Tailoring the Properties of a Ni-Based Superalloy via Modification of the Forging Process: an ICME Approach to Fatigue Performance
Traditionally, material design and property modifications are usually associated with compositional changes. Yet, subtle changes in the manufacturing process parameters can also have a dramatic effect on the resulting material properties. In this work, an integrated computational materials engineering (ICME) framework is adopted to tailor the fatigue performance of a Ni-based superalloy, RR1000. An existing fatigue model is used to identify microstructural features that promote enhanced fatigue life, namely a uniform, fine grain size distribution, random orientation, a distinct grain boundary distribution (specifically high twin boundary density and limited low-angle grain boundaries). A deformation mechanism map and process models for grain boundary engineering of RR1000 are used to identify the optimal thermo-mechanical processing parameters to realize these desirable microstructural features. For validation, small-scale forgings of RR1000 were produced and heat-treated to attain fine grain and coarse grain microstructures that represent the conventionally processed and grain boundary engineered (GBE) conditions, respectively. For each of the four microstructural variants of RR1000, the twin density and grain size were characterized and were in agreement with the desired microstructural attributes. In order to validate the deformation mechanisms and fatigue behavior of the material, high-resolution digital image correlation was performed to generate strain maps relative to the microstructural features. The high density of twin boundaries was confirmed to inhibit the length of slip bands, which is directly attributed to extended fatigue life. Thus, this study demonstrated the successful role of models, both process and performance, in the design and manufacture of Ni-based superalloy disk forgings.
KeywordsGrain boundary engineering Σ3 Twin Fatigue Strain Modeling
Due to their characteristic anomalous strengthening behavior, Ni-based superalloys are well suited for use as structural components in the hot sections of advanced gas turbine engines . Since the 1950s, many decades of engineering advances have contributed to optimizing their thermo-mechanical properties, such as their resistance to creep, fatigue, and corrosion at elevated temperatures . Much of the unique characteristics of Ni-based superalloys can be attributed to their γ-γ′ microstructure, where ordered Ni3Al γ′ precipitates that possess a L12 crystal structure are embedded within a disordered Ni face-centered cubic (FCC) γ matrix. Thus, the excellent high-temperature strength and creep resistance can be attributed to potent levels of precipitate strengthening that restrict dislocation motion during plastic deformation . In many instances, the properties of these alloys can be further tailored or engineered by modifying the grain boundary structure or character distribution [4, 5, 6]. For example, the fatigue properties of high strength, polycrystalline Ni-based superalloys are highly sensitive to both the grain size and grain boundary character distributions [7, 8, 9]. In recent years, demand for increasing performance requirements and operating temperatures for advanced turbine engines has pushed the capabilities of existing polycrystalline Ni-based superalloys toward their limitations. As a result, an improved understanding of the dependence of crack initiation and growth mechanisms on grain boundary character distributions can potentially contribute to the identification and development of optimized microstructures that can effectively extend the capabilities of the material.
Polycrystalline materials possess a grain boundary network, where each individual boundary between neighboring grains corresponds to a specific misorientation angle relative to the orientation of the corresponding lattices. The resulting character of the grain boundary segment greatly influences both the physical and mechanical properties of a material [10, 11, 12]. Depending on the relative orientations of the high-angle grain boundaries (HAGBs), the coherency of the boundary between the individual grains with their neighbors may vary significantly . The coincident site lattice (CSL) is often used to quantify the degree of coherency along grain boundaries and describes the number of positions in the respective lattices of neighbor grains where lattice points coincide . The degree of coincidence and structure of the coincidence sites can be described by the Σ value, used as a parameter defining CSL boundaries. Boundaries with few coincident sites, e.g., low-CSL boundaries typically with Σ > 29, tend to contain large concentrations of crystalline defects and vacancies that serve to both weaken the interface and promote diffusive mechanisms at elevated temperatures . Since grain boundary diffusion and sliding are accelerated along these boundaries, this leads to environmental degradation and poor resistance to creep deformation. On the other hand, adjacent or neighboring grains that exhibit high-CSL boundaries, or Σ < 29, have relatively coherent interfaces that contain fewer crystalline defects that contribute to weakening the boundaries. Moreover, since there are comparatively fewer vacancies and defects along low-Σ interfaces, the mechanisms by which diffusion and mass transport occur along the interfaces also become more sluggish. Among low-Σ boundaries, twin boundaries, Σ9 and Σ27, to name a few, are referred to as special grain boundaries.1 These interfaces typically show lower grain boundary energies compared to those of low-CSL boundaries and HAGBs [15, 16]. Coherent twin or Σ3 boundaries are the most prevalent type of special grain boundaries  and possess the lowest grain boundary energy . Therefore, they are highly desirable features for breaking up the connectivity of the random grain boundary network.
