Influence of Heat Treatment on Fracture Toughness and Wear Resistance of Nicral-Zro2 Multilayered Thermal Barrier Coating

Material preparation

Two kinds of blended powder feedstock were manufactured through ball milling technology before plasma sprayed on the surface of medium carbon steel with a dimension of 16×12×8 mm³. The mixture was milled by using alumina balls with a rotation speed of 180 rpm at the duration of an hour, and the alumina balls to powder mass ratio were 2:1. Blended powders consist of two components in different proportion. The inner layer of the coating was sprayed by the mixed powders of 70 mass% YSZ (ZrO₂-8 mass% Y₂O₃, an average particle size range of −75+38 μm) and 30 mass% NiCrAl (Ni-17Cr-13Al, an average particle size range of −106+45 μm) while the outer layer was of 80 mass% NiCrAl and 20 mass% YSZ.

Coatings preparation and heat treatment

The multilayered coating was deposited with optimal spray parameters by atmospheric plasma spraying (APS). The process conditions for the deposition are summarized in Table 1. Vacuum oil-quenching and gas-quenching in web-strip hot furnace technologies were applied on the as-sprayed multi-coating samples in order to eliminate pores or splat boundaries as well as create metallurgical bond between substrate and multi-coating. In the paper, two kinds of the heat treatment were compared in terms of microstructure, fracture toughness, thermal barrier and wear resistance.

Prior to atmospheric plasma spraying, the medium carbon steel substrates were grit blasted by Al₂O₃ particles to improve the adhesion between the substrate and coating. After the deposition, a group of multilayered coating specimens were put into the high vacuum sintering and heat treatment furnace conducted at 820 ℃ and 1010 ℃. The heat rate was set 8 ℃ /min from the room temperature and each of the duration time was 2 h. Afterwards the specimens were quenched by oil. Another group of specimens were placed in bright annealing furnace in oxygen-free atmosphere at 850 ℃ for 2 h at the same heat rate as vacuum sintering and quenched to room temperature in 60 min. Subsequently, the specimens were marked respectively as follows:

1. as sprayed multilayered coating (AS-MC)

2. gas-quenching multilayered coating (GQ-MC)

3. oil-quenching multilayered coating (OQ-MC)

Coating characterization

Microstructures of different multilayered coatings as well as worn surface were examined with a scanning electron microscopy (SEM, Philips XL30, the Netherlands) in the secondary electrons (SE) mode. The chemical composition of the coating cross-section and worn region was examined with energy-dispersive X-ray spectroscopy (EDX). A series of SEM images were obtained for each sample. Phase compositions of top-coat and powder were analyzed with a standard X-ray diffraction (XRD, Philips X’pert, Netherlands) technique. All XRD patterns were recorded by Cu-Kα radiation (λ=0.1542 nm) between 10 ° and 90° 2θ with a step size of 0.03°.

Vickers microhardness tests were performed on polished cross sections of the coatings under a 1.96 N load and a 15 s dwell time with a digital microhardness tester (MHV-1000Z/V3.0). Over ten indentations were made at different zone of the cross sections and used to calculate the average coating hardness. The fracture toughness of coatings was determined by the Vickers indenter applying 196 N load on the coating cross sections. The distance between each indent was nearly three times longer than the indent diagonal in order to deter the influence of the residual stress from the former indent. Generally cracks measured by optical microscope (KEYENCE/VH-Z500R) were found parallel to the coating/substrate interface. The fracture toughness (Kc) values were calculated by the following equation [12]:

where Kc is interfacial fracture toughness (MPa m^1/2), P is the applied indentation load (N), a is the indentation half diagonal (mm) and c is the crack length from the center of the indent to crack edge (mm). The recommended c/a ratio for use in this equation is 0.6 e crack leng

A reciprocation sliding wear testing machine (SRV, Optimol Instruments Prüftechnik GmbH) with a ball-on-disc configuration under dry sliding conditions was used to measure the coefficient of friction by a piezoelectric load cell below the specimen holder. The normal loads of 50 N, 100 N and 200 N were applied via the counterbody which was a nitrided steel ball mounting from the top, driven by an oscillating actuator. Before the wear test, the specimen and the nitrided steel ball were cleaned by ethanol in order to detach grease and dust impurity. During the sliding wear test, the temperatures of the bottom surface of the specimens were measured by the resistance thermometer equipped in the specimen holder. The wear track was observed by non-contact 3D surface topography instrument (BMT, Breitmeier Messtechnik GmbH).

