Influence of the γ′ Volume Fraction on the High-Temperature Strength of Single Crystalline Co–Al–W–Ta Superalloys

Abstract

Understanding the influence of γ′ and secondary-phase fractions on the mechanical properties of superalloys is very important to optimize these high-strength materials. So far, this has not been systematically investigated for the novel class of Co-based superalloys. In this study, a Co–Al–W–Ta model alloy series was designed with compositions of γ/γ′ on the tie-line and an increasing γ′ volume fraction of up to 70% after heat treatment at 900 °C, while a few alloys are unexpectedly out of γ/γ′ two-phase region with an additional secondary phase fraction of up to 15%. The high-temperature strength and creep properties were evaluated by compression tests up to 1050 °C and compressive creep experiments at 950 °C, respectively. At temperatures of up to 1050 °C, an increasing γ′ volume fraction consistently increased the yield strength, which was not dramatically changed by the presence of secondary phases. Significant work hardening was found in alloys with γ′ volume fractions of 65–70% during compression testing, but not in alloys with either a lower γ′ volume fraction (<50%) or a high fraction of secondary phases (~15%). Similar to the yield strength, the creep strength also increased continuously with the γ′ volume fraction, but was greatly reduced with an increasing fraction of secondary phases. The best creep performance at 950 °C and 200 MPa was found in the alloy with the highest γ′ volume fraction and no secondary phases. At higher creep stresses, rafting contributed significantly to the hardening and, again, the alloy with a high γ′ volume fraction and a small amount of secondary phases exhibited the highest strength.

1. Introduction

Following the report of novel γ′-Co₃(Al,W) precipitates [1], substantial progress has been made in the development of Co- and CoNi-based superalloys via alloy design and processing over the past decade [2,3,4,5,6,7,8,9,10,11]. Alloying effects have been systematically studied to enhance the γ′ solvus temperature [3,4,12], microstructural stability [13,14,15], high-temperature yield strength [2,16], and creep resistance [17,18]. Additionally, W-free alloy systems such as Co–Al–Mo-based [6], Co–Al–V-based [7], and Co–Al–Ta-based [11] alloys have also been derived from the initial Co–Al–W-based system to reduce mass density. Directionally solidified single-crystal alloys without a grain boundary can further improve the high-temperature yield strength [2] and creep properties [3]. More recently, additive manufacturing successfully prepared multicomponent CoNi-based alloys that exhibited both excellent ductility and high ultimate tensile strength [19]. Due to their potential as high-strength, oxidation-resistant, high-temperature engineering materials [19,20,21], it is critical to further our understanding of their deformation mechanism for future applications.

The volume fraction of the strengthening γ′ phase plays an important role in Ni-based superalloys under various loading conditions [22,23,24,25]. A higher γ′ phase fraction and a resulting smaller precipitate spacing leads to a nearly constant 0.2% flow stress from 0 to 700 °C due to the balance between dislocation propagation in the matrix and γ′ phase resistance to dislocation cutting in Ni–14at.%Al. Meanwhile, a lower γ′ phase fraction and a larger precipitate spacing results in decreasing 0.2% flow stress with increasing testing temperature, as deformation mainly occurs through dislocation bowing between precipitates [22]. In a Ni–Cr–Al ternary alloy series designed from the γ/γ′ tie-line, a higher γ′ volume fraction improved the flow stress at a high temperature, such as 950 °C. However, the γ′ volume fraction effects are less pronounced at temperatures below 500 °C, when hyperfine precipitates were formed during cooling, which strengthened the alloy in addition to large precipitates [23]. In multi-component polycrystalline alloys with compositions from the γ/γ′ tie-line of In713C, the alloy with a γ′ volume fraction of 85% exhibited the maximum proof stress from 900 to 1000 °C, but the alloy with a γ′ volume fraction of 65% show the highest strength at 1100 °C [24]. Similarly, a γ′ volume fraction of ~70% led to the best creep performance at low temperatures and high stresses when shearing of γ′ precipitates by coupled dislocations on the {111} planes was dominant [25]. At a lower γ′ volume fraction of 55%, the highest creep strength was reached at high temperature and low stress, where rafting reduced the size of the vertical γ channels and the corresponding dislocation climbing along them [25]. Unlike in Ni-based superalloys, the deformation was mainly dominated by the shearing of γ′ precipitates under the formation of planar faults, such as anti-phase boundaries and stacking faults in Co- and CoNi-based superalloys because of their low stacking fault energy [2,26,27,28,29,30,31,32,33]. It has further been found that increasing the γ′ volume fraction up to 80–90% can lead to even higher YS [16] and creep strength [20,34] in these superalloys. However, systematic studies to understand the deformation mechanisms of these alloys with different γ′ volume fractions remain limited. Furthermore, the influence of secondary phases, which were generally detrimental to the mechanical properties in Ni-based superalloys, as they absorbed strengthening elements from the microstructure and/or facilitate crack initiation and propagation during cyclic loading [35,36], were easily formed because of the narrow γ/γ′ two-phase region in this new type of superalloy. However, their influence is not fully known.