Various grain boundary engineering approaches have been investigated in recent years as increasing the density of coherent twin boundaries has been shown to be an effective strengthening mechanism in polycrystalline materials, while simultaneously maintaining a high degree of ductility . When applied to Ni-based superalloys, grain boundary engineering has been shown to increase resistance to fatigue crack growth and extend the high cycle fatigue life [19, 20]. When either a large grain or a cluster of grains within a polycrystalline microstructure are favorably oriented for planar slip, cyclic deformation conditions may lead to the formation of persistent slip bands that effectively serve as precursors for nucleation of fatigue cracks . Since the length of the persistent slip bands determine the likelihood of crack nucleation, populating the microstructure with a high density of Σ3 twin or high-CSL boundaries will serve to inhibit slip transmission across these boundaries and effectively limit their overall length . Since the length of persistent slip bands are bounded by these Σ3 boundaries, the eventual nucleation of fatigue cracks will likely occur along these boundaries and is consistent with observations in both pure metals [22, 23, 24] and alloys [21, 25, 26]. Furthermore, the tendency of cracks to form along twin boundaries [26, 27, 28, 29, 30] can also be attributed to the high degree of elastic strain anisotropy [31, 32] that exists across these boundaries and induce stress concentrations that lead to strain localization.
In recent years, microstructure explicit fatigue models for polycrystalline Ni-based superalloys have been developed [33, 34, 35, 36, 37, 38, 39, 40] that confirm strain localization around microstructural features, primarily at twin boundaries, prior to nucleation of fatigue cracks. For this particular study, the finite element-based crystal plasticity model was used to inform the design of microstructures and identify desirable meso-scale grain boundary character distributions that could be varied to enhance the fatigue performance of a commercially available Ni-based superalloy RR1000. Following which, innovative hot deformation based grain boundary engineering techniques were used to systematically vary the grain boundary character distributions in small-scale forgings of RR1000. Detailed strain mapping and microstructural characterization of the forged alloy were used to validate the effectiveness of grain boundary engineering on fatigue crack nucleation behavior.
Design for Fatigue Enhancement
A fatigue model was used to inform the design of the microstructure of an existing Ni-based superalloy, RR1000, in order to enhance the fatigue performance. In this material, one of the primary mechanisms that govern the fatigue life at low temperatures is based on failure due to persistent slip bands (PSBs), in which strain is localized within the microstructure. Tanaka and Mura modeled the energy of a PSB , which has been extended to account for the role of grain boundaries by Sangid et al.  and implemented into a polycrystalline formulation . The material of interest, RR1000, has been characterized via electron backscatter diffraction (EBSD), and the microstructural attributes have been used to create a virtual instantiation of the material. A statistically equivalent microstructure was created that matches the distribution of grain size and grain boundary content (including density of twins), possesses a random texture, and is sufficiently large to match the cyclic strength properties of the material (in terms of yield, hardening behavior, and maximum stress during strain controlled cyclic loading) . For more details of the microstructure generation and meshing, please refer to reference . In order to obtain the micromechanical fields relative to the microstructure, a crystal plasticity model was implemented.
