

Results and Discussion
This section is divided into two subsections according to the different characterizations performed in the development of this work. Therefore, first, the microstructural and porosity characterization of the fabricated porous c.p. Ti is presented, and then the bilayer BGs coating on the substrates is evaluated in terms of microstructure, tribomechanical behavior, homogeneity, adherence, microhardness, and estimation of bioactivity by immersion in SBF. Figure 1 shows images acquired with different techniques in which the distribution, size, and morphology of the pores can be appreciated. In all these images, in addition to the porosity due to the spacer particles, micro-pores (<50 µm) could also be observed that were associated with the fabrication procedure. The Sa and S q parameters revealed higher roughness for larger pore sizes, as expected. Figure 1 shows images acquired with different techniques in which the distribution, size, and morphology of the pores can be appreciated. In all these images, in addition to the porosity due to the spacer particles, micro-pores (<50 µ m) could also be observed that were associated with the fabrication procedure. The Sa and Sq parameters revealed higher roughness for larger pore sizes, as expected. The porosity and morphology, sizes, and distribution analyses of the pores were completed using the Archimedes' method, which allowed the measurement of not only density, but also total and interconnected porosity. Table 1 summarizes these parameters and the estimated mechanical properties calculated from them, using the Nielsen equations mentioned above. In view of these results, the efficacy of the spacer-holder technique could be confirmed since the porosity and size of the pores agreed with the volume and pore range sizes of the spacer particles used. On the other hand, it is known that mechanical behavior depends on the content and size of the pores. In fact, it could be observed that the pores' range sizes mainly affected the yield strength. The results obtained were consistent since, as the Ti matrix of the substrate increased, both E and σy increased.  Figure 2 shows macro-images of a fully-dense (reference substrate without pores) and both spacer particles range sizes used for the manufacture of the porous substrates coated with the two BGs. It could be observed that the coating was more homogeneous and had a good structural entirety when pores were present; even the coating replicated the porous structures, which would mean better infiltration and adhesion of the coatings due to the pores, as checked with D-samples ( Figure 3). Fully-dense c.p. Ti substrates showed chipping, corroborating the poor adherence of the coating. The porosity and morphology, sizes, and distribution analyses of the pores were completed using the Archimedes' method, which allowed the measurement of not only density, but also total and interconnected porosity. Table 1 summarizes these parameters and the estimated mechanical properties calculated from them, using the Nielsen equations mentioned above. In view of these results, the efficacy of the spacer-holder technique could be confirmed since the porosity and size of the pores agreed with the volume and pore range sizes of the spacer particles used. On the other hand, it is known that mechanical behavior depends on the content and size of the pores. In fact, it could be observed that the pores range sizes mainly affected the yield strength. The results obtained were consistent since, as the Ti matrix of the substrate increased, both E and σ y increased. Table 1. Experimental parameters measured by Archimedes' (P T , P i ) and IA (P T , D eq , F f ). Mechanical properties (E N , E d , σ y ) estimated from IA data.

Pore Range Size (µm)
Archimedes' Method IA  Figure 2 shows macro-images of a fully-dense (reference substrate without pores) and both spacer particles range sizes used for the manufacture of the porous substrates coated with the two BGs. It could be observed that the coating was more homogeneous and had a good structural entirety when pores were present; even the coating replicated the porous structures, which would mean better infiltration and adhesion of the coatings due to the pores, as checked with D-samples ( Figure 3). Fully-dense c.p. Ti substrates showed chipping, corroborating the poor adherence of the coating.  The characterization of the coated substrates was completed by the analysis of the Dshaped substrates to know the infiltration and thickness of the bilayer BGs coating. As an example, Figure 3 displays SEM and confocal images, as well as EDS-SEM of the porous c.p. Ti substrate of 355-500 µ m porous substrate. The infiltration of the bilayer BG coating into the pores is shown in Figure 3a,b. The SEM image (Figure 3a) revealed a pore completely filled with the coating, while the pore of the CLM image (Figure 3b) was partially filled. Throughout these images, the thickness of the BG layers was measured in a flat area (Ti matrix, without pores. Note that thickness could be higher if under the measure area there was a pore due to the infiltration,), revealing that BG 1393 was a bit thicker than BG 45S5 (73 ± 5 vs. 54 ± 4 µ m), since BG 1393 had already melted at 820 °C, making infiltration easier into the pores. CLM images acquired on the top-view of both pores' range sizes showed the roughness in both cases (Figure 3c,d). It was also checked through compositional analyses performed by EDS-SEM, in which Si, a common element of both BG, 45S5 and 1393, was detected inside the pores, on the Ti matrix (Figure 3e). These measures revealed a 19 wt.% and 27 wt.% of Si and Ti, respectively.

