Author’s Accepted Manuscript Investigation on microstructure evolution and fracture morphology of single crystal nickel-base superalloys under creep-fatigue interaction loading Z.Y. Yu, X.Z. Wang, Z.F. Yue, X.M. Wang www.elsevier.com/locate/msea
PII: DOI: Reference:
S0921-5093(17)30613-5 http://dx.doi.org/10.1016/j.msea.2017.05.018 MSA35034
To appear in: Materials Science & Engineering A Received date: 13 February 2017 Revised date: 7 April 2017 Accepted date: 4 May 2017 Cite this article as: Z.Y. Yu, X.Z. Wang, Z.F. Yue and X.M. Wang, Investigation on microstructure evolution and fracture morphology of single crystal nickel-base superalloys under creep-fatigue interaction loading, Materials Science & Engineering A, http://dx.doi.org/10.1016/j.msea.2017.05.018 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Investigation on microstructure evolution and fracture morphology of single crystal nickel-base superalloys under creep-fatigue interaction loading Z. Y. Yu, X. Z. Wang, Z. F. Yue*, X. M. Wang* School of Mechanics, Civil Engineering Polytechnincal University, Xi’an 710072, China [email protected] [email protected]
Abstract: Creep-fatigue interaction behavior of a single crystal nickel-base superalloy was studied at 980 oC and the microstructure was quantitatively analyzed. A three-stepped plate specimen had been designed to simultaneously observe microstructure changed under three different stresses. It was found that the distribution of γ′ size was basically consistent with lognormal probability scale under all three loading conditions. The average size of γ′ precipitates increased with the increasing of stress at the same interrupted time. Four different layers including oxide layer, γ′-free layer, γ′-reduced layer and γ/γ′ layer can be observed on the cross-section of the specimens. The thickness of γ′-free layer also increased with the increasing of stress. Numerous dimples and square-shaped facets were observed at the fracture surface and furthermore propagation of cracks from micropores to the boundary of the facets could be found, which showed the fracture behaviours of both creep and fatigue damage. Keywords: Nickel-base single crystal superalloy; Creep-fatigue interaction; Microstructure; Fracture morphology
1. Introduction Single crystal nickel-base superalloys have been widely used in hot-section components in modern aircraft engines, such as gas turbine blades and vanes. They exhibit excellent high temperature mechanical strength, creep resistance and low cycle fatigue resistance. During the service, these critical components suffer fatigue damage caused by the repeating start-stop transient loading and the creep damage
from the long term high temperature conditions. Therefore, damage resulting from the creep-fatigue interaction (CFI) should be taken into consideration to better design the components. In order to observe the damage evolution and mechanical behavior of materials under CFI loading, several influence factors are controlled in the experiments, including loading path [1-2], hold time at the peak strain/stress [3-9], initial composition and microstructure [10-11], geometrical features of specimens  etc. According to the loading path, the CFI experiments include continuous cyclic creep (CF), prior fatigue followed by creep loading (F + C) and prior creep followed by fatigue loading (C + F). Under continuous cyclic creep loading, different hold time at the peak strain/stress is introduced to investigate the effect of hold time on the CFI behavior. Under CFI loading, the material damage accumulates as the microstructure evolves. The quantitative analysis of microstructure evolution can be adopted to predict the residual life of materials. It is well known that the microstructure of single crystal alloys consists of a high volume fraction of Ll2 ordered γ′ precipitates coherently embedded in a faced centered cubic γ matrix. These γ′ precipitates have been found to transform into flat shapes under high temperature and external stresses conditions. Under separate condition, i.e. creep loading or fatigue loading in high temperature, plenty of results about microstructure evolution have been reported. Creep tests on sheet specimens of single crystal superalloys had shown the oxide layer near the surface region, and microstructures of four layers were observed symmetrically to a virtual mirror plane: oxide layer, γ′-free layer, γ′-reduced layer and extensive normal γ/γ′ phases [13-15]. The microstructure evolution of each layer was influenced by temperature, loading condition and oxidation. As the γ′ precipitates being the primary strengthening phase of single crystal superalloys, some features were measured to quantitatively analyze the microstructure evolution, such as shape, dimension and volume fraction of γ′ phases [16-18]. During fatigue loading, SEM observations indicated that nearly no change in shape and morphology of γ′ phase occurred far from the fracture surface of the specimens [19-22], while the size and the morphology of γ′ phase changed in some localized areas under high-cycle fatigue
