Superplastic behavior of a powder metallurgy superalloy during isothermal compression

1. Introduction

It is universally accepted that polycrystalline superalloys play paramount roles in the aircraft engines, which are ideal materials as turbine disks [[1], [2], [3], [4]]. No matter these disks are produced by cast & wrought (C&W) or powder metallurgy (P/M) routes, thermo-mechanical processes (TMP) including hot forging are essential for tuning the microstructure and shape of the final components [[5], [6], [7], [8]]. However, as highly alloyed materials, these polycrystalline superalloys are difficult to be produced via conventional TMP techniques [9,10]. As a sufficient method of TMP, superplastic deformation was originally found in Sn-Bi alloy at eight decades ago, and it has been applied to manufacture many kinds of materials including superalloys, considering its enormous potential for industrial hot-working [[10], [11], [12], [13]]. Generally, materials with superplasticity exhibit extremely large plasticity during deformation, for example, elongations over 1000% without cracking are commonly observed during tensile tests, hence these materials can be arbitrarily and safely formed to complex shapes [13,14].

As the superplastic deformation is mainly controlled by grain boundary sliding (GBS) and thermal-activated diffusion, two basic preconditions should be satisfied to perform superplastic forming: ultrafine grains with average size less than 10 μm and high deformation temperature above half of melting point T_m [13,15,16]. However, these two requirements can hardly be compatible in practice, because grains may grow excessively at elevated temperatures, leading to the GBS more difficult. Thereby, the fracture happens frequently at large strain, even though many attempts have been made to inhibit the grain growth rates, such as lowering forming temperature or accelerating strain rate [17,18], adding alloying elements [15,19,20], and precipitating refined second-phase particles into the matrix [21,22].

In general, elongations obtained by tensile tests are used to evaluate superplastic behavior of alloys. It has been observed that the deformation temperature and strain rate strongly influence the final elongation, the grain refinement induced by discontinuous dynamic recrystallization (DDRX) enhances the superplasticity, dislocation proliferation and migration also happen during that deformation, and cavity or necking develops prior to final cracking [[23], [24], [25]]. For example, the elongation over 400% was achieved by superplastic tensile test on fine-grained Inconel 718 [23,24]; the DDRX was thought to be the dominating mechanism controlling the superplastic deformation of Inconel 718, and the shrinking developed gradually until the final fracture with the increase of strain [24]. However, as polycrystalline superalloys during superplastic forging bear compressive loading rather than tensile stress, seldom experiment has been performed to study the superplastic behavior and microstructure evolution via thermal compression, which has been studied in other kinds of materials, and many attempts have been made to relate the grain structure and physical properties of other alloys [[26], [27], [28], [29], [30], [31]].

In this study, a nickel-based P/M superalloy after hot consolidation was chosen to investigate the mechanism of superplasticity via isothermal forging. Several typical stages in superplastic compression was found and connected with the microstructure evolutions.

2. Materials and methods

To study the superplastic behavior during isothermal forging, the nickel-based superalloy A1 containing high level of Co element was prepared by P/M route, as shown in Table 1. In specific, the A1 powders with diameter below 63 μm were firstly produced by argon atomization at Aubert & Duval, and loaded into a steel container. Thereafter, the container was degassed and sealed, followed by hot isostatic pressing (HIP) at 1100 °C and 150 MPa for over 2 h.

The cylinder specimens with diameter of 10 mm and height of 15 mm were compressed at Gleeble 3180D. Moreover, the carbon sheet and lubricant were placed between specimen and dies to better the friction condition and make the deformation more uniform during compression. The specimens were heated up to 1000-1100 °C with rate of 5 °C/s and held for 2 min to homogenize the temperature distribution at initiation. And the shape changes during compression are photographed by digital camera. After deformation, the specimens were immediately quenched in water to freeze the deformed microstructure.

The specimen for γʹ observation was prepared by mechanically polishing, and etching by a solution of 40 ml H₂SO₄ + 12 ml H₃PO₄ + 48 ml HNO₃. The grains in the P/M superalloy is also observed by electron backscatter diffraction (EBSD). To perform EBSD observation on a field-emission scanning electron microscope (SEM, FEI Quanta 650) equipped with an EBSD detector, the sample sections were polished by abrasive papers and 50 nm colloidal silica, followed by vibration polishing for over 5 h. The scan step size during EBSD observation for all samples was uniformly set as 0.10 μm to guarantee enough pixels in each grain. Then the EBSD data were analyzed by HKL Channel5 software. Local misorientation maps were adopted to analyzed the hardening state of individual grains, the detailed description of these maps are presented in another work [2].

