Effects of Laser Power Level on Microstructural Properties and Phase Composition of Laser-Clad Fluorapatite/Zirconia Composite Coatings on Ti6Al4V Substrates

Hydroxyapatite (HA) is one of the most commonly used materials for the coating of bioceramic titanium (Ti) alloys. However, HA has poor mechanical properties and a low bonding strength. Accordingly, the present study replaces HA with a composite coating material consisting of fluorapatite (FA) and 20 wt % yttria (3 mol %) stabilized zirconia (ZrO2, 3Y-TZP). The FA/ZrO2 coatings are deposited on Ti6Al4V substrates using a Nd:YAG laser cladding system with laser powers and travel speeds of 400 W/200 mm/min, 800 W/400 mm/min, and 1200 W/600 mm/min, respectively. The experimental results show that a significant inter-diffusion of the alloying elements occurs between the coating layer (CL) and the transition layer (TL). Consequently, a strong metallurgical bond is formed between them. During the cladding process, the ZrO2 is completely decomposed, while the FA is partially decomposed. As a result, the CLs of all the specimens consist mainly of FA, Ca4(PO4)2O (TTCP), CaF2, CaZrO3, CaTiO3 and monoclinic phase ZrO2 (m-ZrO2), together with a small amount of θ-Al2O3. As the laser power is increased, CaO, CaCO3 and trace amounts of tetragonal phase ZrO2 (t-ZrO2) also appear. As the laser power increases from 400 to 800 W, the CL hardness also increases as a result of microstructural refinement and densification. However, at the highest laser power of 1200 W, the CL hardness reduces significantly due to the formation of large amounts of relatively soft CaO and CaCO3 phase.


Introduction
Biomedical implants typically comprise a thin bioceramic coating deposited on a titanium (Ti) alloy substrate [1][2][3]. Of the various coating materials available, hydroxyapatite (Ca 10 (PO 4 ) 6 (OH) 2 ; HA) is one of the most commonly used since it has the same chemical composition and crystallographic structure as the apatite of living bones, and hence promotes early bonding between the implant and surrounding tissue [4][5][6][7]. HA coatings are generally deposited using a plasma spraying technique due to its short operation time, high accumulation rate, low processing cost, and low heat input. However, HA is intrinsically brittle, and hence the as-sprayed HA coatings generally have poor adhesion with the Ti substrate [8]. Moreover, under high temperature plasma spraying, the HA decomposes into impurities and low-crystallinity phases (e.g., tricalcium phosphate (Ca 3 (PO 4 ) 2 ), tetracalcium phosphate (Ca 4 (PO 4 ) 2 O), and calcium oxide (CaO)). These phases cause the coating to delaminate or flake, and can therefore lead to various medical disorders [9]. Furthermore, the HA coating and Ti substrate have very different thermal expansion coefficients. Consequently, significant residual stress is formed at the interface between the coating and the substrate during the spraying process [9]. This residual stress prompts premature coating degradation and can lead to implant failure after long-term imbedding in the human body. As a result, the need for alternative coating materials with improved structural stability under high temperature conditions has emerged as a pressing concern in recent decades.
The excellent mechanical, electrical, thermal and optical properties of zirconia ceramics render them useful for many structural and functional applications [27][28][29][30]. In the biomedical field, yttria-stabilized tetragonal zirconia (Y-TZP) is widely used for dental restoration work due to its high biological safety and good hydrothermal stability [31]. Various studies have shown that 3 mol % yttria (3Y-TZP) provides an ideal stabilizing effect for zirconia [27,32]. However, even though FA is known to be more thermally and chemically stable than HA, the literature contains little information regarding the mechanical and biological properties of FA coatings mixed with ZrO 2 [4,23,24,31]. Moreover, of those studies which have been performed, the coatings are generally prepared using sintering or plasma spraying techniques. In contrast to such methods, laser beams have high coherence and directionality, and thus have the potential to generate strong metallurgical bonds between the bioceramic coating layer and the substrate [33][34][35][36][37]. Consequently, the present study investigates the microstructural properties and phase composition of composite FA coatings containing 20 wt % yttria (3 mol %) stabilized zirconia (ZrO 2 , 3Y-TZP) deposited on Ti6Al4V substrates using a laser-cladding technique. As in previous studies by the present group [38][39][40], the laser cladding process is performed using an Nd:YAG laser system.
In laser cladding processes, the microstructural properties and phase composition of the coatings are significantly affected by three experimental parameters, namely the specific energy (Es = P/(VˆD), P: power; V: travel speed; D: laser beam diameter), the laser power density (LPD = P/(πD 2 /4)) and the interaction time (t = D/V)). In the present study, the cladding process is performed using three different settings of the laser power and travel speed, namely 400 W/200 mm/min, 800 W/400 mm/min, and 1200 W/600 mm/min, respectively. In other words, the laser-power/travel-speed (P/V) ratio is equal to 2:1 in all of the cladding trials. Moreover, the laser spot diameter is equal to 3 mm (approximately) in every case. Consequently, the specific energy is a constant in the present study. By contrast, the laser power density increases as the laser power is increased from 400 W to 1200 W. Finally, the interaction time reduces as the travel speed increases.

