Introduction

Hydroxyapatite (Ca10(PO4)6(OH)2) (HA) is a phase of calcium phosphate, and belongs to a family of materials called the calcium apatites (Ref 1, 2). Calcium phosphate materials differ in their ratio of calcium to phosphorous, where stochiometric HA has a calcium to phosphorous ratio of 1.67 (Ref 3, 4). The names, chemical formulas, and abbreviations of the various compounds discussed in this research are shown in Table 1. HA is the most stable calcium phosphate phase at ambient temperatures. It has been widely used in the biomedical industry as it has similar properties to the fundamental component of the inorganic part of natural bone (Ref 2, 3, 5). HA is also used in other applications, such as catalysis (Ref 6, 7), fertilizers (Ref 7, 8), and the pharmaceutical industry (Ref 9, 10).

Table 1 Names, chemical formulas, and abbreviations of the different compounds discussed in this study
Full size table

There are a number of methods through which HA may be synthesized, such as wet precipitation (Ref 5, 11,12,13), hydrothermal synthesis (Ref 5, 12, 13), and solid state chemistry (Ref 12, 13). To synthesize HA, a calcium source and phosphate source are needed. HA is a brittle material, with restricted mechanical properties, and as a result is often deposited on substrates, for use in industry (Ref 11, 14). A common method used to deposit HA onto a substrate is plasma spraying (Ref 11). Plasma spraying provides a fast, controlled and economically advantageous method of coating materials onto almost any substrate (Ref 3). However, the extremely high plasma spray temperature leads to the thermal degradation of HA, producing coatings which contain not only HA, but various degradation phases (Ref 15). At high temperatures HA first undergoes dehydroxylation into oxyhydroxyapatite (OHAp) (Ca10(PO4)6(OH)2−2xOx□x) and oxyapatite (OA) (Ca10(PO4)6Ox□x) between 800 and 900 °C, (Eq 1 and 2) (Ref 15,16,17,18,19,20,21,22).

$$ \begin{aligned} & {\text{Ca}}_{10} \left( {{\text{PO}}_{4} } \right)_{6} \left( {{\text{OH}}} \right)_{2} \leftrightarrow {\text{Ca}}_{10} \left( {{\text{PO}}_{4} } \right)_{6} \left( {{\text{OH}}} \right)_{2 - 2x} {\text{O}}_{x\square x} + x{\text{H}}_{2} {\text{O}} \\ & {\text{HA}}\quad \quad \quad \quad \quad \quad \quad \;\;{\text{Oxyhydroxyapatite}} \\ \end{aligned} $$
(1)
$$ \begin{aligned} & {\text{Ca}}_{10} \left( {{\text{PO}}_{4} } \right)_{6} \left( {{\text{OH}}} \right)_{2 - 2x} {\text{O}}_{x\square x} \leftrightarrow {\text{Ca}}_{10} \left( {{\text{PO}}_{4} } \right)_{6} {\text{O}}_{x\square x} + \left( {1 - x} \right){\text{H}}_{2} {\text{O}} \\ & {\text{Oxyhydroxyapatite}}\quad \quad \quad \quad \;\;{\text{Oxyapatite}} \\ \end{aligned} $$
(2)

The □ denotes a lattice vacancy, as during the dehydroxylation of HA, the OH− groups are removed (Ref 23). The 2 OH− groups combine to form a water molecule, leaving a peroxy ion (O2−) behind in the lattice (Ref 18, 24). In the next step of HA degradation OA degrades causing the apatite structure to break down. As the structure is destroyed, P2O5 evaporates leading to the production of tricalcium phosphate (TCP) (Ca3(PO4)2) and tetracalcium phosphate (TTCP) (Ca4(PO4)2O), between 1050 and 1400 °C, (Eq 3) (Ref 15,16,17, 19, 20, 24,25,26).

$$ \begin{aligned} & {\text{Ca}}_{10} \left( {{\text{PO}}_{4} } \right)_{6} {\text{O}}_{x\square x} \leftrightarrow {\text{Ca}}_{4} \left( {{\text{PO}}_{4} } \right)_{2} {\text{O}} + 2{\text{Ca}}_{3} \left( {{\text{PO}}_{4} } \right)_{2} \\ & {\text{Oxyapatite}}\quad \quad \quad \quad {\text{TTCP}}\quad \quad \quad \;\;{\text{TCP}} \\ \end{aligned} $$
(3)

In the final steps of HA thermal degradation, TCP and TTCP degrade into calcium oxide + a liquid melt at 1730 and 1630 °C, respectively, (Eq 4 and 5) (Ref 19, 20, 26). The composition of the melt phase shifts to CaO-rich phases as P2O5 evaporates (Ref 15, 27).

$$ \begin{array}{*{20}l} {{\text{Ca}}_{3} \left( {{\text{PO}}_{4} } \right)_{2} } \hfill & \to \hfill & {3{\text{CaO}}} \hfill & + \hfill & {{\text{P}}_{2} {\text{O}}_{5} } \hfill \\ {{\text{TCP}}} \hfill & {} \hfill & {\text{Calcium oxide}} \hfill & {} \hfill & {\text{Phosphorus pentoxide}} \hfill \\ \end{array} $$
(4)
$$ \begin{array}{*{20}l} {{\text{Ca}}_{4} \left( {{\text{PO}}_{4} } \right)_{2} {\text{O}}} \hfill & \to \hfill & {4{\text{CaO}}} \hfill & + \hfill & {{\text{P}}_{2} {\text{O}}_{5} } \hfill \\ {{\text{TTCP}}} \hfill & {} \hfill & {\text{Calcium oxide}} \hfill & {} \hfill & {\text{Phosphorus pentoxide}} \hfill \\ \end{array} $$
(5)

