Failure Analysis of Thermal Corrosion Cycling of EB-PVD YSZ Thermal Barrier Coatings Exposed to Molten NaCl

Abstract

NaCl is the most abundant salt in seawater, thermal barrier coatings (TBCs) face the challenge of chloride salt hot corrosion in the marine environment. In this work, we focus on evaluating the thermal corrosion cycling of 9 wt.% Y₂O₃-ZrO₂ (YSZ) exposed to NaCl molten salt. YSZ was prepared by electron beam–physical vapor deposition (EB-PVD). The result showed that the thermal cycling life of YSZ coating in the presence of NaCl molten salt was significantly shortened. The shortening of the service life of the coating can be ascribed to the collective effects of the thickening of thermal growth oxide and the degradation of the bond coating.

1. Introduction

Thermal barrier coatings (TBCs) are complex systems. The typical TBCs systems consist of two layers above the substrate: a ceramic top coating (TC) and a metallic bond coating (BC) [1,2,3,4,5]. They can effectively provide heat insulation and high-temperature oxidation resistance, as well as corrosion and wear resistance, to the base metal, thus improving the service temperature of alloy parts and the service life of the base metal as expected [6,7]. The service environment of hot-end components with TBCs is extremely harsh. Thus, under the combined action of heat, force, chemicals, and other loads, coating failure in the form of peeling severely limits the service life of TBCs [8]. Compared with APS, the structure of EB-PVD columnar crystals can withstand higher strain, and coatings prepared by EB-PVD method have good thermal shock resistance, a high surface finish, a dense coating, and improved oxidation and thermal corrosion resistance. Hence, the coating material prepared using this method can better adapt to the harsh service environment with a longer service life [9]. Therefore, 6–8 wt.% Y₂O₃-ZrO₂ prepared by EB-PVD method has advantages of high melting point, high coefficient of thermal expansion, low thermal conductivity, low elastic modulus, high hardness, good corrosion resistance, and adaptability to a harsh service environment. It has been widely used in TBCs for hot-end components of aeroengines [10,11,12,13,14,15].

The erosion of foreign matter, such as molten salt (V₂O₅ and Na₂SO₄) and calcium–magnesium–alumina–silicate (CMAS), in addition to marine atmospheric corrosion, is an important factor affecting the failure of TBCs. The molten salt and CMAS corrosion mechanisms of TBCs are relatively straightforward. Some progress has been made in preventing the corrosion of TBCs [16,17,18,19,20]. In marine atmospheric environments, TBCs are subject to complex corrosion. In offshore areas, seawater droplets containing a large amount of NaCl salt are mixed into the atmosphere when sea spray is dispersed, which results in a high concentration of NaCl particles in the atmosphere. When aeroengines operate in offshore areas, these particles are deposited on TBCs. Due to the high humidity and high-salt environment, TBCs face the risk of hot corrosion from sea salt [21,22,23]. Therefore, there is an urger demand for high-temperature coating technology to resist high-temperature corrosion in the marine atmosphere. It is important to explore the corrosion mechanism of TBC resistance to NaCl molten salt at high temperatures in order to develop high-temperature corrosion-resistant coating technology for the marine atmosphere. Presently, there are two kinds of hot corrosion tests in the marine environment: salt coating and salt immersion methods. Hot corrosion occurs in the temperature range of 800–1000 °C, with a duration of 2–10 h [24,25,26,27].

Zhou et al. investigated the corrosion behaviour of YSZ coatings by exposing them to NaCl and NaCl + water vapour in a high-temperature tubular furnace. The sample was maintained at 1050, 1100, or 1150 °C for 55 min, then cooled in air for 5 min to simulate the actual working conditions of aeroengines. The results showed that the oxidation cycling life of the YSZ coating shortened in the presence of NaCl and NaCl+water vapour as compared to that in air. The failure of the TBCs exposed to NaCl and NaCl+water vapour occurred within the TC and close to the YSZ/thermal growth oxide (TGO) interface [28,29]. Although the corrosion behavior of YSZ coatings exposed to NaCl has been investigated, research has mainly focused on YSZ prepared by the APS method. No studies have successfully explored the corrosion resistance of YSZ coatings prepared by EB-PVD in molten NaCl. Thus, the failure mechanism of the as-deposited coating could not be determined in this case.