Experiments
The cladding trials were performed using Ti6Al4V alloy plates with dimensions of 100 mmˆ60 mmˆ3.8 mm and the chemical composition shown in Table 1. The FA was prepared using Ca 3 (PO 4 ) 2 (β-TCP, β-tricalcium phosphate) and CaF 2 (Sigma-Aldrich, St. Louis, MO, USA) powders in accordance with the solid state reaction 3Ca 3 (PO 4 ) 2 + CaF 2 Ñ Ca 10 (PO 4 ) 6 F 2 [41]. Briefly, the powders were mixed in a stoichiometric molar ratio of 3:1 and milled with ZrO 2 balls in ethanol for 48 h. After drying, the powder was compacted and heated at 1000˝C for 3 h in air to form solid FA cylinders. The cylinders were then ground into powder and reinforced with 20 wt % commercial ZrO 2 powder (3 mol % Y 2 O 3 , EE-TEC Inc., Zhongli, Taiwan). Prior to the reinforcement process, the atomic structures of the FA (JCPDS 15-0876) and ZrO 2 (3Y-TZP, JCPDS 49-1642) powders were examined using XRD ( Figure 1). No separate yttria peak was detected (i.e., the XRD patterns of ZrO 2 (3Y-TZP) and TZP (tetragonal ZrO 2 , t-ZrO 2 ) are very similar). The FA/ZrO 2 powder was mixed with a polyvinyl alcohol binder ((C 2 H 4 O) n ) in a 3:1 ratio (wt %) and stirred until a slurry-like consistency was obtained. As shown in Figure 2, the Ti6Al4V substrates used in the present study were milled with two slots (each with a size of 52 mmˆ44 mmˆ0.8 mm). The FA/ZrO 2 + binder slurry was placed in each slot and the excess quantity removed using a stainless steel scraper. The substrate was then dried in an oven at 100˝C for 30 min under atmospheric conditions. Finally, the specimens were laser clad using an Nd:YAG laser system (ROFIN CW025, 2500 W; Rofin Sinar Technologies Inc., Hamburg, Germany) operating in a continuous-wave mode. For each specimen, the cladding process was limited to a single laser line scan.  Figure 1). No separate yttria peak was detected (i.e., the XRD patterns of ZrO2 (3Y-TZP) and TZP (tetragonal ZrO2, t-ZrO2) are very similar). The FA/ZrO2 powder was mixed with a polyvinyl alcohol binder ((C2H4O)n) in a 3:1 ratio (wt %) and stirred until a slurry-like consistency was obtained. As shown in Figure 2, the Ti6Al4V substrates used in the present study were milled with two slots (each with a size of 52 mm × 44 mm × 0.8 mm). The FA/ZrO2 + binder slurry was placed in each slot and the excess quantity removed using a stainless steel scraper. The substrate was then dried in an oven at 100 °C for 30 min under atmospheric conditions. Finally, the specimens were laser clad using an Nd:YAG laser system (ROFIN CW025, 2500 W; Rofin Sinar Technologies Inc., Hamburg, Germany) operating in a continuous-wave mode. For each specimen, the cladding process was limited to a single laser line scan.     Figure 1). No separate yttria peak was detected (i.e., the XRD patterns of ZrO2 (3Y-TZP) and TZP (tetragonal ZrO2, t-ZrO2) are very similar). The FA/ZrO2 powder was mixed with a polyvinyl alcohol binder ((C2H4O)n) in a 3:1 ratio (wt %) and stirred until a slurry-like consistency was obtained. As shown in Figure 2, the Ti6Al4V substrates used in the present study were milled with two slots (each with a size of 52 mm × 44 mm × 0.8 mm). The FA/ZrO2 + binder slurry was placed in each slot and the excess quantity removed using a stainless steel scraper. The substrate was then dried in an oven at 100 °C for 30 min under atmospheric conditions. Finally, the specimens were laser clad using an Nd:YAG laser system (ROFIN CW025, 2500 W; Rofin Sinar Technologies Inc., Hamburg, Germany) operating in a continuous-wave mode. For each specimen, the cladding process was limited to a single laser line scan.    As stated above, the cladding trials were performed using laser powers (P) and travel speeds (V) of 400 W/200 mm/min, 800 W/400 mm/min, and 1200 W/600 mm/min. The laser spot diameter was equal to approximately 3 mm in every case. Consequently, the specific energy (Es = P/(DˆV)) for each coating was equal to 40 J/mm 2 . Furthermore, the laser power densities (LPD = P/(πD 2 /4)) for the three cases were 5.66, 11.32 and 16.98 kW/cm 2 , respectively. Finally, the interaction times (t = D/V) were 0.9, 0.45 and 0.3 s, respectively. The laser beam was guided to the workstation by an optical fiber with a core diameter of 600 µm and a focal length of 120 mm. The cladding process was performed in an Ar-shielded atmosphere (Ar flow rate: 25 L/min) with a 5˝laser incident angle and a 15 mm positive defocus length. The experimental setup is shown in Figure 3. As stated above, the cladding trials were performed using laser powers (P) and travel speeds (V) of 400 W/200 mm/min, 800 W/400 mm/min, and 1200 W/600 mm/min. The laser spot diameter was equal to approximately 3 mm in every case. Consequently, the specific energy (Es = P/(D × V)) for each coating was equal to 40 J/mm 2 . Furthermore, the laser power densities (LPD = P/(πD 2 /4)) for the three cases were 5.66, 11.32 and 16.98 kW/cm 2 , respectively. Finally, the interaction times (t = D/V) were 0.9, 0.45 and 0.3 s, respectively. The laser beam was guided to the workstation by an optical fiber with a core diameter of 600 µm and a focal length of 120 mm. The cladding process was performed in an Ar-shielded atmosphere (Ar flow rate: 25 L/min) with a 5° laser incident angle and a 15 mm positive defocus length. The experimental setup is shown in Figure 3.  