OHAp is the product of partial dehydroxylation of HA, whereas OA is the product of complete dehydroxylation of HA (Ref 15, 28). Liao et al. speculated that both OA and OHAp are present during HA degradation. OA is stable between 800 and 1050 °C, but can absorb water at 600 °C in a vacuum, due to its highly reactive nature, which links to the idea that it exists with some hydroxyl groups present in the structure (Ref 15, 26, 28,29,30,31,32). As a result, dehydroxylated HA is often described as a solid solution of stoichiometric HA and OA, which is referred to as OHAp (Ref 11, 33, 34). It is difficult to detect oxyapatite through conventional characterization techniques (Ref 15, 26, 35). For example, it is challenging to detect OA through XRD, due to the minimal difference between the crystal structures of HA and OA (Ref 15, 35). The c-axis is slightly larger, and the a-axis slightly smaller in OA (Ref 26, 36, 37). The difference in the c-axis, causes the (002) plane to shift very slightly to smaller diffraction angles (Ref 15, 25, 35, 36). As a result, very precise measurements are required using high resolution techniques, such as synchrotron or neutron diffraction to detect OA. Additionally, OA readily converts back to HA in the presence of moist air during cooling, to lower its free energy (Ref 15, 21, 26, 36, 38). Therefore, HA plasma spray coatings may often contain OA at high temperatures; however, upon cooling to room temperature, the OA transforms into HA.

In plasma spraying, the powder particle temperature and velocity prior to the point of impact determine the production of a dense coating made up of the desired composition (Ref 39, 40). These particle parameters control the heating/melting and spreading of the powder particles (Ref 40,41,42). They are controlled indirectly through changes in the plasma stream temperature and velocity which are affected by variables such as the plasma source, gas type and gas flow rate (Ref 39, 40). The temperature of the powder heating region is related to the gas used to create the plasma, as different gases have different enthalpies. The temperature in turn affects the melting/degradation of the powder particle. The velocity of the powder particle is also influenced by the plasma gas, as different gases have different densities (Ref 39). In addition, the spray distance influences the amount of time the powder particles are exposed to the plasma, which affects the heating and melting of the powder.

Currently, the production of a HA coating is a multi-step process, in which first the HA is synthesized, and often spray dried to produce spherical, flowable particles which are ideal for plasma spraying (Ref 39, 40). However, as the target industry is medical applications, the demand on powder purity and phase composition is extremely high. The HA is then plasma sprayed to produce the coating. As a result, the production of a HA coating is a time consuming and expensive process. Therefore, alternative methods have been used to form HA by combining the synthesis and coating method.

These alternative methods are suspension plasma spray (SPS) and solution precursor plasma spray (SPPS) (Ref 43,44,45). SPS involves delivering a suspension of the coating material into the plasma. The solid materials are often suspended in either water or alcohol, or a mixture of both (Ref 44). The suspension also contains dispersants and stabilizing agents to keep the solid in suspension. Whereas SPPS involves an aqueous or non-aqueous solution containing the reagent materials required to make the final coating material, being fed into the spray gun, rather than a solid powder (Ref 43, 44, 46). There are five main steps involved in the SPPS process (Ref 43). Firstly, the droplets break up, followed by the evaporation of the solvent. Next particle pyrolysis occurs, followed by melting. Lastly, the molten particles are deposited on the substrate (Ref 43, 44). In the liquid droplets, as the liquid vaporizes, the solute precipitates (Ref 47). This is followed by chemical reactions of the reagent salts that lead to the synthesis of the product coating,

Hydroxyapatite coatings have been successfully synthesized by researchers using SPPS (Ref 44, 48,49,50,51,52). Different researchers have used various calcium and phosphate precursor reagents dissolved in different liquid mediums, to form HA coatings. In some cases, these methods have produced coatings which contain HA, and other calcium phosphate phases such as TCP and TTCP, as well as CaO. The formation of these other phases are often explained due to the degradation of HA. To our knowledge this is the closest research has come in attempting to synthesize HA using thermal spray.

Although both SPPS and SPS are promising techniques to synthesize HA they have some disadvantages. SPPS uses a lower solids loading compared to generic plasma spray, due to the saturation limit of the solution and the solutions viscosity limit which leads to a lower deposition rate compared to generic plasma spray (Ref 43). Another disadvantage is that the SPPS process requires a high molarity chemical precursor which has a low enough viscosity to allow for injection, which can be challenging to find. Whereas a disadvantage of SPS, is that it requires the powder to be suspended, and every powder even those of the same composition have different requirements to form a stable suspension.

An alternative method to fabricate cheaper HA coatings for non-medical applications is to spray the reagent powders together using plasma spray. This method has been used to produced FeAl coatings by Zhu et al. they successfully ball milled and spray dried Fe and Al precursor powders together, and then plasma sprayed the powders to fabricate FeAl coatings (Ref 53). To our knowledge, this method has not yet been used to synthesize HA.

This paper is a proof of concept investigation into the inflight reaction of calcium carbonate (CaCO3) and brushite (CaHPO4·2H2O) in spray dried particles to form HA coatings. The main aim was to determine if precursor reagents which are commonly used in the ambient temperature synthesis of HA, could be used to synthesize HA using plasma spray. A second aim was to investigate the effect of plasma spray parameters on the amount of HA produced in the coatings compared to other phases. The effect of the plasma parameters on the formation of HA was discussed to guide in the follow up research which will focus more specifically on the reaction mechanisms.

Experimental Procedure

Powder Preparation

Brushite (Sigma-Aldrich, New Zealand), with a d90 of 38 µm was wet milled for 1.5 h at 33 Hz, using an 01HD attritor (Union Process Inc., USA), to decrease the particle size below 10 micron. In each batch, 30 wt.% brushite (125 g) was wet milled in water (291 g), using 6.4 mm diameter stainless steel balls. Calcium carbonate (99.5% metal basis) (abcr, Germany) with a d50 of 1 µm was used in the as-supplied condition.

