Effect of Hydrothermal Self-Healing and Intermediate Strengthening Layers on Adhesion Reinforcement of Plasma-Sprayed Hydroxyapatite Coatings

the nature of the and The by the consists of free atoms, some neutral atoms, and undissociated diatomic molecules. The temperature of the core of the plasma may up to 30,000 Gas velocity in the plasma spray torch can be varied from subsonic to supersonic using converging-diverging nozzles. Heat transfer in the plasma jet is primarily the result of the recombination of the ions and re-association of atoms in diatomic gases on the powder surfaces and absorption of radiation. Taking advantages of the plasma plume atmosphere, plasma spray can be used for surface modification and treatment, especially for activation of polymer surfaces. I addition, plasma spray can be used to deposit nanostructures as well as advanced coating structures for new applications in wear and corrosion resistance. Some state-of-the-art studies of advanced applications of plasma spraying such as nanostructure coatings, surface modifications, biomaterial deposition, and anti wear and corrosion coatings are presented in this book.


Introduction
Biomaterials employed in calcified hard tissue repair generally serve the purpose of load carrying in cases of fractures, defects and joint replacement. Metallic materials are more suitable for load-bearing applications compared with ceramics and polymeric materials due to their combination of high mechanical strength and fracture toughness. Among generally used metallic biomaterials such as 316L stainless steel and Co-Cr-Mo alloys (ASTM F75), grade II commercial pure titanium (ASTM F67) and Ti-6Al-4V alloys (ASTM F136ELI) exhibit the most suitable characteristics for biomedical applications because of their high biocompatibility, specific strength and corrosion resistance [Niinomi, 2001]. The apparent success of titanium and its alloys in implants has been attributed to the existence of a thin, stable passivation TiO 2 layer. Another advantage of titanium and its alloys for using in hard tissue replacements is their low Young's modulus because a low Young's modulus equivalent to that of human cortical bone is simultaneously required to inhibit stress shielding effect and bone absorption [Pilliar et al., 1979;Kuroda et al., 1998;Niinomi et al., 2002]. Nowadays, they are commonly clinical used in hard tissue implants such as artificial hip prosthesis, knee joints and dental roots. A biological fixation between these hard tissue implants and surrounding bones can be successfully achieved by the bone ingrowth with a mechanical interlocking [Engh et al., 1987;Callaghan, 1993]. However, limitations of metallic biomaterials are the release of toxic metallic ions and corrosion/wear products into surrounding tissues and fluids [Sunderman et al., 1989;Healy & Ducheyne, 1992;Niinomi et al., 1999;Akahori et al., 2004].
In the biomedical applications, another concept to design a bioactive surface fixation has been achieved by the bone apposition method. Bioactive ceramics have often been used as coatings to modify the surface and create a new surface for the bioinert metallic implants. With the same chemical and crystallographic structure as the major inorganic constitute of hard tissues, bioactive hydroxyapatite (Ca 10 (PO 4 ) 6 (OH) 2 , HA) is a widely preferred calcium phosphate bioceramic, which is considered as suitable bone graft substitutes [Holmes et al., www.intechopen.com 1986; Bucholz et al., 1989] in both dentistry and orthopaedics due to its favorable bioactive properties and osteoconductivity [Munting et al., 1990;Jansen et al., 1991;Yuan et al., 2001]. The advantages of HA including: (1) earlier stabilization, rapid fixation and stronger chemical bonding between the host bone and the implant [Jansen et al., 1991;Hench, 1991;Schreurs et al., 1996], and (2) increased uniform bone ingrowth and ongrow the boneimplant interface. In spite of the good biocompatibility and osteoconductivity of HA, the limitations for the usage of the dense HA sintering bulks for bone replacement are their low fracture toughness [van Audekercke & Martens, 1984] and bending strength under loadbearing situations [de Groot et al., 1990;Choi et al., 1998]. Therefore, HA is generally applied as coatings for the purpose of improving the bioactivity of the bioinert metallic implants including the stem and the acetabular cup. The combination of high strength metallic substrates with osteoconductive properties of bioceramic makes HA-coated titanium implants attractive for the load-bearing situations in orthopedic and dental surgery. In addition to promote earlier stabilization of the implant with surrounding bone, another reason for coating HA is to extend the functional life of the prosthesis and to improve the adhesion of the prosthesis to the bone. Studies demonstrated that HA-coated titanium implants show higher push-out strength compared to uncoated titanium implants [Geesink et al., 1988;Wolke et al., 1991;Cook et al., 1992;Wang et al., 1993a], and post-mortem studies reported direct bone contact with implants without a fibrous tissue interface in patients who have had successful HA-coated total hip arthroplasties [Bauer et al., 1991;Lintner et al., 1994]. Moreover, the bone bonding capacity of the HA coatings can help cementless fixation of orthopedic prostheses. It has been shown that the skeletal bonding is enhanced immediately after implantation [Jarcho, 1981;Geesink et al., 1987;Cook et al., 1988].
