Synthesis of Cubic Aluminum Nitride (AlN) Coatings through Suspension Plasma Spray (SPS) Technology

Thermal spraying of aluminum nitride (AlN) is a challenging issue because it decomposes at a high temperature. In this work, the use of suspension plasma spray (SPS) technology is proposed for the in situ synthesis and deposition of cubic-structured AlN coatings on metallic substrates. The effects of the nitriding agent, the suspension liquid carrier, the substrate materials and the standoff distance during deposition by SPS were investigated. The plasma-synthesized coatings were analyzed by X-ray diffraction (XRD), optical microscopy (OM) and scanning electron microscopy (SEM). The results show higher AlN content in the coatings deposited on a carbon steel substrate (~82%) when compared to titanium substrate (~30%) or molybdenum (~15%). Melamine mixed with pure aluminum powder produced AlN-richer coatings of up to 82% when compared to urea mixed with the Al (~25% AlN). Hexadecane was a relatively better liquid carrier than the oxygen-rich liquid carriers such as ethanol or ethylene glycol. When the materials were exposed to a molten aluminum–magnesium alloy at 850 ◦C for 2 h, the corrosion resistance of the AlN-coated carbon steel substrate showed improved performance in comparison to the uncoated substrate.


Background
Aluminum nitride (AlN) is a promising material that can find diverse applications in the aluminum industry [1], for example, in crucibles for handling corrosive chemicals and reaction vessels [2]. In addition, AlN has become the key component of semiconductor equipment, which is appropriate in piezoelectric and electronic applications [3]. Some of these amazing properties of AlN are summarized in Table 1, as high thermal conductivity [4], low thermal expansion coefficient [5], and a large band gap, which is a major factor that determines a solid's electrical conductivity [6], and this makes AlN an electrical insulator. Other characteristics include AlN's chemical and physical stability at fairly high temperature regions, high hardness [7], and high resistance to molten metals, wear and corrosion [5].  Table 3. Summary of different methods used for producing AlN powder.

Method Precursors Reference
Carbothermal reduction Al 2 O 3 + N 2 + C/CH 4 [28] CVD AlCl 3 + NH 3 + N 2 [29] Reaction 500-1000 • C for 5 h Al 2 S 3 + NH 3 [30] Sol-gel C 9  Thermal plasma technology is a well-adapted technology for synthesizing crystalline nano materials because of the fast reaction rates achieved due to the high plasma temperatures [5], its flexibility in the choice of various feedstock materials [38], its high conversion rates and energy efficiency [10] and its rapid quenching [39]. Furthermore, plasma spray technology simplifies the inflight reaction of the precursors and the in situ formation of the coatings in a wide range of thickness. Due to the high reaction rate, a dense coating is produced. Among all the thermal spray methods, plasma spraying is extensively used to produce coatings due to its high enthalpy, with the molten or semi molten particles being propelled to adhere and solidify on the substrate [40]. However, sometimes it is not necessary to melt the particles fully, because it is possible for particles in a semi molten state to be deposited as a coating. Moreover, a thermal plasma process can allow for a wide range of coating thicknesses, stretching from nm to mm [41]. In order for the deposited coating to stick to the substrate, the injected particles must melt or be in semi-molten state when they pass through the hot core of the plasma. The high melting point of AlN (2200 • C) as well as its decomposition at high temperatures does not facilitate the deposition of AlN particles injected through the plasma [41].
Besides ammonia [42], nitrogen gas is usually used as a nitriding agent [32], but recently urea [43], and melamine to produce Si-C-N [44], or AlN ceramics [45] have been proposed for nitriding reactions. Melamine (C 3 H 6 N 6 ), an organic compound containing about 67 wt% nitrogen is a trimer of cyanamide, with a 1,3,5-triazine skeleton, that releases nitrogen at high temperatures [46], and has been used to produce AlN powder by mechanochemical reaction with Al [47]. Direct injection of nitriding powders such as urea or melamine with Al into plasma, unlike N 2 gas, is challenging, particularly when there are different precursor components with various particle sizes and densities [41].
To alleviate the injection problems, suspension liquids have been employed as carrier media of transporting even nano particles into the high temperature zones of plasma, with the added benefits of providing chemical reactions within the suspension liquid. Indeed, powder precursors are dispersed in the liquid media and when injected into plasma, the liquid phase evaporates, while the solid particles melt and vaporize to form clusters. Suspension plasma spray (SPS) was invented to create a homogenous precursor to facilitate direct injection of small particles into plasma by dispersing them in a liquid carrier [48].
In order to produce a durable coating that can withstand thermal shocks, both the coating and the substrate should ideally have linear thermal expansion coefficient that are similar. Substrate materials should have higher melting points and lower thermal expansion coefficients than Al [49]. An interlayer coating can be used to adapt the expansion coefficient difference [50].

