5.1. Characterization of as-spayed coatings
In Fig. 1 the plasma-sprayed coatings generally have lower apatite crystallinity than HA powders used for plasma spray. Besides the broadening of apatite peaks and increased intensity of CaO phase, TCP phases were observed in monolithic HA and graded coatings.
The high temperature involved in plasma spray process had obviously enhanced the decomposition of apatite as well as chemical reactions within HA phase. Since its top-layer
comprised HA phase of roughly 30 μm, it is reasonable that surface phase composition of the graded coatings detected by XRD was almost same as that of monolithic HA coatings.
Similarly, both monolithic HA coating and graded coating should consistently exhibit many absorption bands of apatite structure in its FTIR reflection spectrum (Fig. 2). The broad band at 980-1120 cm-1 in FTIR reflection spectra of the two coatings, probably resulted from overlapping apatite and TCP signals, were attributed to HPO4/PO4 functional groups. The bands between 600 and 560 cm-1 were suggested to result from the vibrational mode of PO4
groups. In contrast to those of HA powder, the decrease of the stretching mode at 3570 cm-1 and the librational mode at 630 cm-1 of OH groups, suggested the OH release of a hydroxylated HA structure that occurred in the two coatings. Additionally, the split sharp bands at 570, 600, 1050, and 1110 cm-1 indicated a well-crystallized apatite powder used for plasma spray.
Despite their different deposition processing, the two coatings had a similar morphology that had patches of smooth and shiny glassy films and irregularly-shaped particles (Fig. 3 (a)).
Randomly distributed pores of different sizes and microcracks were also observed. The surface morphology depended strongly on the sizes and shapes of the feeding powder particles.
During plasma spraying, small size powder was completely melted and formed a glassy phase during fast cooling, while larger size powder, such as the one used in the present study, was only partially melted resulting in a coating morphology comprising the observed irregularly-shaped particles. This observation is consistent with the earlier-discussed XRD results (e.g.
broadening in apatite peaks).
The cross-sectional SEM micrographs showed that the layer defects between the splats within coatings are observed to run roughly parallel to the coating surface, as shown in Fig. 3 (b). Besides, there were a lot of perpendicular defects inside the coatings, e.g., pores and cracks. These plasma spray-induced layer defects were frequently observed in plasma sprayed HA and other coatings. It was found that the microstructure of graded coatings gradually transferred from Ti bond coat to HA topcoat accompanied with the change in composition distribution through coating thickness.
Chemical distribution of three major elements, Ca, P and Ti, in the graded coatings cross-section was studied using SEM-EDS technique, as also shown in Figure 3 (c). In doing through-thickness chemical analysis of the coatings, a series of point analyses were made on the polished cross section of the coating. These elemental point analysis illustrated the graded coatings were with a controlled compositional gradient. The alternating layer within the coating had a gradually decreasing concentration of the Ti toward the outer surface with the exception of initial bond coat, along with a gradually increasing Ca and P contents up to a bioactive top-layer composed of an apatite. The composition changes were so gradual that the interface between the substrate and the coating layer as well as the boundaries among the phases within the coating layer should seem to be tightly bonded, which was the main purpose to be achieved. The measured Ca/P ratio on HA top layer was in the range of 1.9-2.1, higher than the stoichiometric Ca/P ratio (1.67) of HA, and also observed in other plasma-sprayed HA coatings due to the fact that a large quantity of phosphorus was vaporized during the high temperature plasma spray process.
5.2. Characterization of fatigued coatings
The graded HA/Ti composite coatings had a significantly (p < 0.05) greater bond strength of 23.1 ± 3.4 MPa compared to that of monolithic HA coatings (14.2 ± 3.1 MPa). It is believed to be attributable to the Ti affinity and gradient coating processes [9,10]. The existence of an intermixed layer resulted in the achievement of a good adhesion of the deposited coating to the substrate. As is known, the cracks and weakness at HA-Ti interfaces
probably result from mismatch between the coefficients of thermal expansion (CTEs) of these materials. The formation of an intermixed layer in the functionally graded coating may resist the external stress produced by the unmatched CTEs. Hence the graded coatings structure may produce stronger mechanical interlocking, as suggested by Inagaki et al. [10] The major feature of the present paper resided in the facts that the Ti component in the intermixed zone with relatively good affinity to the Ti6Al4V substrate was sandwiched between the metallic substrate and the top biocoating, so that even when the HA surface layer of the composite coating was absorbed in the bone tissue, direct contact between the metallic substrate and the bone tissue can be prevented. Post-test observations indicated that the failure of monolithic HA coatings mostly occurred at the HA-Ti interface. The graded coatings mainly fractured in the HA-rich layer.