To study the effects of the γ′ volume fraction and secondary phases, herein, we focused on a Co-based model alloy series with Ta additions, which is an important alloying element and can significantly improve the high-temperature yield strength [2,11,19] and creep resistance [3,34,37]. These Co–Al–W–Ta model alloys were designed using the lever rule from a pseudo γ/γ′ tie-line of ERBOCo–2Ta (i.e., Co–9Al–7.5W–2Ta) after aging at 900 °C/200 h to produce increasing γ′ volume fractions [38]. Although these alloys exhibited a gradual increase in the γ′ volume fraction, certain alloys with a very high content of γ′-forming elements were off the γ/γ′ two-phase region and generated a high amount of secondary phases such as Co₃W-based χ-D0₁₉, Co₂Ta-based Laves-C14, and Co₇W₆-based μ-D8₅ [38]. Nevertheless, this alloy series allows to study the influence of the volume fraction of the precipitation-strengthening γ′ phase and the presence of additional secondary phases on the mechanical properties with unchanged γ/γ′ phase compositions. In this work, the compressional strength, creep performance, and deformed microstructures are presented and the differences compared to Ni-based superalloys are discussed.

2. Experimental Procedures

The nominal compositions of the investigated alloys are listed in

Table 1. Nominal compositions in at.% of the ERBOCo–VF alloy series.

Alloy	Nominal Composition
	Co	Al	W	Ta
ERBOCo–2Ta	81.5	9.0	7.5	2.0
VF0	86.7	8.8	4.0	0.5
VF20	84.4	8.8	5.7	1.1
VF40	82.1	8.9	7.3	1.7
VF60	79.8	8.9	9.0	2.3
VF70	78.7	9.0	9.8	2.6
VF80	77.5	9.0	10.6	2.9
VF100	75.2	9.0	12.3	3.5

. They were designed based on the tie-line of γ/γ′ compositions at 900 °C in ERBOCo–2Ta (Co–9Al–7.5W–2Ta, at.%) with γ′ volume fractions from 0% to 100%. To obtain alloys with different phase ratios, the composition of the γ matrix and the γ′ precipitate phase of ERBOCo–2Ta were first measured by atom probe tomography (APT). Subsequently, two master melts, VF0 and VF100, were prepared with these two measured γ and γ′ compositions. The alloys VF20, VF40, VF60, and VF80 were then designed by the lever rule to vary the γ′ volume fraction, f_γ′, in steps of 0%, 20%, 40%, 60%, and 80% following the below equation:

$${C_{Alloy}=\left({1-f}_{\gamma \prime }\right)C}_{\gamma }^i+{f_{\gamma \prime }C}_{\gamma \prime }^i$$

