Internal diameter HVOAF thermal spray of carbon nanotubes reinforced WC-Co composite coatings

A novel process, relatively fast and scalable compared to existing ones, has been used to incorporate carbon nanotubes (CNT) 0.5 wt. % onto commercial WC-Co thermal spray powder. Nano enabled WC-Co/CNT coatings were obtained by internal diameter HVOAF thermal spray at three different spray powers (48, 41 and 34 kW). A general increase in microhardness and decrease in fracture toughness has been found when adding CNT. Also, carbide retention index upon spray was improved by lowering the process temperature and adding CNT, reaching 99 % in the low power case with CNT. The interplay of these features has shown an overall better wear performance in the medium power case without CNT; in fact, the addition of CNT has improved the wear performance at high power conditions (reduction of coefficient of friction by half) while reducing the wear performance in medium and low power. A tailored choice of CNT concentration can offer enhanced mechanical or tribological properties according to the needs for different applications. This novel process for nanoparticles incorporation opens the way for the production of large batches of readily usable nano-enabled powder.

internal surfaces as narrow as 70 mm in diameter, without compromising the quality of the coating [9].
The choice of materials to be used as wear-resistant coatings includes mainly cermets [9,10] and often comprises very hard phases like nitrides and carbides such as TiN, TiC, SiC, and WC and a metallic binder. WC is often the best choice among these due to its higher Young's modulus, lower thermal expansion coefficient and hardness stability over a wider range of temperatures [11]. WC usually is sprayed together with Co due to their high wettability; Co acts as a binder yielding a strong bonding to keep together the hard WC phase in forming a hard, wear-resistant coating [12]. This material constitutes one of the industrial gold standards for high wear resistant applications in harsh environments for oil and gas, metallurgy, mining and nuclear industries. Nonetheless, there is an interest in widening the applicability, durability and reliability of WC-Co coatings. Further improvements in wearresistant coatings have recently been explored by powder densification [13] or by adding small fractions (<5%) of nanofillers such as carbon-based materials as graphene nanoplatelets (GNP) and carbon nanotubes (CNT) [14] or other lamellar materials as WS 2 and MoS 2 as reinforcements phases. These carbon allotrope materials are overall characterised by high Young's modulus [15,16] and/or solid lubricating properties [17] and offer enhanced microhardness [18] and reduced friction properties [19] to the materials they are added to. In particular, CNTs are characterised by a fibre-like morphology which allows them to bend and easily adhere on curved surfaces, making them suitable for incorporation in composite coatings. These nano-enabled materials have the potential of pushing forward the current limitations of wear-resistant coatings, such as durability and friction, by providing novel, enhanced tribomechanical properties [20]. At the same time, there is still a need for developing efficient ways to incorporate CNT into coatings and understanding the interplay between their addition and the spray conditions have on the final coating performance.
In this work, CNTs nanofillers have been added to commercial WC-Co powder using a novel ultrasonication-based setup which provided efficient dispersion of CNTs throughout the surface of the powder particles, with promising scale-up possibility compared to existing jar milling methods [21]. The combination of WC-Co and CNT was sprayed using an ID HVOAF gun at three different spray conditions, where the HVOAF gases total flowrate varied to J o u r n a l P r e -p r o o f provide three different flame powers. The aim of the work is producing wear-resistant coatings with enhanced tribomechanical properties for ID application thanks to the addition of CNT nanofillers. The mechanical, structural and tribology properties of the coatings have been analysed and the influence of the three sets of spray parameters and the addition of CNT has been studied. The proposed mixing setup has the potential to be applicable to many different combinations of nanoparticles and commercial powders.

-Materials
A commercial WC-Co 83-17 powder (AMPERIT® 526.059, Höganäs, Germany) with nominal particle size of 30/5 μm and multi-walled CNT (abcr, Germany) with 50 nm nominal diameter and 1-2 μm length, were used as powder feedstock. The CNT powder was suspended in deionised water with the addition of 0.1 wt.% Sodium Dodecyl Sulphate and mixed using an ultrasonicator (Fisher Scientific, United Kingdom) for 1 hour. The WC-Co powder was then added to the suspension, with a final CNT concentration of 0.5 wt. % compared to the WC-Co weight. This process can efficiently handle 50 g WC-Co powder per batch, and needs to be repeated to reach the desired amount of feedstock. The mixture was further ultrasonicated for 1 hour; in this part of the process the WC-Co particles come in contact with the suspended CNTs which adhere on their surface in forming the composite particles. The mixture was then dried in an oven overnight at 60 °C and again mixed using a Turbula shaker mixer (WAB, Switzerland) with fixed rotational speed for 10 minutes. The CNT concentration in the final powder is again 0.5 wt. % compared to the WC-Co weight, as there is no loss during this process.