Feed suspensions for spray drying were prepared by first mixing the dispersant, 2 wt.% poly(acrylic)acid (PAA) (Sigma-Aldrich, New Zealand), with ethanol, followed by vigorous stirring and heating to 50 °C until all the PAA dissolved. The binder, 2 wt.% Poly(vinyl) butyral (PVB) (Sigma-Aldrich, New Zealand) was then added and the solution was stirred and heated until the PVB dissolved, producing a viscous, colorless solution. This solution was then added to a 39 wt.% powder mixture composed of 15.8 wt.% calcium carbonate and 23.6 wt.% milled Brushite powder. More calcium carbonate was added than the stochiometric amount (10.9 wt.%) based on the chemical equation, to reduce the chances of forming calcium deficient HA. Assuming the feedstock components fully reacted to HA and CaO (via Eq 6 and 7), the coating should consist of 25.8 mol.% HA and 74.2 mol.% CaO.

$$ \begin{aligned} & 4{\text{CaCO}}_{3} + 6{\text{CaHPO}}_{4} \cdot 2{\text{H}}_{2} {\text{O}} \to {\text{Ca}}_{10} \left( {{\text{PO}}_{4} } \right)_{6} \left( {{\text{OH}}} \right)_{2}\\ &\quad+ 14{\text{H}}_{2} {\text{O}} + 4{\text{CO}}_{2} {\text{Calcium carbonate}} + {\text{Brushite}} \to {\text{HA}} \\ \end{aligned} $$
(6)
$$ \begin{aligned} & {\text{CaCO}}_{3} \to {\text{CaO}} + {\text{CO}}_{2} \\ & {\text{Calcium carbonate}} \to {\text{Calcium oxide}} \\ \end{aligned} $$
(7)

Spray Drying

Spray drying was conducted to combine calcium carbonate and brushite into a composite powder. A Büchi Mini Spray Dryer B290 (Büchi Labortechnik AG, Flawil, Switzerland) was used in the closed-loop mode with an inert loop B295 accessory, in order to use ethanol as the slurry fluid (Ref 40). Spray drying was performed in co-current mode, using nitrogen (550 kPa) as the drying gas, with an external two-fluid nozzle to atomize the suspension. The inlet gas temperature was fixed at 88 °C and the aspirator was set at the maximum capacity (100%). A gas flow rate of 667 L/h and a pump speed of 60% was used. The feed suspension was continuously stirred during the spray drying experiment, to avoid sedimentation of the solid particles.

Plasma Spray

The calcium carbonate/brushite composite spray dried powders were plasma sprayed using a range of different conditions, Table 2. The different conditions were chosen to investigate their effect on the phases produced in the coatings. Different plasma gases, such as argon, argon–hydrogen, nitrogen–hydrogen, and argon–helium were chosen to alter the temperature of the plasma, (S1-50Ar to S12-50Ar:2.5H2). Different spray distances were used to assess how the particle residence time in the plasma influenced the phases in the coating. Two conditions were chosen with varying amounts of helium, to alter the particle velocity within a low power condition, (S9-40Ar-18He and S10-50Ar:10He). The total gas flow was used as a first approximation to explore the effect of particle velocity (S2-40Ar:6.4H2 to S8-33N2:6.4H2). Multiple argon–hydrogen conditions were chosen to examine the effect of hydrogen content, as well as the variation of particle velocity in a low power condition (S2-40Ar:6.4H2 to S8-33N2:6.4H2, S11-50Ar:4.5H2 and S12-50Ar:2.5H2). Lastly, the effect of powder feed rate on the concentration of phases produced in the coating was examined, (S2-40Ar:6.4H2, S5-50Ar:11H2, S6-73Ar:6.4H2 and S8-33N2:6.4H2 vs S13-40Ar:6.4H2, S14-50Ar:11H2, S15-33N2:6.4H2, and S16-73Ar:6.4H2).

Table 2 A summary of plasma spray parameters used, where all parameters were sprayed at 60, 80 and 100 mm
Full size table

Two different plasma guns were utilized. The SG-100 plasma gun (Praxair Surface Technologies, USA) with a “subsonic” anode nozzle was used in some conditions, and the 3 MB gun (Sulzer Metco, USA) was used for the remainder of the conditions. The plasma guns were mounted on a 6-axis robot, and sprayed samples in a raster pattern. Substrates were composed of (100 mm × 100 mm × 3 mm) mild steel plates which were degreased and grit-blasted prior to spraying. The substrates were mounted on a backing plate at distances of 60, 80, and 100 mm for each trial.

Powder Characterisation

X-Ray Diffraction (XRD)

Phase analysis of the calcium carbonate/ brushite composite spray dried powder and plasma spray coatings was conducted using XRD (Rigaku Ultima-IV, Japan). XRD measurements were collected using monochromated Cu Kα1 radiation (λ = 1.5418 Å), over a 2θ range of 10-80°, at a step size of 0.02°. Samples were rotated to minimize preferred orientation effects.

Qualitative phase analysis was carried out using PDXL, where Rigaku’s ICDD research library was utilized to identify the different phases present in each sample. Semi-quantitative phase analysis was performed using the software package FullProf to carry out Rietveld refinement (Ref 17, 54, 55). Although the coating samples were semi-quantitatively refined, it should be noted that these samples were composed of a mixture of complicated crystalline, semicrystalline, and potentially even amorphous phases, which made the quantitative analysis challenging. As a result, the degree to which the reference patterns were fitted to the experimental data was not expected to produce “perfect refined” values. However, the aim of this semi-quantitative analysis was to achieve an approximation of the relative amounts of each phase in each coating, to gain an understanding of the range of variation of phases in each coating in relation to the plasma spray parameters. This was the sole method by which this information could be assessed.