The microstructures of the clad specimens were observed using SEM (JEOL JSM-6390LV, JEOL Ltd., Tokyo, Japan). In addition, the phase compositions were examined using SEM and EDS. Moreover, the various phases were identified using XRD (Cu Kα radiation, Rigaku D/Max Ш.V, Rigaku Ltd., Tokyo, Japan) with a 2θ scanning range of 20°~70° and a scanning rate of 2° min −1 . Finally, the hardness of the coatings was determined using a micro-Vickers hardness tester under a maximum indentation load of 300 g. Figure 4 presents cross-sectional optical microscope (OM) images of the coating layers (CLs) and transition layers (TLs) of the three laser-clad specimens. As shown, for a constant Es, the depth, width and depth/width ratio (i.e., aspect ratio) of the TL all increase with an increasing laser power (see also Table 2). In addition, cracks are evident within the TLs of the specimens prepared at higher laser powers of 800 W and 1200 W, respectively. The severity of the cracks increases with increasing power.  The microstructures of the clad specimens were observed using SEM (JEOL JSM-6390LV, JEOL Ltd., Tokyo, Japan). In addition, the phase compositions were examined using SEM and EDS. Moreover, the various phases were identified using XRD (Cu Kα radiation, Rigaku D/Max III.V, Rigaku Ltd., Tokyo, Japan) with a 2θ scanning range of 20˝~70˝and a scanning rate of 2˝min´1. Finally, the hardness of the coatings was determined using a micro-Vickers hardness tester under a maximum indentation load of 300 g. Figure 4 presents cross-sectional optical microscope (OM) images of the coating layers (CLs) and transition layers (TLs) of the three laser-clad specimens. As shown, for a constant Es, the depth, width and depth/width ratio (i.e., aspect ratio) of the TL all increase with an increasing laser power (see also Table 2). In addition, cracks are evident within the TLs of the specimens prepared at higher laser powers of 800 W and 1200 W, respectively. The severity of the cracks increases with increasing power. As stated above, the cladding trials were performed using laser powers (P) and travel speeds (V) of 400 W/200 mm/min, 800 W/400 mm/min, and 1200 W/600 mm/min. The laser spot diameter was equal to approximately 3 mm in every case. Consequently, the specific energy (Es = P/(D × V)) for each coating was equal to 40 J/mm 2 . Furthermore, the laser power densities (LPD = P/(πD 2 /4)) for the three cases were 5.66, 11.32 and 16.98 kW/cm 2 , respectively. Finally, the interaction times (t = D/V) were 0.9, 0.45 and 0.3 s, respectively. The laser beam was guided to the workstation by an optical fiber with a core diameter of 600 µm and a focal length of 120 mm. The cladding process was performed in an Ar-shielded atmosphere (Ar flow rate: 25 L/min) with a 5° laser incident angle and a 15 mm positive defocus length. The experimental setup is shown in Figure 3. The microstructures of the clad specimens were observed using SEM (JEOL JSM-6390LV, JEOL Ltd., Tokyo, Japan). In addition, the phase compositions were examined using SEM and EDS. Moreover, the various phases were identified using XRD (Cu Kα radiation, Rigaku D/Max Ш.V, Rigaku Ltd., Tokyo, Japan) with a 2θ scanning range of 20°~70° and a scanning rate of 2° min −1 . Finally, the hardness of the coatings was determined using a micro-Vickers hardness tester under a maximum indentation load of 300 g. Figure 4 presents cross-sectional optical microscope (OM) images of the coating layers (CLs) and transition layers (TLs) of the three laser-clad specimens. As shown, for a constant Es, the depth, width and depth/width ratio (i.e., aspect ratio) of the TL all increase with an increasing laser power (see also Table 2). In addition, cracks are evident within the TLs of the specimens prepared at higher laser powers of 800 W and 1200 W, respectively. The severity of the cracks increases with increasing power.   Given the same laser spot size and thermal diffusivity of the substrate, the size and shape of the weld fusion zone (i.e., weld bead) formed in the laser cladding process depends on the laser power density (LPD). For a low LPD, the weld fusion zone tends to be shallow and bowl-shaped. By contrast, for higher LPDs, the fusion zone is deeper and has a higher aspect ratio [42]. In the present study, even though the LPD increases with an increasing laser power, the specific energy, Es, is fixed. Thus, it seems reasonable to assume that the weld fusion zone should have a similar profile for all three specimens. However, in maintaining a constant Es, the laser travel speed is increased proportionally with increasing laser power. In practice, the laser speed also affects the shape of the weld bead. Thus, as shown in Figure 3, the weld bead profile changes with a changing LPD despite the constant Es in every case.