where C_Alloy is the alloy composition, and C_γ and C_γ′ are the concentrations of the element i (i = Co, Al, W, and Ta) in the γ and γ′ phases, respectively. Afterward, the two master melts VF0 and VF100 were mixed in different ratios and VF20, VF40, VF60, and VF80 were cast. For more details, the reader is referred to reference [38,39]. Henceforth, the alloys are referred to by their intended γ′ volume fraction, e.g., VF60 for the alloy designed to contain a γ′ volume fraction of 60% [38,39]. The SX ingots were prepared in a laboratory-scale Bridgman unit (for further details, see [40]) and casted as rods with a length of 120 mm, a diameter of 12 mm, and an orientation close to [001]. These rods were solution heat-treated at 1300 °C for 12 h to reduce chemical macro-segregation during solidification and subsequently aged at 900 °C for 100 h to generate a γ/γ′ microstructure, followed by furnace cooling. Cylindrical specimens parallel to the [001] orientation were machined from the fully heat-treated rods for compression tests (height of 4.5 mm long and diameter of 3 mm) and compressive creep tests (height of 7.5 mm and diameter of 5 mm). Compression tests were conducted in air from room temperature to 1050 °C at a strain rate of 1 × 10⁻⁴ s⁻¹, and compressive creep tests were performed at 950 °C and various stress levels. After sample preparation with a final polishing step using Stuers OPU, the microstructure of the aged and deformed specimens was characterized by scanning electron microscopy (SEM) using a Zeiss Cross Beam 1540 EsB with secondary electron (SE) and backscattered electron (BSE) detectors. Transmission electron microscope (TEM) foils were cut from interrupted creep specimens perpendicular to the [001] loading axis. Electrolytic thinning of the samples was conducted with a solution consisting of 80 vol.% ethanol, 5 vol.% perchloric acid and 15 vol.% water at about −25 °C using a Struers Double Jet Tenupol-5. The dislocation configurations were characterized using a Philips CM 200 TEM operated at 200 kV. The heat-treated ingot cross-section was examined by electron backscatter diffraction (EBSD) to ensure the single-crystal structure. To evaluate the phase fraction after heat treatment, the area fraction of different phases was measured based on at least five representative images with ImageJ. The estimated γ′ area fraction was stereographically converted to calculate the γ′ volume fraction according to [41], assuming that all precipitates had a rectangular shape, the same size, and were homogeneously distributed.

3. Results and Discussion

3.1. Microstructure

As shown in Figure 1a–e, the volume fraction of VF0–VF70 steadily increased as intended, although the experimentally determined volume fractions were slightly off the designed ones, i.e., 12% for VF20, 40% for VF40, 69% for VF60, and 81% for VF70; for more details, see also [38,39]. However, due to the increasing amount of Ta and W in VF70, a small amount of secondary phases (~2% area fraction) formed in the interdendritic regions in VF70 after the heat treatment, as shown in Figure 1f. For VF80 with an even higher Ta + W content, the volume fraction locally remained 81%, similarly to VF70 (Figure 1g), but the overall area fraction of secondary phases increased to 15% (Figure 1h). The secondary phases in VF70 were likely Co₃W-based χ-D0₁₉, Co₂Ta-based Laves-C14, and Co₇W₆-based μ-D8₅, of the same type as that in VF80 [38]. Given the different phase constituent, the experimental alloys were divided into two groups to study the effects of the influence of the γ′, as well as the secondary phase fractions: Group I consisting of the γ single-phase and γ/γ′ two-phase alloys VF0–VF60 with an increasing γ′ volume fraction, and Group 2 with similar local γ′ volume fractions but an increasing amount of additional secondary phases (VF70 and VF80).

3.2. Compressive Yield Strength and Anomalous Work-Hardening

Figure 2a shows the yield strength of the experimental alloys as a function of temperature. For all alloys, the yield strength decreased as the testing temperature increased. With an increasing γ′ volume fraction from VF0 to VF40, the yield strength significantly increased, except for that at 1050 °C, while an additional increase further promoted the yield strength of VF60 only at temperatures higher than 750 °C. At 1050 °C, the γ′ precipitates in VF20–VF40 were expected to greatly dissolve during testing due to their low solvus temperatures (984–1057 °C) [38], hence the difference in the yield strength between VF0, VF20, and VF40 becoming small.

In order to evaluate the initial work-hardening capacity of the alloy series, the 2% offset stress R_p,2.0 is also plotted in Figure 2b. It became clear that those alloys with a higher γ′ volume fraction exhibited significantly stronger work-hardening. Unlike the insignificant yield strength anomaly, anomalous peaks of R_p,2.0 arose in VF40 and VF60 at 850 °C and in VF70 and VF80 at 750 °C. Since the anomalous work-hardening behavior in VF60 at 950 °C originated from extensive shearing of the γ′ phase and the formation of SFs [33], the stress–strain curves of the VF series at this temperature are compared in Figure 2c to rationalize the influence of the γ′ volume fraction and secondary phases. For the Group I alloys, the work-hardening was less obvious in VF0–VF40 and most significant in VF60. This indicates that only with a sufficiently high γ′ volume fraction can the extensive formation of SFs and their interaction effectively promote work-hardening. With the small amount of secondary phases and an even higher γ′ volume fraction in VF70, the compression strength and peak stress strain (at ~3.5%) remained unchanged, but its work-hardening rate was less distinct than that in VF60. The higher secondary phase area fraction of 15% in VF80 significantly reduced the work-hardening rate and led to indistinct work-hardening in the stress–strain curve.