Advances in coating technology have brought about a new dimension in processing of biomaterials. It is clear that surface modification of metallic biomaterials gives rise to enhance biocompatibility. Many coating techniques have been used for HA coatings preparation onto metallic substrates, including plasma spraying [Wang et al., 1993b;], HVOF spraying [Sturgeon & Harvey, 1995;Lugscheider et al., 1996], chemical vapor deposition (CVD) [Liu et al., 2007], physical vapor deposition (PVD, including the RFsputtering method) [Ozeki et al., 2006], sol-gel coating [Ben-Nissan & Choi, 2006], electrochemical deposition [Peng et al., 2006], electrophoresis method [Wei et al., 2005], and biomimetic coating methods [Kokubo et al., 1987]. The HA coatings obtained from these various techniques differ in chemistry and crystallinity, which will affect the biological responses and their performances. Therefore, in addition to consider biological advantages of fast bony adaptation, firm implant-bone attachment, reduced healing time of HA to surrounding bone, the phase composition, mechanical properties and operation feasibility of HA-coated implants should also be considered for using in long-term load-bearing applications of dental implants and orthopedic prostheses. Compared with these techniques, plasma spraying constitutes the state-of-the-art procedure to improve the biological integration implants and the main industrial process to deposit thick HA coatings. The attraction lies in its easy operation, relatively low substrate temperature, high HA coating efficiency and its ability to deposit tailored HA coatings on implants with complex shapes.
The plasma spraying process was patented in 1960s, and the technical utilization of plasma as a high-temperature source is realized in the plasma torch. The torch operates with a central cone-shaped tungsten cathode and a water-cooled cylindrical copper anode. A typical plasma spraying process is shown in Fig. 1 [Suryanarayanan, 1993]. The principle of plasma spraying is that inducing an arc by a high current density and a high electric potential between the anodic copper nozzle and tungsten cathode. The plasma gases flow is injected into the annular gap between the two electrodes, and an arc is initiated by a highfrequency discharge. Noble gases of helium (He) and argon (Ar) are usually used as the primary plasma-generating gas. Diatomic gases of hydrogen (H 2 ) and nitrogen (N 2 ) can be used as the secondary gas to increase the enthalpy of plasma torch. As the plasma gases pass around the arc created between the electrodes, they are heated and partially ionized emerging from the nozzle with high velocity and high temperature. For atmospheric plasma spraying (APS), the processing temperatures are typically in the range of 1 × 10 4 to 1.5 × 10 4 K depending on the type of the plasma gas used and the power input. Factors influence the degree of particles melting during plasma spraying includes variables which control the temperature of the plasma, such as current density, anode-cathode gap distance and gas mixture. The widely used plasma-generating gas is pure Ar (purity > 99.95 wt. %). Since the thermal conductivity and the heat conduction potential for diatomic gases, such as H 2 and N 2 , are much higher than Ar [Bourdin et al., 1983], a mixed gas composition with Ar and H 2 /N 2 gives a quite hotter plasma torch than 100% Ar gas. Figure 2 displays the variation of heat content and temperature during ionization and dissociation stages of these plasma gases [Ingham & Shepard, 1965]. When well-crystallized HA powders are injected into the high temperature plasma torch, small granules will be evaporated in the torch, and larger particles are melted or partial-melted quickly by the high temperature plasma torch. Then these melted droplets are accelerated to about 200 m/s before impacting the substrate [Fauchias et al., 1992;Pfender, 1994]. The high impact velocity supplies high kinetic energy, which is expended in spreading the molten or semi-molten droplets and creating a lamellar microstructure. In addition, high cooling rate upon impact is estimated to be of the order of 10 6 to 10 8 K s − 1 . Therefore, the large contact area with the substrate and the rapid solidification result in producing amorphous calcium phosphate (ACP) component within coatings, and it is more commonly found at the coating/substrate interface.
Because of the extremely high temperature, high enthalpy of the plasma torch and rapid solidification, a significant phase transformation or decomposition of HA is occurred during the plasma spray coating process. It results in large scale dehydroxylation and decomposition effects of crystalline HA phase into tri-calcium phosphate (Ca 3 (PO 4 ) 2 , TCP), www.intechopen.com tetra-calcium phosphate (Ca 4 P 2 O 9 , TP), calcium oxide (CaO), oxyhydroxyapatite and ACP within the sprayed coatings. Plasma-sprayed HA coatings (HACs) with a higher content of impurity phases and ACP component will display higher dissolution rate than crystalline HA in aqueous solutions and body fluids . It results in some problems with decreasing the structural homogeneity and the degradation of mechanical properties in the firm fixation between the implant and surrounding bone tissue [C.Y. Yang et al., 1995Yang et al., , 1997. Therefore, decreasing the impurity phases and ACP is important for the long-term mechanical and biological stability of plasma-sprayed HACs. The ACP is a thermodynamically meta-stable component and impurity calcium phosphate phases are undesirable for the HACs, studies pointed out that performing appropriate thermal treatments, such as air or vacuum heat treatments, spark plasma sintering (SPS) technique, and hydrothermal treatments, etc., are available methods to significantly promote HA crystallization and to improve the mechanical properties and biological responsibility of HACs [Ji & Marquis, 1993;Wang et al., 1995;Lee et al., 2005;Yu et al., 2003;C.W. Yang & Lui, 2008]. Although a thick, tailored HA coating can easily be applied by the plasma spraying process, a limitation of the HACs for applications is its low cohesion and adhesive bonding strength.