Scanning Electron Microscopy (SEM)

The topography and cross-sectional analysis of plasma spray coatings was conducted using scanning electron microscopy (SEM), (Quanta 200, FEI-ESEM, USA). Cross-sectional samples were prepared by cutting sections of the coating using a dry metal bandsaw and mounting under vacuum in epoxy, followed by grinding and polishing using conventional metallographic techniques. In this case only the water-soluble phases at the exposed cross-sectional surface were at risk of dissolution. This risk was minimized by using ethanol-based lubricant for the fine grinding and polishing steps. All samples were sputter coated with platinum for 1 min, (Quorum Q150RS sputter coater) to reduce charging effects. Back-scatter electron imaging (BSE) and energy dispersive x-ray spectroscopy (EDS) were used to image and qualitatively analyze the different calcium phosphate phases.

Particle Size Analysis

Particle size analysis was carried out on the reagent spray dried powder using a laser diffraction-based particle size analyzer (MasterSizer 2000, Malvern Instrument Ltd, UK) using water as the carrier fluid. Samples were prepared in accordance to Chahal et al. (Ref 40).

Results

Powder Characterisation

The spray dried calcium carbonate and brushite composite powder consisted of spherical particles, composed of small needle-like calcium carbonate crystallites and larger cuboid-like brushite crystallites (Fig. 1a and b). XRD identified only phases of calcium carbonate (PDF: 01-072-1650) and brushite (PDF: 01-072-0713) (Fig. 1c). The Rietveld pattern fitting indicated a composition of 59.0 mol.% calcium carbonate and 41 mol.% brushite. The spray dried composite powder had a d10, d50, and d90 particle size of 9, 23, and 149 μm, respectively.

Fig. 1
figure 1

SEM micrographs of (a) a spray dried calcium carbonate and brushite composite particle and (b) a magnified image of the spray dried particle. Note: the yellow square and red circle are highlighting large cuboid brushite particles and small needle-like calcium carbonate particles within the spray dried particle, respectively. (c) is an XRD pattern of the spray dried calcium carbonate and brushite composite powder

Full size image

Coating Analysis

Phase Analysis

XRD showed that the majority of the coatings contained a mixture of 3 phases: HA (PDF: 01-076-0694), TTCP (PDF:01-0701379), and CaO (PDF: 01-077-2010), (Fig. 2). However, it is noted that there is the possibility of the production of OA instead of HA. However, the crystal structure of OA and HA are almost identical, and since OA readily reacts with water vapour to form HA upon cooling, it will be assumed any detection of the HA crystal structure through XRD is HA. The XRD reference patterns of HA and TTCP overlap in many areas, as a result the analysis of the coating patterns was very challenging. Therefore, specific planes were used to identify the presence of HA and TTCP. Planes at ~ 31.75 2θ, ~ 32.14 2θ, and ~ 32.90 2θ indicated that HA was present in the pattern. Whereas planes at ~ 29.25 2θ, and 29.80 2θ revealed the presence of TTCP in the pattern. Different TTCP and HA reference patterns in the database fit different coating patterns, where the Ca to P ratio was slightly different. However, although during the initial process of peak identification, multiple reference database patterns were used to fit the HA and TTCP in the different coatings, to keep the quantitative analysis consistent, only 1 reference pattern for TTCP and HA were used for the Rietveld fitting.

Fig. 2
figure 2

Representative XRD patterns of plasma spray coatings produced from spray dried calcium carbonate and brushite composite powder. Note: S6-73Ar:6.4H2 and S11-50Ar:4.5H2 contain a large and small amount of HA, respectively

Full size image

The Rietveld quantification performed on each sample showed that each coating contained various amounts of the different phases, Tables 3 and 4. Some coatings were found to contain more HA than others, for example S6-73Ar:6.4H2 compared to S11-50Ar:4.5H2, (Fig. 2). Five plasma spray conditions were found to form an additional phase, β-calcium pyrophosphate (β-pyro) (PDF: 01-071-2123), at 60 and 80 mm, (Tables 3 and 4).

Table 3 Summary of the powder feed rate and semi-quantitative phase analysis of each coating produced by the 3 MB gun, in mol.%
Full size table
Table 4 Summary of the semi-quantitative phase analysis of each coating produced by the SG100 gun, in mol.%
Full size table

The different plasma spray conditions investigated using the 3 MB gun are shown in Fig. 3, (Table 3). Across spray distance, HA and TTCP had a slight decrease in molar concentration, whereas CaO had a slight increase in molar concentration. All coatings produced at 60 mm contained the highest amount of HA, except S1-50Ar (red, square symbol), produced the highest HA at a spray distance of 80 mm. Comparing the results in terms of total gas flow, the higher gas flow rate plasma conditions (S3-40Ar:11H2 to S6-73Ar:6.4H2) (green, circle symbol) produced the highest amount of HA and TTCP, but the lowest amount of CaO, (Fig. 3, Table 3). This is further confirmed by comparing a high (S6-73Ar:6.4H2) and low (S7-35Ar:6.4H2) gas flow in a low power generated plasma.