Morphology and Microstructure of Weld Beads
Previous studies have shown that if the laser cladding parameters are not properly controlled, cracks are readily formed in the coating as a result of the thermal expansion mismatch between the coating and the substrate and relatively low toughness of the coating material [43]. As shown in Figure 4, cracks were not formed in any of the CLs. However, for the specimens prepared using laser powers of 800 W and 1200 W (i.e., a higher travel speed, heating rate and cooling rate), cracking occurred in the TL (see Figure 4b,c). The formation of these cracks is closely related to the residual stress generated during the cladding process [44]. Moreover, the stress increases with an increasing heating and cooling rate. The OM images in Figure 4b,c show that the cracks in the two specimens initiate within the TL, i.e., they do not spread from the CL. It is therefore concluded that the thermal expansion mismatch between the CL and the TL plays no role in prompting crack formation in the TL. In other words, crack initiation is dominated by the thermal stress generated by the rapid cooling rate of the substrate. More specifically, for a higher laser power (a higher laser travel speed), the cooling rate is increased. Consequently, the thermal shock within the TL is enhanced, and hence the crack severity increases. Furthermore, under a higher power level, the weld zone size also increases. As a result, a larger shrinkage stress is induced, and thus crack formation is further enhanced. Notably, however, the OM images in Figure 4 show that crack formation can be controlled through an appropriate setting of the laser power and travel speed parameters. Figure 5 shows the surface microstructures of the CLs in the three laser-clad samples. It is seen that for all three samples, the coating has a fibrous-like morphology. Moreover, as the laser power increases, an increasing number of granular compounds are formed between the fibrous-like structures. The SEM images confirm the absence of microcracks in the CL. The lack of cracks is reasonable since the coefficients of thermal expansion (CTE) of FA and ZrO 2 (3Y-TZP) (i.e., 9.1ˆ10´6 K´1 [45] and 10~12ˆ10´6 K´1 [46], respectively) are quite close to that of the Ti6Al4V substrate (8.8ˆ10´6 K´1 [47]) in comparison to commonly used bioceramic coatings in the past, HA (15ˆ10´6 K´1 [47]). Hence, only limited differential expansion between the CL and the TL occurs. The addition of ZrO 2 particles with good mechanical strength and high fracture toughness further enhances the cohesive strength of the CL and suppresses microcrack formation. Therefore, the FA/ZrO 2 -substrate CTE mismatch is less than that for a HA/ZrO 2 -substrate system. Consequently, it can be inferred that the present FA/ZrO 2 coatings have a better adhesive strength with the Ti6Al4V substrate than HA/ZrO 2 coatings [8,29]. Figure 6 shows the interfacial microstructures of the CL and TL in the three specimens. The EDS analysis results (presented in Section 3.2) show that a significant diffusion of alloying elements occurs between the CL and TL. As a result, a strong metallurgical bond is formed between them [34]. The bonding strength of such an interface is greater than the mechanical bonding strength formed in general coating processes such as plasma spraying, sol-gel, and so on. Figure 7 presents SEM cross-sectional metallographs of the mid-section region of the CL in the various samples.
Compared with the CL surface microstructures shown in Figure 5, the quantity of fibrous structures is reduced. However, a large number of spherical and irregularly-shaped particles are observed. For the sample prepared using a low laser power (400 W), the microstructure contains many flower-like structures. However, as the power increases, the flower-like structures are replaced with spherical-like particles. In general, the SEM images presented in Figures 6 and 7 show that the higher cooling rate associated with an increased laser power results in significant microstructural refinement and densification of the CL. with the CL surface microstructures shown in Figure 5, the quantity of fibrous structures is reduced. However, a large number of spherical and irregularly-shaped particles are observed. For the sample prepared using a low laser power (400 W), the microstructure contains many flower-like structures. However, as the power increases, the flower-like structures are replaced with spherical-like particles. In general, the SEM images presented in Figures 6 and 7 show that the higher cooling rate associated with an increased laser power results in significant microstructural refinement and densification of the CL.

Chemical Composition Analysis of Laser-Clad Coatings
Figure 8a presents an SEM image of the CL in the sample prepared using a laser power of 1200 W. Figure 8b,c shows the EDS analysis results for globular particles A and B in Figure 8a, respectively. As shown in Figure 8b, particle A consists mainly of Ti and P. The presence of TiP compounds (Ti phosphides) [48] suggests a partial thermal decomposition of the FA content in the CL during the high-temperature cladding process, accompanied by thermal-induced melting and diffusion of Ti from the substrate. As the laser power increases, the extent of FA decomposition and Ti melting/diffusion also increases, and hence Ti phosphides of a greater size are formed, as shown in Figure 7b,c. It is noted that the present results are consistent with those of Ye et al. [48], who showed that a large number of Ti phosphides (TixPy) were formed when sintering Ti/FA (1:1) with the CL surface microstructures shown in Figure 5, the quantity of fibrous structures is reduced. However, a large number of spherical and irregularly-shaped particles are observed. For the sample prepared using a low laser power (400 W), the microstructure contains many flower-like structures. However, as the power increases, the flower-like structures are replaced with spherical-like particles. In general, the SEM images presented in Figures 6 and 7 show that the higher cooling rate associated with an increased laser power results in significant microstructural refinement and densification of the CL.    Figure 8a presents an SEM image of the CL in the sample prepared using a laser power of 1200 W. Figure 8b,c shows the EDS analysis results for globular particles A and B in Figure 8a, respectively. As shown in Figure 8b, particle A consists mainly of Ti and P. The presence of TiP compounds (Ti phosphides) [48] suggests a partial thermal decomposition of the FA content in the CL during the high-temperature cladding process, accompanied by thermal-induced melting and diffusion of Ti from the substrate. As the laser power increases, the extent of FA decomposition and Ti melting/diffusion also increases, and hence Ti phosphides of a greater size are formed, as shown in Figure 7b,c. It is noted that the present results are consistent with those of Ye et al. [48], who showed that a large number of Ti phosphides (TixPy) were formed when sintering Ti/FA (1:1) with the CL surface microstructures shown in Figure 5, the quantity of fibrous structures is reduced. However, a large number of spherical and irregularly-shaped particles are observed. For the sample prepared using a low laser power (400 W), the microstructure contains many flower-like structures. However, as the power increases, the flower-like structures are replaced with spherical-like particles. In general, the SEM images presented in Figures 6 and 7 show that the higher cooling rate associated with an increased laser power results in significant microstructural refinement and densification of the CL.    