Comparative analysis of microstructure and selected properties of WC-Co-Cr coatings sprayed by high-velocity oxy fuel on S235 and AZ31 substrates

Two types of metallic materials, structural steel S235JR and magnesium alloy AZ31, with a thickness of 5 mm and a diameter of 25 mm, were used as substrates. Before the deposition process, the substrates’ surfaces were sand-blasted with corundum powder (F40 according to the FEPA standard) and cleaned in an ultrasonic bath with ethanol.

The feedstock material was a commercially available powder Amperit 554.071 (Höganäs): WC-10Co-4Cr (wt%) with an apparent density in the range 4.7 to 5.6 g/cm³. According to the manufacturer’s declaration, the particle size range was −45 + 15 μm, and the delivery condition was agglomerated and sintered. The SEM image confirmed this information (Fig. 1). Additionally, the particle size distribution was assessed by the laser diffraction method using PSA 1190 (Anton Paar). The average particle size of the initial feedstock was equal to 34.2 μm.

Fig. 1.

The manufacturing process of the cermet coatings on the different substrate types was realized using a JP 5000 gun from the TAFA system. This gun operates with kerosene as a liquid fuel medium, and the powder is injected in radial mode. The scheme of this system is given in Figure 2.

Fig. 2.

The process parameters were optimized for steel and magnesium alloy separately (Table 1). As can be seen, the two factors are clearly different. The spray distance for AZ31 is longer because of the flammability of the magnesium, and the oxygen to fuel ratio (λ_O/F) [27] is equal to 1.22 and 1.02 for steel and magnesium alloy substrate, respectively. The ratio is connected with the specific properties of each of the substrate materials.

The topography of manufactured cermet coatings was observed using scanning electron microscopy (SEM) (Supra 35, Zeiss). The surface roughness in the as-sprayed state was measured with the use of the MarSurf PS10 profilometer (Mahr) according to the ISO 4288 standard. The average value of the Ra, Rp, Rv, and Rz indicators as well as standard deviation values from five measurements were calculated. In order to carry out microstructure investigations, samples were cut on a precision cutter, then ground with SiC papers up to a grit size of 2500 and polished with a 3-μm diamond suspension. The microstructure of deposited samples from these prepared cross sections as well as the coating’s thickness were examined using SEM apparatus. Ten measurements were taken, and the average value with standard deviation was calculated. The coating’s porosity was also assessed on the cross sections, according to the ASTM E2109-01 standard. The calculations, based on 20 SEM images at 1000x magnification for each coating, were carried out using ImageJ software (version 1.50i). The phase composition was estimated by X-ray diffraction (XRD) using a D8 Advance diffractometer (Bruker) with a Cu-Kα cathode (wave length equal to 0.154 nm). The operating current and voltage were equal to 40 mA and 40 kV, respectively. The measurements were carried out for a 2θ angle in a range from 20° to 100° with 0.60°/min scan rate and 0.02° scanning step. The phases were fitted and identified using DIFFRAC.EVA software with ICDD PDF#2 database.

The Vickers microhardness test of the coatings was carried out according to the ISO 4516 standard with values of maximum load and dwell time equal to 2.94 N (HV0.3) and 15 s, respectively. For each coating, 15 imprints were made and then the average values with standard deviation were calculated, and in both coatings the fine structure was obtained. For this purpose, the instrumental indentation technique (IIT) was used in order to measure the hardness of the carbide particles and metallic matrix. The instrumental hardness (H_IT) assessment was carried out on the samples’ cross sections, according to the ISO 14577-4 standard, using NHT³ nanoindenter (Anton Paar) with the Berkovich indenter. The value of the maximum load was equal to 50 mN. Ten imprints for carbides and matrix area were carried out for both samples. The Oliver-Pharr method was used in order to calculate the instrumental elastic modulus (H_IT) [28]. Using the same apparatus, it was also possible to assess the instrumental elastic modulus (E_IT) of deposited coatings. The value of maximum load ranged from 100 mN to 500 mN, and E_IT estimation was made based on methodology proposed by Tricoteaux et al. [29] and modified by Łatka et al. [30]. For each sample, five repetitions were carried out. In the case of E_IT assessment for tungsten carbide particles and the metallic matrix, the the maximum load ranged from 25 mN up 100 mN.