To solve these problems, it has been generally recognized that performing post-heat treatments is an effective way to improve the bonding strength of HACs. Pure HACs is brittle, thus, some bioinert ceramics or metals, such as dicalcium silicate (β-Ca 2 SiO 4 ) aluminum (Al 2 O 3 ), partially stabilized zirconia (PSZ), titania (TiO 2 ), titanium and its alloys, have been chosen as the reinforcing additives to fabricate HA/ceramics pre-composite powders [Choi et al., 1998;Zheng et al., 2000;Chou & Chang, 2002a;Sato et al., 2008;Cannillo et al., 2008]. The plasma-sprayed composite coatings made from HA and these reinforcing additives can help to alleviate the brittleness of pure HA and to improve the mechanical properties of HACs, as well as the reinforcements. Since a continuous ACP layer is resulted from rapid solidification of crystalline HA droplets, it is thought of acting a high solubility region and a low energy fracture path. This situation will result in weakening the mechanical integrity of the HA coating/substrate interface and further decreasing the adhesive bonding strength. Therefore, another way to improve the interfacial strength is the application of a stable bioinert intermediate strengthening layer, or www.intechopen.com so-call as bond coat, at the coating/substrate interface to enhance the adhesion of HACs to metallic substrates. The bond coat can reduce the thermal gradient at the interface to decrease significant thermal decomposition of HA. The bond coat can help to prevent the release of metal ions to the surrounding tissue. It can also provide better mechanical interlocking and even establish a chemical bonding between bond coat and HACs. Many attempts have been made to apply the above-mentioned bioinert ceramics or metals as bond coat materials to improve the performance of plasma-sprayed HACs [Lamy et al., 1996;Chang et al., 1997;Kurzweg et al., 1998;Fu et al., 2001;Lu et al., 2004].
Plasma-sprayed HACs with a better bonding strength can be achieved by adding reinforced additives to form composite coatings and applying heat treatments to acquire a higher HA crystallinity level and fewer coating defects as a result of HA crystallization. With different materials preparation, manufacturing and characterization, or different in vitro and in vivo examination methods, however, it is difficult to systematically evaluate the relationship between the HA crystallization, interfacial chemical reactions, biological responses and mechanical properties of the coatings. This chapter represents the crystallization effect on influencing the bonding strength of the plasma-sprayed HACs through performing postvacuum heating and hydrothermal treatments. The benefit of low-temperature hydrothermal crystallization on plasma-sprayed HACs is clarified through the evaluation of crystallization mechanism by the Arrhenius kinetics. Through adding a ceramic and a metallic bond coat, the effects of mechanical interlocking and interfacial chemical reactions at interface on improving the bonding strength of HA/bond coat will be discussed. Since variables associated with implants preparation result in a certain extent of fluctuation for material properties, a strong reliability of implants is required when they are extrapolated to clinical applications. To determine the failure probability and reliability, the failure surface morphologies of HACs and the Weibull model of survival analysis [Weibull, 1951] were used to assess the effects on bonding strength data fluctuation pertaining to microstructural feature and the reliability of HACs.

Processing
Medical grade high purity HA (Sulzer Metco XPT-D-701) powder with particle size ranging from 15 to 40 µm were used in the coating process. Commercial yttria-stabilized zirconia (ZrO 2 , YSZ, Amdry 142F), pure α-titanium (CP-Ti, Amdry 9182) powders were selected as bond coat materials, and Ti-6Al-4V alloys (ASTM F136 ELI) were selected as substrates. Prior to spraying, substrates were grit-blasted with SiC grit to roughen the surface. The average surface roughness (Ra) of grit-blasted substrate was controlled at about 3.9±0.3 µm. The powders were carried by high purity Ar gas to the plasma torch following the spraying parameters as listed in Table 1. The coating thickness of YSZ and CP-Ti bond coats was controlled at about 30 µm. Total coating thickness of 120±10 µm was prepared for HA coatings with and without intermediate layers. Table 2 lists the surface roughness of YSZ, CP-Ti bond coats and HA top coat of composite coatings.