3.3. Compressive Creep Behavior at 950 °C

Based on the yield strength results at 950 °C, stress levels between 25 and 600 MPa were selected to investigate the influence of the γ′ volume fraction and secondary phases on the creep strength. Given the low γ′ solvus temperature in these experimental superalloys, rafting was expected to occur at 200 MPa, similar to our previous study on rafting in ERBOCo–2Ta (Co–9Al–7.5W–2Ta) at 950 °C and 150 MPa [42].

At 200 MPa, the global minimum creep rates of VF40 and VF60 were fairly comparable (1–2 × 10⁻⁸ s⁻¹) and over three orders of magnitude smaller than that of VF20, as shown in Figure 3a. Additionally, VF40 and VF60 exhibited two creep rate minima at approximately ~0.2% and ~1% with a local creep rate maximum in between at ~0.5%. By increasing the stress to 350 MPa, the creep curves only exhibited a single creep minimum for both VF40 and VF60, and VF60 possessed a nearly two orders of magnitude smaller creep rate minimum of 3 × 10⁻⁸ s⁻¹ (Figure 3b). For Group II alloys with the additional secondary phases and a higher γ′ volume fraction of 81%, only a single creep minimum was observed under all creep conditions. At 200 MPa, VF70 exhibited a creep rate minimum close to VF60, while VF80 had a significantly higher minimum creep rate than both VF70 and VF60 (Figure 3a). In contrast, a higher stress level of 350 MPa led to a smaller minimum creep rate for VF70 over VF80 and VF60 (Figure 3b).

To compare the creep properties of the experimental alloys under different loading stresses at 950 °C, a Norton plot is displayed in Figure 3c. The creep strength increased significantly with an increasing γ′ volume fraction from VF0 to VF60. A further increase of the γ′ volume fraction in VF70 led to even lower minimum creep rates at stresses between 250 and 450 MPa. However, at lower stresses and corresponding lower minimum creep rates, VF60 exhibited a higher creep strength than VF70. By further increasing the amount of secondary phases in VF80, the minimum creep rates were significantly higher than VF70 at lower stresses, whereby the detrimental effects of the secondary phases diminished with increasing stress.

Due to the difference in the creep curves of VF60 at 200 and 350 MPa, interrupted creep tests were conducted at the two minima at 200 MPa and the minimum at 350 MPa to understand the stress-dependent deformation behavior, as shown in Figure 4. At 200 MPa, rafting was detected in VF60 perpendicular to the applied stress axis at the first creep rate minimum (~0.2% strain; Figure 4a) and completed at the second creep rate minimum (~0.7%; Figure 4b). TEM observation further showed that γ′ shearing under the formation of stacking faults was very limited in the early creep stages (Figure 4d), but fairly active at the second creep rate minimum (Figure 4e). Under 350 MPa, in contrast, the γ′ precipitate shape in VF60 remained cuboidal in the early creep stages (0.5%; Figure 4c), but already experienced significant γ′ shearing under stacking fault formation (Figure 4f).

With the above observations, it is suggested that rafting was the main reason for the optimum creep strength in VF60 at 200 MPa, since it can close the vertical γ channels and thus limit dislocation movement in the horizontal γ channels to promote a second creep rate minimum [29]. However, it is not clear why VF70 showed no double creep minimum at the same low stress of 200 MPa, given its high γ′ volume fraction. One possible explanation is that the limited size and volume fraction of the γ channels in VF70 were not able to accommodate as much plastic deformation as in VF40 and VF60, so shearing of γ′ easily occurred at a smaller strain and most of the subsequent deformation was confined in the γ′ precipitates [43]. In contrast, at higher stresses such as 350 MPa, shearing of γ′ precipitates readily happened in VF60 within a shorter time than at lower stresses, leading to a limited contribution of strengthening by rafting and the formation of a single creep rate minimum. In the case of VF70, its higher γ′ volume fraction of 81% than that of VF60 with 70% may have enabled more plastic deformation in the γ′ precipitates and a higher chance of SF interaction for hardening [31,33].