loading at elevated temperature [20,23]. It showed that γ′ precipitates tended to dissolve into the γ matrix, and the secondary γ′ precipitates would finally form. On the other hand, for the investigation on microstructure evolution of nickel-base superalloys under creep-fatigue interaction loading, present works usually focus on fracture surfaces, in order to reveal the initiation site of fatal crack and to characterize its propagation mode [5-6]. Little work has been reported to quantitatively analyze the microstructure evolution of γ/γ′ phases. And thus describing the creep-fatigue interaction is helpful to precisely estimate the service life of critical components. This paper aims to observe γ/γ′ morphology under CFI loading. The size of γ/γ′ phases under different stress conditions was measured to quantitatively analyze microstructure evolution to obtain the basic laws under CFI loading. Meanwhile, the microstructure evolution near the oxide layers was also observed to describe the effect of oxidation on the creep-fatigue interaction behavior. The fracture surface was also noted to show the damage caused by creep loading and fatigue loading.
2. Experimental details The material used in the experiment was the single crystal nickel-base superalloy DD5 and the composition of the alloy was (in wt. %): C, 0.059; Cr, 7.00; Co, 7.83; W, 5.01; Mo, 1.52; Ta, 6.51; Al, 6.01; Hf, 0.11; B, 0.004; Re, 3.08; Y, 0.016; and Ni, balance. The axis of the plate specimen was orientated with a maximum deviation of 12o from  orientation. The specimen has a three-stepped geometric profile, with a 2mm thickness. The dimension of three-stepped specimen was shown in Fig. 1(a), which had been designed by the FEM (finite element model) analysis to insure uniaxial loading condition in three gauge lengths. Comparing with the normal specimen, the design of three-stepped specimen was in favour of observing microstructures evolution under three different stress loads without requiring extra specimens for each stress condition. As referred in [1-2], several types of creep-fatigue interaction experiments have been practiced in laboratories, which included prior creep followed by fatigue, prior fatigue followed by creep and continuous cyclic creep (with hold period at peak level)
loads. In order to reflect actual working conditions of these critical components, the creep-fatigue interaction experiments in the paper were conducted by means of stress-controlled continuous cyclic creep at 980 oC, the stress ratio R of the minimum stress to the maximum stress was 0.1, which describes the cyclic loading in fatigue. A cycle unit in the load waveform is from point 1 to point 4, as described in Fig. 1(b). The process from point 1 to 2 indicates the transition from minimum stress to maximum stress. In the experiment, this process lasted 480s and the hold time ∆t at peak level was 3600s. Contrast experiments were adopted to systematically observe microstructures of different stages under creep-fatigue interaction experiments. The detailed test conditions are listed in Table 1. #1 specimen totally fractured after 494.2h and suffered 390 loading cycles in all. In contrast to test #1, test #2 was interrupted at 100 hours and totally experienced 79 loading cycles, which was considered to be the intermediate stage of test #1. Polished and etched samples extracted from #1 and #2 specimens were examined using SEM. To avoid multiaxial stress states, microstructure evolution observation were conducted in the cross section of Section A, Section B and Section C, which were far away from the transition section and fracture surface. The locations of three sections are marked in Fig. 1(a). The γ/γ′ microstructure in the near-surface region was inspected to discuss the effect of oxidation during creep-fatigue interaction behavior, and the size distribution of γ′ precipitates was measured by ImageJ analysis software. Method based on references [24-25] was used to describe irregular morphology of γ/γ′ microstructure in test #1. In addition, fracture surface of the failed samples was examined to analyze the failure mechanism.