Motivated by these challenges, in the present study, YSZ coatings were deposited using the electron beam–physical vapour deposition (EB-PVD) method. The crystal structures, morphologies, and elemental distributions of the coatings were characterized and studied. The corrosion behaviour of YSZ coatings in molten NaCl was systematically investigated. The aim of this work is to provide a more comprehensive understanding of the corrosion failure mechanism of EB-PVD YSZ coatings in a marine environment.

2. Experimental Procedure

2.1. Coating Deposition

TBC systems were used to produce a NiCoCrAlYHf layer (BC) on a nickel-based superalloy substrate (12.5 Co wt.%, 6.3 W wt.%, 5.7 Ta wt.%, 6.3 Al wt.%, 2.2 Re wt.%, 5.8 Cr wt.%, 1.3 Mo wt.%, and Ni as balance) followed by TC. The Ni-based superalloy substrate was handled by water grit blasting before deposition; the BC was prepared by arc ion plating–physical vapor deposition (AIP-PVD; Moscow, Russia). During the preparation of NiCoCrAlYHf, the current was adjusted to 650 ± 50 A, and the bias voltage parameter was 20 ± 5 V. The AIP-PVD was finished under high vacuum lower than 1 × 10⁻² Pa. The composition of the BC was 10–15 wt.% Co, 18–23 wt.% Cr, 8–12 wt.% Al, 0.1–0.5 wt.% Y, 0.2–0.6 wt.% Hf, and Ni as balance with a thickness of approximately 40 ± 10 μm. Before the deposition of TC, the substrate with BC was annealed under a vacuum (<1 × 10⁻² Pa) at 870 °C ± 10 °C for 3 h to enhance its bond strength and elemental diffusion. The specimens’ temperature was adjusted to 850 °C during the deposition of TC by EB-PVD (ICEBT, Paton, Ukraine). Then, a YSZ layer was deposited on the BC surface. The preparation processes and conditions for the YSZ coating are presented in

Table 1. Preparation parameters of YSZ.

Layer	Current of Electron Beam (A)	Voltage (KV)	Rotation Speed (rpm)	Deposition Rate (μm/min)
YSZ	Gun3# 1.55–1.65	20.0	20	3.0–3.5

. A commercial YSZ ingot was employed in the EB-PVD process, with a chemical composition of 9 wt.% Y₂O₃ and 91 wt.% ZrO₂. The content of Y₂O₃ in the ceramic coating deposited by EB-PVD with 9YSZ ingot was approximately 8 wt.%. Finally, the coating was cut into a sample of 1 cm² to satisfy the requirements of subsequent experiments, as shown in Figure 1a.

2.2. Corrosion Tests

NaCl was dissolved in water at a mass ratio of 0.16 g/mL and used as the corrosive medium. It smoothly covered the specimen surface at a density of 8 mg/cm². After drying at 80 °C for 1 h in an oven, NaCl crystals were evenly spread on the specimen surface without touching the edges (a clearance of 3 mm from the sample’s edge was maintained). The samples were heat-treated at 950 °C for 9 h and then slowly cooled to room temperature (~25 °C) in atmosphere for 1 h for thermal cycling (Figure 1b). The coatings were photographed using a camera after each cycle. The salts were spread again on the coating’s surface, and a new cycle was started. Six samples treated with NaCl and six samples treated without NaCl were subjected to the thermal cycle to avoid randomness in the experiment.

2.3. Characterization

The phase and surface morphology of the samples before and after corrosion were analyzed using X-ray diffraction (XRD; Rigaku Smart Lab, Tokyo, Japan) and a field emission–scanning electron microscope (SEM; ZEISS, Oberkochen, Germany). The samples were then infiltrated with epoxy resin under vacuum and cold-inlaid to prevent damage to the coating in the subsequent sample preparation stage (Figure 1c). The sample was cut along the diameter to show the section, then rough- and fine-polished. The polished section was analyzed using SEM. The elements in the cross section of the coating were identified using energy-dispersive spectroscopy (EDS; Bruker, Oberkochen, Germany).

