Influence of the coating brittleness on the thermomechanical fatigue behavior of a 𝛽-NiAl coated R125 Ni-based superalloy (2024)

2.1 Material and specimens

The studied material consists of a NiAl coating deposited onto a René125 Ni-based polycrystalline superalloy by low activity vapor phase aluminization (APVS: Aluminizing Phase Vapor Snecma) followed by a heat treatment at $870\text{\,}\mathrm{\SIUnitSymbolCelsius}$ for 16hours. The coating is an intermetallic compound made up of two characteristic layers: (i)the outer layer composed mainly of the $\beta$-NiAl phase and (ii)the interdiffusion zone (IDZ) composed of the $\gamma^{\prime}$-Ni₃Al phase and topologically compact phase precipitates (TCP). A hyperstoichiometric layer forms in a shallow region below the outer surface[16]. Due to its higher aluminum concentration, this layer is particularly brittle and will be prone to cracking under thermal and/or mechanical stress.

2.2 Isothermal oxidation, thermal cycling and thermo-mechanical fatigue conditions

Thermal aging treatments were performed at $1100\text{\,}\mathrm{\SIUnitSymbolCelsius}$ in air under both isothermal and cyclic conditions. The thermal cycles included heating from 100 to $1100\text{\,}\mathrm{\SIUnitSymbolCelsius}$, then holding for 5min or 50min, and finally cooling to room temperature.

A schematic illustration of the thermal cycle is depicted in Figure1(a).

In order to get a clearer picture of the global behavior of NiAl coated R125, out-of-phase strain controlled thermomechanical fatigue (OP-TMF) cycles were performed. This OP-TMF loading combines a thermal cycle with a 5min dwell at the maximum temperature set to $900\text{\,}\mathrm{\SIUnitSymbolCelsius}$, with a strain cycle from 0 to -0.4%, with a dwell time at -0.4% corresponding to the dwell at the maximum temperature, Figures1(a) and(b).

2.3 Tensile testing and mesoscale analysis

The ductility of the whole system and crack initiation behavior of the external coating were studied on the as-received and aged materials by comparison at different temperatures. To study crack initiation, tensile tests were performed at least up to the brittle-to-ductile transition temperature.Tensile specimens were designed as dog-bone specimens with a rectangular cross-section and cross-sectional variation to drive the maximum stress and strain localization to the center of the specimen, where the cross-section is a square of 2x$2\text{\,}\mathrm{m}\mathrm{m}^{2}$. This specimen geometry is particularly useful for concentrating and tracking deformation/damage events using image acquisition during testing[17]. A lamp furnace and K-type thermocouple allow the specimen temperature to be controlled up to $900\text{\,}\mathrm{\SIUnitSymbolCelsius}$. Incremental tensile loading is then prescribed by displacement steps until cracks are detected in the specimen gauge. The critical local strain required for crack initiation in the coating is obtained with accurate in-situ observations of each side of the specimen using a Keyence VHX 1000 microscope and a CCD camera for further digital image correlation (DIC) with VIC-2D software. In the AR state, the natural roughness induces enough contrast for DIC. After aging, to improve the quality of DIC, speckle pattern was obtained by white paint spots on a black paint used as a background. These observations are also combined with two acoustic emission (AE) sensors to detect and localize events during loading. AE greatly improves the resolution of the method in terms of detecting the first crack to appear (Figure2(a) and(b)). The sketch in Figure2(a) presents the association of AE sensors, surrounding the gauge length of the specimen, in this latter markers are used to track global displacement and speckle is used to monitor strain field through DIC. As an example, the strain field obtained by DIC for the as-received state is shown in Figure2(c). The Figure2(b) combines the prescribed displacement with black markers, the colored markers represent the intensity of AE, analyzed area being highlighted by a grey area: the chosen threshold for detection consisted in AE signal level higher than 70 dB when associated to strain localization observed by DIC.