All joints in the human body are constantly subjected to fatigue loading during the simple actions of everyday living. A particular weakness in the mechanical properties of bioceramic coatings resides in their fatigue resistance. After cyclic fatigue in air for 1 million cycles, the average strength of monolithic HA coatings bonded to substrate significantly (p < 0.05) declined from 14.2 MPa, the as-sprayed strength, down to 10.9 MPa. The stability of monolithic HA coating was apparently affected by the cyclic loading, with a remarkable decrease of 23%. The fatigue-induced degradation in bond strength could arise from a poor bonding between ceramic coating and metal substrate associated with adhesive failure. The graded coatings did not change significantly (p > 0.05) over the 1 million-cycle test, keeping the bond strength around 22 MPa. The graded coatings appeared more stable than monolithic HA coatings as fatigued in air at room temperature.
The open circuit potential over time of the two coatings in HBSS before and after cyclic fatigue is depicted in Fig. 4. The OCP of both as-sprayed samples shifted towards active direction after immersion due to the dissolution that occurred at the coating surface, followed by reaching a steady state. Compared to as-sprayed, the two fatigued samples showed a lower initial potential and attained more negative potential values, indicating an inferior corrosion behavior. In contrast to monolithic HA coatings, when subjected to cyclic fatigue, the graded coatings might exhibit a better corrosion resistance during the OCP measurements.
Fig. 5 shows the typical potentiodynamic polarization curves of the plasma-sprayed coatings without and with fatigue. As corrosion potential was concerned, no significant differences (p > 0.05) were observed between the two as-sprayed coatings. In addition, the corrosion current density of the graded coatings was close to that of monolithic HA coatings because the occurrence of corrosion was the same on the surface for both coatings made of HA top-layer. These results agreed with the OCP curves.
After cyclic fatigue, corrosion potential (-505 ± 69 mV vs. SCE) of the monolithic HA coating decreased significantly (p < 0.05) to the value of -875 ± 78 mV. The HA coatings before and after fatigue had a current density of 91 ± 25 and 480 ± 125 nA/cm2, respectively, revealing the significant difference (p < 0.05). The results confirmed that fatigued HA coatings exhibited a worse electrochemical behavior than corresponding as-sprayed coatings by virtue of more active corrosion potential and greater current, although both curves were characterized by a very similar trend. It is worthwhile to note that monolithic HA coatings with fatigue indicated an obvious breakdown potential of around 0.8 V as compared to the other coatings, possibly due to fatigue-induced delaminated structure that led to the penetration of solution into the coating/substrate interface through the perpendicular and/or parallel defects. The comparative study of the corrosion behavior of the graded specimens elicited that the coatings after fatigue presented a more negative corrosion potential value of -720 ± 99 mV than -576 ± 87 mV obtained from the coatings without fatigue, but did not significantly differ (p > 0.05). The results suggested that graded coatings might be not
apparently susceptible to influence of cyclic fatigue in air for 1 million cycles possibly due to the good cohesion degree of the coating, in accordance with the results from bond strength evaluation. As for current density, it could be found that the average values of the graded coatings of between 142 and 210 nA/cm2 were independent on fatigue. To this end, the fatigue could promote HA coatings to be unstable.
5.3. Characterization of heat-treated coatings
Compared to the as-sprayed coating, heat-treated coatings had higher apatite crystallinity and a smaller amount of TCP and CaO phases (Fig. 6). The 400oC temperature did not affect the phase evolution. When the treatment temperature was raised to 500oC, the peaks became sharper. Using the integrated area of (211), (112) and (300) peaks of apatite phase, the crystallinity of heat-treated coatings was about three-fold greater than that of the as-sprayed coating, and slightly increased with an increase in treatment temperature.
Similarly, after post-deposition treatment in air at 500–700oC with the exception of 400oC, the sharper PO4 stretching mode at 604 and 573 cm-1 and the two more appreciable adsorption bands at 970–1130 cm-1 demonstrated an increased crystallinity. The appearance of the absorption band at 630 cm-1 of the OH group suggested the reestablishment of a hydroxylated HA structure. Atmospheric moisture can react with amorphous oxyapatite so that OH groups are recovered promoting the reconstitution of amorphous oxyapatite into crystalline oxyhydroxyapatite. In addition to enhancing coating crystallinity, the presently used heat treatment could effectively convert non-apatite phases such as TCP into apatite, in agreement with previous studies [14,17]. Although higher-temperature treatment can help hasten the amorphous-crystalline phase transformation, it was practical to use lower heat-treatment temperatures to avoid serious irreversible structural and property changes to the Ti alloy substrate, as well as Ti particles within the graded coatings.