Non-Wetting Properties
In many material processing techniques, for example, in casting processes, molten metals are ordinarily in direct contact with ceramic refractories. Therefore, the properties of the final alloy product are highly affected by the high-temperature properties of the molten metal as well as the interfacial wetting or reaction phenomena. Many factors influence the wettability of the molten alloy-ceramic system such as the alloying elements, defects, impurities and surface roughness of the refractory [51]. Besides the processing parameters, the interaction between the molten metal and the substrate is one of the key factors considered in choosing a suitable material that will be in direct contact with the molten alloys [52]. It has been reported that when a molten metal and a ceramic substrate are in contact, the wetting of the ceramic substrate is usually accompanied by an interfacial reaction between the metal and the ceramic [53]. Hence, a new composition is formed at the interface during the reaction, and this has a strong effect on the magnitude of the interfacial free energy and subsequently alters the contact angle between the molten metal and the ceramic. The contact angle measurement for an Al-ceramic system is not simple for two main reasons: (a) the oxidation of Al where the oxide layer on Al surface modifies the properties and behavior of the alloy, and (b) Al being a highly reactive metal corrodes the ceramic and its wettability changes over time [54].
Ceramics such as AlN are considered as suitable electrolysis cell material because of their non-wettability and non-reactivity towards molten Al or its salts [55]. Some of these ceramics have been synthesized as coatings using solution precursor plasma spray (SPPS) technology, with the authors investigating the effect of different direct current (DC) plasma parameters such as standoff distance, torch power, number of torch-passes, types of solvents and plasma velocity on the superhydrophibicity properties of the final coating [56]. They concluded that a short standoff distance, a high arc current, a low number of torches passes, the addition of ethanol to the solvent and a high plasma velocity result in a dense superhydrophobic coating. In addition, the coated layer demonstrated a hierarchical surface roughness structure.

Research Objectives
The objective of this study is to produce cubic AlN-rich coatings deposited on various substrates to improve their corrosion resistance when in contact with molten Al-5 wt% Mg. SPS technology has been chosen because it provides flexibility in the selection of precursors and substrates. Moreover, since the injection rate of the precursor has a significant effect on the reaction and formation of the coating, the powder mixture can be suspended in a liquid carrier at various mass concentrations and injected into plasma using a peristaltic pump. The hypothesis being tested involves investigating the effect of synthesis parameters such as precursor composition, suspension liquid, substrate materials and standoff distance. Coatings were characterized by microscopy and X-ray methods. The anti-wetting properties of AlN were ascertained by the sessile drop test and a corrosion test using liquid Al-5 wt% Mg.

Materials and Methods
All experiments involving the synthesis of AlN coatings in this study were carried out using a radio frequency (RF) induction suspension plasma spray (SPS) reactor. The SPS system consists of a plasma chamber and atomization probe (of 7-12 µm water droplet size), a plasma torch (supplied by TEKNA Inc., Sherbrooke, QC, Canada) and a plasma power supply (Lepel Co., Edgewood, NY, USA).

Preparation of Substrates
In order to increase surface roughness, the substrates were initially sand blasted by alumina particles to improve the coatings' sticking efficiency to the substrate. However, contamination was a weakness with this approach since the presence of alumina was detected by XRD in the samples. Therefore, SiC paper (180) was used to scratch the substrates in order to obtain the desired surface roughness (~0.5 µm Ra) and this proved to be more effective.

Suspension Feedstock for SPS Injection
Metal Al powder was mixed with nitriding agents (either melamine or urea powder) and dispersed in different suspension liquid carriers such as ethanol, ethylene glycol, hexadecane or light mineral oil. First, the Al powder was dispersed in the liquid carrier using an ultrasonic bath to break the agglomerated particles for 10 min. Then, the melamine or urea powder was added to the suspension and the mixture stirred for another hour. The suspension feedstock was finally injected into the plasma using an atomization probe.

Synthesis of Cubic AlN Coatings by SPS
In preparation, the plasma chamber was first evacuated while purging with an inert gas (Argon, Ar), to obtain the pressure set at 16 kPa. Table 4 provides a summary of the Coatings 2021, 11, 500 5 of 21 plasma conditions employed for the deposition of the AlN coatings. During the injection of the suspension, the sample holder with the substrate was kept at a constant standoff distance in the plasma plume. In order to produce a homogeneous thin film, the coating was deposited in several spray cycles of typically 200 loops where the sample was mounted on a sliding stage that moves back and forth at a speed of 50 mm per second. In addition, since steady spraying can cause the substrates to overheat, therefore, intermittent deposition by loops allows for some cooling. Moreover, the sample holder is water cooled to protect the substrates from melting. One effective way of avoiding agglomeration of un-vaporized Al droplets on the cold substrate is by preheating the surface that supports the coating. Nevertheless, in this work, the deposition commenced without preheating the substrates because AlN formation is an exothermic reaction [17].

Materials Characterization
The coatings were characterized by various analytical techniques including X-ray diffraction (XRD) for phase and crystal structure analysis. Optical microscopy (OM) and scanning electron microscopy (SEM) coupled with energy dispersive X-ray spectroscopy (EDX) was used for morphological and elemental analysis. Contact angle measurements between the coatings and water by sessile-drop test were used to determine the wettability of the coatings.

X-ray Diffraction (XRD) Analysis
The samples were analyzed according to the method described in an earlier article using a "Philips Panalytical X'pert PRO MPD" X-ray diffractometer (PANalytical, Almelo, The Netherlands) [57]. The diffractometer uses Cu Kα1 radiation with a wavelength, λ = 1.54 Å, in the 2θ-angle range from 20 to 100 • , at a scanning speed of 0.04 • 2θ-angle per min and a step size of 0.02 • and step time, 0.5 s. Data collection and data analysis was performed using factory installed JADE software (version 7.8.4 @2020-04-23 © MDI). The amount of AlN in the coating was determined by Rietveld quantitative analysis (RQA) in the region of 20-60 • 2θ-angle. Although there was some degree of amorphicity displayed Coatings 2021, 11, 500 6 of 21 by the coatings, phase quantification by RQA using Equation (1) was attempted in order to determine the various amounts of each species in the coatings [58].
where W ρ = relative weight fraction of phase ρ in a mixture of n phases, S ρ = Rietveld scale factor, Z = number of formula units per cell, M = mass of the formula unit (in atomic mass units) and V = the unit cell volume (in Å 3 ).