Thermal cycling oxidation tests were conducted at 600 ℃, 800 ℃ and 1000 °C for 60 h. The oxidation behavior of each sample was evaluated through measuring the weight gains. In the first 5 h one thermal cycle consisted of 15 min ramp-up, 0.5–2 h isothermal soak at the set temperature, and 45 min cool-down to ambient temperature (~25 °C). Afterwards, the duration time of the isothermal temperature was increased to 5–15 h. The weight gains of the specimens were calculated before and after each cycle using an electronic balance with the precision of 0.1 mg.

Microstructural characterization and phase analysis

The cross-sectional SEM images of as sprayed multilayered coating are shown in Figure 1(a). It can be observed that microcracks and voids are present in both layers. Nonetheless, mechanical anchorage is the main adhesion mechanism of spray coatings, and the interface between outer layer and inner layer is difficult to identify, which indicates strong adhesion bond strength between them. Some voids can be observed clearly on the bottom of coating, which results from the fact that refractory ZrO₂ particles impacting on the substrate cannot form laminar structure. Subsequently the initial geometries of particles were distinctively observed in the inner layer sprayed by 70 mass% YSZ. From the EDX analysis (Figure 1(b)), it can be inferred that the hemispherical ZrO₂ particle was partially melted. A large number of cracks also existed in the ceramic Al₂O₃/YSZ coating especially at top-coat/bond-coat interface because local compressive stress was generated around ZrO₂ particles [5]. Few melted metal particles were squeezed between the “mosaic” ceramic in plasma sprayed functionally graded ZrO₂/NiCrAl coatings [13]. Point B in the short and dark grey stripes shown in Figure 1(a) implies Al₂O₃ forming during plasma spraying process, confirmed by the EDS result in Figure 1(c). Aluminum was precipitated from NiCrAl powder and formed the oxide Al₂O₃ along the lamellae surfaces or boundaries.

Figure 1:

(a) Cross-sectional morphologies of AS-MC, (b) EDS spectra of point A, and (c) EDS spectra of point B.

Figure 2 presents the SEM images of GQ-MC and elements distribution measured with EDX line scanning. It can be observed that the irregular shaped voids become more rounded, even though the number of them seems to make no difference. The distributions of nickel and chromium are almost homogeneous with neighboring aluminum, because Al₂O₃ is formed along the nickel boundaries. C. Zhu pointed out that the formation of Al₂O₃ on the X-ray-detectable layer of YSZ was helpful/supportive to stabilize the m-phase in Al₂O₃ /YSZ top-coat [5]. Zirconium was discovered at the completely opposite locations. In addition, it is noticeable that a certain slope of intensity curves has been detected near the interface between inner layer and substrate. This indicates that ferrum and nickel percolate or diffuse through the interface. The diffusion is more obvious in Figure 3(b) from the decreased slope of metal elements. As can be seen from Figure 3(a), the boundaries of the roughness from foregoing grit blasting are hard to distinguish and many voids or holes have been close because of the high temperature during vacuum oil-quenching process. Zirconium does not dissolve with other metal elements as seen in Figure 2(b). The content of aluminum in OQ-MC decreases perhaps because extra aluminum does not form into Al₂O₃ during plasma spray but has been evaporated in vacuum heating furnace or dissolved in (Ni, Cr, Al) solid solution.

Figure 2:

(a) Cross-sectional morphologies of GQ-MC, (b) EDS line scan profile of GQ-MC.

Figure 3:

(a) Cross-sectional morphologies of OQ-MC, (b) EDS line scan profile of OQ-MC.