Figure 8a presents an SEM image of the CL in the sample prepared using a laser power of 1200 W. Figure 8b,c shows the EDS analysis results for globular particles A and B in Figure 8a, respectively. As shown in Figure 8b, particle A consists mainly of Ti and P. The presence of TiP compounds (Ti phosphides) [48] suggests a partial thermal decomposition of the FA content in the CL during the high-temperature cladding process, accompanied by thermal-induced melting and diffusion of Ti from the substrate. As the laser power increases, the extent of FA decomposition and Ti melting/diffusion also increases, and hence Ti phosphides of a greater size are formed, as shown in Figure 7b,c. It is noted that the present results are consistent with those of Ye et al. [48], who showed that a large number of Ti phosphides (TixPy) were formed when sintering Ti/FA (1:1)  Figure 8a presents an SEM image of the CL in the sample prepared using a laser power of 1200 W. Figure 8b,c shows the EDS analysis results for globular particles A and B in Figure 8a, respectively. As shown in Figure 8b, particle A consists mainly of Ti and P. The presence of TiP compounds (Ti phosphides) [48] suggests a partial thermal decomposition of the FA content in the CL during the high-temperature cladding process, accompanied by thermal-induced melting and diffusion of Ti from the substrate. As the laser power increases, the extent of FA decomposition and Ti melting/diffusion also increases, and hence Ti phosphides of a greater size are formed, as shown in Figure 7b,c. It is noted that the present results are consistent with those of Ye et al. [48], who showed that a large number of Ti phosphides (Ti x P y ) were formed when sintering Ti/FA (1:1) composite powders under temperatures of 1100˝C or 1200˝C. In general, Ti phosphides can have a wide range of compositions, and it is thus difficult to reliably determine the exact composition of Ti x P y by XRD analysis alone [49]. Notably, the EDS results presented in Figure 8c, corresponding to particle B in Figure 8a, show that even under the maximum laser power of 1200 W, the CL matrix still contains a large amount of fluoride. In other words, the potential for FA residue or conversion to fluoride still exists even under high cladding temperatures, as discussed later in Section 3.3.

Chemical Composition Analysis of Laser-Clad Coatings
composite powders under temperatures of 1100 °C or 1200 °C. In general, Ti phosphides can have a wide range of compositions, and it is thus difficult to reliably determine the exact composition of TixPy by XRD analysis alone [49]. Notably, the EDS results presented in Figure 8c, corresponding to particle B in Figure 8a, show that even under the maximum laser power of 1200 W, the CL matrix still contains a large amount of fluoride. In other words, the potential for FA residue or conversion to fluoride still exists even under high cladding temperatures, as discussed later in Section 3.3.  Figure 9 presents the EDS line scanning results for the individual alloying elements of the FA/ZrO2 cladding layer near the CL/TL interface in the specimen prepared using a laser power of 1200 W. It is seen that the Ca and F elements, i.e., the main decomposition components of FA, are confined almost entirely to the CL. Previous studies have shown that in the thermal decomposition of FA, F is either vaporized as HF gas or forms fluoride [24]. Thus, the F content in the CL layer most probably comes from the FA or fluoride, but requires further XRD analysis for confirmation (see Section 3.3). The P and O ions in the coating material have the ability to diffuse rapidly from the CL to the TL due to their small radii and low activation energy. Nonetheless, as shown in Figure 9, some ions still remain within the CL. It is noted, however, that some of the P ions produced in the FA decomposition process simply vaporize in the high-temperature cladding process [37][38][39][40].
Observing the results presented in Figure 9, it is seen that the distributions of Zr in the CL and TL, respectively, are roughly the same. In other words, given an addition of 20 wt % ZrO2 to the FA cladding material, a certain amount of Zr remains within the CL despite the high temperature diffusion effect. Furthermore, the TL retains a very high Ti content; with only a small amount of Ti diffusing to the CL. Overall, the EDS results reveal that while significant diffusion of the alloying elements occurs between the TL and the CL, the alloying elements in the TL are basically similar to the composition of the substrate, while those in the CL are similar to that of the coating material.   Figure 9 presents the EDS line scanning results for the individual alloying elements of the FA/ZrO 2 cladding layer near the CL/TL interface in the specimen prepared using a laser power of 1200 W. It is seen that the Ca and F elements, i.e., the main decomposition components of FA, are confined almost entirely to the CL. Previous studies have shown that in the thermal decomposition of FA, F is either vaporized as HF gas or forms fluoride [24]. Thus, the F content in the CL layer most probably comes from the FA or fluoride, but requires further XRD analysis for confirmation (see Section 3.3). The P and O ions in the coating material have the ability to diffuse rapidly from the CL to the TL due to their small radii and low activation energy. Nonetheless, as shown in Figure 9, some ions still remain within the CL. It is noted, however, that some of the P ions produced in the FA decomposition process simply vaporize in the high-temperature cladding process [37][38][39][40]. Observing the results presented in Figure 9, it is seen that the distributions of Zr in the CL and TL, respectively, are roughly the same. In other words, given an addition of 20 wt % ZrO 2 to the FA cladding material, a certain amount of Zr remains within the CL despite the high temperature diffusion effect. Furthermore, the TL retains a very high Ti content; with only a small amount of Ti diffusing to the CL. Overall, the EDS results reveal that while significant diffusion of the alloying elements occurs between the TL and the CL, the alloying elements in the TL are basically similar to the composition of the substrate, while those in the CL are similar to that of the coating material. composite powders under temperatures of 1100 °C or 1200 °C. In general, Ti phosphides can have a wide range of compositions, and it is thus difficult to reliably determine the exact composition of TixPy by XRD analysis alone [49]. Notably, the EDS results presented in Figure 8c, corresponding to particle B in Figure 8a, show that even under the maximum laser power of 1200 W, the CL matrix still contains a large amount of fluoride. In other words, the potential for FA residue or conversion to fluoride still exists even under high cladding temperatures, as discussed later in Section 3.3.  Figure 9 presents the EDS line scanning results for the individual alloying elements of the FA/ZrO2 cladding layer near the CL/TL interface in the specimen prepared using a laser power of 1200 W. It is seen that the Ca and F elements, i.e., the main decomposition components of FA, are confined almost entirely to the CL. Previous studies have shown that in the thermal decomposition of FA, F is either vaporized as HF gas or forms fluoride [24]. Thus, the F content in the CL layer most probably comes from the FA or fluoride, but requires further XRD analysis for confirmation (see Section 3.3). The P and O ions in the coating material have the ability to diffuse rapidly from the CL to the TL due to their small radii and low activation energy. Nonetheless, as shown in Figure 9, some ions still remain within the CL. It is noted, however, that some of the P ions produced in the FA decomposition process simply vaporize in the high-temperature cladding process [37][38][39][40].