Post-heat treatments were performed in a vacuum heating chamber (Vacuum industries, System VII) with 1.33 × 10 − 3 Pa at 600°C (named as V-HACs) with a heating rate of 10°C/min, held for 3h and then furnace cooling. The hydrothermal treatment was carried out in a hermetical autoclave (Parr 4621, Pressure Vessel) at 150°C for 6h (named as HT-HACs). The heating temperature was maintained throughout the experiments using a www.intechopen.com heater attached to the autoclave and the temperature was precisely controlled by a Parr 4842, PID controller with ± 1°C. The autoclave contained 100 ml deionized water, which was used as the source of steam atmosphere during the hydrothermal treatment, and the saturated steam pressure at 150°C was 0.48 MPa. The specimens were isolated without the immersion in the water. Phase compositions of the YSZ, CP-Ti bond coats and plasmasprayed HACs were identified by X-ray diffraction (Rigaku D/MAX III. V), using CuKα, operated at 30 kV, 20 mA. A commonly used index of crystallinity (IOC) was adopted for the purpose of further quantitatively evaluating the crystallization state of the vacuum and hydrothermally-treated HACs. The IOC data is a ratio of three strongest HA diffraction peaks ((211), (112), (300), JCPDS 9-432) integral intensity of the HACs (Ic) and the asreceived HA powder (HAP, Ip) according to the relationship IOC = (Ic/Ip) × 100%. This method supposes that the IOC of as-received HAP is 100% and the calculated IOC value of the as-sprayed HACs is about 20%. To realize the variation of TCP, TP and CaO impurity phases after applying heat treatments, the internal standard method was used to quantitatively determine these phase content within the heat-treated HACs. The integral intensity of known weight percent pure Si powder added in the specimens was taken as internal standard. The calibration curves for impurity phase content have been established by Wang et al. [Wang et al., 1995]. The main peak integral intensity ratio between TCP, TP and CaO phases from various XRD patterns of V-HACs and HT-HACs were compared to the calibration curves and the concentrations (in wt. %) in these specimens were calculated. Powder carrier gas (l/min) Ar (3) Ar (3) Ar (3) Powder feed rate (g/min) 20 20 20 Surface speed (cm/min) 8000 8000 8000 † 90 µm was prepared for the HA top coat of HA/YSZ, HA/CP-Ti composite coatings, and 120 µm was prepared for plasma-sprayed HA coatings without bond coats. 3.9 ± 0.3 6.4 ± 0.4 6.4 ± 0.2 7.6 ± 0.7 8.7 ± 1.1 8.9 ± 0.7 Note: values are given as mean ± S.D., each value was the average of ten tests (n=10).

Microstructural evolution and biological responses of heat-treated HA coatings
Figure 3(a) shows the phase composition of as-sprayed CP-Ti bond coat. In addition to the diffraction peaks of titanium (α-Ti), the oxidation product of α-Ti within the coating is TiO 1.04 , which represents two different crystal structures: the major oxide is cubic TiO 1.04 (JCPDS 43-1296) and the main peak another oxide of hexagonal TiO 1.04 is observed at 2θ = 36° (JCPDS 43-1295). The as-sprayed YSZ bond coat remains a cubic crystal structure (JCPDS 27-0997) as shown in Fig. 3(b). Figure 3(c) shows the phase composition of as-sprayed HACs. A fairly high content (about 49.3 wt. %) of ACP and impurity calcium phosphate phases, including α-Ca 3 (PO 4 ) 2 (α-TCP), β-Ca 3 (PO 4 ) 2 (β-TCP), Ca 4 P 2 O 9 (TP) and CaO, are identified in the as-sprayed HACs besides the desired HA phase.  According to the phase diagram of CaO-P 2 O 5 system, since there is a lack of ambient partial water vapor pressure during vacuum heating, TCP and TP phases are stable phases and they cannot be eliminated without the replenishment of hydroxyl groups (OH − ). The CaO remained within the V-HACs because it cannot easily be converted into HA if the ambient heating atmosphere without abundant H 2 O molecules Cao et al., 1996]. It is worth noting that these TCP, TP and CaO impurity phases are significantly eliminated www.intechopen.com after hydrothermal treatment as shown in Fig. 3(e). The impurity phase content within HT-HACs is about 12.1 wt. %. The sharpening of three strongest HA main peaks and the flattening of the diffraction background (2θ at about 28º to 34º) mean that the plasmasprayed HACs further crystallized and the content of ACP significantly decreased by the 150°C hydrothermal treatment in an ambient saturated steam pressure system. The IOC of HT-HACs is about 66%, which is close to the high-temperature vacuum heat treatment.