Fig. 3
figure 3

Semi-quantitative mol.% of each phase in S1-50Ar to S8-33N2:6.4H2 coatings produced using the 3 MB gun where (a) is HA, (b) is TTCP, and (c) is CaO. Note: The Argon condition has been colored red and has a square symbol (S1-50Ar), the Argon–Hydrogen trials have a circle symbol (S2-40Ar:6.4H2 to S7-35Ar:6.4H2), and the Hydrogen–Nitrogen trial has been colored purple and has a triangle symbol (S8-33N2:6.4H2). Additionally, the high and low total gas flow rates within the Argon–Hydrogen trials have been colored green (S3-40Ar:11H2 to S6-73Ar:6.4H2) and blue (S2-40Ar:6.4H2 and S7-35Ar:6.4H2), respectively

Full size image

Three different plasma gases were used on the 3 MB gun, argon (S1-50Ar), argon–hydrogen (S2-40Ar:6.4H2 to S7-35Ar:6.4H2), and nitrogen–hydrogen (S8-33N2:6.4H2) (purple, triangle symbol). These results show that all the plasma spray conditions used on the 3 MB gun produced HA, however the coatings contained mainly CaO, followed by TTCP, and lastly HA. The addition of hydrogen to the plasma (comparing S2-40Ar:6.4H2 to S7-35Ar:6.4H2 with S1-50Ar) (all circle symbol graphs versus the square symbol graph) had a negative effect on the production of HA, as S1-50Ar seems to have a larger HA content than all other trials except S6-73Ar:6.4H2, (Fig. 3 and Table 3). However, the TTCP and CaO content were increased due to the addition of hydrogen. The hydrogen content was also investigated, by comparing S2-40Ar:6.4H2 to S3-40Ar:11H2 and S4-50Ar:6.4H2 to S5-50Ar:11H2. S2-40Ar:6.4H2, S4-50Ar:6.4H2 and S3-40Ar:11H2, S5-50Ar:11H2 are low and high hydrogen content conditions, respectively. These conditions show the opposite result, where the addition of hydrogen has increased the HA and TTCP content but decreased the CaO content. The nitrogen–hydrogen plasma spray condition (S8-33N2:6.4H2) produced a similar amount of HA compared to the argon–hydrogen condition (S7-35Ar:6.4H2), (Fig. 3). Both conditions produced a low amount of HA. The highest HA content produced by the 3 MB gun, was in the argon–hydrogen plasma condition, with the highest gas flow rate and powder feed rate (S16-73Ar:6.4H2), at a spray distance of 60 mm.

Figure 4 shows the mol.% of each phase from coatings produced using the SG100 gun (Table 4). In the argon-helium conditions (square symbol graphs), the condition with the higher amount of helium (S9-40Ar-18He) (blue, square symbol graph) produced the highest HA and TTCP content, but lowest CaO content, (Fig. 4). In terms of spray distance, HA content decreased, TTCP was consistent, and CaO increased with increasing spray distance. All coatings produced at 60 mm contained the highest amount of HA, except S11-50Ar:4.5H2, which produced the highest HA at a spray distance of 100 mm.

Fig. 4
figure 4

Semi-quantitative mol.% of each phase in S9-40Ar-18He to S12-50Ar:2.5H2 coatings produced using the SG100 gun, where (a) is HA, (b) is TTCP, and (c) is CaO. Note: the helium and hydrogen trials have been illustrated by squares and circles, respectively. High and low total gas flow rates have been colored blue and red, respectively

Full size image

In the argon–hydrogen conditions (circle symbol graphs), both hydrogen flowrates produced similar HA contents (Fig. 4). However, the lower hydrogen condition (red, circle symbol graph) produced a higher amount of TTCP, and lower amount of CaO. Overall, the helium trials produced a higher amount of HA, and TTCP, but lower amount of CaO compared to the hydrogen trials (Table 4). The highest HA content was produced by the SG100 gun, was in the argon-helium plasma condition with the largest helium content (S9-40Ar-18He) at a spray distance of 60 mm, (Tables 3 and 4).

Four trials were repeated at a higher powder feed rate of 15 g/min, (Fig. 5 and Table 3). These trials were trial S2-40Ar:6.4H2, S5-50Ar:11H2, S6-73Ar:6.4H2, and S8-33N2:6.4H2, and the new trials were S13-40Ar:6.4H2, S14-50Ar:11H2, S16-73Ar:6.4H2, and S15-33N2:6.4H2, respectively (Fig. 5). Increasing the powder feed rate produced a higher amount of HA and TTCP but a decreased amount of CaO for each parameter setting (Fig. 5 and Table 3).

Fig. 5
figure 5

Comparison of semi-quantitative mol.% of each phase in samples with different powder feed rates where (a) is HA, (b) is TTCP, and (c) is CaO

Full size image

The relationship between plasma power and the amount of HA, TTCP, and CaO produced in each coating was also investigated (Fig. 6). As the power was increased, the amount of HA, and TTCP increased, whereas the amount of CaO decreased. These trends occurred across the various plasma spray parameters, and spray distance.

Fig. 6
figure 6

Graphs showing the relationship between plasma power and the semi-quantitative mol.% of each phase in each of the coatings produced by the range of plasma spray parameters at a spray distance of 60, 80, and 100 mm, where (a) HA, (b) TTCP, and (c) CaO

Full size image

Microstructure Analysis

All coatings appeared to be composed of molten splats (Fig. 7). There also appeared to be a lot of cracking in the coatings. The large cracks may be due to sample preparation (Fig. 7a). The shrinkage cracks observed within the individual splat, rather than between splats, indicates that the splats are well bonded (Fig. 7b). The macroscopic porosity appeared to result from pull-out of splat segments during metallographic preparation. This was particularly evident in pockets with a high concentration of small spherical particles assumed to be poorly adhered dust particles (Fig. 8). SEM images showed that the coatings contained many spherical particles which we are referring to as “dust” particles (Fig. 8b). An alternative mechanism is that the small spherical particles were produced when the molten splats hit the substrate surface. These spherical particles impact the surface in a solid form and so do not deform and adhere to the coating but are only entrapped by the deposition of subsequent molten droplets, because they are not well adhered to the coating, they are prone to be pulled out during sample preparation. The spherical porosity within the splats is postulated to be gas bubbles, possibly formed from gases dissolved in the liquid particles inflight, on encapsulation of a gaseous reaction product. These porosity features were observed within all the coating cross sections.