Optical Microscopy (OM)
The Keyence VHX-5000 optical microscope (KEYENCE CORPORATION OF AMER-ICA, Itasca, IL, USA) was used to investigate the surface properties of coatings in relation to their corrosion resistance when in contact with molten the Al-Mg alloy.

Scanning Electron Microscopy (SEM)
Microscopic imaging by SEM coupled with energy dispersive X-ray spectroscopy (EDX) were used to investigate the morphology, structure, particle size and composition of the materials. In this work, SEM imaging and analysis was done on a "Hitachi S-4700" Field-Emission Scanning Electron Microscope (Hitachi High-Technologies Corporation, Tokyo, Japan), equipped with an EDX X-Max Oxford spectrometer (Abingdon, UK) [57].

Surface Roughness Test and the Coating Thickness Measurements
The surface roughness was measured by Mitutoyo Surftest SJ-210-Series 178-Portable surface roughness tester (Mitutoyo Corporation, Kawasaki-shi, Kanagawa, Japan). The measurement range is automatically set with a resolution between 0.0016 and 0.0256 µm. The stylus material is diamond with a tip radius of 2 µm and the force utilized is 0.75 mN. The surface roughness of the substrate was measured before and after the deposition of the coatings. In addition, the coating thickness of samples was measured by Mitutoyo digital thickness gauge equipped with absolute digimatic indicator.

Materials Testing
(a) Wetting test: Wettability of the samples was studied using a drop shape analyzer, the Krüss Advance goniometer model DSA25E (KRÜSS, Hamburg, Germany). In order to validate the analysis method, the static contact angle was calibrated at ambient conditions by depositing a droplet of water or ethylene glycol on the surface of Teflon (polytetrafluoroethylene, PTFE). From the manufacturer's database, the contact angle of water on PTFE is known to be 113.7 • , which we validated to be 112.1 • as indicated in Figure 1a, while that of ethylene glycol was found to be 98.1 • as shown in Figure 1b against 94.9 • of the database [59]. A liquid on a solid surface with contact angle below 90 • is considered to be a wetting liquid, while a contact angle above 90 • indicates nonwetting properties [60]. Superhydrophobic surfaces with the 'Lotus Effect' that exhibit an extremely high contact angle with water (>150 • ) have been broadly investigated and can be considered as a perfect example in anti-corrosion applications [61].
Coatings 2021, 11, x FOR PEER REVIEW 7 of 22 indicates non-wetting properties [60]. Superhydrophobic surfaces with the 'Lotus Effect' that exhibit an extremely high contact angle with water (>150°) have been broadly investigated and can be considered as a perfect example in anti-corrosion applications [61]. (b) Corrosion test: The fresh coatings synthesized by SPS technology were tested for corrosion resistance by contacting them directly with a molten Al-alloy containing 5% magnesium (Al-5 wt% Mg) for 2 h at 850 °C. OM was used to observe the surface properties of the coatings to investigate their corrosion resistance when in contact with  (b) Corrosion test: The fresh coatings synthesized by SPS technology were tested for corrosion resistance by contacting them directly with a molten Al-alloy containing 5% magnesium (Al-5 wt% Mg) for 2 h at 850 • C. OM was used to observe the surface properties of the coatings to investigate their corrosion resistance when in contact with molten Al-Mg alloy.

FactSage™ Modelling for the In Situ Synthesis of AlN in Plasma
In plasma, at temperatures above 3000 • C, FactSage™ (Montreal, QC, Canada) can predict the composition of materials at thermodynamic equilibrium in vapor phase. At lower temperatures, or when multiphase materials are present, the kinetics of the reactions may be considered, but FactSage™ in our context is used to simulate or predict the outcome of targeted reactions during synthesis and the performance of materials in realistic industrial applications. FactSage™ modelling predicted that using a feedstock of hexadecane, metallic Al and melamine in the presence of plasma gases (including hydrogen and nitrogen), AlN begins to form at temperatures below 2300 • C as portrayed in Figure 2, noting that not all the reactions and species occurring in the simulation results are published on the graph. (b) Corrosion test: The fresh coatings synthesized by SPS technology were tested for corrosion resistance by contacting them directly with a molten Al-alloy containing 5% magnesium (Al-5 wt% Mg) for 2 h at 850 °C. OM was used to observe the surface properties of the coatings to investigate their corrosion resistance when in contact with molten Al-Mg alloy.