Each PSB in the virtual polycrystal was monitored and the critical PSB is defined as the PSB that is the first to reach its stable value. A detailed sensitivity analysis, uncertainty quantification, and uncertainty propagation study for the fatigue model are available in ref. .
Grain Boundary Engineering
Typical grain boundary engineering (GBE) processes known today were first introduced in the mid-1980s and consist of cold rolling/working at strains ranging from 5 to 20% followed by a short annealing time at high temperature, in order to promote the formation of annealing twins [55, 56]. Subsequent iterations of deformation/anneal are often performed in order to obtain a sufficient fraction of twin boundaries . The iterations also allow for the interaction of Σ3n (n = 1,2,3) boundaries to induce multiple twinning, the formation of incoherent twins, or triple junctions. From a commercial manufacturing perspective, room temperature deformation of high-strength Ni-based superalloys, such as RR1000, is impractical as these materials can withstand only limited deformation without cracking. Additionally, forging die materials cannot withstand the stresses required to deform the material and forging equipment would limit the size and complexity of potential component configurations. Furthermore, the short annealing times employed to limit recrystallization and grain growth are not compatible with the considerable thermal inertia inherent in large structures. Finally, the multiple deformation and thermal processing cycles would add manufacturing lead time and cost even if the aforementioned limitations could be overcome. As such, the current approaches used for GBE are not ideally suited for the fabrication of physically large and complex-shaped Ni-based superalloy components for propulsion and power generation applications.
Conventional isothermal forging of RR1000 utilizes deformation parameters that fall within the superplastic flow region of the deformation mechanism map of Fig. 3. In order to increase the density of twin boundaries in the microstructure of the material following deformation and annealing, the deformation parameters were modified such that the dominant mechanism for accommodating strain was shifted toward the regime where dislocation-based plasticity becomes operative.
Following deformation, the average grain size was smaller in the GBE condition at 2.6 μm compared to 4.1 μm in the baseline material. Similarly, the length fraction of Σ3 boundaries was lower in the GBE condition at 13% compared to 21% in the baseline forging. Additional details on the calculation of the density and length fraction of Σ3 boundaries as well as the generation of intragranular misorientation maps can be found in Ref. .
Processing conditions for the four variants of RR1000, including isothermal forging condition (temperature and strain rate) and heat treatment. Each heat treatment was performed for 4 h with a controlled cooling rate of 1 °C/s
Conventionally forged (baseline)
Grain boundary engineered (GBE)
Fine grain (FG)
Forged: 1100 °C and 0.003 s−1
Sub-solvus annealed: 1115 °C
Forged: 1020 °C and 0.05 s−1
Sub-solvus annealed: 1115 °C
Coarse grain (CG)
Forged: 1100 °C and 0.003 s−1
Super-solvus annealed: 1170 °C following 1115 °C
Forged: 1020 °C and 0.05 s−1
Super-solvus annealed: 1170 °C following 1115 °C
It is well known that Ni-based superalloys deform with high degrees of heterogeneity , manifesting high levels of deformation along slip bands . In order to understand the effects of GBE on this material, it is important to capture the discrete slip bands and their interactions with grain boundaries. To capture material heterogeneity during different stages of deformation, the surface strain is characterized spatially relative to the microstructural features using digital image correlation. Digital image correlation (DIC) is a technique that uses a random surface pattern to track displacements from an initial state. As the sample is deformed, these random surface features are displaced, mirroring the underlying microstructural surface behavior. Image correlation between initial and deformed states yields in-plane surface displacements, which are then integrated into in-plane strain components . In this experiment, the initial state is correlated with deformed states at 1 and 10 fatigue cycles at room temperature. Four different conditions of RR1000 were investigated during this experiment, as previously summarized in Table 1.
