First creep results on thin-walled single-crystal superalloys

First creep results on thin-walled single-crystal superalloys

Materials Science and Engineering A 510–511 (2009) 307–311 Contents lists available at ScienceDirect Materials Science and Engineering A journal hom...

376KB Sizes 0 Downloads 11 Views

Materials Science and Engineering A 510–511 (2009) 307–311

Contents lists available at ScienceDirect

Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea

First creep results on thin-walled single-crystal superalloys R. Hüttner a,∗ , J. Gabel b , U. Glatzel a , R. Völkl a a b

Metals and Alloys, University Bayreuth, D – 95440 Bayreuth, Germany MTU Aero Engines, Dachauer Str. 665, D – 80995 Munich, Germany

a r t i c l e

i n f o

Article history: Received 28 November 2007 Received in revised form 1 October 2008 Accepted 8 October 2008 Keywords: Creep Thin-walled Single-crystal nickel-base superalloy René N5 Resistant heating Coating

a b s t r a c t The knowledge of creep behavior of structural materials for high temperature applications is prerequisite for lifetime predictions. In order to optimize both the cooling efficiency and the weight of fast rotating turbine blades a general trend to reduce the wall thickness of the hollow investment casting parts is observed. In order to determine the influence of wall thickness on creep properties, constant-load tensile creep tests of samples with different thickness of the single-crystalline nickel-base superalloy René N5 are performed at high temperatures. The test equipment uses resistance heating to achieve fast heating and cooling rates. Creep strain is measured with an accuracy better than 0.1% by a non-contact imaging technique. The tests were performed at 1253 K at different stress levels. Specimens were tested uncoated and aluminized. © 2009 Elsevier B.V. All rights reserved.

1. Introduction The effect of section thickness (also called section debit) on creep deformation and rupture plays an important role in the design and durability. An overview for conventionally cast (CC), columnargrain (CG) and single-crystalline (SC) PWA 1483 was given by Duhl [1] and is illustrated in Fig. 1. Different reasons for the reduction of lifetime in thin sections were pointed out. Stress-assisted diffusion of oxygen along the grain boundaries causes oxygen embrittlement, especially for uncoated specimens. Intergranular creep fracture occurs in equiaxed alloys as a result of cavity nucleation, cavity growth and coalescence. Deformation constraint and deformation anisotropy may change with section thickness. In reference [1] only deformation constraint and anisotropy account for lifetime reduction of thin single-crystalline nickel-base superalloys. Doner and Heckler [2] come to the conclusion that the decrease in rupture lifetime for thin section specimens of CMSX-3 is only important at 1255 K if the applied stress was below 275 MPa. They also found by comparison of creep data of coated and uncoated samples that only for uncoated samples a thickness influence exists and suggested that the difference denote an environmental effect. On the other side Seetharaman [3] described a reduction in rupture strength of 30–40% if the wall thickness of coated samples was reduced from 1.52 mm to 0.25 mm.

For René N5, a second generation single-crystalline nickel-base superalloy, no data about the wall thickness influence is available in literature. The current work aims to investigate the wall thickness influence on creep properties of René N5. Also the influence of aluminized specimens, denominated as-coated is investigated. 2. Experimental 2.1. Material The René N5 material was delivered from the MTU Munich as heat-treated. The chemical composition of the material is listed in Table 1 and is found elsewhere [4]. The aluminide coating of the specimens was also applied at MTU in Munich by an aluminum pack cementation process with an subsequent heat-treatment for 24 h at 1003 K. The coating of an originally 0.2 mm thick sample is presented in Fig. 2. The aluminide scale (1) is about 40 ␮m thick. A diffusion zone (2) of about 40 ␮m was found between the bulk material (3) and the aluminide (1). Hence the coating process reduces the thickness of the René N5 substrate by about 130 ␮m. The overall thickness, including aluminide scale, diffusion zone and substrate is about 0.23 mm and 1.03 mm for thin and thick coated specimens, respectively. 2.2. Specimen design and manufacture

∗ Corresponding author. Tel.: +49 921 555573; fax: +49 921 555561. E-mail address: [email protected] (R. Hüttner). 0921-5093/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2008.10.057