2.4 Micro-testing and OHR-DIC at room temperature

Interrupted tensile tests were conducted at room temperature on flat micro-tensile specimens in the as-received and aged states. Ex-situ optical high resolution-digital image correlation techniques (OHR-DIC) were conducted using an OLYMPUS LEXT OLS5100 laser scanning confocal microscope (LSCM) to document onset of cracking. LSCM allowed for focused image acquisition regardless of the surface topography. A gold speckle pattern similar to Ref.[18] was applied to track surface kinematics after deformation. Large mosaic of images were acquired before and after deformation (13$\times$7images with an overlap of 20$\%$ and a pixel size of 205nm). Non-periodic optical distortions were corrected similarly to Ref.[18]. DIC calculations were performed using fast optical flow algorithms[19].

3.1 Impact of aging on microstructure evolution and damage

As a first intent, the evolution of microstructure from as-received (AR) state to different agings was herein described (see Fig.3).

3.2 Strain-to-failure evolution with temperature and aging

3.2.1 The mesoscale DIC and acoustic emission

The coating brittleness of AR and aged coated systems was evaluated as a function of the temperature using interrupted tensile tests paired with mesoscale DIC and acoustic emission (AE). The strain-to-failure corresponds to the macroscopic strain at which the first crack is observed within the external coating at a given temperature.

The as-received coating was found to be particularly brittle from room temperature to $700\text{\,}\mathrm{\SIUnitSymbolCelsius}$, Figure 4. The associated cracks were observed to reach the whole width of the specimen in a single step of loading. This is evidenced by strain localization patterns observed by DIC, see the arrow Figure 2(c). For temperature equal or higher than $700\text{\,}\mathrm{\SIUnitSymbolCelsius}$, the ductility increase is obvious, for test carried out at $900\text{\,}\mathrm{\SIUnitSymbolCelsius}$ no crack was observed, this maximum strain being used to rationalize the experimental strain values for each test ($\varepsilon_{c\max}$ $900\text{\,}\mathrm{\SIUnitSymbolCelsius}$).

After thermal aging, a low number of thermal cycles with long dwell time at maximum temperature does not impact the room temperature ductility (see green curve corresponding to 42cycles for 1-h dwell at $1100\text{\,}\mathrm{\SIUnitSymbolCelsius}$ in Figure4). However, a slight gain in ductility is observed for this condition of aging for temperature of $600\text{\,}\mathrm{\SIUnitSymbolCelsius}$ and higher.

For longer high temperature exposure at maximum temperature or shorter dwell time, the ductility at room temperature has been clearly increased, as illustrated in Figure4 (gray and red curves, corresponding to 500cycles for 5-min dwell and 84cycles for 1-h dwell at $1100\text{\,}\mathrm{\SIUnitSymbolCelsius}$, respectively). It is noteworthy that the time spent at maximum temperature is of about 42-h for 500 cycles of 5-minutes, that is to say that the dwell time impacts the ductility of the considered material.

Thermal aging alters the microstructure of the outer layer of the coating, resulting in a significant increase in strain to failure in room temperature tensile tests. In order to gain insight into the crack evolution within the coating and its relationship to the microstructure, $i.e.$, virgin or aged, $postmortem$ sections were observed using SEM after discontinuous tensile tests at room temperature. These micrographs allow a direct comparison between the as-received and thermally aged systems. It is important to note that the cracks are modified by the microstructure.

For AR state, the cracks easily join the substrate to IDZ interface despite some local deviations (two cracks are shown in Figure5(a)). In addition to very limited ductility, the through ”coating plus IDZ” thickness cracking are very detrimental for the integrity of the system, directly impacting the failure of the substrate. For 42cycles of 1h, crack opening is more limited, and the crack path is clearly more tortuous (Figure5(c)). Furthermore, crack deviations are observed at the external coating/IDZ interface. This could impede the crack growth even though in the given observation, the crack in the IDZ reaches again the IDZ/substrate interface. For 84cycles of 1h, the crack path seems to be connected to $\beta/\gamma^{\prime}$ interfaces in the external coating (Figure5(d)). Again this should limit the crack extension even though the density of crack is higher for 84cycles of 1h aging than for 42cycles of 1h aging. For 500 cycles of 5 min, crack paths are also driven by $\beta/\gamma^{\prime}$ interfaces, but could also be observed in a pure $\beta$ region, associated with both $\beta$ grain boundaries and not (Figure5(b)). However, cracks are tortuous for both cases. The brightest particles in the external coating do not seem to impact cracking. The role of surface roughness associated with rumpling on the coating cracking is not obvious from these observations. This point will be addressed in a follow-on paper.