Region of EDS line scanning
Observing the results presented in Figure 9, it is seen that the distributions of Zr in the CL and TL, respectively, are roughly the same. In other words, given an addition of 20 wt % ZrO2 to the FA cladding material, a certain amount of Zr remains within the CL despite the high temperature diffusion effect. Furthermore, the TL retains a very high Ti content; with only a small amount of Ti diffusing to the CL. Overall, the EDS results reveal that while significant diffusion of the alloying elements occurs between the TL and the CL, the alloying elements in the TL are basically similar to the composition of the substrate, while those in the CL are similar to that of the coating material.   For the 1200 W specimen, the CL contains large quantities of CaO, CaCO3 and TTCP, but a lesser amount of CaTiO3. Furthermore, the quantity of t-ZrO2 increases, while that of m-ZrO2 and CaZrO3 decreases. Notably, the XRD pattern also indicates the presence of trace amounts of several unknown compounds in the CL. In general, the XRD results indicate a greater tendency toward compound phase formation in the CL as the laser power is increased.    For the 1200 W specimen, the CL contains large quantities of CaO, CaCO3 and TTCP, but a lesser amount of CaTiO3. Furthermore, the quantity of t-ZrO2 increases, while that of m-ZrO2 and CaZrO3 decreases. Notably, the XRD pattern also indicates the presence of trace amounts of several unknown compounds in the CL. In general, the XRD results indicate a greater tendency toward compound phase formation in the CL as the laser power is increased.  The FA powder used in the present study was prepared via a solid state reaction of Ca 3 (PO 4 ) 2 (TCP) and CaF 2 at 1000˝C for 3 h. However, in the cladding process, the FA reverts to Ca 3 (PO 4 ) 2 and CaF 2 in accordance with Reaction (1) if the temperature remains sufficiently high for a sufficiently long period of time [23,31]. For example, Nasiri-Tabrizi and Fahami [23] reported that a partial decomposition of FA to Ca 3 (PO 4 ) 2 and CaF 2 occurs at 900˝C for 1 h in the presence of zirconia. However, at higher temperatures (i.e., greater than 1100˝C), CaF 2 transforms to CaO through hydrolysis, as shown in Reaction (2) [50]. Thus, the high CaO content in the CL of the present specimen prepared using a high laser power of 1200 W is most likely the result of the thermally-induced hydrolysis of CaF 2 .

XRD Patterns of CL Surface
Ca 10 pPO 4 q 6 F 2 Ñ 3Ca 3 pPO 4 q 2 pTCPq`CaF 2 (1) In high temperature processes, the decomposition of FA powder without oxide addition can be described by Reaction (3). (Note that Reaction (3) is equivalent to Reaction (1) + Reaction (2)). In other words, in high temperature processes, FA decomposes as TCP and CaO [24,50]. In the present study, ZrO 2 is also added to the coating. The additional reactions which therefore take place under high density laser power irradiation are described by Reaction (4) below.
Ben and Bouaziz [24] examined the decomposition of FA doped with ZrO 2 , and showed that given sufficient time and temperature, the decomposed TCP and CaO react with the ZrO 2 to produce TTCP and CaZrO 3 . In addition, many researchers have analyzed the reaction between ZrO 2 and CaO in accordance with Reaction (5) [4,24,29,51].
In the XRD patterns in Figure 10, a peak corresponding to the original ZrO 2 powder (3Y-TZP, XRD pattern similar to t-ZrO 2 ) is absent for the 400 W specimen. Moreover, only small quantities of ZrO 2 are detected in the 800 W and 1200 W samples. In other words, most of the ZrO 2 particles melt and undergo phase transformation during the laser cladding process. However, FA is still present in all three samples. Thus, it is inferred that even though the original FA powder melts completely during the laser-cladding process, the high thermal stability of the FA powder and the short residence time of the powder at high temperature result in only a partial decomposition of the FA to TTCP and CaO. However, as the laser power increases, the extent of FA decomposition also increases. Thus, for the specimens prepared using higher laser powers of 800 W and 1200 W, respectively, the intensity of the TTCP and CaO peaks in the XRD patterns increases.
Under the high temperatures produced during the laser cladding process, the ZrO 2 particles reside in a molten state and the diffusion of Ca ions into the ZrO 2 particles is enhanced. Consequently, CaZrO 3 is formed during cooling in accordance with Reaction (5) above. Furthermore, CaO is unstable and reacts with CO 2 to form calcium carbonate (CaCO 3 ) in air [31]. As the temperature increases, more of the original FA powder decomposes as CaO, and hence the quantity of CaCO 3 increases. In addition, Al 2 O 3 is produced via the reaction between Al atoms diffused from the substrate and the coating material atoms or environmental O atoms. Finally, CaTiO 3 is produced through a reaction between the decomposed or melted FA and the Ti6Al4V substrate.