The fracture toughness (K_IC) of the manufactured coatings was estimated using the indentation method, with the Vickers indenter, according to the ASTM C1421 standard. The value of the maximum load was equal to 98.1 N, and 10 imprints were made for each sample. The average value of K_IC was calculated with Equation (1): 1 KIC=0.079⋅(Pa3/2)⋅log(4.5⋅ac) \[{{K}_{IC}}=0.079\cdot \left( \frac{P}{{{a}^{3/2}}} \right)\cdot \log \left( \frac{4.5\cdot a}{c} \right)\]

In Equation (1) the elements are as follows: P is maximum load, N; a is indentation half diagonal, m; c is crack length from the center of the indent, m. This model was proposed by Evans and Wilshaw [31] and adapted to HVOF coatings by Liu et al. [32].

The topography of manufactured coatings is presented on the top-surface SEM images shown in Figure 3. In general, it is a typical morphology for HVOF coatings [33, 34]. The detailed observations confirmed a generally smooth surface with some minor irregularities. The presence of the defects could be related to phenomena occurring in the spray process. Carbide particles remain solid in the flame because of their high melting point and relatively high heat capacity, whereas metallic matrix becomes liquid [35]. Moreover, some partially melted or even unmelted fine particles could be seen. This could be related to a lower particle inertia moving them on the flame’s periphery [4].

Fig. 3.

The results of surface roughness measurements expressed by the most important parameters are collected in Table 2. The data obtained are in good agreement with the literature [36, 37]. It should be stressed that in the vast majority of applications the surface is mechanically treated. On the other hand, when the surface roughness for as-sprayed surfaces is lower it facilitates the finish machining of coatings.

The microstructure observation on SEM images was carried out at low (500x) and high (10,000x) magnification. In Figure 4, it can be seen that coatings exhibit dense, compact and uniform structure that is typical for the HVOF process [38, 39]. Moreover, at the interface the tightly adhering coating can be seen. It is the result of proper substrate preparation and reflects the advantages of the HVOF process, which is the high kinetic energy of the molten particles and/or highly plasticized powder particles [40, 41]. It strengthens mechanical bonding between coating and substrate and improves adhesion.

Fig. 4.

The coatings’ microstructure observed at high magnification (Fig. 5) also confirms a good quality of deposits. Uniformly distributed carbides particles in metal matrix are wettable by the Co- Cr mixture. Nevertheless, there are some minor material discontinuities, such as cracks and pores. These defects could be a result of thermal stresses that occurred during the deposition process and also the fact that the feedstock particles stayed in the flame for a relatively short time [42, 43].

Fig. 5.

On the coatings’ cross sections, additional analyses were performed in order to estimate porosity. The results show that MBS exhibits a slightly lower porosity value (2.3 ± 0.4 vol%) than that for SBS (2.8 ± 0.7 vol%). For both samples, these values are similar to the data fouind in the literature [44, 45]. Another issue is the dimensions of the observed pores. Mostly, they are less than 1 μm. Such fine pores in combination with compact structure could be beneficial from the point of view of an improvement in corrosion resistance [46].

The analysis and results above are strongly connected with process parameters (Table 1). In the HVOF method, increasing the oxygen and fuel flow rate results in an increase of the particle velocity and temperature. This trend, however, is only for “critical O₂ flow rate,” which is about 850 L/min. Then, with increasing flow rate, the particle temperature decreases while particle velocity continues to increase. In case of kerosene, the “critical flow rate” is about 0.385 L/min, but after crossing that point, the changes differ, as in case of oxygen flow rate. Particle temperature decreases, but particle velocity remains constant [47]. Taking into account the kerosene–oxygen mixture, the excess oxygen cools down the flame, thus reducing particle temperature. On the other hand, particles move in the flame at a higher velocity (for greater oxygen flow rate), which results in reducing the residence time in the HVOF flame [48]. This could provde an explanation for why the porosity of the SBS sample is slightly higher than that of MBS.