Transmission electron microscope (TEM) observation was performed on the field-emission TEM (Tecnai G² F20, FEI), under 200 kV accelerating voltage, to detect the microstructure evolution during superplastic deformation. The TEM samples with diameter 3 mm and thickness around 50 μm were sectioned from center of the compressed specimens, then the slices were twin-jet electropolished in the corrosive solution of 90% ethanol and 10% perchloric acid at -25 °C and 20 V. In this work, the microstructure observation on the samples was focused on the center regions of the sections along compressing direction.

3. Results and discussion

3.1. Determination of superplastic region

Strain rate and temperature are two essential factors affecting the flow behaviors for a specific polycrystalline superalloy, and the superplastic regions considering these two aspects are predicted via the flow data obtained from the preceding compression tests [9]. Basically, the true strain ε and true stress σ can be related by following equation [32,33]:

σ=k⋅$\dot{ε}^m$ (1)

where k is a materials constant, $\dot{ε}$ is the strain rate, m is strain rate sensitivity. Thereafter, m can be determined by:

$m=\frac{∂ln(σ)}{∂ln(\dot{ε})}\lvert_{ε,T}$ (2)

Based on experimental data, Woodford [34] connected the measured ductility with strain rate sensitivity m and found that the superplastic ductility can be reached when m excesses a critical value, specifically, 0.5, as concluded by Langdon [13]. The critical m value 0.5 has also been adopted to predict plasticity of other superalloys [10,35,36].

To get the distributions of strain rate sensitivity at different hot-working conditions, the thermal compression tests were conducted at temperatures in the range of 1000-1100 °C and strain rates of 1-10^-3 s^-1. Moreover, the flow data have been corrected considering friction and temperature variations, as described in previous work [9]. Specifically, cubic spline interpolation is used to calculate he flow stress values at finer intervals, using the experimental data points as knots, as expressed by the following equation:

lnσ=k₁+k₂ln$\dot{ε}$+k₃(ln$\dot{ε}$)²+k₄(ln$\dot{ε}$)³ (3)

Where the coefficients of k₁, k₂, k₃ and k₄ can be calculated by four sets of lnσ- ln$\dot{ε}$ at specific ε and temperature T. Thereafter, the value of strain rate sensitivity m in each subinterval can thereby be determined via following equation according to Eq. (2):

m=k₂+k₃(ln$\dot{ε}$)+k₄(ln$\dot{ε}$)² (4)

The contour maps at different strain levels have been finally constructed, showing the m distributions at different strain rates and temperatures, as shown in Fig. 1. The superplastic area is not constant with the changes of the strain, which is related with the dynamic evolutions of microstructure [9,36]. In most cases, the m value over 0.5 is located at low temperature and strain rate regions around 1000 °C/10^-3 s^-1. Thereafter, specific isothermal tests on cylinder specimens are performed at 1000 °C/10^-3 s^-1 and 1025 °C/10^-3 s^-1, to exam the superplastic flow and study the deformation mechanisms behind that.

View original graphic| Download| PPT slide

Fig. 1.   Contour maps showing strain rate sensitivity m at different strain levels.

3.2. Superplastic behavior and microstructure evolution

3.2.1. Initial microstructure

The initial microstructure parallel to compression direction of the P/M superalloy sample is shown in Fig. 2. The ultrafine grains with average grain size of 2.45 μm are distributed uniformly and no significant texture is observed, the area fraction of γ′ precipitates is detected to be around 50%, wherein a great number of the primary γ′ phases have equivalent size excessing 1 μm.

View original graphic| Download| PPT slide

Fig. 2.   Initial microstructure of grain and γ′ precipitates in as-HIPed A1 sample: (a) EBSD IPF image; (b) γ′ precipitates morphology.

3.2.2. Flow behaviors of superplastic compression

The shape changes after compression with different height reductions are illustrated in Fig. 3, there is no macro-crakes at the surface after the samples have been isothermally compressed with 80% height reduction at the predicted superplastic region. In addition, bulging at cylinder surface caused by die-workpiece friction is observed, which is significant at relatively low strain levels. As shown in Fig. 4, obvious bulging at cylinder surface is identified at initial compression. With more materials flow to the ends of the specimen, the contact area at die-workpiece interface gets larger gradually, which may increase the die-workpiece friction [[37], [38], [39]].

View original graphic| Download| PPT slide

Fig. 3.   Shape changes of P/M superalloy samples after compression with different height reductions.

View original graphic| Download| PPT slide

Fig. 4.   Photos of specimen with different height reductions during isothermal compression at 1025 °C/10^-3 s^-1 condition.