For the 400 W sample, the XRD pattern indicates the presence of both m-ZrO 2 and CaZrO 3 compounds. For the sample prepared with a higher power of 800 W, the intensity of the m-ZrO 2 and CaZrO 3 peaks increases and a trace amount of t-ZrO 2 emerges. However, for the sample prepared using the highest power of 1200 W, the intensity of the t-ZrO 2 peak increases, but that of the m-ZrO 2 and CaZrO 3 peaks decreases. In other words, the original ZrO 2 (3Y-TZP) powder melts more completely under high-energy laser irradiation. During the cooling and solidification phase, an allotropic transformation of the ZrO 2 compound takes place from the cubic phase (c-ZrO 2 2370-2680˝C (melting point)) to the tetragonal phase (t-ZrO 2 , 1170˝C-2370˝C), and finally to the monoclinic phase (m-ZrO 2 , room temperature-1170˝C) [32]. For laser cladding at 400 W, the t-ZrO 2 phase has sufficient time to transform to m-ZrO 2 due to the lower cooling rate (i.e., the lower travel speed). As a result, the ZrO 2 exists almost entirely in the monoclinic phase. However, for higher laser powers of 800 W and 1200 W, respectively, insufficient time exists for t-ZrO 2 transformationto m-ZrO 2 and consequently the quantity of t-ZrO 2 increases while that of m-ZrO 2 decreases. Notably, no c-ZrO 2 is observed in any of the XRD patterns in Figure 10. This finding suggests that the temperature required for c-ZrO 2 formation is not maintained for a sufficient length of time during the cooling process, and hence the quantity of c-ZrO 2 formed is too low to be detected via XRD analysis.
Ramachandra Rao and Kannan [26] and Nagarajan and Rao [51] showed that the CaO released during the sintering of HA/ZrO 2 at temperatures of 1150˝C and above stabilizes the m-ZrO 2 via a solid solution reaction, and prompts the formation of t-ZrO 2 . However, with the release of excess CaO through the further decomposition of FA, the solubility of Ca in ZrO 2 exceeds the maximum solid solution range and hence CaZrO 3 is formed in preference to t-ZrO 2 [52,53]. Heimann and Vu [52] showed that when CaO is added to HA/ZrO 2 composite sintered samples, the surplus CaO is effectively fixed by the ZrO 2 , which acts as a sink for the Ca 2+ ions. Consequently, either t-ZrO 2 or CaZrO 3 is formed. Furthermore, according to the ZrO 2 -CaZrO 3 phase diagram [54], the formation of CaZrO 3 depends on the extent of CaO diffusion into ZrO 2 . Hence, a fuller decomposition of FA promotes the production of CaZrO 3 . However, in the present study, although the quantity of CaO increases significantly with an increasing laser power, that of t-ZrO 2 increases only slightly. Furthermore, the quantity of CaZrO 3 reduces under the highest laser power of 1200 W. By contrast, the quantity of CaCO 3 increases significantly with an increasing laser power. This feature suggests that given a sufficiently high laser energy, the tendency of CaO and CO 2 to react and form CaCO 3 is higher than that of CaO and ZrO 2 reacting to form CaZrO 3 . However, further investigation is required to confirm this inference and to clarify the related underlying mechanisms. Figure 11 shows the cross-sectional hardness profiles of the various specimens from the CL (thickness approximately 0.2-0.3 mm) through the TL and into the substrate. For all three samples, the TL has a higher hardness than the CL, which in turn has a greater hardness than the substrate. Comparing the CL hardness values of the three samples, it is seen that the hardness increases from 1100 HV 0.3 to 1300 HV 0.3 as the laser-cladding power is increased from 400 W to 800 W, but then reduces to around 800 HV 0.3 as the laser-cladding power is further increased to 1200 W. From inspection, the CL hardness of the three specimens is around 2~3 times higher than that of the Ti6Al4V substrate. allotropic transformation of the ZrO2 compound takes place from the cubic phase (c-ZrO2 2370-2680 °C (melting point)) to the tetragonal phase (t-ZrO2, 1170-2370 °C), and finally to the monoclinic phase (m-ZrO2, room temperature-1170 °C) [32]. For laser cladding at 400 W, the t-ZrO2 phase has sufficient time to transform to m-ZrO2 due to the lower cooling rate (i.e., the lower travel speed). As a result, the ZrO2 exists almost entirely in the monoclinic phase. However, for higher laser powers of 800 W and 1200 W, respectively, insufficient time exists for t-ZrO2 transformationto m-ZrO2 and consequently the quantity of t-ZrO2 increases while that of m-ZrO2 decreases. Notably, no c-ZrO2 is observed in any of the XRD patterns in Figure 10. This finding suggests that the temperature required for c-ZrO2 formation is not maintained for a sufficient length of time during the cooling process, and hence the quantity of c-ZrO2 formed is too low to be detected via XRD analysis. Ramachandra Rao and Kannan [26] and Nagarajan and Rao [51] showed that the CaO released during the sintering of HA/ZrO2 at temperatures of 1150 °C and above stabilizes the m-ZrO2 via a solid solution reaction, and prompts the formation of t-ZrO2. However, with the release of excess CaO through the further decomposition of FA, the solubility of Ca in ZrO2 exceeds the maximum solid solution range and hence CaZrO3 is formed in preference to t-ZrO2 [52,53]. Heimann and Vu [52] showed that when CaO is added to HA/ZrO2 composite sintered samples, the surplus CaO is effectively fixed by the ZrO2, which acts as a sink for the Ca 2+ ions. Consequently, either t-ZrO2 or CaZrO3 is formed. Furthermore, according to the ZrO2-CaZrO3 phase diagram [54], the formation of CaZrO3 depends on the extent of CaO diffusion into ZrO2. Hence, a fuller decomposition of FA promotes the production of CaZrO3. However, in the present study, although the quantity of CaO increases