The cross-sectional pictures of 400oC-treatment coatings, similar to the as-sprayed coating, showed that the layer defects between the splats within coatings, as indicated by arrows, were observed to run roughly parallel to the coating surface. Additionally, there were a lot of perpendicular defects inside the coatings, e.g., pores and cracks. These plasma spray-induced layer defects were frequently observed in plasma sprayed HA and other coatings, which were determined to be amorphous with micro-Raman spectroscopy and nano-indentation techniques [24]. After post-deposition heat treatment at 500–700oC, layer defects appearing within plasma-sprayed coatings were deceased. The differences in the CTE (coefficient of thermal expansion) caused by the annealing might more or less contribute to microstructure changes because different phases possess different CTEs (HA: 11.5 × 10-6oC-1, TCP: 14.2 × 10-6oC-1, Ti: 8.5 × 10-6oC-1) [25, 26]. In addition, this can be explained in terms of the heat treatment during which the atoms gained high kinetic energy and diffused much faster than at room temperature [16]. The faster diffusion of the atoms speeded up the phase transition and affected the coating appearance.
The bond strength of the as-sprayed coatings was 23.1 ± 3.4 MPa. After heat treatment at 400, 500, 600 and 700oC, the value became 21.1, 22.0, 25.6 and 14.1 MPa, respectively, indicating there was significantly different (p < 0.05). Clearly, the bond strengths of the specimens heat treated at 400–600oC were not significantly higher than that of the as-sprayed coating (p > 0.05). Statistical analysis using Scheffe’s multiple comparison test showed that the bond strength of 700oC-treatment coatings significantly declined by about 39%. The 600oC-treatment coating had greater bond strength than the as-sprayed coating, although there was not significantly different (p > 0.05). The differences in bond strength can be explained in terms of stress relief and microstructural changes. When heat-treated at higher temperatures such as 700oC, the regions near the substrate may result in a larger amount of cracks caused by the large volume shrinkage during crystallization and phase transformation.
In turn, it is detrimental to the bonding strength of the coating, although it possesses a higher
degree of crystallinity.
The OCP-time plots of the coating samples heat-treated at different temperatures as a function of time along with the sprayed coating are shown in Fig. 7. It seems that the as-sprayed coating was in a steady state possibly due to the apatite precipitation, although with a more negative initial potential of -0.47 V. On the contrary, all heat-treated samples except the coating heat-treated at 400oC showed a higher initial OCP, tending to a decreased potential ranging from -0.10 V to -0.25 V after OCP scanning for 5 h. Fig. 7 also shows the typical potentiodynamic polarization curves of the plasma-sprayed coatings without and with heat treatment in deaerated HBSS at 37oC. All electrochemical parameters, including corrosion potential (Ecorr), current density (icorr), and polarization resistance (Rp), of coating samples are also compiled in Table 1. Concerning the corrosion potential, there are significant differences (p < 0.05) among all test samples. After heat treatment at 400, 500, 600 and 700oC, corrosion potentials of the heat-treated samples were found to be -598, -518, -464 and -692 mV (vs.
SCE), respectively. Our results confirmed that the 600oC-treatment coating exhibited a significantly better corrosion-resistance than all the other coatings by virtue of a more noble corrosion potential, although all polarization curves were characterized by a very similar trend.
The corrosion current density and polarization resistance of the samples were determined from the potentiodynamic polarization curves using the Tafel extrapolation method. As for the current density, it was determined that the average values of the heat-treated coatings of between 83 and 153 nA/cm2, which were comparable to that of the as-sprayed coating (142 nA/cm2), were dependent on treatment temperature, revealing the significant difference (p < 0.05). It is worthwhile to note that heat-treated coatings at 700oC indicated a lower corrosion potential and larger corrosion current density as compared to the 500oC- and 600oC-treated coatings, possibly due to higher temperature-induced structure changes that led to the penetration of solution into the coating/substrate interface through perpendicular and/or parallel defects. In contrast to the current density, there was an increase in the Rp values by about a factor of two, illustrating that heat treatment endowed the coatings better corrosion resistance except for the treatment at 400oC.