3. Results and discussion. 3.1 Morphology evolution of γ/γ′ structures Through pictures selected from the near-surface region, the oxide layers can be observed, as shown in Fig. 2. Under high temperature and loading condition, the investigation of oxide layers is very important to analyze the load carrying capacity of
the cross section, especially for thin specimens. Both Fig. 2(a) and Fig. 2(b) show four layers of the microstructure. These features were also examined in M. Bensch and A. Srivastava’s works [13-14] in creep tests. It indicated from Fig. 2(b) that γ/γ′ microstructure occurred topological inversion at 494.2 hours, and the γ matrix was embedded in γ′ phases. As the distribution of γ/γ′ structure in Fig. 2(b) is irregular, the quantitative analysis of γ/γ′ structure is achieved by calculating the specific connectivity number of γ′ phase NA(γ′) solely. Therefore, the measurement and analysis of γ/γ′ structures are taken in both #1 and #2 specimens, in order to investigate morphology evolution of γ/γ′ structures under three different stresses and different time. The first layer next to the oxide layer is the γ′-free layer, no γ′ precipitates can be observed, while large microvoids present in this layer. The thickness of the γ′-free layer was measured along the near-surface boundary, several representative SEM images are used here to accurately reflect the variation tendency of the thickness. The average thickness of γ′-free layer in #2 at 146.67MPa, 176MPa and 220MPa are 5.59μm, 5.69μm and 5.88μm respectively. Comparing the average thickness under three different stresses, it is shown that larger stress condition leads to more evident oxidation effect during creep-fatigue interaction behavior at 980 oC. As high volume fraction of γ′ precipitates improve the mechanical property of single crystal superalloy, the existence of γ′-free layer influences the life of creep-fatigue specimen. This influence can also be seen in M. Bensch’s works [13-14]. The extensive central part of the specimen is the normal γ/γ′ phases. The γ′ size distribution of the normal γ/γ′ phases in test #2 was characterized to quantitatively analyze the microstructure change under three different stress conditions. Several representative SEM images were selected as statistical samples, which were displayed in Fig. 3. In these pictures, most γ′ precipitates remain quadrate, and no obvious rafting is observed. By contrast, under 146.67MPa and 176MPa at 494.2 hours, γ′ precipitates in the central part of the specimen had experienced complete rafting and then topological inversion occurred. These features are shown in Fig. 4, which can also be observed at the accelerated creep stage in .
In Fig. 5, the size distribution of γ′ precipitates is presented by a cumulative distribution function. According to the cumulative distribution function, we can get the average size of γ′ precipitates at a frequency of 0.5. Fig. 5 shows the γ′ precipitates distribution under three different stresses: 146.67MPa, 176MPa, and 220MPa. It can be found that γ′ size mainly ranges from 0.16μm to 0.8μm. The γ′ size distribution at 146.67MPa is smaller than that at 176MPa and 220MPa, and the average γ′ size is 0.345μm. Comparing with average γ′ size at 176MPa and 220MPa, it is nearly the same, which is about 0.380μm. However, many larger sizes of γ′ precipitates exist at 220MPa than that at 176MPa, and it can reach as large as 2μm. Fig. 6 indicates that the distribution of γ′ size under three different stresses basically accords with the lognormal probability scale. In Table 2, the function expression and parameters are listed. At the same time and temperature, we can find that the value of parameters xc, w and A under three stresses has little difference, but the difference of parameter y0 is obvious, which is guessed to depend on the stress condition. The external stress is considered to serve as one of the driving force for rafting. It affects the internal stress distribution in γ/γ′ structure, which has influence on the diffusion of alloying elements . Therefore, stress conditions influence the microstructure evolution, which is reflected by the change