FactSage™ Modelling for the In Situ Synthesis of AlN in Plasma
In plasma, at temperatures above 3000 °C, FactSage™ (Montreal, QC, Canada) can predict the composition of materials at thermodynamic equilibrium in vapor phase. At lower temperatures, or when multiphase materials are present, the kinetics of the reactions may be considered, but FactSage™ in our context is used to simulate or predict the outcome of targeted reactions during synthesis and the performance of materials in realistic industrial applications. FactSage™ modelling predicted that using a feedstock of hexadecane, metallic Al and melamine in the presence of plasma gases (including hydrogen and nitrogen), AlN begins to form at temperatures below 2300 °C as portrayed in Figure 2, noting that not all the reactions and species occurring in the simulation results are published on the graph. Since H2 and N2 gases dissociate to atomic H and N above 5000 °C, the presence of a significant quantity of hydrogen and other gas species in the chamber helps to delay the Since H 2 and N 2 gases dissociate to atomic H and N above 5000 • C, the presence of a significant quantity of hydrogen and other gas species in the chamber helps to delay the recombination of nitrogen to N 2 at lower temperatures. Gaseous nitrogen is in fact, not a good nitriding agent because when it decomposes to atomic N, it prefers to recombine to N 2 instead of reacting with other plasma species. Therefore, NH 3 is deemed as a better nitriding agent because once it decomposes at high temperature, it forms N radicals that can react with Al to form AlN. When melamine is injected into the hot core of plasma, it decomposes to nitrogen, hydrogen, ammonia and HCN, besides forming some radicals such as CH˙, NH˙, CN˙, and ions including C 2 N 2+ , C 3 N 2+ and C 3 N 3+ that facilitate the reaction with Al to produce AlN.
FactSage™ results in Figure 2 show that at temperatures below 2700 • C, solid carbon starts forming, and at about 700 • C most of the carbon in the chamber is drastically consumed by the formation of CH 4 . Formation of AlN starts at~2300 • C through the reaction between Al and HCN. Atomic nitrogen reacts with atomic hydrogen and carbon to form HCN at 4100 • C. As the temperature drops, production of HCN increases to the maximum at 3000 • C. It is confirmed by some authors that the formation of HCN gas plays an important role in producing AlN [62]. It is clear that the role of N 2 in the formation of AlN is negligible, but the CN species seems to facilitate the nitridation of Al. Other researchers have demonstrated that at the stoichiometric reactant ratios, the reaction between Al and melamine is mainly governed by the polymerization of melamine and the formation of a carbon nitride (CNx) phase [45]. The presence of oxygen in urea prohibits the nitriding reaction below 1500 • C as predicted from FactSage™ modelling in Figure 3 because it generates alumina (Al 2 O 3 ). From the quantities produced between the AlN and Al 2 O 3 , it is clear that the two products are in direct competition. Therefore, besides using an oxygen-free nitriding agent, the chamber must be evacuated of oxygen to facilitate the reaction between Al and nitrogen. For this reason, suspension liquids with a lot of oxygen such as ethanol and ethylene glycol performed poorly. With urea, the production window of AlN is narrow (1200-2000 • C), while with melamine the AlN's stability window is wider stretching from 2300 • C to ambient temperature. reaction between Al and HCN. Atomic nitrogen reacts with atomic hydrogen and carbon to form HCN at 4100 °C. As the temperature drops, production of HCN increases to the maximum at 3000 °C. It is confirmed by some authors that the formation of HCN gas plays an important role in producing AlN [62]. It is clear that the role of N2 in the formation of AlN is negligible, but the CN species seems to facilitate the nitridation of Al. Other researchers have demonstrated that at the stoichiometric reactant ratios, the reaction between Al and melamine is mainly governed by the polymerization of melamine and the formation of a carbon nitride (CNx) phase [45].
The presence of oxygen in urea prohibits the nitriding reaction below 1500 °C as predicted from FactSage™ modelling in Figure 3 because it generates alumina (Al2O3). From the quantities produced between the AlN and Al2O3, it is clear that the two products are in direct competition. Therefore, besides using an oxygen-free nitriding agent, the chamber must be evacuated of oxygen to facilitate the reaction between Al and nitrogen. For this reason, suspension liquids with a lot of oxygen such as ethanol and ethylene glycol performed poorly. With urea, the production window of AlN is narrow (1200-2000 °C), while with melamine the AlN's stability window is wider stretching from 2300 °C to ambient temperature.

SEM Analysis
The freshly synthesized coatings showed that the splats produced in the plasma were in the mean size range of about 2 µm. A sample secondary electron image (SEI) of AlN coating produced from melamine is presented in Figure 4a,b depicts the mixed color EDX elemental mapping of Al and N, while the individual EDX elemental mappings of Al and N are indicated in Figure 4c,d, respectively. The highly dispersed and concomitant presence

SEM Analysis
The freshly synthesized coatings showed that the splats produced in the plasma were in the mean size range of about 2 µm. A sample secondary electron image (SEI) of AlN coating produced from melamine is presented in Figure 4a,b depicts the mixed color EDX elemental mapping of Al and N, while the individual EDX elemental mappings of Al and N are indicated in Figure 4c,d, respectively. The highly dispersed and concomitant presence of Al and N indicates the success of producing AlN using SPS technology. The EDX spectrum is displayed in Figure 4e. The atomic percentage of each elements calculated by EDX were Al (40%), N (35%) and O (23%) for the coating deposited on carbon steel substrate, and the rest of analysis are summarized in Table 5 for Ti and Mo substrates. Since these values are semi quantitative, they should be used conservatively because of the high surface roughness of the coatings, which normally lowers the signal intensity in some areas under analysis. of Al and N indicates the success of producing AlN using SPS technology. The EDX spectrum is displayed in Figure 4e. The atomic percentage of each elements calculated by EDX were Al (40%), N (35%) and O (23%) for the coating deposited on carbon steel substrate, and the rest of analysis are summarized in Table 5 for Ti and Mo substrates. Since these values are semi quantitative, they should be used conservatively because of the high surface roughness of the coatings, which normally lowers the signal intensity in some areas under analysis.   Figure 5 provides the surface view of the coating deposited on a carbon steel substrate. The morphology and structure of the coating tend towards agglomerated spherical particles. In addition, Figure 6 displays the cross-sectional view of the same coating showing good bonding between the coating and the substrate. The coating and the bonding remain intact when placed in ultrasonic bath suspended in ethanol for 15 min. Moreover, after exposure to molten Al-Mg alloy, no peeling was observed.

XRD Analysis
Various parameters were used to optimize the production of AlN coating as the target product. Figure 7 shows the XRD patterns of the two major peaks that were used to semi-quantitively determine the amount of the cubic AlN in the coatings, where the main peak at 44.85° 2θ-angle represents the (200) plane, followed by the smaller (111) peak at approximately 38.59° 2θ-angle. A comparatively higher intensity of the (111) peak than the (200) peak implies the presence of metallic Al, while an intense (200) peak accompanied by a smaller (111) peak implies the presence of AlN.