Deposition of the BG Bilayer and Characterization of the Coated Porous c.p. Ti Substrates
In addition, the tribomechanical properties of the coated substrates were evaluated through P-h curves and scratch tests on the surface (top view) of the samples. The P-h curves on both coated porous substrates are shown in Figure 4. The parameters calculated  The characterization of the coated substrates was completed by the analysis of the Dshaped substrates to know the infiltration and thickness of the bilayer BGs coating. As an example, Figure 3 displays SEM and confocal images, as well as EDS-SEM of the porous c.p. Ti substrate of 355-500 µ m porous substrate. The infiltration of the bilayer BG coating into the pores is shown in Figure 3a,b. The SEM image (Figure 3a) revealed a pore completely filled with the coating, while the pore of the CLM image (Figure 3b) was partially filled. Throughout these images, the thickness of the BG layers was measured in a flat area (Ti matrix, without pores. Note that thickness could be higher if under the measure area there was a pore due to the infiltration,), revealing that BG 1393 was a bit thicker than BG 45S5 (73 ± 5 vs. 54 ± 4 µ m), since BG 1393 had already melted at 820 °C, making infiltration easier into the pores. CLM images acquired on the top-view of both pores' range sizes showed the roughness in both cases (Figure 3c,d). It was also checked through compositional analyses performed by EDS-SEM, in which Si, a common element of both BG, 45S5 and 1393, was detected inside the pores, on the Ti matrix (Figure 3e). These measures revealed a 19 wt.% and 27 wt.% of Si and Ti, respectively.
In addition, the tribomechanical properties of the coated substrates were evaluated through P-h curves and scratch tests on the surface (top view) of the samples. The P-h curves on both coated porous substrates are shown in Figure 4. The parameters calculated The characterization of the coated substrates was completed by the analysis of the D-shaped substrates to know the infiltration and thickness of the bilayer BGs coating. As an example, Figure 3 displays SEM and confocal images, as well as EDS-SEM of the porous c.p. Ti substrate of 355-500 µm porous substrate. The infiltration of the bilayer BG coating into the pores is shown in Figure 3a,b. The SEM image (Figure 3a) revealed a pore completely filled with the coating, while the pore of the CLM image (Figure 3b) was partially filled. Throughout these images, the thickness of the BG layers was measured in a flat area (Ti matrix, without pores. Note that thickness could be higher if under the measure area there was a pore due to the infiltration,), revealing that BG 1393 was a bit thicker than BG 45S5 (73 ± 5 vs. 54 ± 4 µm), since BG 1393 had already melted at 820 • C, making infiltration easier into the pores. CLM images acquired on the top-view of both pores range sizes showed the roughness in both cases (Figure 3c,d). It was also checked through compositional analyses performed by EDS-SEM, in which Si, a common element of both BG, 45S5 and 1393, was detected inside the pores, on the Ti matrix (Figure 3e). These measures revealed a 19 wt.% and 27 wt.% of Si and Ti, respectively. In addition, the tribomechanical properties of the coated substrates were evaluated through P-h curves and scratch tests on the surface (top view) of the samples. The P-h curves on both coated porous substrates are shown in Figure 4. The parameters calculated from these curves are summarized in the table included in the Figure 4, comparing to other percentages of porosity and pores range size substrates coated in the same way (BG bilayer) already evaluated by the authors [41]. The micro-hardness of the coatings depends on several aspects such as: (i) the porosity of BG 45S5 (on the top), (ii) the hardness of the layer of BG 1393 (between the c.p. Ti substrate and BG 45S5), and (iii) the porosity of the c.p. Ti substrate.   