Assessment of Fatigue Resistance of Aluminide Layers on MAR 247 Nickel Super Alloy with Full-Field Optical Strain Measurements

The load analysis of the constituents of aviation engines such as turbine blades shows that material durability is dependent on crack initiation and development in thermo-mechanical fatigue, high cycle fatigue, low-cycle fatigue or creep conditions. Fatigue resistance of Ni alloys (such as CMSX, Rene, Inconel) is in the scope of interest of numerous research centers. Many authors analyzed influence of protecting layers on fatigue resistance (Ref 1-7). However, these studies were limited to determination of influence of the layer thickness on material/layer system durability. In some of them, the change in number of cycles to break of layer deposited on material in the mechanical notch location has been investigated (Ref 4-6). This approach enabled localization of the crack initiation area and its development observation; nonetheless, this strategy does not concern changes of the layer quality in a notch area and the real loading conditions are not fully replicated. Moreover, crack initiation and development observations require subsequent interruptions of fatigue tests (for example every 200,000 cycles) (Ref 4-6) for microscopic observations. This results in change of the loading conditions—due to the need of removing samples out of grips—and may have negative influence on the tests repeatability and accuracy, thus limiting its reliability. The usage of optical methods of strain field measurements gives an opportunity to observe whole samples’ gauge sections. They are fixed in loading device during testing, and this allows to determine the areas of crack localization occurring naturally. The most popular methods of strain field measurement are Electronic Speckle Pattern Interferometry (ESPI) and Digital Image Correlation (DIC). The main advantage of ESPI method is its high displacement measurement resolution (about 10⁻⁶). However, the method is vibrations sensitive and it enables resolving only a limited amount of displacement between consecutive loading steps, inducing the requirement for stopping tests and conducing measurements under the static load. On the other hand, DIC measurements are much less sensitive to vibrations at the expense of lower displacement measurement accuracy.

In this paper application of both methods is presented, especially considering their performance in fatigue tests of MAR 247 alloy with aluminides layers of 20 and 40 µm thickness.

The results presenting the influence of aluminide coating thickness on the fatigue strength of investigated samples are presented in Table 1 and Fig. 4. Samples with thinner coatings revealed slightly better performance.

Fractures images presented in Fig. 5 depict the layers thickness and quality. Obtained layers have uniform thickness even at the edges of samples, and there were no discontinuities observed between layers and core material.

Numerous crack presence was identified in aluminides layers in close neighborhood of fracture surfaces oriented perpendicularly to specimen’s surface (see Fig. 6). These cracks are initiated and propagate through the layer thickness at quite early stage of fatigue tests. Several of them transferred to the core material, followed by much slower crack growth rate.

The maps of longitudinal component of strain tensor for selected loading cycles for sample with 20 µm coating are presented in Fig. 7. The sample was loaded with maximal stress in cycle of 650 MPa. Sample has been damaged after 46,364 cycles and most of the damage development is observed in the last few hundreds of cycles.

First traces of strain localization were observed after 45 thousand of loading cycles. It is probably related with the formation of the crack detectable by DIC method already in the core material of the length close to the critical. For presented investigation, this example is an only one for which crack commencement has not been observed in the sample corner and may be explained by a larger damage of the protective layer induced in the production processes. Fracture surface image (see Fig. 8) confirms the location of crack origin in the middle part of front surface.

Second presented example is the sample with 40 µm thick coating also loaded with 650 MPa. The selected strain maps are presented in Fig. 9.

Strain localization in the case of sample with thicker coating was clearly detectable from the first loading cycles (see the strain map for 10000th cycle in Fig. 7). During whole test, the area of the largest strain values is located at the same place. Starting from ~34,000 cycles, the spot of the highest strain area has been revealed located near the sample edge on the left side, which may be interpreted as the beginning of final stage of crack development. The sample has been damaged during the next ~1100 cycles following this strain spot detection. Similarly, as in the case of 20 µm layer fracture surface image (see Fig. 10) confirms the location of crack origin, but this time at the sample corner.