of γ′ size. It has been shown that γ′ precipitates have larger size at 220MPa in the interrupted creep-fatigue interaction experiment, and the larger stress promotes the directional coarsening of cubical γ′ precipitates, comparing with the 176MPa condition. G.M. Han  and D. Chatterjee  obtained similar results when investigating the creep property and microstructure evolution of a single crystal superalloy. Due to the irregular morphology of γ/γ′ structure in test #1, the topological inversion of γ/γ′ microstructure can be quantified by measuring the variation with time of the specific connectively number of the γ′ phase NA(γ′). The γ′ phase NA(γ′) is calculating using the following expression: NA(γ′) = (NT - NTP)/2S
where NT is the termination number and NTP the triple point number of the γ′ phase, which are marked by black dots and black triangles respectively in Fig. 7. S, the
analyzed field, is about 1000μm2 in the paper. Fig. 7 is derived from Fig. 4. In Fig. 7(a), NT is 44, NTP is 60, and the application of equation (1) gives -0.008μm−2 . Similarly in Fig. 7(b), NT is 52, NTP is 76, and the application of equation (1) gives -0.012μm−2 . According to the reference , topological inversion occurs when NA(γ′) becomes negative, and the strain rate increase simultaneously, which corresponding to strain accelerated stage. 3.2 Fracture surface observation The fracture surface of test #1 specimen is shown in Fig. 8. The stress ratio R of the creep-fatigue test was 0.1, and the dwelling time of maximum stress was one hour. Therefore, the single crystal superalloy was subjected to tension stress all along, and the fracture morphology was similar to the pure-creep condition. There are numerous dimples and square-shaped facets on the surface, which shows a ductile mode of failure. However, the dimples are shallower and sparsely distributed, which indicates the existence of fatigue damage influence. Square-shaped facets shown in Fig. 8(b) is interconnected by ductile tearing ridges, each of the individual tearing ridge converges and forms the river pattern, and this fracture morphology exhibits quasi-cleavage behavior. In the centre of square-shaped facets there exists one pore, and the micropore would cause local stress concentration, which eventually acts as the fatigue crack initiation. Fig. 8(c) shows the cracks propagation from the micropore to the boundary of the facet. However, at high temperature and low stress, i.e. 220MPa/980 oC, the oxidation phenomenon is very severe, meanwhile cracks exposed in the air would be oxidized at the same time. And this can promote the closure of cracks and finally slow down the crack propagation.
4. Conclusions A three-stepped plate specimen of DD5 single crystal superalloy had been designed to observe the microstructure change and fracture morphology on creep-fatigue interaction behavior at 980 oC. It was found that γ′ precipitates had larger size at 220MPa, comparing with the γ′ precipitates size at 146.67MPa and
176MPa. The distribution of γ′ size under three different stresses basically accords with lognormal probability scale. In the near-surface region, the oxide layer can be observed and the thickness of γ′-free layer is related to the stress condition. It is revealed that larger stress condition contributes the effect of oxidation during creep-fatigue interaction behaviour at 980 oC. Numerous dimples and square-shaped facets as well as crack propagation from the micropore to the boundary of the facet were observed at the fracture surface, which reflected the fractography of both creep and fatigue behaviours.
Acknowledgements This work is supported by National Natural Science Foundation of China (Grant No. 51375387 and 51210008)