Inset table: maximum penetration depth and relative elastic recovery for different percentages of porosity and pores range size. Figure 5 illustrates schemes of the behavior of porous substrates depending on the percentage of porosity and the pores range sizes, observed from the top-view and crosssection of the coated porous c.p. Ti substrates. On the one hand, for a fixed pores range size (i.e., 100-200 µ m), the greater the mean free path among pores, the greater the influence the titanium matrix had, since the behavior could be similar to a fully-dense substrate; the effect of the porosity increased, so a higher penetration was observed in the Ph curves. On the other hand, for a fixed percentage of porosity (i.e., 50 vol.%), the mean free path decreased as the average pore size increased, so harder substrates were detected for smaller pore range sizes. Note that the interpretation of these results was not straightforward because of the variation in the thickness and mechanical properties of both BGs deposited on the porous substrates. In addition, the behavior would also be different depending on whether the indentation got the substrate: Ti matrix or pore. Under these considerations, the results of P-h curves and scratch tests were consistent since penetration depth is related to the thickness of the coating (bilayer BGs). When it is a tenth of the thickness, results are influenced for the next layer, which in our case is BG 1393 (which is more fragile and it can be microcracked after heat treatment) and/or the porous c.p. Ti matrix. It could explain the fact that, for example, on the substrate 50 vol.% and 100-200 µ m, there is higher E for larger pores, since the penetration depth is higher and the measure should be influenced by the Ti substrate.  Figure 5 illustrates schemes of the behavior of porous substrates depending on the percentage of porosity and the pores range sizes, observed from the top-view and crosssection of the coated porous c.p. Ti substrates. On the one hand, for a fixed pores range size (i.e., 100-200 µm), the greater the mean free path among pores, the greater the influence the titanium matrix had, since the behavior could be similar to a fully-dense substrate; the effect of the porosity increased, so a higher penetration was observed in the P-h curves. On the other hand, for a fixed percentage of porosity (i.e., 50 vol.%), the mean free path decreased as the average pore size increased, so harder substrates were detected for smaller pore range sizes. Note that the interpretation of these results was not straightforward because of the variation in the thickness and mechanical properties of both BGs deposited on the porous substrates. In addition, the behavior would also be different depending on whether the indentation got the substrate: Ti matrix or pore. Under these considerations, the results of P-h curves and scratch tests were consistent since penetration depth is related to the thickness of the coating (bilayer BGs). When it is a tenth of the thickness, results are influenced for the next layer, which in our case is BG 1393 (which is more fragile and it can be microcracked after heat treatment) and/or the porous c.p. Ti matrix. It could explain the fact that, for example, on the substrate 50 vol.% and 100-200 µm, there is higher E for larger pores, since the penetration depth is higher and the measure should be influenced by the Ti substrate. Furthermore, from the P-h curves, using Oliver and Pharr methods, the Vickers hardness and Young's modulus were calculated. The results are displayed in Figure 6, which compares not only the substrate studied in this work, but also other porosities and pores range sizes [63]. As discussed before, the hardness was larger for lower porosity and, for the same porosity, like 50 vol.% bigger pores. The results obtained were consistent with the microporosity of the BG 45S5.  Furthermore, from the P-h curves, using Oliver and Pharr methods, the Vickers hardness and Young's modulus were calculated. The results are displayed in Figure 6, which compares not only the substrate studied in this work, but also other porosities and pores range sizes [63]. As discussed before, the hardness was larger for lower porosity and, for the same porosity, like 50 vol.% bigger pores. The results obtained were consistent with the microporosity of the BG 45S5.