-Thermal Spray
Three sets of spray flowrate parameters were used in order to obtain three different flame powers, labelled as high power (HP), medium power (MP) and low power (LP). The corresponding normalised flowrate values are presented in Table 1. The various flowrate regimes have an effect on the combustion process inside the HVOAF gun and of particular importance is the air flowrate, which has an effect in lowering the flame temperature since only part of it will actively take part to the combustion process. The different flame power values will increase both the flame temperature and velocity as the power increases: this J o u r n a l P r e -p r o o f has the dual effect of increasing the temperature of the particles but also decreasing their residence time in the flame due to a higher flame velocity. Throughout the paper the aspect of temperature will be mainly considered as it is the major effect of the change in power, but the interplay of higher temperature-shorter residence time is always present and must be kept in mind. The spray setup is the same as described in our previous work [9] with an ID HVOAF gun [22]. The powder was fed using a volumetric disc-based powder feeder, with slight adjustments in RPM and carrier gas flowrate to keep the powder feed rate constant in all runs. The coatings were deposited on AISI 416 stainless steel discs (nominal composition: 12-14% Cr, 1.25% Mn, 0.15% C, 0.15% S, 0.6% Mo, 0.06% P, 1% Si in wt. %) of 38.1 mm diameter and 6 mm thickness. The discs were grit blasted and cleaned with acetone prior to spray.

-Wear tests
Wear tests were carried out using a Ball-on-disc tribometer (Ducom Instruments, The The wear mass loss of both coating and counterbody ball was measured using a precision scale with 5 significant figures, and specific wear rates were calculated by accounting for load and total distance. The specific wear rate was then calculated by dividing volume and mass loss by the applied load and wear test distance.

-CNT incorporation on WC-Co powder
The sonication process of the WC-Co and CNT mixture in water had a dual effect in breaking the CNT agglomerates and allowing mixing of the WC-Co particles, which over time came in contact with the CNT. The outcome of the process is presented in Figure 1, the CNTs offers an effective van der Waals attraction which keeps them attached throughout the process from drying up to the spray stage. The outcome of this process is a hybrid powder that includes both the WC-Co and CNT materials and is readily usable as a powder feedstock for thermal spray, with controlled WC-Co to CNT ratio and little reduction in flowability. This novel process has the benefit of being relatively fast and scalable compared to existing CNT incorporation processes for thermal spray [26], opening the way for the production of large batches of readily usable nano-enabled powder. to dense (white arrow).

-Cross-sections and mechanical properties characterisation
The polished cross-sections of the coatings in SEM SE image are presented in Figure 2. The deposited thickness per pass was previously measured and found to decrease with decreasing flame power, from 12.5 μm for HP, to 10.5 μm for MP and 9 μm for LP, with no difference with or without CNT. All coatings show a well-bonded interface between the coating and the substrate with no interfacial porosity. The porosity within the coatings is reported in Table 2, with very low values (< 1%) in all the coatings except in the HP WC-Co case, where porosity is higher than 4%. Different flame powers appear not to affect the level of porosity globally; however, the 4% porosity case happens to be at high power, where the flame temperature is higher. Overall, the presence of CNTs, which ensures better heat dissipation, and lower power, which lowers the flame temperature, help keep the coating temperature lower during deposition and then hinder the formation of porosity.
The high magnification microstructure of the coatings is shown in the SEM BSE images in Figure 3. Here it is possible to observe the typical WC-Co microstructure, with WC particles (bright phase) embedded in a Co binder matrix (grey phase). The porosity appears as black whereas additional features characterised by a higher atomic mass density, such as elemental W or W 2 C phases, appear as the brightest phase; these can be seen at the WC/Co interface at the outer edge of the WC particles, as marked by circles in both MP images and in the HP WC-Co image. These features are most likely present in the coatings obtained at all powers, with a different frequency from one coating to another, as they are a by-product of the high temperature spray process itself. One additional interesting feature that occurs upon spray is the WC particles size refinement, of which one clear example is marked by the square in MP WC-Co where a cracked WC grain is shown. This does not compromise the quality of the coatings and can be conversely beneficial for their performance [27]. CNTs are not visible in these images as they are below the resolution threshold for SEMs in crosssection, and furthermore, the cross-section sample preparation is too aggressive for these nanoparticles. Similarly to microhardness, a general descending trend is seen for the fracture toughness of the coatings containing CNTs as the power is decreased, possibly suggesting a less compact coating. If WC/Co samples are considered, again no trend is identifiable and the highest fracture toughness is found in the MP case; however, comparable within the error with the HP case, around 8 MPa·m 1/2 . Note that the coatings were deposited using a different number of passes, therefore the resulting difference in thickness as evident in Figure 2. The comparison of their mechanical properties is, however, reliable as their thickness is always higher than 150 μm, which is much larger than the size of the indents, hindering the contribution from the substrate or the embedding resin.