The XRD diagrams for deposited coatings are presented in Figure 6. In both cases, there are two main phases: WC and W₂C, except for some minor amounts of Co_0.9W_0.1 and Co₃W₉C₄ that were detected. Similar results were obtained by Pulsford et al. [49] and Komarov et al. [50].

Fig. 6.

Additional information, which can be extracted during XRD diagram analysis, is the carbide retention index (CRI). More details about this indicator can be found in articles by Liphout and Sato [51] and Górnik et al. [52]. This factor is calculated with Equation (2): 2 CRI=IWCIWC+IW2C+IW \[CRI=\frac{{{I}_{WC}}}{{{I}_{WC}}+{{I}_{{{W}_{2}}C}}+{{I}_{W}}}\] where the intensity I of the diffraction peaks corresponds to WC, W₂C, and W measured at the positions 2θ = 35.7°, 2θ = 39.8°, and 2θ = 40.5°, respectively [53]. The calculated values of the CRI are collected in Table 3. The results are similar to findings reported by Picas et al. [54].

Determination of the crystallite sizes was carried out on the basis of the width of the diffraction curves of the crystal plane measurements and Scherrer formula [55] according to Equation (3): 3 Dp=0.94⋅λβ⋅cosθB \[{{D}_{p}}=\frac{0.94\cdot \lambda }{\beta \cdot cos{{\theta }_{B}}}\]

In Equation (3), the elements are as follows: Dp is the crystallite size; λ is the wavelength of X-ray radiation; β is the full width at half maximum (FWHM) of the peak intensity; θ_B is the Bragg angle of reflection of a specific crystal plane.

Figure 7 presents the calculated crystallite size for the main phases (WC and W₂C). The values are slightly lower for the SBS sample. However, the differences are within the value of the standard deviation. In the case of WC, the crystallite size is less than 90 nm, whereas for W₂C it is less than 30 nm. These values are similar to those obtained by Tillmann et al. [56].

Fig. 7.

The results of microhardness measurements of the manufactured coatings confirmed the established hypothesis that proper selection of HVOF parameters results in no change in the mechanical properties, regardless of the substrate material. The results obtained are highly correlated with the porosity level and are as follows: 1192 ± 114 HV0.3 and 1238 ± 105 HV0.3 for SBS and MBS, respectively. These values are close to the those found in the literature [57, 58].

The IIT results for carbide particles and the metallic matrix are presented in Table 4. The differences are within the range of the standard deviation. Values obtained are similar to those in the literature [59, 60]. Slightly lower E_IT values for tungsten carbides could be a result of the spraying process and, in part, the decarburization phenomenon [61]. On the other hand, the E_IT values for sprayed coatings was practically identical, and it was equal to 344 ± 11 GPa. Results that were congruous with these were obtained by Culha et al. [62] and Matikainen et al. [63].

In order to ensure high wear resistance, which is a fundamental property of the WC-Co-Cr coatings, not only high hardness but also high fracture toughness (K_C) is required. The results for deposited coatings are as follows: 4.72 ± 0.68 MPa · m^1/2 and 4.57 ± 0.75 MPa · m^1/2 for MBS and SBS, respectively. These values are in strong agreement with literature data [64, 65]. The differences in KC values for the coatings investigated could be a result of different process parameters. It is well known that fracture toughness depends on, among other things, oxidation decarburization [66]. In current studies, the flame for the SBS sample exhibits a more oxidative nature. The flame used for MBS limits the formation of a brittle W₂C phase.