3. Results and Discussions

3.1. Preparation and Characterization of YSZ Coating

The phase composition, surface morphology, cross-sectional morphology, and elemental distribution of the YSZ coating were characterized, as shown in Figure 2a,b. The YSZ coating showed diffraction peaks at 2θ = 30.12°, 35.04°, 50.32°, 59.90°, 62.74°, and 74.18° corresponding to the (101), (110), (112), (211), (202), and (400) planes of the Zr_0.9Y_0.1O_1.95 phase (tetragonal phase of YSZ, JCPDS Nos. 82-1241), respectively. The peaks were narrow and sharp, and the TC was highly crystalline. The XRD patterns of YSZ powder and the as-sprayed coatings are shown in Figure 3a. Compared with the XRD patterns of YSZ powder, the preferred growth orientation of the tetragonal structure of as-sprayed coating occurs in the (110) crystal direction due to the stabilizing effect of Y element and the high vapor pressure over an overheated molten pool of ingot in the EB-PVD process. The XRD patterns of BC powder are shown in Figure 3b. The BC from the β-NiAl and γ-Ni(CoCr) consists of two phases. The γ-Ni(CoCr) phase provides the toughness of the alloy, and the β-NiAl phase can guarantee the strength of the alloy [30,31]. Y and HF, as active elements, are present in low concentrations and can be completely dissolved in the alloy matrix.

From a macro view, the surface was flat and uniform (Figure 1a). In contrast, the micro surface showed a classic pyramidal crystal, and the top of the pyramidal crystal was sharp, with a crystal size of approximately 1–2 μm (Figure 2a). Obvious three-layer structures were observed in the coating (Figure 2b). According to the results of EDS maps, the TC was YSZ with a thickness of 80–90 μm. The YSZ layer had a typical columnar crystal structure of TBCs prepared by EB-PVD, and there were some gaps between columnar crystals, with deposited BC in the middle and nickel-based alloy in the lower layer. The YSZ interfaces were perfect, indicating that the process parameters were reasonable.

The porosity of the coating affects the penetration of molten salt. Thus, it is necessary to carry out studies on the porosity of the resulting YSZ coating. The porosity (P) was measured and calculated by direct measurement principle:

$$\rho =\frac{{\Delta }_m}{{\Delta }_v}=\frac{m-m_s}{v-v_s}$$

(1)

$$P=\left(1-\frac{{\rho }_0}{\rho }\right)\times 100\%$$

(2)

where m and m_s are the total quality of coating and quality of the substrate, respectively; v and v_s are the total volume of the coating and the volume of substrate, respectively; and ρ and ρ₀ are the actual density and theoretical density of the ceramic coating, respectively. According to our calculation, the porosity of the YSZ ceramic material was 53.11%.

Generally, a YSZ coating can be used in subsequent thermal corrosion experiments.

3.2. Corrosion Behavior of YSZ Coating

3.2.1. Selected Visual Images

The selected YSZ coating samples are shown in different corrosion stages in Figure 4. The specific areas corroded using salt were marked with black lines. In the presence of NaCl, the top surface of the YSZ coating did not seem to be significantly damaged until eight corrosion cycles. After 10 corrosion cycles, the damage began from the edge of the sample and gradually spread to the center of the coating. In comparison, there was no visible damage to the blank sample after 14 thermal cycles. There may have been differences in micromorphology, although there was no obvious damage was observed during visual inspection. The failure mechanism of coatings exposed to molten NaCl can be explored by investigating changes in micromorphology during the thermal cycling process. Thus, samples were selected for additional in-depth analysis after four and eight cycles.

3.2.2. Surface Morphologies and Phase Composition

The morphologies of the top surface of YSZ coatings thermally processed after four and eight cycles are shown in Figure 5 and Figure 6, respectively. No changes were observed for the crystal of the YSZ coating in the absence of NaCl. The pyramidal grains at the top of the YSZ column were arranged neatly and densely, with some gaps around them. As shown in Figure 6, there were no changes in the crystals of the YSZ coatings in the presence of NaCl.

The XRD spectra of the top surface of the YSZ coatings after different numbers of cycles are shown in Figure 7. After four and eight thermal cycles, characteristics peak were observed at 30.16°, 35.20°, 59.98°, and 74.30° corresponding to tetragonal zirconia. This indicates that the major phase of the TC was a tetragonal YSZ solid solution. The phase of the YSZ coatings was consistent with that of the as-deposited coating in the original spraying state. Figure 7b is a partially enlarged XRD diagram from 28° to 32°. A possible monoclinic phase peak was observed at approximately 29° after eight cycles in the presence or absence of molten NaCl. This result indicates that YSZ coatings may undergo structural transformation or changes in the monoclinic M phase at this time, although it is independent of exposure to NaCl molten salt.