The critical factors influencing the corrosion behavior of HA-based coatings are the quality (crystallinity, purity, residual stress and ion substitution in the apatite lattice) and structure (porosity and cracks) [27]. The porosity is a characteristic of plasma sprayed coatings and strongly affects their corrosion-resistance. Generally speaking, the corrosion rate increases with increasing porosity of the coatings. The electrolyte infiltrates into the inner portion of the coating through structural imperfections such as pores and cracks or pinholes existing in the coating, and it comes into contact with the deeper portion of the coating [27], causing corrosion. The obtained polarization resistance (Rp) can be used to determine the porosity that corresponds simply to the ratio of the polarization resistance of the uncoated substrates and the coated samples [28]. Using a modified equation, the ratio of Ps/Phs = Rp,hs/Rp,s can be used to represent the change in porosity, where Ps and Phs are the porosity of the as-sprayed and heat-treated coatings, and Rp,s and Rp,hs are the polarization resistance of the as-sprayed and heat-treated coatings, respectively. Substituting the obtained Rp values into the above-mentioned equation, it is obvious that Ps/Phs is approximately two for the samples heat-treated at temperature greater than 400oC, indicating that there was a positive effect on the reduction of the porosity occurred in those samples. This was because the heat treatment apparently reduced plasma spray-induced layer defects, as described earlier in the morphology change, resulting in heat-treated coatings possessing higher corrosion-resistance.
More importantly, the in vitro electrochemical test results indicated that treatment at 600oC had a more beneficial and desired effect on corrosion behavior than the as-sprayed and the other three heat-treated samples at 400, 500 and 700oC from the viewpoints of Rp and corrosion potential.
5.4. In vitro drug release
Gentamicin loading onto coatings is a clinically relevant concept in the context of total joint arthroplasty and dental surgery. We used antibiotic-soaked coatings in an in vitro drug release study. The appearances of the heat-treated coating surfaces were similar to those of as-sprayed coatings, which had well-flattened splats and shiny glassy films and irregularly shaped particles, as shown in Fig. 8a. Randomly distributed pores of different sizes as well as microcracks were also observed. Gentamicin might incorporate into the pore within the plasma sprayed coating. In contrast to the image in Fig. 3a, the gentamicin-loaded surface became much smoother and quite uniform, but fractures were visible possibly due to drying (Fig. 8b). After drug release, coating morphologies are similar to those without drug loading (Figs. 8c, d).
Gentamicin release profiles from the two coatings in PBS as a function of time are shown in Fig. 9. Generally, the release curves can be separated into an initial fast release, followed by a slow release pattern. The fast release is mainly caused by the dissolution of the drug that is physically adsorbed on coating implants, and the slow release may be attributed to chemically adsorbed drugs. It can be seen that during the fast release, the rate of the untreated coatings was larger than that of the heat-treated coatings. It is noted that, at the initial 1 h, the two coatings released almost entirely gentamicin. Osteoconductive coatings have the potential to serve as drug carriers to prevent infection in the setting of total joint arthroplasty and dental therapy [29].
5.5. RGD grafting
An ideal biomaterial for bone repair and replacement would administer the appropriate signals to direct the processes of osteogenesis, such as cell attachment, proliferation, differentiation, matrix deposition and ultimately mineralization of extracellular matrix. As aforementioned, RGD peptide can induce cell attachment. The MTT assay indicated there are no significant differences between all samples with or without RGDC peptide. The effect of heat treatment on cell attachments also elucidated similar results. The morphology of attached cells remained rounded morphology and had no differences among all samples after incubation for 1 day. Interestingly, cells were better spread on heat-treated coatings without RGDC grafting surface after 4 h incubation. It seems the presence of extracellular matrix secretion. The morphology of cells attached to the samples surface would not be affected by RGDC physical immobilization. This might be attributed to the degradation of physically adhered RGDC on samples surface after contacting with culture medium [30]. Two important design factors, when using RGD, are addressed: (1) the spatial distribution or concentration of peptides incorporated into biomimetic materials and (2) the spacer for peptide modification, which enables peptide sequences to freely extend outward of the Network [21]. Sawyer et al.
suggested a potential therapeutic benefit for functionalizing HA with RGD, however such a benefit will likely depend upon the RGD density [23]. Hern et al. showed that cells adhered to the RGD modified surface with a PEG spacer (MW 3400) at low surface density (0.01 pmol/cm2), although the surface modified with RGD without the spacer exhibited limited cell adhesion even at a higher surface density (1 pmol/cm2) [31]. To clarify the role of RGD
suggested a potential therapeutic benefit for functionalizing HA with RGD, however such a benefit will likely depend upon the RGD density [23]. Hern et al. showed that cells adhered to the RGD modified surface with a PEG spacer (MW 3400) at low surface density (0.01 pmol/cm2), although the surface modified with RGD without the spacer exhibited limited cell adhesion even at a higher surface density (1 pmol/cm2) [31]. To clarify the role of RGD