Furthermore, scratch tests and curves were performed and are shown in Figure 7 as the main parameters related to the mechanical properties, which are summarized in the figure. It could be observed that, when the applied load increased, the penetration was also higher, as the upper layer of the coating was BG 45S5, which was microporous. The black arrow marks the change in behavior (Figure 7a), which implied 81.07 µm of penetration depth and a load of 1.05 N. In the case of the 30 vol.%, the change of behavior was observed at 0.65 N for the smaller pores (Figure 7b). In this case, the elastic recovery was greater because of the effect of micro-porosity. As penetration continued, it started testing BG 1393, which was harder; therefore, the penetration depth decreased. As the application of load continued, the effect of c.p. Ti substrate started to influence the results. The table included in this figure summarizes the average permanent penetration depth and the absolute elastic recovery for the two areas marked with the black arrow in Figure 7a. On the right side, both values were higher. The results are compared to porous substrates fabricated under the same protocol but with different percentages of porosity. It should be noted that when the load test increased, depending on the penetration depth, the response was noticed by the upper layer of the coating (BG 45S5), second layer, BG 1393, and/or the c.p. Ti substrate (as it was described before for the P-h curves, see Figure 5). In general, the scratch resistance depends on the micro-porosity of the BG 45S5 layer, the hardness of each layer, and/or the type of Ti substrate (metal matrix size). As shown in Figure 7c, it is evident that as the porosity of the substrate increased, the penetration depth also increased. Also, comparing the penetration depth of 50 vol.% (both range sizes), as it was mentioned for the P-h curves, it decreased (higher resistance to scratch), which could be related to the fact of a higher c.p. Ti matrix substrate. Furthermore, from the P-h curves, using Oliver and Pharr methods, the Vickers hardness and Young's modulus were calculated. The results are displayed in Figure 6, which compares not only the substrate studied in this work, but also other porosities and pores range sizes [63]. As discussed before, the hardness was larger for lower porosity and, for the same porosity, like 50 vol.% bigger pores. The results obtained were consistent with the microporosity of the BG 45S5.   Furthermore, scratch tests and curves were performed and are shown in Figure 7 as the main parameters related to the mechanical properties, which are summarized in the figure. It could be observed that, when the applied load increased, the penetration was also higher, as the upper layer of the coating was BG 45S5, which was microporous. The black arrow marks the change in behavior (Figure 7a), which implied 81.07 µ m of penetration depth and a load of 1.05 N. In the case of the 30 vol.%, the change of behavior was observed at 0.65 N for the smaller pores (Figure 7b). In this case, the elastic recovery was greater because of the effect of micro-porosity. As penetration continued, it started testing BG 1393, which was harder; therefore, the penetration depth decreased. As the application of load continued, the effect of c.p. Ti substrate started to influence the results. The table included in this figure summarizes the average permanent penetration depth and the absolute elastic recovery for the two areas marked with the black arrow in Figure 7a. On the right side, both values were higher. The results are compared to porous substrates fabricated under the same protocol but with different percentages of porosity. It should be noted that when the load test increased, depending on the penetration depth, the response was noticed by the upper layer of the coating (BG 45S5), second layer, BG 1393, and/or the c.p. Ti substrate (as it was described before for the P-h curves, see Figure 5). In general, the scratch resistance depends on the micro-porosity of the BG 45S5 layer, the hardness of each layer, and/or the type of Ti substrate (metal matrix size). As shown in Figure 7c, it is evident that as the porosity of the substrate increased, the penetration depth also increased. Also, comparing the penetration depth of 50 vol.% (both range sizes), as it was mentioned for the P-h curves, it decreased (higher resistance to scratch), which could be related to the fact of a higher c.p. Ti matrix substrate. It cannot be forgotten that instrumented microindentation tests (P-h curves) are static and localized tests (reduced indentation area, local phenomena), which may differ from macro behavior, while scratch tests are dynamic tests and are also performed along a larger area (scratch path). Therefore, parameters such as elastic recovery could not follow a clear relationship since the behavior of the coating under different load conditions and rates may be different due to cracks, chipping, etc. In addition to the intrinsic effect of the porosity and nature of the coating, as well as the type of substrate (percentage of porosity It cannot be forgotten that instrumented microindentation tests (P-h curves) are static and localized tests (reduced indentation area, local phenomena), which may differ from macro behavior, while scratch tests are dynamic tests and are also performed along a larger area (scratch path). Therefore, parameters such as elastic recovery could not follow a clear relationship since the behavior of the coating under different load conditions and rates may be different due to cracks, chipping, etc. In addition to the intrinsic effect of the porosity and nature of the coating, as well as the type of substrate (percentage of porosity and pores size, distribution, morphology), the appearance of cracks and chipping also depends on the conditions of the experiments (static and dynamic), type of indenter (pyramidal and conical), and applied load, among others.