3.2.3. Cross-Sectional Morphologies

The morphologies of the YSZ coated without NaCl after four and eight thermal cycling are shown in Figure 8a–d. The microstructure of TC was dense, the number of cracks perpendicular to the BC was small, and the size was not significant. Continuous, uniform, and dense TGO layers were formed between TC and BC. The thickness of the TGO increased with increased cycling. As shown in the red areas of Figure 8c,d, a large amount of TGO was generated after eight thermal cycles, which caused damage to the BC.

The morphologies of the YSZ coated with NaCl after four and eight corrosion cycles are shown in Figure 9a–d. As a result of thermal corrosion, the YSZ coating became loose and porous. It cracked along the gap around the columnar crystals (indicated by the red arrow in Figure 9a “I”). YSZ coating spalled off from the BC at the edge with additional corrosion cycles. There was a large gap between the TC and BC after eight corrosion cycles. BC degradation is clearly indicated by the red dashed area in Figure 9b.

Comparison of the results between the two samples (in the absence or presence of NaCl), showed that the longitudinal cracks in the cluster of columnar crystal gaps were obviously increased when there were bulges in the BC. Longitudinal and transverse cracks contributed to the spalling of the YSZ coated with molten NaCl, especially at the edge of the coating. Additionally, the BC degraded to a certain extent for in the center of the sample coated with NaCl.

The detailed cross-sectional morphologies of the YSZ coating thermally processed with NaCl after eight cycles are shown in Figure 10. The overall damage to the coating is shown in Figure 10a. Figure 10b provides an enhanced view of the coating. Figure 10a shows that the reaction of molten NaCl with the TC and BC resulted in severe damage, developing gradually in the coating from the edge toward the center. This damage resulted in the separation of the TC and BC. The gap between the TC and BC decreased from the edge toward the center.

A high proportion of YSZ coating spalled from the BC. Some residues of the YSZ coating adhered to the BC in area “I”. The energy spectrum of EDS can be observed at the top-right side of Figure 10a, indicating that separation occurred between the TC and BC. The separation of the TC and BC can still be observed at the center of the coating in Figure 10a. The degree of separation was low, as shown in in Figure 10b. As shown in Figure 10b, we observed a transverse crack similar to that shown in Figure 10a. This crack can be observed in the enlarged image shown in Figure 10c in area “II,” occurring near the TGO according to EDS maps (Figure 10d).

This implies that the spalling and damage were mainly caused by the propagation of transverse cracks caused by TGO. Overflowing of the salts to the side surfaces occurred due to the inert nature of molten NaCl in contact with the TC, which resulted in damage development from the edge to the center. Thus, the degree of separation was most significant at the edge and decreased gradually from edge to center.

The degradation parts of the BC in the center of the coating were analyzed. BC degradation in a small area at the center of the coating can be observed in Figure 11. This may have occurred because a small amount of molten NaCl penetrated the BC through the gap due to long-term corrosion cycling. This resulted in the fluxing of TGO. Nevertheless, fluxing of the alumina scale damaged the TGO, and the salts reached the BC. The corrosion of the BC resulted in the internal oxidation of alumina, along with the presence of some voids, as indicated [32].

3.3. Corrosion Mechanisms

It is well known from existing studies that failure of coatings mainly occur as a result of cracking in the TGO layer at the TC/BC interface. The overall thickness of the TGO layer increases during cyclic corrosion cooling, and TGO growth contributes to the increase in σ11, σ22, and σ12 stress near the TC/BC interface. The TC and BC all off because of the difference between thermal mismatch stress and strain tolerance caused by the thermal expansion coefficient [33,34,35,36]. TGO formation is related to the diffusion of aluminum and oxygen, which is parabolic in the case of enhanced oxygen diffusion, which is demonstrated in Figure 12. Compared with the TBCs exposed to air, there accelerated oxidation occurred in the TBCs exposed to NaCl molten salt, and the thickness of the aluminum loss zone followed an almost linear trend [33].