By plotting the required time to reach 1% plastic strain at 200, 250, and 350 MPa in the different alloys (see Figure 5a), it became apparent that the beneficial effect of rafting in VF60 at an applied stress of 200 MPa led to the highest creep strength. For higher stress values and corresponding shorter test periods, rafting did not occur quickly enough and the higher γ′ volume fraction in VF70 then led to the best creep performance. The decreased strength in VF80 at all stress levels was a direct consequence of its higher fraction of secondary phases, which decreased the total γ′ volume fraction, created local stress concentrations due to their morphology, and depleted the surrounding microstructure from W and Ta.

To compare the creep behavior with Ni-based SX superalloys, the normalized creep time with a maximum of 1000 h was plotted as a function of the measured γ′ volume fraction in VF20–VF70, as shown in Figure 5b. Note that it is not a direct comparison between Co- and Ni-based alloys, because their creep test conditions are significantly different, e.g., tensile stress and creep rupture time for the Ni-based superalloys TMS-75 and its γ/γ′ tie-line alloys [25] compared to compressive applied stress and time to 1% plastic strain for the experimental VF alloys. However, it sheds some light on the different influences of the γ′ volume fraction. Similar to Ni-based SX superalloys under high-temperature and low-stress conditions (1100 °C/137 MPa), rafting can improve the creep strength of Co-based alloys and lead to an optimum γ′ volume fraction in the present alloys. Interestingly, consistent with observations in earlier studies [44], the optimum γ′ volume fraction in the Co-based alloy series is 15% higher than in the Ni-based superalloy series investigated by Murakumo et al. [25]. As Tanaka et al. investigated tensile creep behavior [44] while we investigated compressive creep on similar alloys, the observation of a higher optimum γ′ volume fraction in Co-based alloys is apparently not related to the different types of rafting (e.g., parallel or perpendicular to the stress axis), but to a general beneficial effect of rafting on inhibiting the glide and climb motion of matrix dislocations.

Unlike Ni-based SX alloys, under intermediate temperature and stress conditions where rafting is not in effect (900 °C/392 MPa for Ni-based alloys), an increasing γ′ volume fraction up to 80% continuously promoted the creep resistance in Co–Al–W–Ta alloys, which is again approximately 15% higher than in the Ni-based alloys (see Figure 5c). While their principle microstructure is similar, differences in their general deformation behavior seem responsible for the distinct optimum γ′ volume fraction, with superlattice stacking fault shearing in γ′ in experimental Co-based superalloys and the γ′ shearing by APB-coupled dislocation pairs in Ni-based alloys in this creep regime [45]. The higher γ′ volume fraction and accordingly higher chance of planar fault interaction inside γ′ precipitates likely contributed to a stronger creep resistance in the Co-based alloys over the Ni-based SX superalloys. Detailed deformation mechanism investigations are warranted for a better understanding of these differences. A systematic study of γ′ volume fraction effects in Ni-based SX superalloys would complement the knowledge of precipitate strengthening in a low-temperature and high-stress creep regime where γ′ shearing with SF formation occurs.

4. Conclusions

Temperature-dependent compression and creep performance was investigated in a single-crystal Co–Al–W–Ta model superalloy series with varying γ′ volume fractions and additional secondary phases at 950 °C. The following conclusions can be drawn from this study:

• A 0.2% yield stress increased with increasing γ′ volume fraction from 0% up to ~70% in the VF0–VF60 alloys from room temperature to 1050 °C. The additional formation of secondary phases with a volume fraction of up to 15% did not significantly decrease the yield strength.
• Strong strain-hardening was found in the VF60 and VF70 alloys with a high γ′ volume fractions at 950 °C, but not in the low-γ′ volume fraction VF0–VF40 alloys during compression tests. The additional 2% area fraction of the secondary phases in VF70 did not decrease the compression strength, but led to a longer, less steep work-hardening behavior, while the 15% area fraction of the secondary phases in VF80 suppressed the anomalous work-hardening behavior.
• The creep strength consistently improved with an increasing γ′ volume fraction from 0% to 70% in VF0–VF60 at 950 °C. However, the creep properties in VF70 and VF80 gradually decreased at 200 MPa, while VF70 exhibited the best creep resistance at higher stresses of 250 and 350 MPa.
• A creep curve with two creep rate minima was found in the VF40 and VF60 alloys with γ′ volume fractions from 50% to 70% at 950 °C and 200 MPa. The second minimum was associated with rafting at early creep stages. At higher applied stresses from 250 to 350 MPa, a single minimum creep behavior was evident in all experimental alloys, because extensive γ′ shearing occurred in the early creep stages.