Specimen Preparation and Fatigue Testing Conditions
Dog bone specimens were produced via EDM from forged disks of baseline and GBE RR1000. Specimens had gauge sections measuring 10 mm by 3 mm by 1.25 mm thick. A cross-section of the forged disk, through the thickness, was taken, and EBSD scans were conducted on these areas, in order to identify and avoid the surface discontinuities produced by the forging process. Forged surfaces were removed, such that a consistent microstructure, specifically the twin density, twin length fraction, and grain size, was present in all specimens. Specimens were prepped via mechanical grinding and polishing. Final polishing was completed with 0.3 and 0.05 μm alumina and colloidal silica, respectively. Fiducial markers were placed in a rectangular array in the middle of each specimen defining the area of interest. The microstructure was characterized via EBSD scans, in order to acquire the spatial crystallographic orientations that were used to reconstruct the spatial position of the grain boundaries.
Digital Image Correlation Using Electron Imaging
A 100 × 75 μm2 area of interest was observed in the fine grain specimens, while a larger area of 200 × 200 μm2 was observed in the coarse grain specimens, in order to capture a representative number of grains during the characterization of each specimen. Image correlations were conducted using Correlated Solutions VIC-2D™. Regions of interest for the analysis of the fine grain material consist of four stitched images; each image was correlated using a subset of 0.42 × 0.42 μm2 and a step size of 0.03 μm. The strain maps for the coarse grain samples consist of six stitched images, two wide by three long, and each image was correlated using a subset of 1.34 × 1.34 μm2 and a step size of 0.06 μm. A larger subset was selected for the coarse grain material to capture the discrete slip bands; smaller subsets could not capture the relative large displacements that occurred along slip bands resulting in poor correlations.
Strain Map Results and Discussion
Digital image correlation using images obtained by electron microscopy was conducted on the four aforementioned cases. Strain maps are shown at 10 cycles in Fig. 12. The strain component along the loading direction of the specimen is plotted, εxx. Strain maps at 10 cycles were selected to demonstrate the role of cyclic hardening, due to the presence of strain accumulation at grain boundaries and along slip bands, but well before diffuse strain images, whereas out-of-plane deformation obscures individual slip bands within grains. The white areas in Fig. 12 correspond to a low confidence value of strain during the digital image correlation process, and thus, the subsequent strain value is not reported in the strain map.
Average strain accumulation in the CG baseline and GBE microstructures in the unloaded state, Fig. 12a, b, are 0.73 and 0.87% after 10 loading cycles, respectively. The baseline structure demonstrates a larger mean distance between slip bands, as a consequence almost all slip activity captured is part of a saturated slip band, where the strain value measured is 3.5% or greater in the loading direction. Meanwhile, the CG GBE structure does contain more strain, Fig. 12b. On average, many of the grains in the CG GBE structure show finer spacing between slip bands, as compared to the CG baseline; this results in strain not localizing on fewer slip bands but rather accommodating the deformation over more slip bands. All slip bands observed were straight, characteristic of shearing γ′ precipitates  in Ni-based superalloys and indicative of planar slip .
Accumulated average plastic strain states of each RR1000 variant after 1 and 10 fatigue cycles in the unloaded state
Baseline fine grain
GBE fine grain
Baseline coarse grain
GBE coarse grain
As previously noted, for the FG microstructures (in both the baseline and GBE cases), the individual slip bands are obscured, such that strain appears as a continuous field. By viewing the strain field, strain manifests along 45° macroscopic bands relative to the specimen’s loading direction with distinct patterning around microstructural features. Moreover, the higher stress imposed on the FG material results in the activation of multiple slip systems within individual grains, as discussed by Boettner et al.  and Kocks and Mecking [78, 79]. These additional slip systems coupled with the smaller slip band spacing explain the inability to resolve slip band in the FG material. Shyam and Milligan  investigated the dislocation arrangement in these FG structures via transmission electron microscopy and found that deformation within smaller grains were driven by an isolated movement of dislocations during deformation, not planar slip bands as in coarse grain material. This resulted in highly homogenous deformation, where homogenous deformation was defined as highly decreased spacing between slip bands and activation of multiple slip systems .