Fig. 7
figure 7

SEM images of representative coatings, where (a) is showing large cracks within the coating, and (b) is showing shrinkage cracks within a splat. Note: cracks have been highlighted by a red rectangular box. (c) is a graph showing the EDS Ca/P mol ratio from the cross-sectional general areas of representative coatings and the reagent powder

Full size image
Fig. 8
figure 8

SEM micrographs of representative coatings illustrating (a) porosity within a cross-sectional coating (highlighted by yellow squares), and (b) “dust” particles in the topographical coating (highlighted by red circles)

Full size image

The interface between the substrate and the coating was continuous in all the samples. All the coatings appear to be well adhered to the substrate. The Ca/P mol ratio was calculated using the EDS data collected on a general cross-sectional area of each sample analyzed (Fig. 7c). The Ca/P ratio reflects the composition of the splats but does not necessarily reflect the Ca/P mol ratio of the HA within the coating, especially given the low HA concentrations measured through XRD. In this way, the Ca/P ratio is distinctly different from that traditionally presented for coatings using a pure HA feedstock powder.

The Ca/P mol ratio data showed that the Ca/P ratio for S6-73Ar:6.4H2-60 mm and S9-40Ar-18He-60 mm were similar giving a Ca/P ratio of 2 (Fig. 7c). However, coatings produced by S7-35Ar:6.4H2 and S11-50Ar:4.5H2 had a higher Ca/P mol ratio of 2.2 and 2.3, respectively. This indicates that these latter coatings contained a larger mol.% of calcium compared to other coatings. In general, some coatings have a loss in calcium and others have a gain in calcium, when comparing the Ca/P ratios of the coatings to the reagent powder. In terms of the Ca/P mol ratio across spray distance, all the ratios were the same for all trials analyzed, indicating, that the spray distance had a limited effect on the different phases present in the coating.

There are very subtle differences in the greyscale within each splat because the backscatter coefficient values for the different phases are very close together (Fig. 9). The backscatter coefficient is related to the image produced by the SEM, and it is the average weight percent of different atoms within a particular material. In general, atoms which have a higher molecular weight have a higher backscatter coefficient. Since the backscatter coefficient values are close together, it is not possible to define the individual phases. This indicates that the backscatter electron contrast is low, which is what causes the greyscale to appear homogeneous. The minimal variation in greyscale also implies that the phases were well molten and mixed inflight during the plasma spray process, forming splats with a complex composition.

Fig. 9
figure 9

SEM micrograph of a select cross-sectional coating illustrating the difference in greyscale throughout the coating (a). (b) is a graph showing the backscatter coefficient for different phases, where CaO is calcium oxide, TTCP is tetracalcium phosphate, TCP is tricalcium phosphate, pyro is calcium pyrophosphate, CaCO3 is calcium carbonate, and HA is hydroxyapatite

Full size image

Discussion

HA was successfully produced through the inflight reactions of calcium carbonate and brushite during the plasma spray process. To the best of the authors knowledge, this is the first time this reaction pathway has been published. The coatings were composed of well-bonded molten splats which were a homogenous mix of HA, TTCP, and CaO. The reaction mechanisms occurring in this research are considered to be different compared to those observed in SPPS. This is because in SPPS the reagents are in a liquid medium to begin with, whereas in this research, the reagents are in the form of a solid powder. Therefore, the reagent mixtures are likely to be reacting under different circumstances. The overall reaction describing the formation of HA is hypothesized to occur by Eq 6. However, the reaction mechanism is postulated to occur via a series of intermediate reactions. It is hypothesized that, the precursor powders thermally decompose, where calcium carbonate degrades into CaO and brushite degrades into monetite (CaHPO4), which further decomposes into calcium pyrophosphate (Ca2P2O7), (Eq 7-9) (Ref 56,57,58).

$$ \begin{array}{*{20}l} {{\text{CaHPO}}_{4} \cdot 2{\text{H}}_{2} {\text{O}}} \hfill & \to \hfill & {{\text{CaHPO}}_{4} } \hfill & + \hfill & {2{\text{H}}_{2} {\text{O}}} \hfill \\ {{\text{Brushite}}} \hfill & \to \hfill & {{\text{Monetite}}} \hfill & {} \hfill & {} \hfill \\ \end{array} $$
(8)
$$ \begin{array}{*{20}l} {{\text{CaHPO}}_{4} } \hfill & \to \hfill & {{\text{Ca}}_{2} {\text{P}}_{2} {\text{O}}_{7} } \hfill & + \hfill & {{\text{H}}_{2} {\text{O}}} \hfill \\ {{\text{Monetite}}} \hfill & \to \hfill & {\text{Calcium pyrophosphate}} \hfill & {} \hfill & {} \hfill \\ \end{array} $$
(9)

As a result, the degradation products of the precursor powders could also be reacting to form HA (Eq 10-11) (Ref 59).

$$ \begin{array}{*{20}l} {4{\text{CaO}} + 6{\text{CaHPO}}_{4} \cdot 2{\text{H}}_{2} {\text{O}}} \hfill & \to \hfill & {{\text{Ca}}_{10} \left( {{\text{PO}}_{4} } \right)_{6} \left( {{\text{OH}}} \right)_{2} + 14{\text{H}}_{2} {\text{O}}} \hfill \\ {{\text{Calcium oxide}} + {\text{Brushite}}} \hfill & \to \hfill & {{\text{HA}}} \hfill \\ \end{array} $$
(10)
$$ \begin{array}{*{20}l} {4{\text{CaO}} + 6{\text{CaHPO}}_{4} } \hfill & \to \hfill & {{\text{Ca}}_{10} \left( {{\text{PO}}_{4} } \right)_{6} \left( {{\text{OH}}} \right)_{2} + 2{\text{H}}_{2} {\text{O}}} \hfill \\ {{\text{Calcium oxide}} + {\text{Monetite}}} \hfill & \to \hfill & {{\text{HA}}} \hfill \\ \end{array} $$
(11)

The presence of TTCP and CaO could be due to a number of postulated mechanisms:

  1. 1.