Since the hydroxyl groups promote the reconstitution of ACP into crystalline hydroxyapatite , therefore, the saturated steam pressure atmosphere of autoclaving hydrothermal treatment can effectively improve HA crystallization and effectively eliminate the ACP and impurity phases of plasma-sprayed HACs with the replenishment of hydroxyl groups.
Figure 4(a) shows the typical surface morphology of the plasma-sprayed HACs, which represents an accumulated molten splats feature with a fair amount of pores and thermalinduced microcracks at the rapid cooling stage. After applying 600°C vacuum het treatment, Fig. 4(b) also shows a surface cracking feature for the V-HACs specimens. Different from the thermal contraction cracking during plasma spraying, however, these cracks are resulted from the significant crystallization-induced contraction effect [C.W. Yang et al., 2006] during hightemperature crystallization of HACs. Figure 4(c) displays the typical coating surface of HT-HACs, which provides evidence in microscopic surface features different from that of assprayed HACs and V-HACs. It is worth noting that nano-scale crystalline growth, indicated by the circle in Fig. 4(c), is observed on the surface of the HT-HACs specimen. These particles can be attributed to HA crystallites, which crystallized from the hydroxyl-deficient structure of plasma-sprayed HACs through the replenishment of hydroxyl groups. In addition, since the ACP is more soluble than crystalline HA phase in an aqueous environment, part of the newgrowth crystalline HA might have formed through a dissolution-recrystallization process. The nano-scale HA crystalline experiences further grain growth with a larger crystal size in the vicinity of microcracks as indicated by the arrow. The reduction of coating defects for hydrothermally-treated HACs can be recognized as the self-healing effect of the hydrothermal treatment [C.W. Yang & Lui, 2008]. As a result of crystallization during heat treatments, the contraction-induced cracking and the self-healing phenomena will significantly influence the bonding strength and the failure mechanism of HACs.  HACs specimen shows a coating structure with many vertical, apparent contractioninduced cracks as indicated by the arrow in Fig. 5(b). The defects content of V-HACs coating is about 5.3%. As shown in Fig. 5(c), the HT-HACs possess significant fewer microcracks and lower defects content (about 2.6%). It displays a much denser microstructure than the as-sprayed HACs and V-HACs. The self-healing effect of hydrothermal crystallization resulted from the new-growth HA crystallites can be recognized to diminish the coating defects and further increase the densification of plasma-sprayed HACs. Figure 5(d) and 5(e) shows the cross-sectional features of HA/CP-Ti and HA/YSZ composite coatings, respectively. The HA top coat shows a similar microstructure to the as-sprayed HACs. Since the bond coat can reduce the thermal gradient at HA/substrate interface, the HA top coat in Fig. 5(d) and 5(e) shows less thermal-induced microcracks for these composite coatings. The coating thickness of both CP-Ti and YSZ bond coats is about 30 μm, and the Ti-6Al-4V substrate is fully covered by the bond coat. In addition, the CP-Ti and YSZ bond coats provide a rougher surface than the substrate (refer to Table 2) to the HAC top coat. This can help to improve the mechanical interlocking between HA/substrate interface and further increase the adhesive bonding strength of the HA coatings.

www.intechopen.com
In addition, the biological responses of plasma-sprayed HACs, V-HACs and HT-HACs are quantitatively evaluated in vivo using the Chinese coin implant model in the femoral of a goat, and details of the experimental procedure have been well described in the previous reports [C.Y. Yang et al., , 2009. The osteoconductivity of the implants is evaluated quantitatively in terms of the new bone healing index (NBHI), which defined as the (area of new bone/area of surgical defect region) × 100%. The ability of osseointegration of implants is addressed as apposition index (AI), which defined as the (length of direct bone-implant contact/total length of bone-implant interface) × 100%. This method can help to determine the success or failure of an implant by evaluating the interaction occurring at the bonebiomaterial interface. The mean NBHI and AI data listed in Table 3 indicated that the crystallized HACs with applying heat treatments have a statistically higher extent of new bone healing and apposition index compared to the as-sprayed HACs after 12 weeks of implantation. Figure 6 shows the amount of new bone increased within the surgical defective bone regions of as-sprayed HACs, V-HACs and HT-HACs after 12 weeks of implantation. Since the phase composition and crystallinity of post-heat-treated HACs remain stable at 12 weeks, it provides better treated-HACs-to-bone contact area, which can provide the firm bone/implant fixation compared to as-sprayed HACs. Considering the crystallized coatings of V-HACs and HT-HACs, hydrothermally-crystallized HACs show better in vivo biological responses at 12 weeks and show the potentiality to provide biological fixation than the other condition in the present results.