References  Pierce C J, Palazotto A N, Rosenberger A H, Materials Science and Engineering: A. 527(29) (2010) 7484-7489.  Hu D, Wang R, Materials Science and Engineering: A. 515(1) (2009) 183-189.  Chen G, Zhang Y, Xu D K, et al, Materials Science and Engineering: A. 655 (2016) 175-182.  Vacchieri E, Costa A, Riva A, et al, MATEC Web of Conferences, EDP Sciences. 14 (2014) 19002.  Hu D, Ma Q, Shang L, et al, Materials Science and Engineering: A. 670 (2016) 17-25.  Billot T, Villechaise P, Jouiad M, et al, International Journal of fatigue. 32(5) (2010) 824-829.  Liu H, Bao R, Zhang J, et al, International Journal of Fatigue. 59 (2014) 34-42.  Shi D, Liu J, Yang X, et al, Materials Science and Engineering: A. 528(1) (2010) 233-238.  Zrn k J, Semeňák J, Vrchovinský V, et al, Materials Science and Engineering: A. 319 (2001) 637-642.  Le Graverend J B, Cormier J, Jouiad M, et al, Materials Science and Engineering: A. 527(20) (2010) 5295-5302.  Tsang J, Kearsey R M, Au P, et al, Materials at High Temperatures. 27(1) (2010) 79-88.  Silva J M, Cláudio R A, Branco C M, et al, Procedia Engineering. 2(1) (2010)
1865-1875.  Bensch M, Konrad C H, Fleischmann E, et al, Materials Science and Engineering: A. 577 (2013) 179-188.  Bensch M, Preußner J, Hüttner R, et al, Acta Materialia. 58(5) (2010) 1607-1617.  Srivastava A, Gopagoni S, Needleman A, et al, Acta Materialia. 60(16) (2012) 5697-5711.  Murakumo T, Kobayashi T, Koizumi Y, et al, Acta Materialia. 52(12) (2004) 3737-3744.  Meissonnier F T, Busso E P, O'Dowd N P, International Journal of Plasticity. 17(4) (2001) 601-640.  Mukherji D, Rösier J, Zeitschrift für Metallkunde. 94(5) (2003) 478-484.  Zhou H, Ro Y, Harada H, et al, Materials Science and Engineering: A. 381(1) (2004) 20-27.  Liu Y, Yu J J, Xu Y, et al, Materials Science and Engineering: A. 454 (2007) 357-366.  Shi Z X, Li J R, Liu S Z, et al, Transactions of Nonferrous Metals Society of China. 21(5) (2011) 998-1003.  Shi Z, Wang X, Liu S, et al, Progress in Natural Science: Materials International. 25(1) (2015) 78-83.  Liu Y, Yu J J, Xu Y, et al, Materials science forum, Trans Tech Publications. 546 (2007) 1219-1224.  Epishin A, Link T, Brückner U, et al, Acta materialia. 49(19) (2001) 4017-4023.  Caron P, Ramusat C, Diologent F, Superalloys 2008. (2008) 159-167.  Kamaraj M, Sadhana. 28(1-2) (2003) 115-128.  Han G M, Yu J J, Hu Z Q, et al, Materials Characterization. 86 (2013) 177-184.  Chatterjee D, Hazari N, Das N, et al, Materials Science and Engineering: A. 528(2) (2010) 604-613.
(b) Fig. 1. (a) Geometry of the specimen used in creep-fatigue test (unit: mm), and (b) Schematic of loading condition for the load-controlled creep-fatigue tests on DD5 single crystal nickel-base superalloy.
Fig. 2. SEM images from the near-surface region showing various oxide layers at 146.67MPa/980 o
C, and the samples were selected from section A: (a) t = 100h (b) t = 494.2h.
Fig. 3. Microstructure of DD5 of three different stresses at 100 hours, most γ′ precipitates remain quadrate, and no obvious rafting is observed: (a) σ = 146.67MPa (b) σ = 176MPa (c) σ = 220MPa.
Fig. 4. Microstructure of DD5 under two different stresses at 494.2 hours, γ/γ′ microstructure occurred topological inversion, and the γ matrix was embedded in γ′ phases: (a) σ = 146.67MPa (b) σ = 176MPa.
Fig. 5. Cumulative distribution of γ′ size under three different stresses at 100 hours.
Fig. 6. γ′ size distribution of three different stresses at 100 hours: (a) γ′ size distribution at 220MPa (b) γ′ size distribution at 176MPa (c) γ′ size distribution at 146.67MPa (d) γ′ size distribution of three different stresses shown in uniform Cartesian coordinate system.
Fig. 7. SEM images of γ/γ′ microstructure derived from Fig. 4. The area is 1000μm2 in both pictures. Counted NT and NTP of the γ′ phase are marked by black dots and black triangles respectively. (a) σ = 146.67MPa (b) σ = 176MPa.
Fig. 8. Fracture morphology of DD5 single crystal nickel-base superalloy at 220MPa/980 oC after creep-fatigue interaction behavior.
Table 1. Test conditions of creep-fatigue interaction experiments.
Test condition Stress level in Stress level in Stress level in #(code) Section A (MPa) Section B (MPa) Section C (MPa) #1 146.67 220 176 #2
Test end state Fracture 494.2 hours Interrupted 100 hours
Table 2. The fitted curve function of γ′ size distribution of three different stresses at 100 hours.
√2 Value 0.00254 0.50831 0.35037 0.25052 0.00615 0.51951 0.33912 0.24276 0.00468 0.4824 0.32006 0.24998
Standard Error 0.00136 0.00159 0.00213 0.00142 0.00471 0.00554 0.00749 0.00476 0.00351 0.00385 0.00505 0.00361