3.1. Powder Deposition
shows the particle size distribution of the feedstock powders. It can be observed that the prealloyed composition (powder 1) was characterized by the normal distribution that was centred at a mean size of 30 µm. On the contrary, the experimental self-decomposed and SHS powders were of non-symmetrical distribution. Powders 2 and 3 contained a larger amount of fine particles with d
10 = 3 µm/d
90 = 56 µm and d
10 = 7 µm/d
90 = 68 µm. The flowability of powders 2 and 3 was quite low.
Each of the three powders revealed an irregular morphology. The BSE-SEM micrographs of the cross sections showed the uniform composition of powder 1, whereas the other two presented the varied degree of greyness (Figure 2
); the distribution of the phases in powders 2 and 3 from one particle to the other was considerably different (Figure 2
b,c), especially in the last one, where some particles exhibited a laminar structure.
shows the XRD of the powders. The peaks of powder 1 were identified as the fundamental lines of the FeAl intermetallic (h
= even), which appeared in the disordered lattice; no superlattice lines (h
= odd) were observed. The occurrence of broad peaks was related to the fine grain size and microstrains resulting from the milling. Powder 2 showed the presence of different intermetallic Fe-Al phases, mainly Fe2
, together with some SiO2
, while powder 3 contained FeAl, FeAl2
, and FeAl2
. In [35
], the self-decomposing process was analyzed in detail; these authors put forward a hypothesis about the self-decomposition of the Fe-Al-C-Me alloys (Me = Ni, Mn, Cr, Mo, V, B, Si). The powder was composed of iron aluminide intermetallic base with different Al content and thin oxide films as particle shells; this allowed for the obtainment of nanocomposite coatings.
From each powder, the same coatings were deposited. Figure 3
shows the as-sprayed cross sections of the powders 1, 2, and 3, with 103 ± 9, 76 ± 13 and 93 ± 11 µm of coating thickness respectively, after spraying nine layers. The coating obtained from the prealloyed powder (powder 1) was quite uniform in thickness whereas the other two were less homogeneous; the roughness of the coatings was found to be Ra
= 3.06 ± 0.6, 4.3 ± 0.3, and 6.8 ± 0.4 µm, respectively.
The detailed inspection of the HVOF coating microstructures revealed the occurrence of the typical lamellar structure of thermal sprayed coatings with light and dark contrasts between splats and porosity (Figure 4
). In coating 1 (Figure 4
a), these were identified as iron-rich and oxide zones by EDS, respectively, and it was observed that heating the particles in-flight time resulted in the depletion of aluminium [36
]. XRD confirmed the presence of oxidation (Figure 5
); the additional peaks identified as FeAl corresponded to the superlattice lines in accordance with the ordering process when reaching the melting temperature [37
The grey region in coating 2 consisted mainly of FeAl2
, while the lightest phases at the intersplats were Fe-rich areas (Figure 4
b). Some porosity was observed but not much oxidation, as detected in coating 1. In coating 3, the lightest regions were poorer in aluminium than the middle-grey ones that were identified by the Fe3
Al phase (Figure 4
The heterogeneous structure in the sprayed coatings may affect different properties, i.e., the occurrence of Al-depleted regions may lead to susceptibility to corrosion or oxidation and intersplat oxides may result in an increase of coating hardness and wear resistance; all of these features can be controlled by process variables i.e., spraying distance, oxygen and fuel flow rates, and particle size.
3.2. Coatings Performance
As it can be observed in Table 2
, the microhardness values of coatings 1 and 2 are quite similar despite exhibiting different porosity levels, while the properties of coating 3 differ significantly, displaying a lower value due to its higher porosity and the amount of cracks in the as-sprayed condition. Actually, Figure 6
shows two indentations that were made in coatings 2 and 3 and illustrates the occurrence of many cracks as the result of applied force. It was observed that some of these cracks propagated along the intersplat boundaries and can be explained by the embrittlement caused by the occurrence of the Al-rich phases, namely Fe2
In many cases, the hardness values were essential to predict the wear trend among different materials, according to Archard’s equation [39
], but it was also related to the characterized type of wear; therefore, when performing such an analysis, caution must be taken. By this mean, the abrasive performance can be predicted more accurately than sliding wear when hardness is taken into consideration. In addition, at high temperatures, the damage was increased due to the additional oxidation effect [40
]. Figure 7
presents the XRD of the samples after the wear tests at 800 °C, showing the oxide products beside the wear tracks. For coatings 1 and 3, the formation of the fast growing oxide Fe2
at 800 °C, was observed; coating 2, however, showed similar phases as under the sprayed conditions, but with narrower peaks, which might be attributed to grain growth.
shows the variation of friction coefficient versus sliding distance at a normal load rate of 5 N for the different coating systems as well. It is was found that the friction coefficients fluctuate, which is probably due to plowing/delamination/oxidation mechanisms [12
]; observations from Zhang et al. [12
] are in agreement with that. Coating 1 presented a similar final COF as coating 3, while coating 2 peaked at a higher value; in addition, coating 1’s curve was in fact quite smooth compared to coating 3, which presented ups and downs along the test; coating 2 however, started decreasing and then progressively increased afterwards, reaching a more stable value during the last few meters of the test.