To summarize these results, the ductility can be increased after significant aging inducing microstructure evolution.

3.3 Thermomechanical fatigue behavior

Thermomechanical tests were conducted to evaluate the fatigue life of the coated system for out-of-phase thermomechanical fatigue (OP-TMF) conditions, and the impact of prior high temperature aging, implying modification of the coating, interdiffusion and substrate, on the TMF lifetime. The chosen OP-TMF cycles induced a complex behavior considering the stress-strain hysteresis plot. For the first cycle, the compressive stress monotonically increased up to a deformation of $\varepsilon$=-0.3%, followed by a stress plateau and a final compressive stress decrease due to the very high temperature (Figure7(a)). At the maximum compressive strain, the stress relaxation is about $100\text{\,}\mathrm{M}\mathrm{P}\mathrm{a}$. Then, when the strain goes back to zero along with the temperature decrease, the stress relaxation induces large tensile stresses at the lowest temperature (500MPa at the first cycle). It is worth noting that the stress relaxation decreased in amplitude with further cycling (Figure7(b)). However, both stress relaxation and progressive evolution of stress-strain loops by cyclic kinematic hardening, that could be associated to the so-called Bauschinger effect[20], increased the maximum tensile stress (610MPa at the $\#$100cycle). This point is critical considering the observed low ductility of the coating for temperatures below $600\text{\,}\mathrm{\SIUnitSymbolCelsius}$, as reported in the previous sections using tensile testing.

The impact of pre-aging on the durability within such TMF loading will be discussed in a follow-on paper.

4.1 Impact of microstructure evolution of the coating on strain-to-failure

Thermal aging, in an isothermal or cyclic way, lead to microstructural evolutions of the coated system due to the combined effect of the oxidation and diffusion with the Ni-based superalloy substrate (See Figure3). The main evolutions consisted in $\beta$ to $\gamma^{\prime}$ transformation in both the external coating and in the IDZ matrix. In the external coating, $\gamma^{\prime}$ precipitation is mainly localized at grain boundaries, promoted by thermal cycling. These microstructural evolutions are controlled by the diffusion front related to the accumulated time spent at maximum temperature. In addition, thermal cycling promotes surface undulation, the so-called rumpling phenomenon, and damage associated with oxide pegs, that are potential sites for crack initiation. Repetitive short-dwell thermal cycling was found to be particularly detrimental for the coated systems due to the occurrence of short cracks within the external coating system. Such microstructural evolutions are in good agreement with the literature on aluminide coating systems[21, 12, 22]. The major difference from previous studies for (Ni,Pt)Al systems is the large amount of TCP phases within the outer coating and within the substrate after long-term exposure[12].

The effect of these microstructural changes on the brittleness of the coated system was investigated from room temperature to $900\text{\,}\mathrm{\SIUnitSymbolCelsius}$ using different approaches. Incremental tensile tests paired with mesoscale DIC and AE aimed at detecting the initiation of the first crack as a function of the macroscopic stress/strain level for different temperatures (Figure4). This approach aimed at documenting the effect of thermal aging on both the strain-to-failure capability of the coated system and its ductile-to-brittle transition temperature (DBTT). $\beta$ to $\gamma^{\prime}$ transformation in the outer layer lead to an increase in ductility of the coated system, the coating being particularly brittle when the coating is purely composed of $\beta$ phase and in the presence of hyperstoichiometric layer (i.e., without significant $\beta$ to $\gamma^{\prime}$ transformation). Thus, the aging should be long enough to promote hom*ogenization of the composition, and a significant volume fraction of $\gamma^{\prime}$ precipitates to gain ductility. However, it is also obvious that the time spent at high temperature modifies the behavior of the coating but also of the superalloy due to modification of the $\gamma$/$\gamma^{\prime}$ microstructure. The DBTT was not significantly affected by the thermal aging despite the microstructural evolutions.

