Influence of Grain Size and Volume Fraction of η/δ Precipitates on the Dwell Fatigue Crack Propagation Rate and Creep Resistance of the Nickel-Base Superalloy ATI 718Plus

Critical materials for applications in jet engines or stationary gas turbines have to show outstanding mechanical properties, such as high temperature stability, creep resistance, and crack resistance at extreme conditions. One alloy that meets these requirements by simultaneous good processability is the polycrystalline Ni-base superalloy ATI718Plus® (hence called 718Plus) which can developed on the basis of the commonly used Ni-base superalloy Inconel 718.^[2‐8] 718Plus has two principle precipitate phases: γ′ as the main hardening phase and δ/η phase to control grain growth during processing. It has a higher thermal stability than Inconel 718 due to the utilization of the γ′ precipitate phase instead of the less stable γ″ phase in Inconel 718. Depending on the processing and heat treatment procedure, either the δ or η phase forms as main grain boundary precipitate phase. The δ phase can precipitasuste inside η particles or vice versa.^[9‐12]

Investigations by Huenert et al. have shown that the orientation of the plate-like η/δ phase can be adjusted by optimizing the material flow.^[1] The orientation of the η/δ plates has a high impact on crack propagation properties. In Reference 1, a “good” orientation, where the η/δ precipitates are aligned perpendicular to the direction of crack propagation, is distinguished from a “bad” orientation with aligned η/δ precipitates that are parallel to the growing crack. The good orientation slowed the crack propagation enormously in comparison with the bad orientation.^[1] Since the targeted alignment of η/δ precipitates in components with complex geometry can be challenging in certain applications, the knowledge about the impact of other microstructural features is also of great importance.

In this study, the grain size, the γ′ precipitate size, and the η/δ volume fraction are varied systematically in different samples with randomly oriented η/δ plates. Since an improved crack propagation resistance is typically accompanied with a reduced creep strength, the creep properties of the microstructure variants are also investigated. These findings are finally compared to the orientation effect of the η/δ plates as measured by Huenert et al.^[1] as well as the commercially used Ni-base superalloys such as IN718, UDIMET720Li, and WASPALOY in order to classify the different microstructural parameters regarding their impact on crack propagation.

The material used in this study was provided by Rolls-Royce Deutschland Ltd & Co Kg (RRD, Deutschland), which was processed by Allegheny Technologies Incorporated (ATI, USA) into billets and hot formed at Otto Fuchs KG (Germany) into a near turbine disk form. The nominal composition of the alloy is given in Table I.

Specimens were cut from the provided material and further heat treated to create different microstructures. The standard heat treatment is based on the recommendations of ATI^[13]: (1) pre-solution heat treatment (PSHT) between 843 and 871 °C for 16 h, (2) air cooling or raising to the solution heat treatment (SHT) between 954 and 982 °C up to 1 h, air cooling or faster cooling, and (3) γ′ age hardening at 788 °C for 2–8 h, furnace cooling + 704 °C for 8 h air cooling. The variations of the different heat treatment steps to generate different microstructures are shown in Table II. The IDs of the investigated conditions describe the properties of the targeted microstructure. Different grain sizes, different η/δ volume fractions, as well as one additional condition with an increased γ′ diameter were realized. Each sample is labeled with an ID for its grain size (small grain size: SG; medium grain size: MG; large grain size: LG) and its η/δ volume fraction. It is important to note that the η/δ precipitates show no preferred alignment in all conditions. Additionally, the same texture in all conditions is guaranteed as the different conditions were generated from the same initial material. Further, no difference in the twin density was confirmed by analyzing the BSD images of the different conditions. Accordingly, the mechanical properties are solely affected by the microstructural features mentioned in Table III.

To investigate the different microstructures, the specimens were mechanically polished and examined using backscattered electrons (BSE) in a Zeiss Crossbeam 1540 scanning electron microscope (SEM) with an acceleration voltage of 20 kV at a working distance of 7.5 mm. With the same equipment, energy-dispersive X-ray spectroscopy (EDS) was conducted to determine the chemical composition of the γ / γ′ compound in the grain interior. To measure the γ′ precipitate diameter, dark field transmission electron (TEM) images were made using a Philips CM200. The dark field images were taken at 200 kV using the installed 2 k × 2 k CMOS camera F216 (TVIPS). For this, thin foils were mechanically polished and further thinned by electropolishing with an electrolyte solution containing 8 pct perchloric acid and an applied voltage of 40 V at a temperature of − 30 °C.

The grain size as well as the γ′ precipitate diameter was determined using the linear intercept method and the freeware image tool ImageJ.^[14] The η/δ area fraction was determined using the free ImageJ plugin “Weka Trainable Segmentation.”^[15]

To calculate the γ′ volume fraction, the lever-rule (1) was used according to Blavette et al.^[16]:where f _γ′ accounts for the γ′ volume fraction and c_n is the chemical composition of the γ/γ′ compound in the grain interior without η/δ phase. For the chemical compositions of the γ and γ′ phase (c_γ and c_γ′ accordingly), atom probe tomography (APT) data from an earlier study on 718Plus were used.^[17] Since the microstructure also contains η/δ precipitates, the determined f_γ′ must be corrected by the η/δ volume fractionwith the assumption, that the η/δ area fraction equals the η/δ volume fraction f_η/δ.

To characterize the mechanical properties of the different microstructures, fatigue crack propagation tests and compressional creep experiments were conducted. Compact tension (CT) specimens according to ASTM E 647-13^[18] were used to determine the fatigue crack propagation rate. The samples were heated to 650 °C in air for the actual crack propagation test. A trapezoid signal with 1-120-1-1 was used with 120 seconds dwell time at maximum stress till the ratio of crack length a to sample width w reached 0.6. The crack length was measured using the potential drop method. The specimens for the compression creep experiments were cylindrical with a height of 6 mm and a diameter of 4 mm. The samples were tested in a pneumatic testing machine at 650 °C and different applied constant stresses ranging from 800 to 1100 MPa.