XRD Analysis
Various parameters were used to optimize the production of AlN coating as the target product. Figure 7 shows the XRD patterns of the two major peaks that were used to semi-quantitively determine the amount of the cubic AlN in the coatings, where the main peak at 44.85 • 2θ-angle represents the (200) plane, followed by the smaller (111) peak at approximately 38.59 • 2θ-angle. A comparatively higher intensity of the (111) peak than the (200) peak implies the presence of metallic Al, while an intense (200) peak accompanied by a smaller (111) peak implies the presence of AlN.  Quantification of the various phases in the AlN coatings by RQA using XRD data is exemplified in Figure 8, which indicates the percentage of the AlN formed in the coatings alongside other by-products. The ICDD cards used to identify the phases include: [00-039-1950] [ [70] for Fe. This approach was used to analyze all the other samples, where the raw data extracted from the RQA curve fitting are summarized in Table 6. Quantification of the various phases in the AlN coatings by RQA using XRD data is exemplified in Figure 8, which indicates the percentage of the AlN formed in the coatings alongside other by-products. The ICDD cards used to identify the phases include: [00-039-1950] [ [70] for Fe. This approach was used to analyze all the other samples, where the raw data extracted from the RQA curve fitting are summarized in Table 6.
Quantification of the various phases in the AlN coatings by RQA using XRD data is exemplified in Figure 8, which indicates the percentage of the AlN formed in the coatings alongside other by-products. The ICDD cards used to identify the phases include: [00-039-1950] [63] [70] for Fe. This approach was used to analyze all the other samples, where the raw data extracted from the RQA curve fitting are summarized in Table 6. Further, the quantity of AlN (%) was plotted versus the standoff distance, the substrate materials, the liquid suspension carriers and the nitriding agents as provided in Figure 9. Figure 9a illustrates the impact the substrate materials have on the formation of AlN in the coatings, while Figure 9b shows that comparatively, melamine is a stronger nitridation agent than urea. The presence of oxygen in urea seems to have a negative  Table 6. Measured amount of the various phases in the plasma-synthesized coatings by RQA using XRD data.

Cubic-AlN hcp-AlN Al Al 4 C 3 Others
Substrates Further, the quantity of AlN (%) was plotted versus the standoff distance, the substrate materials, the liquid suspension carriers and the nitriding agents as provided in Figure 9. Figure 9a illustrates the impact the substrate materials have on the formation of AlN in the coatings, while Figure 9b shows that comparatively, melamine is a stronger nitridation agent than urea. The presence of oxygen in urea seems to have a negative impact on nitridation reaction because Al has a higher affinity for oxygen than nitrogen. In addition, melamine by having a higher concentration of nitrogen per unit mass than urea increases the chance of the reaction between Al and nitrogen in plasma. Figure 9c shows the effect of standoff distance on the formation of AlN, with the optimum distance being found to be 8 cm. On one hand, decreasing the standoff distance to 6 cm saw the formation of AlN decrease drastically, because of insufficient residence time for the species to react and produce AlN. On the other hand, increasing the standoff distance to 10 cm equally lowered the production of AlN, although there was enough time for the materials to react. However, by lengthening the residence time, two other negative phenomena emerged: (a) the in situ produced AlN decomposed and could not make it into the coating, and (b) the AlN particles formed and hardened in such a way that they could no longer produce splats onto the substrate. It is necessary that coating materials should be in molten state in order to stick on the substrate. Figure 9d, shows that at standoff distance of 8 cm, using melamine and a carbon steel substrate, the hexadecane liquid carrier produced the highest amount of AlN in the coatings as compared to the other liquid carriers.

Substrates
Ti   Figure 9c shows the effect of standoff distance on the formation of AlN, with the optimum distance being found to be 8 cm. On one hand, decreasing the standoff distance to 6 cm saw the formation of AlN decrease drastically, because of insufficient residence time for the species to react and produce AlN. On the other hand, increasing the standoff distance to 10 cm equally lowered the production of AlN, although there was enough time for the materials to react. However, by lengthening the residence time, two other negative phenomena emerged: (a) the in situ produced AlN decomposed and could not make it into