    Firstly, they could be degradation products of the HA formed inflight, (Eq 1-5).

  2. 2.

    Alternatively, they could have been formed from or left over from unreacted reagents or the degradation products of the reagents (CaO, monetite, and calcium pyrophosphate), and so were competing with the formation of HA, (Eq 5, 7 and 12-13) (Ref 60).

    $$ \begin{array}{*{20}l} {2{\mathrm{CaHPO}}_{4} + 2{\mathrm{CaCO}}_{3} } \hfill & \to \hfill & {{\mathrm{Ca}}_{4} \left( {{\mathrm{PO}}_{4} } \right)_{2} {\mathrm{O}}} \\ &\quad+ 2{\mathrm{CO}}_{2} + {\mathrm{H}}_{2} {\mathrm{O}} \hfill \\ {{\mathrm{Monetite}} + {\text{Calcium carbonate}}} \hfill & \to \hfill & {{\mathrm{TTCP}}} \hfill \\ \end{array} $$
    (12)
    $$ \begin{array}{*{20}l} {2{\mathrm{CaHPO}}_{4} + 2{\mathrm{CaO}}} \hfill & \to \hfill & {{\mathrm{Ca}}_{4} \left( {{\mathrm{PO}}_{4} } \right)_{2} {\mathrm{O}} + {\mathrm{H}}_{2} {\mathrm{O}}} \hfill \\ {{\mathrm{Monetite}} + {\text{Calcium oxide}}} \hfill & \to \hfill & {{\mathrm{TTCP}}} \hfill \\ \end{array} $$
    (13)
  3. 3.

    They may also be a combination of formation and degradation products, where for example some of the TTCP was formed from the reaction of certain reagents, and the rest of the TTCP was a degradation product of HA, (Eq 1-5 and 12-13).

To address this uncertainty, future trials are planned to spray combinations of the feedstock materials and their respective degradation phases to establish the reaction pathway to HA formation.

The different plasma spray parameters had different effects on the quantity of each phase produced in the coating. Particle velocity appeared to have the largest effect on the amount of HA produced. It was found that the plasma spray parameters which provided the powder particles with the highest velocity produced coatings which contained the highest amount of HA. As the particle velocity was increased, the particle was exposed to the hot plasma for a reduced time, leading to less degradation, and an increase in HA content, according to postulated mechanism 1. In particular, increases in the total gas flow rate, which increased the particle velocity, led to an increase in the amount of HA.

Increases in the plasma power led to increases in the concentration of the desired phases HA and TTCP, and a decrease in CaO. This occurred across all combinations of secondary gases and gas flows and remained consistent with spray distance. The plasma power is determined by the current setting, secondary gas composition and secondary gas flowrate, all of which have complex effects on the plasma temperature, velocity, conductivity, and density (Ref 61). This confounding of parameter effects makes it difficult to attribute the increase in HA content to any one particular variable. However, studies on the effect of plasma power on coating quality have generically reported an increase in particle melting with increasing plasma power (Ref 62,63,64,65). If it assumed that increasing plasma power generated increased particle temperatures, then postulated mechanism 1 would predict a decrease in HA content with increasing power due to a greater rate of phase degradation. The fact that the opposite trend was observed potentially supports the second postulated mechanism, or a complex combination of both mechanisms, i.e., postulated mechanism 3, in terms of observing an increase in TTCP. However, the increase in HA with increasing power is difficult to explain based on the current formation mechanisms, therefore further research is required to understand this relationship.

Different plasma gases were used to provide different plasma temperatures, and velocities. The argon plasma condition produced more HA than the argon–hydrogen plasma. This is postulated to be due to the following reasons. Firstly, it was a low power generated plasma condition, which led to decreased HA degradation. Additionally, the argon plasma condition had a middle-range total gas flow rate, which increased the velocity of the particles. Lastly, the addition of hydrogen to the plasma increased the enthalpy and thermal conductivity of the plasma, which in turn increased the plasma temperature. This increase in plasma temperature meant that the particles were exposed to higher degrees of heating compared to an argon plasma, leading to increased degradation of HA. All these effects combined, aided the argon plasma condition in producing a large amount of HA compared to the argon–hydrogen plasma conditions. However, the argon-only trial (S1-50Ar) is postulated to be an anomaly based on what is observed through the other plasma spray trials. As a result, further investigation is required.

However, there was one exception, S6-73Ar:6.4H2, which produced the highest HA, even though it contained hydrogen. This is because it had the highest gas flow rate, therefore highest particle velocity. This indicated that the hydrogen flow rate effect was secondary to the total gas flow rate. This is further confirmed when comparing trials based on hydrogen content. The higher hydrogen content conditions were also the higher total gas flow conditions and were found to produce a higher amount of HA. The total gas flow also impacted the TTCP content in a similar way. However, in terms of CaO content, the opposite effect was observed, whereby a higher total gas flow produced a lower amount of CaO, due to decreased HA degradation.