XRD pattern in Figure 4(a) shows that NiCrAl powder is composed of γ-(Ni, Cr) and pure aluminum phases. Blended powders are composed of tetragonal, cubic and monoclinic YSZ phases according to XRD pattern in Figure 4(b). XRD pattern in Figure 4(c) indicates that a few Ni₃Al precipitates (γ′) dispersed in solid-solution, chromium-enriched nickel matrix (γ). The volume fraction of the γ′ precipitates can be controlled by aging time [14]. The fact that nickel and chromium distributed uniformly throughout the coating according to X-ray mapping indicates that they co-exists together in the core of the splat [15]. During the process of thermal spray the monoclinic ZrO₂ in the original powder had transformed into cubic phase. Then most of the melted particles turned into the non-transformable tetragonal t’ phase of zirconia rather than monoclinic ZrO₂ due to the high quenching rate (10⁶–10⁷ K/s) of the molten droplets [16]. The existence of t’-ZrO₂ is favorable for the stability of the coating under high temperature. Aluminum alloyed with internal NiCr or forms into Al₂O₃ increased the difficulty in detecting as reported in Ref [17]. XRD pattern in Figure 4(e) is almost the same as that in Figure 4(a). Nevertheless, the narrower full width at half maximum intensity in Figure 4(d) indicates alloy phase becomes coarser after heat treatment because of the slower cooling rate.

Figure 4:

XRD patterns of original powder of NiCrAl (a), blended powders of outer layer (b) and AS-MC (c), GQ-MC (d), OQ-MC (e).

Hardness and fracture toughness assessment

Since the microstructure of multilayered coatings depends on the distance from the surface, Figure 5 shows the hardness of multilayered coatings before and after heat treatment as a function of displacement into surfaces. The effects of gas quenching and oil quenching on hardness of coating are different. The increase of the cooling rate in oil-quenching process increased the hardness of the substrate. Moreover, even with the high scatter in the measurement points, a clear trend can be seen with increasing hardness of the whole coating except the interface between inner layer and substrate. This phenomenon could be attributed to the mutual diffusion of elements as well as the existence of transition layer. After the coating was annealed in gas-quenching process, the residual stress caused by rapid cooling of molten particles during thermal spray was relieved which could have a negligible influence on the hardness values.

Figure 5:

The hardness curves of multilayered coatings before and after heat treatment.

Deviations and uncertainty of the crack lengths of thermal sprayed cermet coatings in the indentation fracture toughness measurement seem inevitable for their brittleness and inhomogeneity. Nevertheless, the proposed model for determining the fracture toughness of sprayed coatings, where cracks other than Palmqvist or half-penny/radial- median cracks were developed, was established [18]. According to the indentation test results, the specimens did not fracture until the indentation load reached 196 N. Unlike the WC–12%Co and APS Al₂O₃ specimens, whose cracking pattern includes edge cracks, ring cracks and spallation, radial cracks of the multilayered coatings originated from indentation corners or near the center of the indentation edges, and their fracture toughness is listed in Table 2. Two radial cracks presented in Figure 6(a) developed from the center position of the indentation edges where the highest tensile stress occurred as indicated by Baung et al [19]. Moreover, similar result was found for the Al₂O₃–3TiO₂ and ZrO₂–8Y₂O₃ coating [20]. This probably results from the microstructure of the coatings that depended on the thermal spray process such as powder composition, spray distance, torch traverse speed. Figure 6(a) shows the indenter impressing at the interface of the outer and inner layers. It is worth noticing that the shape of two wavy cracks on the left side of the indent corner corresponds to the laminated structure of the thermal coating, which suggests that poor bonding strength exists at the interface between the outer and inner layers. In addition, crack is more easily to develop in the brittle ZrO₂ material as revealed in Figure 6(a) and Figure 6(b). It can be deduced that gas-quenching make no improvement on cohesion bonding strength of multilayered coating. No obvious crack appears in the outer layer owing to the slower cooling rate compared to as-sprayed process resulting in decreasing residual stress. Thermal stress caused by different thermal expansion coefficients increased through bulk heat treatment. However, cracks became shorter in Figure 6(c) or even did not appear at some dents after the process of oil-quenching. The possible reason is that the NiCrAl particles remelted and increased the density of the coating.

Figure 6:

Indentation test imprints: (a) AS-MC (b) GQ-MC (c) OQ-MC.