Effect of Nitriding Agent on AlN Formation
In this study, two nitriding agents were used, namely melamine (C 3 H 6 N 6 ) and urea (CH 4 N 2 O). From the XRD analysis using RQA, it was observed that melamine produced a higher percentage (82%) of AlN in the coatings in comparison to urea (25%). Melamine having more nitrogen in its structure (67% by mass N) than urea (47% by mass N), introduced more nitrogen into the system, thereby increasing the chance of Al reacting with nitrogen. Since the mass transfer in plasma is characterized by the laminar flow, mixing is restricted and any reactant that would increase the chance of reaction is crucial. Figure 10 shows XRD spectra of coatings produced using different nitriding agents. It was observed that in materials where the (200) peak was more intense than the (111) peak, then the material was richer in AlN. However, the presence of metallic Al was characterized by an intense (111) peak accompanied by a smaller (200) peak.
When urea was used as a nitriding agent, there was a dismal conversion of Al to AlN, but melamine produced more AlN under similar plasma conditions. The dominant phase in the AlN was the cubic crystal structure (Fm-3m space group). In effect, urea did not react well with Al, but melamine performed comparatively better. This confirmed what a number of authors have suggested in using melamine as a nitriding agent to generate nano-structured AlN powder [71]. It has been reported that melamine releases NH 3 to form a melem (C 6 N 10 H 6 ) intermediate at temperatures of about 550 • C, and at higher temperatures to produce chemically reactive hydrogen-, carbon-or nitrogen-containing atomic species such as CN 2+ and NCNH + [72]. CN group is so reactive that it easily produces charged nitriding species in plasma due to the instability of the triple bond between C and N, especially with the presence of the π electron bonds. a higher percentage (82%) of AlN in the coatings in comparison to urea (25%). Melamine having more nitrogen in its structure (67% by mass N) than urea (47% by mass N), introduced more nitrogen into the system, thereby increasing the chance of Al reacting with nitrogen. Since the mass transfer in plasma is characterized by the laminar flow, mixing is restricted and any reactant that would increase the chance of reaction is crucial. Figure 10 shows XRD spectra of coatings produced using different nitriding agents. It was observed that in materials where the (200) peak was more intense than the (111) peak, then the material was richer in AlN. However, the presence of metallic Al was characterized by an intense (111) peak accompanied by a smaller (200) peak. When urea was used as a nitriding agent, there was a dismal conversion of Al to AlN, but melamine produced more AlN under similar plasma conditions. The dominant phase in the AlN was the cubic crystal structure (Fm-3m space group). In effect, urea did not react well with Al, but melamine performed comparatively better. This confirmed what a number of authors have suggested in using melamine as a nitriding agent to generate nano-structured AlN powder [71]. It has been reported that melamine releases NH3 to form a melem (C6N10H6) intermediate at temperatures of about 550 °C, and at higher temperatures to produce chemically reactive hydrogen-, carbon-or nitrogen-containing atomic species such as CN 2+ and NCNH + [72]. CN group is so reactive that it easily produces charged nitriding species in plasma due to the instability of the triple bond between C and N, especially with the presence of the π electron bonds. Figure 11a shows SEM analysis of the AlN coating produced by urea. Figure 11b is the EDX spectrum of the coating, while Figure 11c Figure 11a shows SEM analysis of the AlN coating produced by urea. Figure 11b is the EDX spectrum of the coating, while Figure 11c-f represents the respective EDX elemental analysis for N, Al, Ti and oxygen. When compared, the coating deposited on Ti substrate using urea, shown in the EDX spectrum of Figure 11b, contained more oxygen and less AlN than the sample derived from melamine, as seen in Figure 4. The atomic percent of elements calculated by EDX were Al (39.9%), N (5.5%), Ti (5.8%) and O (51.4%).

Effect of Standoff Distance on AlN Formation
In testing our hypothesis, it was postulated that the standoff distance can affect the quality of the coating. It was observed that an optimum distance destined for the maximum formation of AlN in plasma exists, which is about 8 cm as shown in Figure 9c. The XRD data from which this information was derived are provided in Figure 12. The two scenarios that

Effect of Standoff Distance on AlN Formation
In testing our hypothesis, it was postulated that the standoff distance can affect the quality of the coating. It was observed that an optimum distance destined for the maximum formation of AlN in plasma exists, which is about 8 cm as shown in Figure 9c. The XRD data from which this information was derived are provided in Figure 12. The two scenarios that were observed to influence the quality of the coating in relation to standoff distance were: (i) Long standoff distance: where after the nitridation, AlN failed to stick to the substrate, which was too far because either the AlN decomposed before reaching the substrate, or the AlN condensed and reached the substrate after hardening. (ii) Short standoff distance: characterized by low residence time in the reactor because the materials did not have enough time to react and produce AlN.

Effect of Suspension Liquids on AlN Formation
It was observed that with the application of RQA using the XRD data shown in Figure 13, utilizing hexadecane produced the highest percentage of AlN in the coatings (~82%) in comparison to using other suspension liquid carriers. Initial experiments of this project were conducted by injecting powder precursors directly into plasma. However, the difficulty with powder injection involved irregular powder flow due to high particle friction, clogging, and particle agglomeration, which led to poor coating quality. Nevertheless, by suspending the powder in an oily liquid, lubrication and therefore the smooth flow of the feedstock was enhanced. With the SPS method, we overcame the problem of irregular powder injections. Several liquids were used to create the powder suspensions, and it was observed that suspension liquids richer in oxygen (i.e., ethanol, ethylene glycol, and light mineral oil) inhibited the formation of AlN, because the oxygen preferentially reacts with Al to form Al2O3 instead of AlN. These observations imply that at a standoff distance of 8 cm, the materials have enough residence time to react during nitridation, and with high velocity at suitable particle temperature, the coatings are deposited on the substrate more effectively. At the extreme standoff distances of, for example, 10 cm there were many by products formed as shown by the many peaks in the XRD spectrum, while at a standoff distance of 6 cm the Al and melamine remained unreacted. The melamine peaks are visible below the 2θ angle of 30 • .

Effect of Suspension Liquids on AlN Formation
It was observed that with the application of RQA using the XRD data shown in Figure 13, utilizing hexadecane produced the highest percentage of AlN in the coatings (~82%) in comparison to using other suspension liquid carriers. Initial experiments of this project were conducted by injecting powder precursors directly into plasma. However, the difficulty with powder injection involved irregular powder flow due to high particle friction, clogging, and particle agglomeration, which led to poor coating quality. Nevertheless, by suspending the powder in an oily liquid, lubrication and therefore the smooth flow of the feedstock was enhanced. With the SPS method, we overcame the problem of irregular powder injections. Several liquids were used to create the powder suspensions, and it was observed that suspension liquids richer in oxygen (i.e., ethanol, ethylene glycol, and light mineral oil) inhibited the formation of AlN, because the oxygen preferentially reacts with Al to form Al 2 O 3 instead of AlN.