The high initial values and the immediate drop can be attributed to the asperity contact given by the increased surface roughness initiated by oxide nodules. Coatings 1 and 3 tested at 800 °C, both formed Fe2
nodules (Figure 7
and Figure 9
), but the wear rates that were observed in Table 3
were quite different, i.e., negligible in coating 1 and high for coating 3. Given the positives tests of coating 1, this was also examined at 400 °C and proved a higher friction coefficient than at 800 °C, convergent with what was found at room temperature [31
]. Such differences clearly indicated a transition of friction regimes at both temperatures influenced by the predominance of different wear mechanisms. Zhu et al. [41
] reported a decrease of wear rate of undoped Fe-40Al with increasing temperature up to 500 °C. Other authors have reported that more rapid oxidation at 800 °C than at 400 °C may stabilize the adhesion of the oxide layer reducing friction [42
At 800 °C, the removal of oxide through sliding contact and reoxidation in coatings 1 and 3 was rapid enough to avoid Fe2O3 detachment as debris particles; it was also assumed that splat cohesion in coating 1 was good enough to prevent delamination mechanisms, which explained its low wear rate. Nevertheless, the presence of brittle Fe2Al5 and FeAl2 phases in sprayed coating 3 might induce subsurface crack nucleation and fatigue mechanisms that would result in accelerated delamination. Coating 2 originally showed the presence of the Al-rich intermetallic phases, but the oxidation was much less extensive, and, although the same phases were identified at 800 °C, the friction coefficients followed different trends.
Not only was the mean average of the friction coefficient value presented in Table 3
, but also the value of wear tracks widths, depths and wear rates. Coating 1 displayed the lowest rates at both tested temperatures. The higher wear at 400 °C when compared to 800 °C was in agreement to the highest friction rate and might be attributed to the particulate adhesive wear debris observed during the examination of the wear track that remained between the contacting surfaces during sliding (Figure 9
]. Additionally, the two-dimensional (2D) profiles of the wear tracks presented in Figure 10
confirmed the smoothness of the wear track of the coating tested at 800 °C; the roughness profile was indicative of the oxide layer formation.
Coating 1 showed better wear rates than high velocity arc sprayed (HVAS) coatings tested at room temperature by Xu et al. at 7 N, with a Si3
ball (55.4 × 10−5
/7 N = 7.91 × 10−5
]. After the examination of the wear tracks, at 800 °C, coating 1 presented small Fe2
nodules formed by continuous oxidation of the freshly deposited coating (top left inset in Figure 9
b); the top right inset showed the morphology of the oxidized surface with iron oxide nodules and needles. Coating 2 showed severe wear with patches of delaminated coating (Figure 9
c). Coating 3, however, was formed by patches of fresh and oxidized coating (Figure 9
d). The wear mechanism of coating 3 at 800 °C consisted of continuous formation and removal of a thick “glaze” tribofilm that was formed by smearing and sintering of iron oxide debris on the sliding surface. The high testing temperature was sufficient to promote the occurrence of such a dense glaze layer without the additional flash temperature between contacting surfaces.
The high wear rates at 800 °C of coatings 2 and 3 stemmed from the larger contact area, created by the lack of splat cohesion; the reported nanocomposite features that were reported by Senderowski et al. [38
] did not seem to positively impact the wear performance. Moreover, it was proved that low intersplat cohesion favored oxygen penetration, which weakened the boundaries. The failure in coating 3 is well illustrated in Figure 11
through the cross section of the wear track, where the substrate was reached; this is in agreement to the predominance of iron oxide in the XRD analysis and the formation of the tribofilm layer, although both of the coating compositions were quite different, the performance appeared to be related to the microstructure, occurrence of brittle phases, subsequent delamination, and rapid oxidation. Other microstructural heterogeneities within the coating, such as pores, intersplat oxides, and other imperfections serve performed as stress concentrators. This lead to favorable conditions for crack initiation.