It was also interesting to study the results regarding the roughness of the surface before and after scratch tests. Figure 8 shows them, and the table summarizes the main results. For the same percentage of porosity, 50 vol.%, substrates with larger pores were rougher. First, the roughness results evaluated in the scratch path were consistent with the roughness tendency evaluated with the CLM (Figure 1). However, it is more interesting to see the differences with the roughness of the scratch groove; the results showed a clear increase in R a and R y , which could be related to the presence of debris and chipping inherent to the intrinsic fragility of the coating layers. For the same percentage of porosity, as the size of the spacer particles increased, the coating was rougher (R a = 7.77 vs. 10.64 µm when comparing smaller and larger pores range sizes, respectively), since it replicated the macro-structure of the substrate. Furthermore, the analysis of the scar revealed the presence of debris and chips, which were greater in the larger pores (see R z and R y in the table for the larger pores range size), since the Ti matrix among the pores was also larger, so adherence was poorer.
Coatings 2022, 11, x FOR PEER REVIEW 11 of 16 and pores size, distribution, morphology), the appearance of cracks and chipping also depends on the conditions of the experiments (static and dynamic), type of indenter (pyramidal and conical), and applied load, among others. It was also interesting to study the results regarding the roughness of the surface before and after scratch tests. Figure 8 shows them, and the table summarizes the main results. For the same percentage of porosity, 50 vol.%, substrates with larger pores were rougher. First, the roughness results evaluated in the scratch path were consistent with the roughness tendency evaluated with the CLM (Figure 1). However, it is more interesting to see the differences with the roughness of the scratch groove; the results showed a clear increase in Ra and Ry, which could be related to the presence of debris and chipping inherent to the intrinsic fragility of the coating layers. For the same percentage of porosity, as the size of the spacer particles increased, the coating was rougher (Ra = 7.77 vs. 10.64 µ m when comparing smaller and larger pores range sizes, respectively), since it replicated the macro-structure of the substrate. Furthermore, the analysis of the scar revealed the presence of debris and chips, which were greater in the larger pores (see Rz and Ry in the table for the larger pores range size), since the Ti matrix among the pores was also larger, so adherence was poorer. Finally, to estimate the bioactivity induced by the bilayer, the potential generation of HA was investigated by immersing the substrates in SBF for 21 days. The chemical composition of the medium was obtained by ICP, measuring the concentration of Ca, P, Si, and Ti ions before and after 21 days of immersion. According to Table 2, there was an important reduction in the concentration of Ca and P ions, suggesting the formation of calcium phosphate salts. In addition, a significant increase in the Si ions was detected. This increase is very important when considering the generation of HA. As previously demonstrated in the literature [21,64], the formation of HA in a BG layer occurs in five steps ( Figure 9). The process starts with the leaching of Ca 2+ and Na + from BG, which is replaced by H + from the medium. Acidification of BG causes the breakage of some Si-O bounds, generating Si(OH)4 molecules. The silicic acid can dissolve and repolymerize, creating a Si-rich interface on the BG surface. Finally, the incorporation of Ca 2+ and PO4 3-ions into this interface leads to the generation of an amorphous calcium phosphate layer that crystallizes to form HA after the addition of CO3 2− from the medium. Therefore, the increase in Si concentration in SBF from values lower than 0.04 mg/L in freshly prepared SBF to values of 17.5 and 24.7 mg/L in samples with 100-200 and 355-500 μm, respectively, after 21 days of immersion, was clear evidence of HA formation on the substrate surface. However, the discrepancies in the Ca/P ratio with natural HA obtained by these types of substrates with a porosity of 50 vol.