An existing fatigue model, based on persistent slip bands (PSBs) [20, 40, 42, 43], is used to identify microstructural features that are beneficial for achieving an enhanced fatigue life. A PSB of longer length is attributed to more stored strain and a higher stress concentration, therefore leading to a shorter fatigue life. Therefore, a small, uniform grain size is preferred without the presence of texture. Further, the grain boundaries play a critical role in the fatigue performance, low-angle GBs, promoting slip transmission and PSB forming across grain clusters, should be avoided. Finally, twin boundaries are beneficial in their role of impeding the length of PSBs that form.
Conventional GBE approaches are not suited to the processing of Ni-based superalloys for turbine disk applications. Models for the formation and grain size dependence of twin boundaries were used to identify conditions favorable to the improvement of the grain boundary network during isothermal forging. A process map was developed for RR1000, and GBE was achieved using hot deformation parameters that trigger dislocation-based plasticity mechanisms. The residual strain energy stored within the microstructure promoted the formation of twin boundaries during subsequent annealing via SIBM.
Small-scale forgings of RR1000 were produced using the conventional forging parameters and the modified forging parameters to achieve GBE. Furthermore, two heat treatments, respectively, sub-solvus and super-solvus, were performed to produce fine grain (FG) and coarse grain (CG) microstructures. The density of twin boundaries was almost doubled in the GBE condition compared to the conventionally processed forging, in the FG variants. The CG only showed a modest increase in the density and length fraction of twin boundaries.
High-resolution DIC was performed on the four variants of RR1000. Each strain map displayed heterogeneous deformation along 45° macroscopic bands relative to the specimen’s loading direction and accumulating around microstructural features, especially twin boundaries. Individual strain bands could be readily observed within the strain maps of the CG material. The length of these slip bands was shorter across the GBE material, which, as the fatigue model predicts, is attributed to a longer fatigue life.
It is noted that a complete description of the GB requires five degrees of freedom (three for orientation and two for the GB normal). While the GB networks discussed herein do not represent a unique description of each GB, they are sufficient to demonstrate correlations for the resulting material performance.
Financial support for this work was provided by the National Science Foundation (grant number CMMI 13-34664 and 13-34998) and Rolls-Royce Corporation. The authors would like to thank Randy Helmink, Robert Goetz, and Eugene Sun of Rolls-Royce Corporation for useful discussions about this work and Saikumar Yeratapally for assistance with the simulations. Martin Detrois is appointed to the U.S. Department of Energy (DOE) Postgraduate Research Program at the National Energy Technology Laboratory administered by the Oak Ridge Institute for Science and Education.
ST and MDS conceived the project. The deformation mechanism map, process models for GBE of RR1000, and GBE characterization were performed by MD under the supervision of ST. MH provided technical expertise to produce the small-scale forgings. MDS originated and supervised the fatigue modeling. The DIC strain maps and mechanical behavior analysis were performed by JR under the supervision of MDS. Each author wrote their respective section and the group iterated on the document to converge upon the final form of the manuscript. All authors have read and approved the final manuscript.
Compliance with Ethical Standards
The authors declare that they have no competing interests.