Affected factors on crystallization during heat treatments
The experimental results demonstrate that the autoclaving hydrothermal treatment can actually promote significant crystallization to improve the phase purity, crystallinity, microstructural homogeneity and biological responses of plasma-sprayed HACs. Previous studies have indicated that the kinetics of crystallization and chemical reactions during heat treatments are significantly related to heating temperatures, which is recognized as a main factor for promoting HA crystallization [Chang et al., 1999;Campos et al., 2002;Roeder et al., 2006]. Since the IOC value represents the degree of crystallization for heat-treated HACs, it can be recognized of the conversion ratio from ACP to crystalline HA under different heating conditions. Considering the theory of chemical reaction kinetics and the definition of IOC for crystallized HACs, the HA crystallization process should follow the Arrhenius equation [Chang et al., 1999;Huang et al., 2000;Liu et al., 2001; as represented in Eq. (1). Based on the reaction kinetics of Arrhenius equation, the rate constant (k) can be thought as the reaction rate, and it represents the crystallization rate during heat treatments. The reaction rate and the activation energy of HA crystallization within vacuum can be quantitatively evaluated by the IOC of each specimens and HA crystallization under the vacuum heating follows the second-order Arrhenius reaction kinetics.
However, a significant crystallization of hydroxyl-deficient HACs requires at least 600°C Feng et al., 2000;Lu et al., 2003], and high heating temperatures tend to undermine the structural integrity and cause phase decomposition of crystalline HA. In addition, the effect of the ambient heating atmosphere is another factor that should be considered to affect the reaction rate for HA crystallization. Referring to the phase diagram of CaO-P 2 O 5 at 500 mmHg partial steam pressure (P vapor ), HA is a stable phase and water vapor is a significant factor to promote HA crystallization. The low-temperature hydrothermal treatment system with a surrounding saturated steam pressure can help to diminish the ACP, impurity phases within the plasma-sprayed HACs and significantly promote HA crystallization. This is a result of replenishment of missing OH − groups with surrounding H 2 O molecules . Thus, the influence of ambient water vapor during the autoclaving hydrothermal treatment should be considered to evaluate the kinetics of the hydrothermal crystallization at lower heating temperatures.
Since the dehydroxylation is a result of hydroxyl groups (OH − ) broken away from HA crystal structure during plasma spraying process, the ACP with a reduced crystallinity of HA occurred in the coating layers. When the hydrothermal treatment is applied to promote the crystallization of plasma-sprayed HACs, the water is vaporized. The ionized water vapor molecules contain H + and OH − groups within the hermetical autoclave and the content of H + and OH − groups increases with increasing the temperature .
Experimental evidence confirmed that the ambient saturated steam pressure plays an important role in lowering heating and reactions temperatures. This results in a significant microstructural self-healing effect through the grain growth of HA nanocrystallite [C.W. Yang & Lui, 2009], as shown in Figs. 4(c), within the hydrothermally-crystallized HACs, which also shows a statistically higher extent of new bone apposition [C.Y.  essential in the initial fixation of implants in clinical applications.
The XPS analysis results clarify the replenishment of OH − groups and the reduction of the dehydroxylation state of hydroxyl-deficient HACs during the hydrothermal treatment. Figure 7 shows the high resolution XPS O 1s spectra of HA coatings with curve-fittings, which resulted from the Gaussian peak-fitting routine. The corresponding O 1s band of the as-sprayed HACs presented in Fig. 7  In contrast with the as-sprayed HACs, XPS O 1s spectra of hydrothermally-treated coatings (HT-HACs) with relatively large integration area of the POH bonding peak but without the adsorbed H2O peak is shown in Fig. 7(b). The surface residual adsorbed H 2 O molecules can be seen to be reduced. The hydroxyl-deficient state of the as-sprayed HACs is significantly improved with the abundant replenished OH − groups from the hydrothermal treatment, especially under a higher saturated steam pressure atmosphere. In Fig. 7(c), the O 1s spectra obtained at the HT-HACs/Ti-substrate interface are fit with four peaks: the abovementioned PO, POH peaks of HA, the Ti-O peak at 529.6 eV and the Ti-OH peak at 532.5 eV. The Ti-O peak can be attributed to the surface oxide ion of Ti-substrate, and the peak at ΔBE about 3.0 eV from Ti-O peak can be assigned to the chemisorbed OH − groups of Ti-OH [Healy & Ducheyne, 1992;Takadama et al., 2001]. Since the rapid solidification of molten HA droplets during plasma spraying induces the formation of ACP within the as-sprayed HACs, XPS analysis results demonstrate that the hydrothermal treatment helps to promote the HA crystallization through the replenishment and the chemisorption of OH − groups. The presence of Ti-OH bonding can enhance the bioactive properties of the HA coating by promoting the osteointegration process [Massaro et al., 2001].