The true stress-true strain curves of isothermal compression at 1000 and 1025 °C with 10^-3 s^-1 strain rate are plotted in Fig. 5, wherein the negative values of strain and stress reflect the compression direction. In general, the true stresses continuously increase to the peak (σ_P) induced by strain hardening at initial, then the stresses gradually decrease to a relatively steady state till -0.9 true strain (near 60% height reduction). Many works concerning the flow behavior of normal hot-forging on metallic materials concluded that the initial strain hardening is dominated by the dislocation accumulation, and thereafter materials softening is triggered by dynamic recovery and recrystallization as some critical conditions are satisfied, the state status is mainly realized by the balanced materials softening and hardening competition [[40], [41], [42], [43]]. While in superplastic deformation on alloys, the grain boundary sliding plays a significant role in the materials flow as well [2,14,19,44]. Based on the microstructure evolution, the mechanisms of superplastic deformation on this P/M superalloy will be discussed in the ensuing part.

View original graphic| Download| PPT slide

Fig. 5.   True stress-true strain curves of isothermal compression at (a) 1000 °C and (b) 1025 °C with strain rate of 10^-3 s^-1.

After 60% height reduction, the obtained stresses keep climbing to -1.6 true strain (near 80% height reduction), which may be mainly caused by the die-specimen friction rather than the microstructure evolution [37]. Basically, the average friction coefficient μ during compression can be estimated via the change of specimen geometries, according to the following expression [37,45]:

μ=$\frac{\frac{R}{H}⋅b}{\frac{4}{\sqrt{3}}-\frac{2b}{3\sqrt{3}}}$ (5)

where R is the radius of the sample after frictionless compression, and H is the height of the deformed sample, so, R=$R_0\sqrt{H_0/H}$, R₀ and H₀ represent initial radius and height of cylinder specimen respectively. The barreling factor, b, can be determined by:

b=$\frac{4(R_m-R_1)}{R}⋅\frac{H}{(H_0-H)}$ (6)

where R₁ and R_m are minimum and maximum radius of the bulged sample, respectively. Thereby, the changes of average friction coefficient μ can be calculated, based on above equations. Thereafter, the flow curves obtained by compression tests could be corrected by following equations [38,46,47]:

σ=$\frac{C^2F}{2[exp(C)-C-1]}$ (7)

wherein,

C=$\frac{2μR_0}{H_0}$ (8)

where σ and F mean the corrected and experimental flow stress considering friction, R₀ and H₀ are the original radius and height of the initial specimens correspondingly. Finally, the friction-corrected flow curves are presented in Fig. 6, which indicates that the true stresses at high strain level keep nearly steady with strain.

View original graphic| Download| PPT slide

Fig. 6.   Experimental and corrected flow curves of isothermal compression at (a) 1000 °C and (b) 1025 °C with strain rate of 10^-3 s^-1.

3.2.3. Microstructure evolution during superplastic compression

To avoid possible friction induced strain gradient and microstructure inhomogeneity, the microstructure of all specimens at the center region in the section along compressing direction is characterized. The EBSD inverse pole figures (IPF) of specimens deformed at different conditions are shown in Fig. 7.

View original graphic| Download| PPT slide

Fig. 7.   IPF images of specimens deformed at different conditions.

As shown in Fig. 8, the grains in specimens compressed at both 1000 °C/10^-3 s^-1 and 1025 °C/10^-3 s^-1 conditions keep fine during deformation. The average grain size fluctuates slightly during superplastic compression, which decreases at the initial compression, then increases at 60% height reduction, and drops again at 80% height reduction. Specifically, the grain size distribution in samples with 10% reduction, where the softening is just initialed, is close to the original state. As true strain reaches -0.35 (30% height reduction), approximating to the steady status, the number of tiny grains below 2 μm rises, which leads to the decrease of average grain size. At true strain of -0.92 (60% height reduction), more grains larger than 2 μm are detected, which are reduced at higher true strain of -1.61 (80% height reduction). It is clear that the grain growth is enhanced at higher deformation temperature, as the average grain sizes of specimens deformed to specific strains at 1025 °C are generally greater than their counterparts deformed at 1000 °C.

View original graphic| Download| PPT slide

Fig. 8.   Distribution of equivalent grain size of specimens deformed at (a) 1000 °C and (b) 1025 °C with strain rate of 10^-3 s^-1 and (c) average grain size of specimens deformed with different height reduction.

The grain evolution during thermal deformation is generally controlled by the dynamic recovery, recrystallization, and grain growth [41,[48], [49], [50], [51]]. To trigger DRX, adequate energy should be stored in materials by dislocation accumulation, hence the strain hardening is significant in specimens compressed at 1000 °C/10^-3 s^-1 with 10% and 30% reduction, as illustrated by the local misorientation maps in Fig. 9. As the temperature increased to 1025 °C, the strain hardening gets relieved by dynamic recovery (DRV) at higher temperature, which also facilitates the DRX via subgrain formation [2,52]. With the true strain reaches -0.35, corresponding to the steady state in the corrected flow curves, general misorientation in alloys keeps at a relatively low level, which reflects that the dislocation accumulation and annihilation are balanced.