- 3.Sims CT, Stoloff NS, Hagel WC (1987) Superalloys II: high-temperature materials for aerospace and industrial power. Wiley, New YorkGoogle Scholar
- 4.Durand-Charre M (1998) The microstructure of Superalloys. CRC Press, Boca RatonGoogle Scholar
- 11.Randle V (1996) The role of the coincidence site lattice in grain boundary engineering. The Institute of Materials, LondonGoogle Scholar
- 13.Palumbo G, Aust KT (1992) Special properties of Σ grain boundaries. In: Wolf D, Yip S (eds) Materials interfaces: atomic-level structure and properties. Chapman & Hall, London, pp 190–207Google Scholar
- 14.Kronberg ML, Wilson FH (1949) Secondary recrystallization in copper. AIME TransGoogle Scholar
- 30.Alam Z, Eastman D, Weber G, Ghosh S, Hemker K (2016) Microstructural aspects of fatigue crack initiation and short crack growth in René 88DT. In: Hardy M, Huron E, Glatzel U, Griffin B, Lewis B, Rae C, Seetharaman V, Tin S (eds) Superalloys 2016: proceedings of the 13th international symposium of Superalloys. Wiley, Hoboken, pp 561–568. https://doi.org/10.1002/9781119075646.ch60 CrossRefGoogle Scholar
- 32.Neumann P (1999) Analytical solution for the incompatibility stresses at twin boundaries in cubic crystals. In: Wu XR (ed) Proceedings of the 7th International Fatigue Conference (FATIGUE’99). 1:4Google Scholar
- 37.Zhang T, Jiang J, Britton B, Shollock B, Dunne F (2015) Crack nucleation using combined crystal plasticity modelling, high-resolution digital image correlation and high-resolution electron backscatter diffraction in a superalloy containing non-metallic inclusions under fatigue. Proc R Soc A 472:0792. https://doi.org/10.1098/rspa.2015.0792 Google Scholar
- 39.Cerrone A, Stein C, Pokharel R, Hefferan C, Lind J, Tucker H, Suter R, Rollett A, Ingraffea A (2015) Implementation and verification of a microstructure-based capability for modeling microcrack nucleation in LSHR at room temperature. Model Simul Mater Sci Eng 23(3):035006. https://doi.org/10.1088/0965-0393/23/3/035006 CrossRefGoogle Scholar
- 40.Yeratapally SR, Glavicic MG, Hardy M, Sangid MD (2016) Microstructure based fatigue life prediction framework for polycrystalline nickel-base superalloys with emphasis on the role played by twin boundaries in crack initiation. Acta Mater 107:152–167. https://doi.org/10.1016/j.actamat.2016.01.038 CrossRefGoogle Scholar
- 44.Armstrong PJ, Frederick CO (1966) A mathematical representation of the multiaxial Bauschinger effect. Report RD/B/N, 731, Central Electricity Generating Board, Berkeley, UKGoogle Scholar
- 53.Yeratapally SR, Glavicic MG, Argyrakis C, Sangid MD (2017) Bayesian uncertainty quantification and propagation for validation of a microstructure sensitive model for prediction of fatigue crack initiation. Reliab Eng Syst Saf 164:110–123. https://doi.org/10.1016/j.ress.2017.03.006 CrossRefGoogle Scholar
- 56.Watanabe T (1984) An approach to grain boundary design for strong and ductile polycrystals. Res Mechanica 11(1):47–84Google Scholar
- 61.Detrois M, McCarley J, Antonov S, Helmink RC, Goetz RL, Tin S (2016) Comparative study of high-temperature grain boundary engineering of two powder-processed low stacking-fault energy Ni-base superalloys. Mater High Temp 33:310–317. https://doi.org/10.1080/09603409.2016.1155689 CrossRefGoogle Scholar
- 62.Detrois M, Rotella R, Goetz RL, Helmink RC, Tin S (2016) The influence of the starting grain size during high-temperature grain boundary engineering of Ni-Base Superalloy RR1000. In: Hardy M, Huron E, Glatzel U, Griffin B, Lewis B, Rae C, Seetharaman V, Tin S (eds) Superalloys 2016: proceedings of the 13th international symposium of Superalloys. Wiley, Hoboken, pp 459–468. https://doi.org/10.1002/9781119075646.ch49 Google Scholar
- 73.Mello A, Book T, Nicolas A et al (2017) Distortion correction protocol for digital image correlation within a scan electron microscope: emphasis on long duration and ex-situ experiments. Exp. Mech 57:1395–1409. https://doi.org/10.1007/s11340-017-0303-1
- 74.Stoltz RE, Pineau AG, Materiaux C (1978) Dislocation-precipitate interaction. 34:275–284Google Scholar
- 77.Petch NJ (1953) The cleavage strength of polycrystals. J Iron Steel Inst 174Google Scholar
- 79.Mecking H (1981) Deformation of polycrystals: mechanisms and microstuctures. p 73Google Scholar