Effect of the strengthening bond coats on the adhesive bonding strength of composite HA coatings
The most commonly used method of determining tensile bonding strength for the plasmasprayed coatings is the criterion ASTM C633-01. The roughness of substrates is an important factor in achieving high bonding strength of plasma-sprayed HACs because the bonding of the HACs to metallic substrates appears to be mechanical interlocking in nature. There is less degree of chemical bonding in as-sprayed HACs. Ti-6Al-4V cylindrical rods with dimensions of 25.4 mm (φ) and 50 mm (l) are used as substrates for the tests. Each test specimen is an assembly composed of a substrate fixture, to which the HACs of 120 ± 10 µm are applied, and a loading fixture. The loading fixtures are also grit-blasted and attached to the surface of the HACs top coat using adhesive glue with an adhesive strength of about 60 MPa. After curing, the assemblies are subjected to tensile tests at a crosshead speed of 1 mm/min until failure. For the statistical significance of the following Weibull analysis, 20 specimens are tested for bonding strength measurements. Fig. 8 shows that the bonding strength of plasma-sprayed HACs is improved (ANOVA statistical analysis, p < 0.05) with applying the reinforced bond coats and post-heat treatments. The highest bonding strength about 39.9 ± 2.4 MPa is acquired for the HA/YSZ composite coating, and the HT-HACs shows a fairly high bonding strength of 38.9 ± 1.0 MPa. It can be recognized that the adhesive bonding strength is significant improved with applying the hydrothermal treatment and adding the YSZ bond coat compared to the other conditions. Referring to the cross-sectional features shown in Fig 5(b), the 600ºC vacuum heat-treated specimen shows many vertical and apparent large cracks, which are resulted from crystallization-induced contraction of coating by the thermal dilatometry measurements . Figure 4(c) represents cracks are obviously healed with the crystalline HA grains within the HT-HACs, which display a dense microstructure as shown in Fig. 5(c). It can be seen that the microstructural homogeneity with a self-healing effect occurred from the hydrothermal crystallization throughout the whole HA coating layers under the abundant saturated steam pressure environment. Although the bonding strength is improved with the crystallization of coating layers during vacuum heat treatments, however, it can be recognized that a detrimental effect of contraction-induced cracks accompanied with HA crystallization may result in the deterioration of bonding strength.
The representative failure surfaces of these coatings are shown in Fig. 9. According to ASTM C633-01, the variation of bonding strength in situ is suggested to be governed by the cohesive strength of coatings and the adhesive strength of a coating to a metal substrate. The affecting factors of the adhesive strength of a coating and substrate interface include the surface roughness of substrate and the residual stress. As for the cohesive strength of coating, the factors include the crystallinity and the densification of a coating, which appearing on the Young's modulus of a coating. A large area fraction of cohesive failure (co) can be commonly observed for high strength coatings [Kweh et al., 2000;C.W. Yang & Lui, 2008]. The cohesive failure is dominated by the microstructural features such as crystallinity, defects and lamellar texture. Compared with the failures of as-sprayed HACs shown in Fig.  9(a), the failure morphologies of HT-HACs ( Fig. 9(b)) represent homogeneity and display a larger area fraction of cohesive failure, since strengthened coatings resulted from the selfhealing effect of hydrothermal crystallization. In contrast, the decreased area fraction of adhesive failure (ad) indicates that the adhesion of HT-HACs to Ti-6Al-4V substrate is improved. Referring to the evidence from the XPS analysis as shown in Fig. 7(c), the hydrothermal treatment helps to promote the interfacial crystallization through the replenished and the chemisorbed OH − groups, which results in a significant chemical bonding of HA coating to Ti-substrate interface. Considering the HA/CP-Ti and HA/YSZ composite coatings, the decreased area fraction of adhesive failure (Fig. 9(c) and 9(d)) represents that the adhesion of HA top coat to substrate is also enhanced by adding YSZ and CP-Ti bond coats. The significant improvement of bonding strength for HA/CP-Ti and HA/YSZ composite coatings can be recognized that a higher surface roughness of CP-Ti and YSZ layers than grit-blasted Ti-6Al-4V substrate ( Table 2) to provide better interfacial mechanical interlocking. The idea to further increase interlocking is to establish a chemical bonding between HA coating and bond coat. The evident shift of XPS binding energy of Ca 2p peak for the HA/YSZ interface compared with the as-sprayed HACs (Fig. 10) indicated that there is a significant interfacial diffusion [Vincent, 2000] for Ca ions at the interface of the HA top coat to the YSZ bond coat. However, there is no interfacial chemical reaction between the HA coating and the CP-Ti bond coat. Therefore, it can be related to the fact that the diffusion of Ca ions from HA matrix into the YSZ bond coat and the formation of chemical bonding of Ca-ZrO 2 Chou & Chang, 2002b]. The interfacial chemical bonding can help to improve the bonding of HA/YSZ composite coatings.

Evaluation of failure behaviors with statistical techniques
To characterize the strength data fluctuation, reliability, failure probability and failure mechanism of materials, a powerful statistical distribution function, which is called as the Weibull distribution function, was invented by Waloddi Weibull in 1937 and delivered his hallmark American paper in 1951 [Weibull, 1951]. He claimed that this model can be applied to a wide range of problems, and the Weibull models have been used in many different applications for solving a variety of problems from many different disciplines. Equation (3) shows the general form of the Weibull distribution function.