DIC-based strain field measurements were relatively easy to implement in the fatigue strength investigation of samples with aluminides coatings of different thickness. However, limited strain measurement resolution allowed only for the detection of damages at the last stage of crack development, especially for sample with thinner layer. The improvement of measurements resolution would require limiting of the area of observation at the expense of resigning from the observation of the whole sample surface or using a camera with sensor of larger resolution.

The next presented example is fatigue test accompanied by ESPI measurements for sample with 40 µm coating. Sample was cyclically loaded with maximum stress of 600 MPa. In Fig. 11, strain maps for the selected moments of testing (1, 40, 60, 70, 71, 72, 75 and 80 thousand cycles) are presented. Due to manual loading procedure application, stopping the test after 80,000 loading cycles was possible. This was the moment, when the advanced crack development was detected, without total damage of sample. It enabled usage of eddy current method for crack detection and made metallographic observation of almost fully developed fatigue cracks.

In Fig. 11, strain maps with automatic scaling are presented while in Fig. 12—with scaling locked for strain values span taken from the last map. In case of automatic scaling strain, localization is revealed in ~6 places while in the case of locked scaling 3 main localization areas on the sample surface are discernible after ~70,000 cycles.

Eddy current measurements confirmed existence of cracks or layer discontinuities in locations predetermined by ESPI measurements (Fig. 13a). The size of defects was estimated basing on comparison to the signal received from reference sample with artificial flaws of known geometry (Fig. 13b and c). Final confirmation of crack presence was done by metallographic observations of appropriate cross sections presented in Fig. 14.

Metallographic inspection of places predetermined by ESPI measurements revealed change in crack direction after crossing layer/core material interface. In aluminides layers, cracks propagated perpendicularly to sample surface, while in core material cracks were deflected by approximately 45° (sometimes with the change of deflection direction to −45°) to the initial direction. This crack behavior may explain considerably early detection of strain localization areas (especially in case of ESPI measurements). This was caused by the emergence of the largest layer discontinuities oriented perpendicularly to the surface and followed by the period of small changes in strain maps related with deflected crack propagation in the core material.

The last presented results of ESPI-assisted fatigue tests are for 20 µm coating deposited on sample loaded with maximal stress of 650 MPa. Similarly to the case of the sample with 40 µm aluminides coating, automatic and locked scaling of strain values was applied (see Fig. 15 and 16). The sample was damaged after 61,968 cycles, just after registering the last set of images for ESPI processing. It is worth noticing that in case of the last presented sample, a few areas of crack initiation were detected starting from the sample edge (similarly to the previous example); however, the width of these areas in most cases covered approximately 50% of sample width, while in case of 20 µm coating approximately 25%. The extension of this range is plausibly related with both lager maximal load and larger thickness of coating for the last sample presented. Moreover, the presence of larger crack zones explains general lower fatigue resistance of samples with 40 µm layers.

High-quality aluminides layers on MAR 247 alloy have been obtained in CVD process. Fatigue strength of samples with layers of 20 and 40 µm thickness was examined.

The results of fatigue tests showed slightly better performance of samples with 20 µm layers. It might be explained by the much faster strain localization in samples coated with 40 µm layers revealed by both, DIC and ESPI measurements. Nonetheless, for the room temperature fatigue conditions the most crucial aspect for the investigated material/layer system seems to be the initiation and propagation of the crack in core material.

The analysis of strain field maps obtained from DIC and ESPI measurements during fatigue tests of samples with aluminides layers on nickel alloy allowed investigation of the influence of coating thickness and stress level based on the identification of the strain localization and its development during tests. The areas of initial localization of strain showed where cracking of coating commenced because of the supposed structural notches or origins related to surface roughness. Further increase of registered strain values and it localization in smaller areas showed the last stage of deflected cracks development in core material.

Each of the analyzed non-contact strain measurement techniques requires different testing procedures. They also deliver measurements results with different level of resolution. In cases of investigations requiring lower resolution, more vibration tolerant DIC method might be applied without the need of pausing fatigue tests. On the other hand, when higher resolution of measurements is required ESPI might detect smaller changes on observed surfaces, but more complicated testing procedure has to be applied. Nonetheless, application of any optical strain field measurements in fatigue tests of modern coated materials leads to an enhancement of the understanding of their degradation mechanisms.