Effect of Substrate on AlN Formation
Several substrates were used to deposit the AlN coatings by SPS technology and carbon steel performed better than Mo or Ti. Their performance is related to the substrates' thermal conductivity, whose values are 143, 54 and 24 W/m K for Mo, carbon steel and Ti, respectively [73]. We suppose that Mo with a high thermal conductivity leads to a much colder surface of the substrates that quenches the nitridation reaction. On the other hand, Ti having the lowest thermal conductivity exhibits the hottest surface temperature, which prevents the deposition of AlN. The carbon steel with a medium thermal conductivity is neither too hot nor too cold for nitridation reaction. Controlling the temperature of plasma and the substrate is therefore an important parameter in the formation of AlN coating [17]. The overlaid XRD spectra of the AlN coatings on Ti, Mo and carbon steel substrates are provided in Figure 14. Several observations were made: (a) The surface of the Mo substrate was exposed and detectable by XRD analysis because very little of the coating was formed. (b) Ti reacted with Al to produce several TiAl alloys. In addition, the AlN produced on the Ti took after the hexagonal (hcp) crystal structure of the substrates. (c) Carbon steel had the highest amount of cubic AlN coating, which is typical of the cubic (fcc) crystal structure of carbon steel substrate at high temperature. (d) Prolonged deposition periods led to the substrates overheating, consequently resulting in less AlN deposition and enhancing surface carbon formation.

Effect of Substrate on AlN Formation
Several substrates were used to deposit the AlN coatings by SPS technology and carbon steel performed better than Mo or Ti. Their performance is related to the substrates' thermal conductivity, whose values are 143, 54 and 24 W/m K for Mo, carbon steel and Ti, respectively [73]. We suppose that Mo with a high thermal conductivity leads to a much colder surface of the substrates that quenches the nitridation reaction. On the other hand, Ti having the lowest thermal conductivity exhibits the hottest surface temperature, which prevents the deposition of AlN. The carbon steel with a medium thermal conductivity is neither too hot nor too cold for nitridation reaction. Controlling the temperature of plasma and the substrate is therefore an important parameter in the formation of AlN coating [17]. The overlaid XRD spectra of the AlN coatings on Ti, Mo and carbon steel substrates are provided in Figure 14. Several observations were made: (a) The surface of the Mo substrate was exposed and detectable by XRD analysis because very little of the coating was formed. (b) Ti reacted with Al to produce several TiAl alloys. In addition, the AlN produced on the Ti took after the hexagonal (hcp) crystal structure of the substrates. (c) Carbon steel had the highest amount of cubic AlN coating, which is typical of the cubic (fcc) crystal structure of carbon steel substrate at high temperature. (d) Prolonged deposition periods led to the substrates overheating, consequently resulting in less AlN deposition and enhancing surface carbon formation. Figure 15 shows the SEM images of the freshly synthesized coatings using Al powder and melamine on different substrates, where Figure 15a is for carbon steel, Figure 15b represents Ti and Figure 15c is for Mo. Although surface roughness was evident after coating deposition, the carbon steel sample comprised the highest AlN content (80% cubic phase + 2% hcp), while Mo had the least amount of AlN (15% cubic, with no hcp phase). Ti substrates consisted of 17% cubic and 13% hcp.  Figure 15 shows the SEM images of the freshly synthesized coatings using Al powder and melamine on different substrates, where Figure 15a is for carbon steel, Figure 15b represents Ti and Figure 15c is for Mo. Although surface roughness was evident after coating deposition, the carbon steel sample comprised the highest AlN content (80% cubic phase + 2% hcp), while Mo had the least amount of AlN (15% cubic, with no hcp phase). Ti substrates consisted of 17% cubic and 13% hcp.  Table 7 summarizes the results obtained from roughness and thickness tests. It was observed that surface roughness increased after the deposition of the coatings, with Mo recording the highest roughness.    Figure 15 shows the SEM images of the freshly synthesized coatings using Al powder and melamine on different substrates, where Figure 15a is for carbon steel, Figure 15b represents Ti and Figure 15c is for Mo. Although surface roughness was evident after coating deposition, the carbon steel sample comprised the highest AlN content (80% cubic phase + 2% hcp), while Mo had the least amount of AlN (15% cubic, with no hcp phase). Ti substrates consisted of 17% cubic and 13% hcp.  Table 7 summarizes the results obtained from roughness and thickness tests. It was observed that surface roughness increased after the deposition of the coatings, with Mo recording the highest roughness.   Table 7 summarizes the results obtained from roughness and thickness tests. It was observed that surface roughness increased after the deposition of the coatings, with Mo recording the highest roughness. It was observed that samples containing a high percentage of AlN demonstrated an increased contact angle as illustrated in Figure 16a. Materials with a contact angle of less than 90 • are normally considered to be more hydrophilic, while those with a contact angle higher than 90 • are perceived to be more hydrophobic [60]. For example, the sample containing~82% AlN displayed a contact angle of 136 • which confirms the SPS synthesized coatings is hydrophobic, while the sample with a lower quantity of AlN (~25%) displayed a decreased contact angle to 73 • and it is hydrophilic as seen in Figure 16b. The measured water contact angle on PVD-deposited AlN by [74] was 119 • which shows hydrophobic properties of their coatings. Furthermore, it has been observed that the contact angle between molten Al and AlN surface is 136 • at 850 • C, which confirms the non-wettability of AlN by Al [55]. synthesized coatings is hydrophobic, while the sample with a lower quantity of AlN (~25%) displayed a decreased contact angle to 73° and it is hydrophilic as seen in Figure 16b. The measured water contact angle on PVD-deposited AlN by [74] was 119° which shows hydrophobic properties of their coatings. Furthermore, it has been observed that the contact angle between molten Al and AlN surface is 136° at 850 °C, which confirms the non-wettability of AlN by Al [55].