In Eq. (3), σ represents the bonding strength, and at least twenty specimens (n=20) are tested for the purpose of statistical significance of the analysis. F(σ i ) is the cumulative failure probability corresponding to a bonding strength σ i (i is the ranking of specimens (3), the failure probability density function f(σ) curves of the as-sprayed HACs, HA/CP-Ti and HA/YSZ composite coatings are plotted in Fig. 11(a). The cumulative failure probability F(σ i ) is estimated using the Benard's median rank of Eq. (4), which is a very close approximated solution of a statistical function [Faucher & Tyson, 1988]. The reliability function R(σ i ) with a relation of R(σ i ) = 1−F(σ i ) is defined as the survival probability. Fig.  11(b) shows the natural logarithmic (ln) graphs for the cumulative failure probability at each corresponding bonding strength σ i (i=1-20) of the specimens, it can graphically evaluate the Weibull modulus (m) from the slope of a least-squares fitting method of Eq. (5) at a maximum coefficient of determination (R 2 ). Since the Weibull distribution function is used to model the reliability and the failure behavior of materials, a failure rate function λ(σ i ) shown in Eq. (6) at each corresponding bonding strength is defined for evaluating the failure behaviors [Burrow et al., 2004]. The examination of the Weibull modulus listed in Table 4 represents that HACs are reliable materials with a wear-out failure model (m > 1) of increasing failure rate (IFR). Figure 12 shows the failure rate function (λ(σ)) and reliability function (R(σ)) curves of the as-sprayed HAC, HA/CP-Ti and HA/YSZ composite coatings. These curves start from the minimum strength (σ o ), which implies the failure probability of HACs less than this strength is zero and the reliability of HACs is 1.0. The minimum strength can be recognized as the safe loading level for the plasma-sprayed HACs. Meanwhile, knowledge of the Weibull distribution function can provide further explanation for the strengthening effect of reinforced YSZ and CP-Ti bond coats on the bonding strength of as-sprayed HACs, and it can be used to determine which coating has higher uniformity and reliability. The Weibull modulus is also a measure of the variability of the data, which being larger as the degree of bonding strength fluctuation decreases. It is evident that the failure probability density function ( Fig. 11(a)) and the failure rate ( Fig. 12(a)) curves shift to the higher bonding strength and produce a concentrated data distribution for the HA/YSZ composite coating. The YSZ bond coat effectively enhances the bonding strength of plasma-sprayed HAC, and helps to acquire more stable HACs with less reliability decrease ( Fig. 12(b)) while the loading exceeds the minimum strength.

Conclusion
The evolution of microstructural features, biological responses, crystallization kinetics, tensile mechanical properties and failure behaviors of post-spray heat-treated HACs by the vacuum heating and the hydrothermal treatment are evaluated. Applying heat treatments is an effective way to improve the crystallization state, biological responses and adhesive bonding strength of plasma-sprayed HACs. Compared with these two heat treatments, the hydrothermal treatment is more favorable to eliminate the impurity phases and ACP than high temperature heat treatments in vacuum. Hydrothermal crystallization, which proceeded within a saturated steam environment, significantly improves the microstructural homogeneity, coating density and HA/Ti-substrate interfacial reaction of plasma-sprayed HACs through the self-healing effect with the grain growth of the crystalline HA. In addition, hydrothermally-treated HACs display a higher new bone healing and apposition index than the as-sprayed and vacuum heat-treated HACs. The crystallization of plasmasprayed HACs during heat treatments is second-order Arrhenius reaction kinetics. The effect of ambient heating atmosphere with a saturated steam pressure is an important factor for the hydrothermal treatment to further promote HA crystallization rate at lower heating temperatures. The addition of CP-Ti and YSZ as intermediate strengthening bond coats between HA/Ti-substrate can significantly enhance the adhesive bonding strength of plasma-sprayed HACs. In addition to the higher mechanical interlocking between HA top coat and bond coat interface, a chemical bonding resulted from the interfacial diffusion at the HA/YSZ bond coat interface can further improve the adhesive bonding strength for the HA/YSZ composite coating. Fractures with less area percentage of interfacial failure are indicative of a better adhesion of a coating. According to the results of Weibull model analysis, plasma-sprayed HACs represent a wear-out failure behavior (the Weibull modulus, m > 1) with increasing failure rate. The adhesion and survival probability of plasma-sprayed HACs are improved by adding the YSZ bond coat to form a HA/YSZ composite coating.

Acknowledgment
This study was financially supported by the National Science and Council of Taiwan (Contract No. NSC 100-2221-E-150-037) for which we are grateful.