Friction and wear results

Coefficient of friction (COF), which is affected by the parameters of plasma and HVOF spraying [21], depends on blended powder containing solid lubricant. As can be seen from Figure 7, apart from the heat treatment of oil-quenching, no significant influence on coefficient of friction is observed, which is also mentioned in Ref [9]. The COF of the coupons decreases with increasing loads and the measured values are more or less similar under the same load, excluding coefficients of OQ-MC maintain low values with small amplitude under 50 N and 100 N forces. The temperature measured at the bottom surface of the specimens is depicted in Figure 8. It is clear to see that the variation tendency well agrees with the trend of the COF presented in Figure 7. The measured temperature increases in the duration time as the COF slightly rises. The elevated temperature of the coupons, treated by different technology, increases with the load values, but shows almost no distinction under the same load. However, in the case of oil-quenching coating, the temperature remains low and steady under the loads of 50 N and 100 N, but increases rapidly under the load of 200 N after certain time and the temperature has a corresponding increase. This is probably because of the direct contact between the steel ball and the inner layer which can retain the original morphology for its high melting point. Therefore, more friction energy arises from higher COF when the steel ball slides against the rough surface of the inner layer with no ductility. The remelting can enhance the cohesion of the outer layer, which leads to positive influence on thermal barrier and sliding wear resistance. The temperature increments of bottom surfaces are proportional to the loads and the COF.

Figure 7:

COF of specimens against different loads.

Figure 8:

Temperature of bottom surface.

Worn surface (Figure 9) reveals wear resistance of the multilayered coating at a normal load of 200 N after different post treatments. There are two images of line scan across the center of the wear track parallel and perpendicular to the sliding direction on the right side of each three-dimensional graphic, and the sliding wear made the coating smoother. Some salient points existing at the circular form of the wear track suggest the hard particles were pulled out during wear duration. Wear depths of all the specimens are summarized in Table 3. It is observed that the oil-quenching multilayered coatings have the shallowest wear depth, followed by the gas-quenching multilayered coating. It is shown that the applied loads have a major impact on wear resistance as the wear depths increase more rapidly, and AS-MC exhibits the poorest wear resistance.

Figure 9:

Surface topographies produced by profilometry (a-c) AS-MC, (d-f) GQ-MC, (g-i) OQ-MC.

Tribo-oxidation was most often reported as the main wear mechanism [9]. The wear rate depends on the number of oxides stripped out of the friction system. The differences between the wear mechanisms of selected coatings – AS-MC, GQ-MC and OQ-MC – are displayed in Figures 10, 11 and 13. It can be perceived in Figure 10(a), (c) and (e) that the width and length of the wear track do not show much difference but the wear morphology changes greatly. Figure 10(a) illustrates indistinct scratch marks caused by the grinding against the nitrided steel ball exist in the uniformly worn surface. Some oxidized particles with a size of several microns are dark grey or bright surface, as indicated by the arrows in Figure 10(b). Nickel can be detected as well, which indicates the newly formed iron oxides have not fully covered the worn surface. Deep wear ploughs can be observed, which can be attributed to the abrasion of wear debris and hard asperities in Figure 10(c). In addition, detailed observation of the worn surface can be characterized by ferric oxide with grooves formed around the long axis of the elliptic wear area. A crack normal to the sliding direction inside the coating splats (Figure 10(d)) implies that the tribological failure mechanism consists of both fatigue wear and abrasive wear. The wear scars became wider as the applied load increased, but the shear stress was high enough to remove the dark gray oxide film near the center line of worn area. The hard ZrO₂ ceramic particles with edges and corners observed in Figure 10(f) also made the abrasive wear more serious. The above results well agree with the wear depth data shown in Table 3.

Figure 10:

Worn surface morphologies of AS-MC (a-b) 50N, (c-d) 100N, (e-f) 200N.

Figure 11:

Worn surface morphologies of GQ-MC (a-b) 50N, (c-d) 100N, (e-f) 200N.

Figure 12:

Energy dispersive X-ray spectrometry of the points marked “1” and “2” in Figure 11(b).

Figure 13:

Worn surface morphologies of OQ-MC (a-b) 50 N, (c-d) 100 N, (e-f) 200 N.

As shown in Figure 11(a), more ferric oxide films are formed compared to the coating demonstrated in Figure 10(a). Local analysis (see Figure 12) of the points marked “1” and “2” implies that they are iron oxides for the other elements such as nickel and chromium are much less. A large scale of oxide film is formed near the center line of the worn surface in Figure 11(b), but the scratches are more shallow and less in number. Oxide has a protective effect on the wear surface. However, the wear debris, such as unoxidized and oxidized iron scurf with spheroidal shape, was stuck upon the oxide film, which showed that three body abrasion was the second wear mechanism. During the wear test, the spalling from the fatigue crack and oxidation carried out alternately in the oxide layer. Similar to the coating in Figure 10(c), a portion of oxidation film was peeled off as indicted in Figure 11(e), even though its composition was still dominated by iron oxide. The transferred iron oxide was plastically deformed and became plate-like during continued action of sliding. The transferred layer detached in cyclic loading when the bond between the oxide layer and the coating had been locally weakened. The gas-quenching multilayered coating can maintain quite a part of undamaged oxidation film, which lessens the amount of wear debris. By contraries exfoliation is a little more serious in the as sprayed multilayered coating with higher hardness and poor adhesion bonding, which probably accounts for the differences in the wear depths.