Two test series were performed with uncoated samples. Specimens of the test series denominated René N5 16◦ have a misorientation to [0 0 1] of up to 16◦ and were also tested with an

308

R. Hüttner et al. / Materials Science and Engineering A 510–511 (2009) 307–311

Fig. 3. Comparison of temperatures measured with the dualwavelength pyrometer Impac-300 and type K thermocouple. Fig. 1. Comparison of the thickness influence of conventional cast (CC), columnargrain (CG) and single-crystalline (SC) alloys at 1255 K [1]. Table 1 Nominal composition of René N5 (wt.%). Cr

Co

Mo

Re

W

Al

Ta

Hf

Ni

7.0

8.0

2.0

3.0

5.0

6.2

7.0

0.2

Bal.

aluminide coating. Specimens of the second test series denominated René N5 have a misorientation below 5◦ and were tested up till now without coating. For each test series two specimen geometries of 95 × 3 × 1 mm3 and 95 × 4 × 0.2 mm3 are cut from cast and heat-treated single-crystal plates by wire electrical discharge machining (EDM). Handling of 0.2 mm thick specimen turned out to be very difficult, therefore the thickness was increased to 0.3 mm for René N5 test series (specimen geometry for René N5 test series: 95 × 4 × 0.3 mm3 ). Four small ridges around the specimen centers mark the gauge length of 10 mm for the video extensometer. Specimens were tested as-cut. Stresses for uncoated samples were calculated by neglecting the recast layer produced during the EDM process. For the coated samples, the stresses were calculated referring to the original cross-sections. 2.3. Mechanical testing Constant-load tensile creep testing was performed at 1253 K at different stress levels till rupture. The proprietary test equip-

Fig. 2. Aluminide coated, originally 0.2 mm thick sample of René N5.

ment uses resistance heating to achieve fast heating and cooling. A typical heating rate is 100 K/s and a typical mean cooling rate is 50 K/s with somewhat faster cooling at high temperatures and slower cooling when the temperature falls below 773 K. Testing can be performed under vacuum, inert gas or air in order to determine in-situ oxidation effects on creep behavior. For measuring the creep strain with accuracy better than 0.1%, a self developed, noncontact video extensometer is used. Functionality and the efficiency experiments are found in literature [5]. Temperature measurements for René N5 16◦ 1 mm specimens were done with a TC-2000 dual-wavelength pyrometer from LEIA, with an accuracy of ±10 K. Remaining tests were measured with an Impac-300 series dualwavelength pyrometer with an accuracy of ±5 K at 1253 K. Correct measurement of the temperature was proofed with a Gleeble 3500 from Dynamic Systems using type K thermocouples (Ni/Ni–Cr). A 0.3 mm René N5 sample was used for calibration (Fig. 3). 3. Results 3.1. Creep testing of René N5 16◦ series Comparing the graphs in Figs. 4 and 5 it is obvious that 0.2 mm thick specimens generally show shorter rupture lifetimes and shorter stationary creep than 1.0 mm thick specimens. The creep rupture times are reduced by a factor of 4–7 and rupture strains were reduced by a factor of 6–9, too. Figs. 6 and 7 compare testing series on coated specimens of René N5 16◦ . The nominal stresses refer to the original cross-section area without taking the thickness of the aluminide coating into account as a load-bearing zone [6]. The 0.2 mm thick coated René N5 16◦

Fig. 4. Creep curves of 0.2 mm thick specimens of the test series RenéN5 16◦ at 1253 K under different loads.

R. Hüttner et al. / Materials Science and Engineering A 510–511 (2009) 307–311

Fig. 5. Creep curves of 1 mm thick specimens of the test series RenéN5 16◦ at 1253 K under different loads.