Figure 4 -Coatings Vickers microhardness (a) and Fracture toughness (b) of the coatings.
A comparison of the mechanical and microstructural properties of the coatings in this work with results reported for coatings obtained with HVOF of similar powders confirms that the best performing coatings here reported are positioned around the high-end specification for microhardness, fracture toughness and porosity [29][30][31], notwithstanding the miniaturised gun for narrow apertures and very short stand-off distance.

-XRD phase characterisation and carbide retention
J o u r n a l P r e -p r o o f XRD was performed on the WC-Co powder and on the coatings to study phase changes and decarburisation. The by-products of decarburisation are the W 2 C and elemental W phases, which are undesirable as they can worsen the performance of the coating [32]. The XRD spectra are shown in Figure 5a, where all the different phases found are marked. The WC-Co powder spectrum shows the two expected sets of WC and Co peaks and a faint amorphous region around 2θ = 30°. In all the spectra of the coatings, the WC peaks are still present, but the Co peaks disappear. This is due to the melting of the crystalline Co into the amorphous matrix that binds the WC particles together in forming the coating. Minor dissolution of C and W in the Co matrix in forming Co6W6C yielded the partially amorphous region centred at 2θ = 43°, which has more crystalline character in LP. Additional peaks appear upon spray in the spectra of the coatings, namely those corresponding to W 2 C and W phases: these peaks appear since WC decarburised due to the high temperature of the spray process.
Consequently, these peaks are more pronounced at high power and are instead very weak or non-identifiable at lower power. Quantitatively, the carbide retention index was calculated and is presented in Figure 5b.

increases. Evidence of decarburisation is shown in high and medium power as two main additional peaks from W 2 C and W appear at around 40°. b) Carbide retention index calculated for the different coatings based on the WC, W 2 C and W XRD peak intensities
according to the method in [23].

-Raman spectroscopy and CNT degradation
The low concentration of CNTs (0.5 wt. %) makes them non-detectable with XRD, where normally only phases above 2 wt. % can be properly identified. A more efficient way of detecting and studying CNTs is Raman spectroscopy. A Raman spectrum from the as received CNT powder is presented in Figure 6a, showing the D, G and 2D bands which are typical of carbon allotropes, and an additional higher-order D+D' band. The relative change in height and shape of these bands yields information on the structural degradation CNTs may have undergone [33], in particular within the single tube wall (D band) or between the different tube walls (2D) [34], in analogy with graphene layers stacking [35]. degradation, as reported from previous works involving carbon allotropes [18,36].
Furthermore, WC-Co powder has been shown to act as a carrier and to provide thermal shielding to CNTs when sprayed using HVOF, plasma spray and cold spray [37], and is confirmed in this work for HVOAF in ID configuration.

-Wear performance
Wear tests were carried out on the coatings to assess their wear resistance and friction properties. The specific wear rates in terms of volume and mass are presented in Figure 7a for the coating and 7b for the counterbody. The presented results are averages from two repeated tests, with standard error used as associated uncertainty. All the trends from the mass measurements agree with the volume measurements, validating the relevance of the results. Concerning the coating, interesting trends emerge from Figure 7a, where all the coatings containing CNT appear to have a higher specific wear rate than their WC-Co only counterparts, by approximately a factor 2. As the power changes, it can be noticed how HP and MP perform similarly, with slightly lower specific wear rate for MP, whereas LP presents higher specific wear rate, of approximately a factor 2 compared to HP and MP, which results from the overall lower microhardness as presented in Figure 4.
A different picture emerges from the counter body specific wear rate in Figure 7b. Here, the presence of CNT is associated with a significant reduction of specific wear rate in HP, whereas a slight increase is seen in MP and LP, in line with the coating behaviour.
Concerning the power, a general decreasing trend in the specific wear rate is seen as the power decreases, with LP presenting the lowest counter body specific wear rate for both WC-Co and WC-Co/CNT. The HP high specific wear rate in the WC-Co case can be explained by the much higher porosity presented by this coating as in Table 2, which can favour a more abrasive wear behaviour, particularly detrimental for the counter body. Considering the combined coating and counter body specific wear rate, MP WC-Co is the coating providing the most efficient wear resistance in these test conditions. A comparison with a similar work [14], noteworthy thanks to the small differences in the spray setup and  Additional insight into the wear tests is provided by the friction information, presented in Figure 8. Figure 8a shows the coefficient of friction over distance for the various coatings. It   Considering now the MP coatings presented in Figure 10, it is again possible to detect a more evident wear track in WC-Co/CNT compared to the WC-Co case, where only a very faint mark is visible. As a support to this, in Figure 10c it is shown only an early onset of oxidation whereas in 10d extended oxidised patches with cracks are visible. Abrasive wear is here not evident, as little debris is present and no ploughing marks can be found. The LP coatings shown in Figure 11 offer a very different picture. The wear tracks in Figure   11a and 11b are both very evident, and also narrower compared to the previous coatings obtained in HP and MP conditions. Here, the coating is wearing to a great extent compared to the ball, leading to a deep and narrow track, as confirmed by the previous wear rate analyses. The BSE images in Figure 11c and 11d both show a high amount of oxidation and cracking, in line with the higher amount of wear occurred and higher than all the coatings analysed so far. Some debris is present though not as much as in HP WC-Co/CNT case, however, ploughing marks can be easily seen as a higher fraction of the wear track has here J o u r n a l P r e -p r o o f transformed into the softer oxidised patches, more subject to ploughing marks.
An SEM-EDX characterisation, not shown here, of an oxidised patch from the wear track of sample HP WC-Co chosen as an example, has revealed both Co and W oxides are present and therefore form upon wear testing. Tribo-oxidation of both Co and W is therefore one of the mechanisms of wear.