As far as the brittleness of the coating is concerned, cross-sectional observations (Figure5) and microscale DIC techniques aimed at evaluating differences in crack path depending on the thermal aging condition (Figure6). It is worth mentioning that long-term isothermal oxidation can significantly improve the cracking resistance of the coating system (here 308h at $1100\text{\,}\mathrm{\SIUnitSymbolCelsius}$). This is in line with in-depth study of NiAl evolution with aging as developed for example in[16]. Indeed, the coated system experienced cracking at 250MPa in its as-received state compared to 600MPa for the long-term isothermal state. In addition to the onset of crack development, the crack pattern also sheds light on the coating brittleness. Through coating plus IDZ cracking and fully transverse cracking are signs of high brittleness. Shorter cracks and crack arrests/bifurcations within the IDZ testify to the influence of microstructural change on improving the ductility of the coated system. It is worth noting that the surface crack pattern observed with OHR-DIC (Figure 6) is fully consistent with the thermal cycling crack observed through the cross-section (Figure 5): the cracks are mostly limited by either $\beta/\gamma^{\prime}$ interfaces or $\beta$ grain boundaries.

This result is particularly important, showing that prior aging can be a solution to improve the overall ductility of such coated systems under the DBTT. Another message is the fact that the ductility of the coated system will present a significant gain during service if $\beta$ to $\gamma^{\prime}$ transformation occurs. As aforementioned, the severity of the thermal cycling using short-dwell cycling can introduce damage and early cracking of the coating. This damage feature should explain that the ductility reaches a sort of asymptotic regime at 5minutes dwell, because the benefit of the $\gamma^{\prime}$ transformation is balanced by the introduction of damage at these short cycles.

4.2 Impact of aging on thermomechanical fatigue failure

To clarify the way aging could impact the behavior of the studied system, TMF loading were tested after prior aging at high temperature.Pre-aged specimens under thermal cycling conditions were also tested in similar out-of-phase thermomechanical fatigue (OP-TMF) conditions. Since short cycles are detrimental in terms of coating damage, we preferred to use 42 cycles of 1h dwell at $1100\text{\,}\mathrm{\SIUnitSymbolCelsius}$ as pre-aging.

Stress-strain hysteresis was plotted in Figure7(c) for the first cycle. The stress-strain-temperature evolution is very close to the AR condition, an increase of the compressive stress is found as a function of the temperature increase. However, less pronounced stress-relaxation during the strain dwell led to moderate tensile stresses at the lowest temperature (200MPa lower than AR condition). The Bauschinger effect is yet more pronounced between the $\#$6 and $\#$55 cycles for the aged specimen, but the maximum tensile stress after $\#$300cycles (and even $\#$600cycles) is lower than the maximum tensile stress experienced during the OP-TMF test in the AR condition. Interestingly, the aged specimen had a two times greater lifetime under the same OP-TMF conditions compared to the AR specimen.

As a first observation, it is obvious that the mechanical behavior has been modified by this pre-aging: this should be related to the fact that the pre-aging should have affected the microstructure of both the coating and the substrate, since the maximum temperature during aging was $1100\text{\,}\mathrm{\SIUnitSymbolCelsius}$ which exceeds the $870\text{\,}\mathrm{\SIUnitSymbolCelsius}$ post-coat heat treatment. Further investigation is underway to determine the extent to which pre-aging has modified the superalloy microstructure (in particular the $\gamma-\gamma^{\prime}$ ratio and the precipitations of TCPs), the coating microstructure (in particular the phase transformation) and morphology (in particular its roughness).

Last but not least, the life extension associated to pre-aging demonstrates that optimal heat treatment should consider the coating system in the sense that a given diffusion coating is impacted by the choice of the superalloy. This offers new guidelines to obtain a full optimization of the coated superalloy due to the drastic impact of the composition of this latter on this effect [23].