The nitrogen–hydrogen and argon–helium plasma conditions were also influenced by the effect of velocity. The nitrogen–hydrogen trial produced a similar amount of HA compared to the argon–hydrogen trial, although the nitrogen–hydrogen plasma had an increased temperature, and so was expected to produce less HA due to increased degradation. Both trials produced low amounts of HA. This may be because they had low gas flow rates, which decreased the particle velocity, causing the particles to be exposed to the plasma for an increased time, leading to increased degradation of HA into TTCP and CaO. The increased plasma temperature appeared to have a limited effect on the amount of HA, TTCP, and CaO produced. Instead, perhaps the velocity plays a greater role in the relative amounts of each phase produced. In terms of the argon–helium trials, helium gas produces a dense plasma, which increases particle velocity, while reducing the residence time of the particles. This in turn means the particles will be heated less, and so are less likely to degrade. The trial with the highest helium content produced the most HA. However, this trial also had the lowest total gas flow rate, indicating that perhaps the addition of helium overshadowed the total gas flow rate, in relation to the particle velocity. Comparing hydrogen and helium trials, the helium trials produced more HA, due to the increased particle velocity as well as the lower temperature, which reduced HA degradation.

The time the powder particles were exposed to the plasma was also manipulated using the spray distance, with longer distances exposing the particles to the hot plasma for a longer time. The decrease in HA and TTCP over spray distance may link to the degradation of these phases as spray distance is increased. This seems plausible as HA degrades into TTCP, which degrades into CaO. This mechanism would occur to a greater extent with increasing spray distance. The high retention of HA under high velocity conditions and short spray distances implies that the reaction between the feedstock components to form HA occurred extremely quickly upon injection into the plasma, within the first 60 mm. Once formed, it appears that the HA rapidly degraded.

Additionally, considering the EDS data of the general areas of the coatings selected, S6-73Ar:6.4H2-60 mm and S9-40Ar-18He-60 mm were found to have a loss of calcium, compared to the initial reagent powder, and were also found to produce a high amount of HA. In contrast, the opposite was observed with S7-35Ar:6.4H2 and S11-50Ar:4.5H2, whereby, they were found to have an excess amount of calcium compared to the reagent powder but produced low amounts of HA. This perhaps indicates that that in the case of S6-73Ar:6.4H2-60 mm and S9-40Ar-18He-60 mm plasma spray parameters, more of the calcium was consumed by the various reactions to produce HA compared to the S7-35Ar:6.4H2 and S11-50Ar:4.5H2 plasma spray parameters. Alternatively, samples which contained excess Ca may have had a greater inclusion of CaO potentially dissolved into solution, while those with a lower ratio potentially had lost CaCO3 (or its degradation product, CaO) inflight, leading to a Ca-poor composition.

It is noticeable that the same trends were observed for the concentration of TTCP, implying that this also formed extremely quickly. The rapid formation and subsequent degradation of HA and TTCP, relative to that seen in conventional HA feedstock spraying, suggests that these phases may have formed as surface films at the interface between reacting feedstock phases. The thin layer of HA or TTCP would then be highly susceptible to undergoing further degradation very quickly. This surface reaction type mechanism would account for the observed effect of the spray parameters, but not the high amounts of HA and TTCP formed. In this instance unreacted feedstock phases within the core of the particles would be expected to be encapsulated within the coating. However, this was not observed. It is notable that calcium carbonate and brushite thermally degrade into CaO and calcium pyrophosphate, respectively, without forming a stable molten phase. In contrast, both degradation products do melt, at 2613 and 1353 °C, respectively (Ref 66). The presence of well molten splats in the coating indicates impact of liquid droplets, implying that HA and TTCP formation occurred via reaction of these molten precursors. However, such a mechanism would be expected to favor high plasma temperatures, low plasma gas flow rates and long spray distances to maximize the time of melting and reaction.

In short, this is the first work to show that HA can be formed from the precursor feedstock phases calcium carbonate and brushite during plasma spraying. However, the interpretation of HA formation as a function of plasma deposition conditions remains unclear, due to uncertainty over the exact mechanism/s taking place inflight (postulated mechanism 1-3 above). This is complicated further by the relative rates of the multiple steps potentially required in HA formation, such as the degradation of calcium carbonate and brushite, and the subsequent reaction of the degradation products. There is the potential for each step to favor different plasma conditions, at different points in-flight. Additionally, complicating this interpretation further, is the fact that the feedstock powders were not sintered prior to thermal spray. This was in order to avoid thermal decomposition of the feedstock phases. However, the weak internal bonding of the spray dried composite particles raises the potential that they physically broke down into smaller particle fragments in flight prior to phase reaction and liquid phase formation. Follow up trials are underway to address these uncertainties and more clearly establish the reaction pathways to HA and TTCP formation using the most promising plasma spray parameters highlighted in this work.

Conclusion

Coatings composed of molten splats containing a homogenous mixture of HA, TTCP and CaO were successfully synthesized through the in-flight reaction of calcium carbonate and brushite using plasma spray for the first time. Various plasma spray parameters were used to investigate their effect on the coating composition and the quantity of HA produced in the coating. The parameters which increased the particle velocity (total gas flow rate, helium plasma gas) were found to increase the amount of HA produced in the coating. In addition, parameters that decreased the plasma temperature (increasing spray distance, changes in plasma gases) were also found to increase the amount of HA produced. However, the effect of velocity overshadowed the effect of temperature in terms of the condition which produced a higher amount of HA. The highest amount of HA was produced at a spray distance of 60 mm, in the argon-helium plasma condition with the largest helium content (S9-40Ar-18He), by the SG100 gun. Whereas the highest amount of HA produced by the 3 MB gun was in the argon–hydrogen plasma condition, with the highest gas flow rate and powder feed rate (S16-73Ar:6.4H2), at a spray distance of 60 mm. The HA was hypothesized to be made through reactions between the precursor reagents or their degradation products. In contrast TTCP and CaO were thought to be either degradation products of HA or formed through the precursor reagents or their degradation products. Further investigations into the mechanism of HA formation needs to be explored using optimal spray parameters on different combinations of the precursor reagents and their degradation products to further understand the reaction mechanism of HA formation using plasma spray.