View original graphic| Download| PPT slide

Fig. 9.   EBSD local misorientation images of P/M superalloy samples compressed at 1000 °C/10^-3 s^-1 and 1025 °C/10^-3 s^-1 with different height reductions, wherein different colors in rang of 0° to 5° represent different misorientations as indicated by the color bar.

To further investigate the mechanisms of microstructure evolution during superplastic deformation, TEM observation is performed. As indicated in Fig. 10, the dislocation accumulation is significant at relatively low temperature and strain, especially at the serrated grain boundary and triple junctions (TJs) in specimens, as the plastic incompatibility induces stress concentrations there [53,54]. Subgrains isolated by dislocations are also found in these regions. As the temperature increases or strain reaches a certain high level, the DRX grains or the initial grains with few dislocations, classified as “undeformed grains”, mix with deformed grains, and some twins are also formed at TJ regions, as depicted in Fig. 10(h). Interestingly, the GBS is detected in the deformed specimens, since the scratch offset, which is commonly thought as the evidence of GBS [14,55,56], is also observed in the superplastic deformed P/M superalloy. As shown in Fig. 11, the grain boundary between grain A and B is piled up with dislocations in the sample compressed at 1025 °C/10^-3 s^-1, and a series of dislocations near the grain boundary are regularly arranged, showing the trace of the scratching during GBS, as indicated by the arrows in Fig. 11. Basically, the GBS is schematically illustrated in Fig. 12, wherein the grain cores in regions away from the grain boundary contain less or barely dislocation.

View original graphic| Download| PPT slide

Fig. 10.   TEM images of P/M superalloy samples compressed at 1000 °C/10^-3 s^-1 and 1025 °C/10^-3 s^-1 with different height reductions.

View original graphic| Download| PPT slide

Fig. 11.   Montaged TEM image of P/M superalloy sample compressed at 1025 °C/10^-3 s^-1 with 80% height reduction.

View original graphic| Download| PPT slide

Fig. 12.   (a) Global and (b) local schematic images showing sliding of grain boundary between grains A and B.

In terms of the superplastic tensile deformation, the cavities and cracks are broadly observed at triple junctions and interfaces between matrix and second-phase particle [18]. In contrast, there is little cavity or crack in the P/M superalloy samples after superplastic compression with large strains. The superplastic compression comes to a steady statue when the strain excesses a certain value, and the accumulation and annihilation of dislocation are relatively balanced. Specifically, as the strain concentrated at TJs, the nucleation of DRX or twins is triggered to deduced the global stored energy. In addition, since the P/M nickel-based superalloy is mainly strengthened by γ′ precipitates interactions between dislocations and γ′ precipitates are also characterized, as presented in Fig. 13. Clearly, γ′ precipitates hinders the motion of dislocations in the γ matrix, and the dislocations stack and curve at the γ/γ′ interface, as shown by the arrows in Fig. 13. Some of the dislocations bypass the obstacles by diffusion-controlled climbing, and some dislocations shear through the particles. As the dislocations inner the grains are much less that these along grain boundaries, and many of them could be annihilated via gliding, climbing and cross-slipping at high temperatures, the dislocation aggregation at the γ/γ′ interface is not severe enough to form cavity or crack [2,14].

View original graphic| Download| PPT slide

Fig. 13.   Montaged TEM image showing interactions between dislocations and γ′ precipitates in sample compressed at 1025 °C/10^-3 s^-1 with 80% height reduction.

4. Conclusions

In this work, the flow behaviors and microstructure evolution during superplastic compression on the P/M nickel-based superalloy A1 have been studied. Accordingly, following points can be concluded:

(1) The predicted superplastic region is generally located at relatively low temperature and strain rate domains, specifically around 1000 °C/10^-3 s^-1, based on the location of the estimated strain rate sensitivity m over 0.5. Additionally, the friction-corrected flow curves indicate that the true stresses primally increase to the peak, then the stresses gradually decrease to a relatively steady state.

(2) The dislocations accumulated initially are mainly annihilated by DRV and DRX, and grain growth is also enhanced at higher deformation temperature. Moreover, the existence of grain boundary sliding (GBS) is found in the deformed alloy, based on the scratching offsets in TEM images.

(3) The cavities and cracks at triple junctions and interfaces between matrix and second phase particle are barely observed in the superplastically compressed samples, which is different from superplastic tensile deformation.

Acknowledgements

This work was supported financially by the National Key Research and Development Program of China (No. 2016YFB0701404), the National Natural Science Foundation of China (No. 91860105) and the Fund from Innovation and Entrepreneur Team Introduced by Guangdong Province (No. 201301G0105337290).