Corrosion Test
The samples with coated substrates were exposed to molten Al-Mg alloy at 850 °C for 2 hrs and their images depict after exposure to the molten Al alloy as shown in Figure 17.  All the coated samples indicated better corrosion resistance to the hot molten alloy, and it means that the coatings provided some form of protection to the substrates. However, in the Ti and Mo based samples, the presence of AlN coating was relatively lower with significant amount of metallic Al. Some forms of islands or micro cracks were witnessed in the Ti and Mo samples, which could be related to poor thermal shock resistance. Since the coating materials are richer in metallic Al (linear thermal expansion coefficient, αL ≈ 22 ×10 −6 K −1 ), the difference in the thermal expansion of the coatings and

Corrosion Test
The samples with coated substrates were exposed to molten Al-Mg alloy at 850 • C for 2 h and their images depict after exposure to the molten Al alloy as shown in Figure 17.
synthesized coatings is hydrophobic, while the sample with a lower quantity of AlN (~25%) displayed a decreased contact angle to 73° and it is hydrophilic as seen in Figure 16b. The measured water contact angle on PVD-deposited AlN by [74] was 119° which shows hydrophobic properties of their coatings. Furthermore, it has been observed that the contact angle between molten Al and AlN surface is 136° at 850 °C, which confirms the non-wettability of AlN by Al [55].

Corrosion Test
The samples with coated substrates were exposed to molten Al-Mg alloy at 850 °C for 2 hrs and their images depict after exposure to the molten Al alloy as shown in Figure 17.  All the coated samples indicated better corrosion resistance to the hot molten alloy, and it means that the coatings provided some form of protection to the substrates. However, in the Ti and Mo based samples, the presence of AlN coating was relatively lower with significant amount of metallic Al. Some forms of islands or micro cracks were witnessed in the Ti and Mo samples, which could be related to poor thermal shock resistance. Since the coating materials are richer in metallic Al (linear thermal expansion coefficient, αL ≈ 22 ×10 −6 K −1 ), the difference in the thermal expansion of the coatings and Figure 17a-c represent the freshly plasma-synthesized coatings for Ti, carbon steel and Mo, respectively. Figure 17d-f presents the coated samples after the corrosion test for Ti, carbon steel, and Mo, respectively.
All the coated samples indicated better corrosion resistance to the hot molten alloy, and it means that the coatings provided some form of protection to the substrates. However, in the Ti and Mo based samples, the presence of AlN coating was relatively lower with significant amount of metallic Al. Some forms of islands or micro cracks were witnessed in the Ti and Mo samples, which could be related to poor thermal shock resistance. Since the coating materials are richer in metallic Al (linear thermal expansion coefficient, α L ≈ 22 ×10 −6 K −1 ), the difference in the thermal expansion of the coatings and the substrates (Ti, α L ≈ 9 × 10 −6 K −1 ; Mo, α L ≈ 5 × 10 −6 K −1 ) [75], could generate some cracks in the coatings. Nevertheless, in the carbon steel sample (α L ≈ 12 × 10 −6 K −1 ), there was less cracking on the surface of the coating, perhaps because of the enhanced presence of AlN (82%) with α L ≈ 5.3 × 10 -6 K −1 . However, in the Ti and Mo based samples, the presence of AlN coating was relatively lower with significant amount of metallic Al. Some forms of islands or micro cracks were witnessed in the Ti and Mo samples, which could be related to poor thermal shock resistance. As thermal resistance (R θ ) can be calculated by dividing the thickness of the samples by the product of thermal conductivity (W/m K) and cross-sectional area (m 2 ), it was observed that the R θ of Mo, carbon steel and Ti substrates were 0.114, 0.29 and 0.65 (K·W −1 ) respectively. This implies that the samples on the two extremes (either too high, that is Ti, or too low, which is Mo) performed poorly as a substrate.

Conclusions
AlN has attracted a lot of attention for different applications in industry due to its unique properties. In the Al industry, AlN coatings may find use when in direct contact with molten Al alloys due to their non-wetting and corrosion resistant properties. Therefore, this work proposes in situ production of AlN-rich coatings by SPS technology, which has successfully produced up to approximately 80% cubic AlN coating on a carbon steel substrate with about 2% AlN in the hexagonal phase. Temperature control for AlN synthesis is an important parameter and it is demonstrated by changing standoff distance and substrate thermal conductivity. The thermal resistance of Mo, carbon steel and Ti substrates are 0.114, 0.29 and 0.65 (K·W −1 ), respectively, and it was observed that the samples on the two extremes (either too high or too low) performed poorly. It suggests that surface temperature of the substrates may play a role in the production of the AlN-rich coatings, and this is influenced by both the standoff distance during spraying and the substrates' thermal conductivity. The optimum plasma conditions included a standoff distance of 8 cm, the use of hexadecane liquid carrier, and reacting metallic Al powder with melamine as opposed to urea. Results show that favorable in situ AlN synthesis conditions by SPS include a nitrogen-rich precursor (e.g., melamine), a system without oxygen in the suspension liquid carrier (e.g., hexadecane), and depositing AlN on a carbon steel substrate.

Data Availability Statement:
The data presented in this study will be publicly available in the PhD of Faranak Barandehfard (2021) at Université de Sherbrooke, which doesn't have a DOI.