-Effect of flame power and CNT addition
The change in flame power yields several noticeable effects in the coatings and their performance, although it is not always possible to detect a monotone trend in the properties of the coatings as the power changes. As reported in Table 1 The addition of CNT has shown a variety of effects on the structure, composition and tribomechanical properties of the coatings. CNT thermal conductivity has been reported in the 2800-6000 W/mK range [38][39][40], which is one order of magnitude higher than that reported for WC-Co (60-160 W/mK) [41,42]. It is worth noticing that wear performance, when adding CNTs, generally decreases in the other two cases: medium and low power. CNT are found here to generally decrease the fracture toughness of the coatings, as does lowering the power from high to low. The combined effect of these two parameters brings the fracture toughness of MP CNT and LP CNT below the threshold of 5 MPa·m 1/2 , leading to a poorer wear performance and therefore hiding any possible benefit of CNT addition to wear performance, which in fact emerges only in the HP CNT case. Overall, the coatings showing a higher fracture toughness are also those showing a better wear performance (lower specific wear rate). Since one of the main failure mechanisms is cracking, with following material removal and debris J o u r n a l P r e -p r o o f formation, the coatings with higher fracture toughness are better at reducing this wear mechanism.
The three different flame power parameters also influence the CNTs in-flight. Even if CNTs are known to be able to survive thermal spray [14], as are other 2D materials both as nanocomposites [19] and alone [17], a certain amount of degradation should be foreseen due to the high temperature and oxygen presence in the flame. This degradation has not been detected in this work, where CNTs in the coatings showed pristine Raman spectra, but may be detected by a comparison of the concentration of CNT in the final coatings compared to the starting powders, which is beyond the scope of this work. The heat resistance of CNTs is remarkable even at high temperatures of thermal spray, especially due to their high thermal conductivity and strong covalent bonded structure. Regarding oxidation resistance, the HVOAF flame can be tuned to provide different stoichiometries of fuel to oxygen which in this work are around 2.1% for HP conditions and 1.7% for MP and LP. This yields a slightly reducing atmosphere for HP and oxidising atmosphere for MP and LP, which suggests an overall lower degradation of CNTs upon thermal spray and adds up to the explanation why HP coatings are the most benefited by the addition of CNTs. Another aspect that goes beyond the scope of his work is the change in CNT concentration. The addition of a higher concentration of CNT has been shown to improve the tribology and wear properties of CNT composites; this at the expense of microhardness, which in turn is reduced [43], and fracture toughness. Therefore, it is advisable to tailor the CNT concentration according to the aimed application and optimise tribological or mechanical properties.

Conclusion
WC-Co/CNT coatings have been deposited with HVOAF and the effect of changing spray parameters as well as the effect of CNT addition has been investigated.
CNTs were successfully incorporated using a novel setup in commercial WC-Co powder for thermal spray, undergoing no or minimal degradation upon ID-HVOAF thermal spray as confirmed by Raman spectroscopy. Low power and the presence of CNTs helps hinder decarburisation by keeping the process temperature lower and favouring heat dissipation.
The HP conditions are the most benefited by the addition of CNT, with a reduction in coefficient of friction by half, an increase in microhardness and a decrease in porosity and J o u r n a l P r e -p r o o f counter body wear rate. However, not all the samples benefited by the addition of CNT as the decrease in fracture toughness can disrupt wear performance.
The applicability of this process extends beyond the scope of this work offering a practical way to develop various combinations of commercial powders and nanofillers.