The worn areas of the oil-quenching multilayered coating are much smaller at the applied load of 50 N and 100 N. Granular particles are the mixture of zirconia and iron oxide, which dwell in the concaves of the rough surface and help to decrease the friction coefficient. The formed film is composed of the mixture of iron oxide and nickel chromium alloy. In spite of obvious grooves observed in Figure 13(b) and (d), the oxide film is supported by the coating with greater bond strength. A mild form of oxidation wear is dominated and even the abrasion still can not be avoided at all. Carbon is distributed universally on the worn surface, mainly because the activated carbon atoms were decomposed from quenching oil during vacuum heating and generated carburized layer on the workpiece surfaces. Severe carbonization could likely occur with the increase of the quenching temperature. At the loads of 50N and 100N, OQ-MC alleviated the abrasive wear when the protuberances on a hard surface ploughed or cut through the counterpart and showed lower mass loss. Both of adhesive film and fatigue crack being visible in Figure 13(e) demonstrate dynamic equilibrium between oxidation and spalling. After the outer layer worn out at the load of 200 N, the wear loss was comparable to that of AS-MC due to fatigue wear of the inner layer.

Thermal cycle test

The oxidation kinetic curves at the experimental temperatures of 600 °C and 800 °C, are showed respectively in Figure 14. The average weight gains of GQ-MC and OQ-MC samples are represented by dotted lines. Two samples of AS-MC are displayed by solid line respectively because the data are too scattered. Based on the referred literatures [22, 23], the weight gains of the coatings can be expressed as following equations from the Wagner theory of oxidation [24], which is helpful to the quantitative prediction of oxidation rates. Divergent data of TGO thickness was recorded at the experiment of thermal cycling fatigue [25]. It is believed that grain-boundary diffusion provides an initially high oxidation rate [26]. The defects and vacancies in the interface of different particles can consume a large number of oxygen atoms and further enhance the barrier effect of the ceramic layer from external oxygen. The weight gain of one AS-MC specimen was less than others at the temperatures of 600 °C in thermal cycle test, while the weight gain of another AS-MC specimen was the highest. Most intertwining crevices inside the coating were the channels through which the air diffused [26]. External oxidation will become dominated if the crevices are filled up by internal oxidation product. However, wider crack observed on the surface of AS-MC contributes the dramatic increase of the weight gain. Similar static oxidation kinetic curves of CoCrAlY coating without irradiation mentioned in Ref [22] also confirmed that porous structure may promote the inward penetration of oxygen. The splashing of partially or fully molten particles sprayed onto the substrate surface was rapidly quenched, resulting in high hardness. During thermal cycle test the coating was cracked caused by internal stress originated from the temperature gradient of the whole coating system, thermal stress mismatched by different thermal expansion coefficient of different materials, thermal growth stress and phase transformation stress of TGO. The GQ-MC and OQ-MC exhibit steady mass growth at the experimental temperatures of 600 °C and 800 °C respectively. Defects such as pores, cracks, and uneven top surfaces can be reduced by remelting technology [27, 28]. Dense and homogeneous remelted layers significantly improve the high-temperature oxidation resistance. It can be observed that oxidation kinetic curves of OQ-MC exhibit nearly parabolic growth behavior. The formation of oxide scale was accelerated at the first several hours. Subsequent the kinetic curves become flattened which suggests low mass gain. It implies that the generation of remelting coating can impede oxygen from diffusing into grain boundaries for the gaps between them are almost extinct. The pale green surface of specimens conducted at 1000 °C bulged like a drum with deeper cracks inferring that chromium oxide was formed. But the coating was easy to peel off, as no intermediate layer existing to retard the stress growing out of different thermal expansion coefficients, and it ultimately resulted in a bad anti-pressure ability.

Figure 14:

Static oxidation kinetic curves of multilayered coatings with different temperature (a) 600 °C, (b) 800 °C.