NaCl reacts with oxygen at a high temperature to release chlorine according to reaction (3) [34]. The released chlorine can react with Al or Cr in the BC to form their respective volatile chlorides (according to reaction (3)). When encountering oxygen with higher partial pressure, it is reoxidized to form oxides of aluminum and chromium (according to reaction (5)) [35,36]. This process rereleases chlorine, and the chlorine continues to react with Al/Cr according to reaction (4).

$$4\mathrm{NaCl}+{\mathrm{O}}_2\left(\mathrm{g}\right)=2\mathrm{NaO}+2{\mathrm{Cl}}_2\left(\mathrm{g}\right)$$

(3)

$$2\left(\mathrm{Al},\mathrm{Cr}\right)+3{\mathrm{Cl}}_2\left(\mathrm{g}\right)=2\left(\mathrm{Al}, \mathrm{Cr}\right){\mathrm{Cl}}_3$$

(4)

$$2\left(\mathrm{Al}, \mathrm{Cr}\right){\mathrm{Cl}}_3+3{\mathrm{O}}_2\left(\mathrm{g}\right)={\left(\mathrm{Al}, \mathrm{Cr}\right)}_2{\mathrm{O}}_3+3{\mathrm{Cl}}_2\left(\mathrm{g}\right)$$

(5)

In this study, molten NaCl overflowed to the edge of the coating, which resulted in the dissolution of the dense protective TGO layer. Because the original TGO changed from a dense to loose state, new Al in the BC was exposed. New TGO was rapidly generated under the action of Cl₂, which resulted in an increase in the overall thickness of the TGO layer. Figure 13 shows the thickness of the TGO layers more intuitively at five random positions of the edge part of the coating. After four cycles, the average thickness of the TGO at the edge of the coating corroded by NaCl and the coating not corroded by NaCl was 3.874 μm and 3.0375 μm, respectively. After eight cycles, the average thickness of the TGO at the edge of the coating corroded by NaCl and the coating not corroded by NaCl was 5.499 μm and 3.8125 μm, respectively. The thickness of the TGO at the edge of the coating corroded by NaCl was much greater than that of the uncorroded coating. Figure 14 shows the thickness of the TGO layers more intuitively at five random positions of the center part of the coating. After four cycles, the average thickness of the TGO at the center of the coating corroded by NaCl and the coating not corroded by NaCl was 3.199 μm and 3.06175 μm, respectively. After eight cycles, the average thickness of the TGO at the center of the coating corroded by NaCl and the coating not corroded by NaCl was 4.324 μm and 3.848 μm, respectively. The thickness of the TGO at the center of the coating was only slightly affected by molten NaCl. The stress increase was caused by different thermal expansion coefficients and strain tolerances during cyclic corrosion cooling, which resulted in the propagation of cracks and the TC spalling from the BC.

Secondly, molten NaCl directly reacted with the BC according to reactions (4) and (5) at the edge areas of the coating, as shown in Figure 10 and Figure 11, which accelerated the degradation of the BC (as shown in Figure 15) and the interface separation of the BC and TC.

The degradation of the BC at the edge and the spalling off of the TC from the edge resulted in damage to YSZ coating. The damage was initiated at the edge and gradually spread, although the damage at the edge of the YSZ coating was the most serious. Then, the center areas became prone to damage.

4. Conclusions

The presence of NaCl led to more serious peeling and damage of the TBC systems, which significantly shortened the service life of the coatings.

First, molten NaCl accelerated the degradation of the BC at the edge and damaged the integrity of the TC. Second, molten NaCl resulted in alumina dissolution and promoted the rapid growth of alumina at edge of the YSZ coating. This, in turn, increased the thermal stress during cooling. The shortening of the service life of the coating can be ascribed to the collective effects of the thickening of thermal growth oxide and the degradation of the bond coating. Additionally, a small part of the BC in the center of the YSZ coating degraded in the process of cyclic corrosion, which also caused some damage to the coating.

In this study, the failure mechanism of YSZ coatings exposed to molten NaCl was determined, which can provide some ideas for improving the corrosion resistance of YSZ coatings. Protecting the edge of coatings and reducing porosity are possible solutions, e.g., placing an Al₂O₃ layer on the YSZ or produce a modified layer on the surface by laser glazing.