% were very similar to the one presented by the freshly prepared SBF although with a slightly reduced proportion of Ca/P. This fact could be explained by the amount of Ca and P ions required for the formation of HA: five Ca cations Finally, to estimate the bioactivity induced by the bilayer, the potential generation of HA was investigated by immersing the substrates in SBF for 21 days. The chemical composition of the medium was obtained by ICP, measuring the concentration of Ca, P, Si, and Ti ions before and after 21 days of immersion. According to Table 2, there was an important reduction in the concentration of Ca and P ions, suggesting the formation of calcium phosphate salts. In addition, a significant increase in the Si ions was detected. This increase is very important when considering the generation of HA. As previously demonstrated in the literature [21,64], the formation of HA in a BG layer occurs in five steps ( Figure 9). The process starts with the leaching of Ca 2+ and Na + from BG, which is replaced by H + from the medium. Acidification of BG causes the breakage of some Si-O bounds, generating Si(OH) 4 molecules. The silicic acid can dissolve and repolymerize, creating a Si-rich interface on the BG surface. Finally, the incorporation of Ca 2+ and PO 4 3− ions into this interface leads to the generation of an amorphous calcium phosphate layer that crystallizes to form HA after the addition of CO 3 2− from the medium. Therefore, the increase in Si concentration in SBF from values lower than 0.04 mg/L in freshly prepared SBF to values of 17.5 and 24.7 mg/L in samples with 100-200 and 355-500 µm, respectively, after 21 days of immersion, was clear evidence of HA formation on the substrate surface. However, the discrepancies in the Ca/P ratio with natural HA obtained by these types of substrates with a porosity of 50 vol.% were very similar to the one presented by the freshly prepared SBF although with a slightly reduced proportion of Ca/P. This fact could be explained by the amount of Ca and P ions required for the formation of HA: five Ca cations consumed against three P atoms for each HA unit, which would lead to the reduction of the initial Ca/P proportion. In addition, these results were in agreement with those calculated for samples with a porosity of 30 vol.% and 60 vol.%, already described by the authors [21]. The combination of these results pointed to the generation of HA on top of the BG bilayer. consumed against three P atoms for each HA unit, which would lead to the reduction of the initial Ca/P proportion. In addition, these results were in agreement with those calculated for samples with a porosity of 30 vol.% and 60 vol.%, already described by the authors [21]. The combination of these results pointed to the generation of HA on top of the BG bilayer.  HA formation onto BG-coated samples after 7 and 14 days of immersion was already investigated by the authors [21,41,63]. In all cases, the Ca/P discrepancy was very high after 14 days, which would mean that the HA is still in the process of formation. Probably, the HA generation mechanism is in the third step, where a Si-rich interface is formed, as the ICP result points out the increment in the Si concentration. Therefore, from our previous experience, carrying out further investigations on samples after 14 days of immersion in SBF would not show relevant information from the final scenario, which is the one that would be potentially used as an application.