significantly with an increasing laser power, that of t-ZrO2 increases only slightly. Furthermore, the quantity of CaZrO3 reduces under the highest laser power of 1200 W. By contrast, the quantity of CaCO3 increases significantly with an increasing laser power. This feature suggests that given a sufficiently high laser energy, the tendency of CaO and CO2 to react and form CaCO3 is higher than that of CaO and ZrO2 reacting to form CaZrO3. However, further investigation is required to confirm this inference and to clarify the related underlying mechanisms. Figure 11 shows the cross-sectional hardness profiles of the various specimens from the CL (thickness approximately 0.2-0.3 mm) through the TL and into the substrate. For all three samples, the TL has a higher hardness than the CL, which in turn has a greater hardness than the substrate. Comparing the CL hardness values of the three samples, it is seen that the hardness increases from 1100 HV0.3 to 1300 HV0.3 as the laser-cladding power is increased from 400 W to 800 W, but then reduces to around 800 HV0.3 as the laser-cladding power is further increased to 1200 W. From inspection, the CL hardness of the three specimens is around 2~3 times higher than that of the Ti6Al4V substrate.  The hardness of laser-clad coatings is related to both their microstructural characteristics (e.g., porosity and density) and their phase constituents [41,43,55]. For the present samples, the CL microstructure exhibits a greater refinement and densification effect as the laser power (travel speed) increases (see Figure 6). Consequently, the hardness increases. The phase constituents of the CL can be ranked in order of decreasing hardness as ZrO 2 > CaF 2 > CaO > CaCO 3 [56]. As shown in the XRD patterns in Figure 10, the sample prepared using a laser power of 1200 W has a high CaO and CaCO 3 (low hardness) content. Thus, the microstructure-induced hardness enhancement is outweighed by the softening effect of the CaO and CaCO 3 phases, and consequently a reduction in the CL hardness occurs.

Micro-Hardness Evaluation
Chien et al. [57] showed that the average hardness of the CL formed in the laser cladding of pure FA on Ti6Al4V substrates was equal to 617 HV 0.3 for a laser power of 740 W and 750 HV 0.3 for a laser power of 1150 W. It is noted that these hardness values are significantly lower than those obtained in the present study. The laser source and laser cladding parameters are similar in both cases. Hence, it is inferred that the higher CL hardness of the present specimens is due to the addition of ZrO 2 to the FA matrix. Kim et al. [41] prepared sintered FA samples containing 20 and 40 vol % ZrO 2 powder, respectively, and found that the hardness increased with an increasing ZrO 2 content. Ouyang et al. [43] used a laser-cladding process to deposit yttria partially stabilized ZrO 2 (7 wt %) ceramic coatings doped with 2.5 wt % TiO 2 on aluminum alloy substrates. The cladding layer was found to have a hardness of 1415~1575 HV 0.1 . This value is greater than that observed for the present coatings. Kim et al. [41] conducted sintering trials using HA and FA powders doped with 20 vol % ZrO 2 . The results showed that the FA-ZrO 2 composites had a greater hardness (~8 GPa) than the HA-ZrO 2 samples (~1 GPa). Various studies have attributed the greater hardness of laser-clad ZrO 2 or ZrO 2 composite coatings to the absence of porosity and a fine-grained structure [41,43]. However, the present results suggest that the hardness is actually determined by a competition process between the microstructural hardening effects and the phase composition softening effects.

Conclusions
The present study has deposited composite coatings consisting of fluorapatite (FA) and 20 wt % yttria (3 mol %) stabilized zirconia (ZrO 2 , 3Y-TZP) on Ti6Al4V substrates using a laser cladding process with laser powers and travel speeds of 400 W/200 mm/min, 800 W/400 mm/min, and 1200 W/600 mm/min, respectively. The experimental findings support the following main conclusions.

1.
The depth, width and depth/width ratio of the TL increase with an increasing laser power (travel speed). For the 800 W and 1200 W specimens, cracks are formed in the TL due to the greater cooling rate and larger weld zone (i.e., greater shrinkage stress). However, no cracks are formed in the CL due to the addition of ZrO 2 to the FA powder and the relatively small CTE mismatch between the FA/ZrO 2 powders and the substrate.

2.
A significant diffusion of alloying elements occurs between the CL and the TL. As a result, a good metallurgical bond is formed between them. Overall, the alloying elements of the TL are close to the composition of the substrate, while the alloying elements of the CL are close to the composition of the coating material.

3.
The CL of the 400 W specimen consists mainly of FA, TTCP, CaF 2 , CaZrO 3 , CaTiO 3 , m-ZrO 2 and a small amount of θ-Al 2 O 3 . For the 800 W specimen, the CL also contains CaO, CaCO 3 and trace amounts of t-ZrO 2 . For the highest laser power of 1200 W, the CaO, CaCO 3 and TTCP contents of the CL increase significantly. The t-ZrO 2 content also increases. However, that of m-ZrO 2 and CaZrO 3 reduces. In general, the tendency to form composite phases increases as the laser power increases.

4.
For all of the specimens, the TL has a greater hardness than the CL. Moreover, the CL hardness is around 2~3 times higher than that of the Ti6Al4V substrate. As the laser power increases from 400 W to 800 W, the CL hardness increases due to a microstructural refinement and densification effect. However, under the highest laser power of 1200 W, the hardness reduces significantly due to the formation of CaO and CaCO 3 phases with relatively low hardness.