An overview of the different influencing factors on the crack resistance is shown in Figure 7. The results from this study are compared to the effect of the orientation of η/δ plates, as studied by Huenert et al.^[1] as well as to the commercial Ni-base Superalloys IN718^[21‐24] UDIMET720Li (hence called U720 Li) ^[25] and WASPALOY.^[5]

The condition SG-0 pct with an median grain size of 8.1 µm shows a similar crack propagation resistance as IN718.^[21‐24] The fatigue crack propagation rate can be reduced by about one order of magnitude by strongly increasing the η/δ volume fraction or grain size, as can be seen by the crack propagation rates of SG-17 pct and LG-0 pct. With the precipitation of the η/δ phases at the grain boundaries, the crack does not propagate exclusively along grain boundaries but also along the η/δ plates. Accordingly, the grain boundary η/δ precipitates lead to crack blocking and deflection and thus an increased crack length, which varies strongly with the orientation of the η/δ precipitates compared to the crack propagation direction. Therefore, a higher η/δ volume fraction leads to a higher fatigue crack propagation resistance. Additionally, the crack driving force decreases with higher fractions of η/δ precipitates due to a more pronounced stress relaxation in the microstructure, according to Reference 26.

The grain size seems to have no significant effect on the dwell fatigue crack propagation in the range between 8.1 and 15.5 µm. Only after significantly increasing the grain size to 126.7 µm, a better dwell fatigue crack propagation resistance is obtained. This fits to the results from Chang et al.,^[27] where no pronounced difference in the crack growth rate of IN718 at elevated temperatures between a grain size of 10 and 50 µm could be measured as well as other studies, which showed that an strongly increased grain size enhances crack propagation resistance.^[28,29] An increase in grain size reduces the density of grain boundaries and therefore reduces possible favorable pathways for the crack to propagate, leading to a deflection of the crack along the grain boundaries. This effect seems to only appear in air, which hints to an embrittlement of the grain boundaries due to high temperature oxidation.^[28]

The combination of a higher η/δ volume fraction and larger γ′ precipitate size has a stronger positive effect on the crack propagation resistance than an increase of the volume fraction alone as seen in Figure 5(a), the condition SG-7 pct shows better fatigue crack propagation resistance than the condition SG-10 pct with a higher η/δ content but with smaller γ′ precipitates. A similar effect can be assumed by comparing Figures 5(a) and (b), which shows that a combination of a bigger grain size and higher η/δ volume fraction most likely increases the crack propagation resistance more than solely increasing the η/δ volume fraction or solely increasing the grain size. A positive effect of a combination of different microstructural parameters like η/δ precipitates at grain boundaries in combination with medium grain sizes was also stated by References 26 and 30 where twins played a role, too. Obviously, the increase in grain size, η/δ volume fraction, and γ′ size has a positive impact on the crack propagation resistance of 718Plus. However, in comparison with the effect of the η/δ plate orientation, these improvements are much smaller, which is clearly demonstrated by the blue lines which represent the conditions with microstructures of aligned η/δ plates. The condition labeled as “bad” (upper blue dashed line) corresponds to a η/δ plate orientation of 0° in relation to the crack propagation direction. The condition labeled as “good” shows the results of microstructures with η/δ plates having an orientation of 90° in relation to the crack propagation direction. More detailed information about the different microstructures with aligned η/δ plates can be found in Reference 12 These two groups of conditions with differently aligned η/δ plates demonstrate that beneficially oriented η/δ precipitates result in a strong decrease of the crack propagation rate by at least two orders of magnitude. However, the improvement based on the η/δ plate orientation depends strongly on the crack propagation direction. On the contrary, the improvements due to grain size, η/δ volume fraction, and the γ′ size in conditions with unaligned η/δ plates are independent of the crack propagation direction and therefore should have an advantage under certain conditions.

The creep resistance can be improved by a combination of different microstructural properties, too. Figure 6 shows that a combination of big grains with small γ′ precipitates and a small amount of η/δ phase, which goes along with a high fraction of strengthening γ′ precipitates, enhances the creep resistance at 650 °C. However, as stated before, a decrease of η/δ volume fraction worsens the fatigue crack growth resistance^[31‐35] as shown in Figure 8. Apparently, there seems to be a correlation between an increased fatigue crack propagation resistance and a lower creep resistance and vice versa. This could be attributed to a stress relaxation in front of the crack tip by creep deformation during the dwell time. A lower creep resistance and therefore easier plastic deformation lead to a stronger blunting of the crack tip, which reduces the crack propagation per cycle.^[31,33] Besides, the reason for the higher creep strength at a lower fatigue crack growth resistance could be that a lower η/δ volume fraction results in a higher fraction of the strengthening γ′ phase (see Figure 4) and thus higher creep strength, but also in a reduced crack deflection by a lower content of the η/δ plates.

The impact of different microstructural parameters on the dwell fatigue crack properties as well as the creep properties of the Ni-base superalloy 718Plus was studied at 650 °C. Therefore, various microstructures were generated by proper adjustments of the heat treatments that lead to variations in the η/δ volume fraction from 0 to 17 pct, in the grain size from 8 to 127 µm, and in γ′ size from 21 to 46 nm. All η/δ containing microstructures showed a random orientation of the η/δ plates. The following two main conclusions on the correlation between microstructure and mechanical properties can be drawn:

These results allow to design the desired microstructures that meet the particular demands of the given application regarding crack propagation and creep resistance.