Formation of HA on porous c.p. Ti was clear, as shown in Figure 10 for the porous c.p. Ti of the larger pore range sizes. Surface changes were more than evident, since the mentioned substrate was all covered with a rough surface that was different after SBF treatment, in which kinds of characteristic oval pools were formed by salt deposition and crystallization. Furthermore, the XRD diffraction pattern showed peaks that were consistent with the formation of Ca5(PO4)3(OH) and Ca3(PO4)2 characteristics of the HA. HA formation onto BG-coated samples after 7 and 14 days of immersion was already investigated by the authors [21,41,63]. In all cases, the Ca/P discrepancy was very high after 14 days, which would mean that the HA is still in the process of formation. Probably, the HA generation mechanism is in the third step, where a Si-rich interface is formed, as the ICP result points out the increment in the Si concentration. Therefore, from our previous experience, carrying out further investigations on samples after 14 days of immersion in SBF would not show relevant information from the final scenario, which is the one that would be potentially used as an application.
Formation of HA on porous c.p. Ti was clear, as shown in Figure 10 for the porous c.p. Ti of the larger pore range sizes. Surface changes were more than evident, since the mentioned substrate was all covered with a rough surface that was different after SBF treatment, in which kinds of characteristic oval pools were formed by salt deposition and crystallization. Furthermore, the XRD diffraction pattern showed peaks that were consistent with the formation of Ca 5 (PO 4 ) 3 (OH) and Ca 3 (PO 4 ) 2 characteristics of the HA. Indeed, no typical peaks from crystalized BG 45S5, corresponding to Na 2 CaSi 2 O 6 (already published by the authors [21,63]), were observed. Note that the SEM images and XRD patterns were acquired in the top-view and, therefore, on the BG 45S5 layer.
Indeed, no typical peaks from crystalized BG 45S5, corresponding to Na2CaSi2O6 (already published by the authors [21,63]), were observed. Note that the SEM images and XRD patterns were acquired in the top-view and, therefore, on the BG 45S5 layer.

Conclusions
This work summarizes the procedure for manufacturing porous c.p. Ti substrates by the space-holder technique as an efficient way to control the percentage of porosity and characteristics of the pores (morphology, size, and distribution). To enhance biofunctionality, they were coated with a combination of BG 45S5 and BG 1393 layers. Since substrates with 50 vol.% porosity were already described to exhibit good biomechanical balance, here, two different pores range sizes, 100-200 and 355-500 µ m substrates, were coated with the BG. With respect to the results obtained, coated porous substrates presented an improved behavior compared to that of fully-dense reference samples. In addition, compared to previous results, these substrates contain higher porosity and larger pores range size, which could be more recommended since they not only perform better tribomechanical balance (porous c.p. Ti substrate with better mechanical properties, harder coatings, and greater scratch resistance), but also and at the same time, implement biofunctionality (coating infiltrated in larger pores and bioactivity of the coating should be higher). In conclusion, the combination of the bilayer coating BG 45S5-BG 1393 on porous c.p. Ti substrates is a very promising tandem for achieving a desired biomechanical and biofuncional balance.

Conclusions
This work summarizes the procedure for manufacturing porous c.p. Ti substrates by the space-holder technique as an efficient way to control the percentage of porosity and characteristics of the pores (morphology, size, and distribution). To enhance biofunctionality, they were coated with a combination of BG 45S5 and BG 1393 layers. Since substrates with 50 vol.% porosity were already described to exhibit good biomechanical balance, here, two different pores range sizes, 100-200 and 355-500 µm substrates, were coated with the BG. With respect to the results obtained, coated porous substrates presented an improved behavior compared to that of fully-dense reference samples. In addition, compared to previous results, these substrates contain higher porosity and larger pores range size, which could be more recommended since they not only perform better tribomechanical balance (porous c.p. Ti substrate with better mechanical properties, harder coatings, and greater scratch resistance), but also and at the same time, implement biofunctionality (coating infiltrated in larger pores and bioactivity of the coating should be higher). In conclusion, the combination of the bilayer coating BG 45S5-BG 1393 on porous c.p. Ti substrates is a very promising tandem for achieving a desired biomechanical and biofuncional balance.