Fig. 6. Creep curves of 0.2 mm thick coated specimens of the test series René N5 16◦ at 1253 K under different loads.

specimen loaded with 230 MPa showed an unexpected short creep rupture life and rupture strain due to failure outside the gauge length. However, the deduced minimum creep rate of this test is unaffected. Rupture lives were reduced by a factor of 50–200 compared to thick coated specimens. The rupture strains were reduced by a factor of 10–20 reducing the specimen thickness from 1 mm to 0.2 mm. A comparison between coated and uncoated specimen reveals an improvement in rupture lifetime for 1 mm thick samples at a stress level of 300 MPa. In Fig. 8 the creep rates of uncoated specimens are presented. At lower stresses of 230 MPa and 270 MPa a gap of more than one decade between thin and thick wall specimens is observed. For tests

Fig. 7. Creep curves of 1 mm thick coated specimens of the test series René N5 16◦ at 1253 K under different loads.

309

Fig. 8. Minimum creep rates vs. stress for René N5 16◦ uncoated samples; n(1 mm) = 14.3 and n(0.2 mm) = 5.3.

Fig. 9. Minimum creep rates vs. stress for René N5 16◦ coated samples; n(1 mm) = 2.4 and n(0.2 mm) = 7.4.

performed at 300 MPa the difference gets smaller. A Norton exponent n of 5.3 for 0.2 mm thin wall tests and 14.3 for thick wall tests was calculated. For coated samples in Fig. 9 a gap of more than one decade was found. The 0.2 mm thin wall samples show a large scatter. Norton exponents of 7.4 for 0.2 mm thin samples and 2.4 for 1.0 mm samples were calculated. 3.2. Creep testing of René N5 series Figs. 10 and 11 compare the testing series of René N5 [0 0 1] specimens with a nominal thickness of 0.3 mm and 1 mm. Influence of specimen thickness according to creep rupture life is only

Fig. 10. Creep curves of 0.3 mm thick specimens of the test series René N5 at 1253 K under different loads.

310

R. Hüttner et al. / Materials Science and Engineering A 510–511 (2009) 307–311

Fig. 11. Creep curves of 1 mm thick specimens of the test series René N5 at 1253 K under different loads.

Fig. 12. Minimum creep rates vs. stress for René N5 for uncoated samples; n(1 mm) = 8.2 and n(0.3 mm) = 6.8.

significant for tests at 230 MPa. The rupture life was reduced by a factor of more than 2. Rupture life of tests performed at 270 MPa and 300 MPa did not show any significant differences. With decreasing wall thickness the rupture strain also decreased by a factor of 1.5–2 depending on the stress level. Minimum creep rates in Fig. 12 did not exhibit as large a gab for 0.3 mm and 1 mm thick samples compared to René N5 16◦ at different stresses. For 0.3 mm thin specimens a Norton exponent of 6.8 and for 1 mm thick wall specimens a Norton exponent of 8.2 was calculated. 4. Discussion For René N5 samples with misorientation to [0 0 1] of less than 5◦ , a drop of the rupture lifetime by the reduction of the specimen thickness from 1.0 mm to 0.3 mm was distinctive only at the lowest stress level of 230 MPa. For 270 MPa and 300 MPa no significant difference could be observed, which is in good agreement with the findings of Doner and Heckler [2]. The minimum creep rates showed no thickness effect at all. In contrast to these findings, 0.2 mm thick René N5 16◦ samples with a misorientation of up to 16◦ showed shorter rupture lifetimes than 1.0 mm thick samples at all investigated stress levels. Hence, even in view of the limited number of tests and possible measurement errors, the crystal orientation seems to influence the thickness effect. René N5 samples with a misorientation to [0 0 1] of less than 5◦ had shorter rupture lifetimes than the René N5 samples with misorientation of up to 16◦ . This was unexpected since literature [7] showed only a very weak influence of the single-crystal orientation

on the creep behavior of CMSX-4 at 1253 K. Explanations for the differences could be the varying pyrometer types used for these series. However, scatter of the limited number of creep tests make a final judgment difficult. The creep tests of uncoated samples at stress levels below 300 MPa showed a reduction of the rupture strains with sample thickness. The rupture strain was reduced from over 20% for 1 mm thick René N5 16◦ and René N5 to around 10% for 0.3 mm thick René N5 and to below 5% for 0.2 mm thick René N5 16◦ . The rupture strains of uncoated samples tested with a load of 300 MPa are always somewhat lower but show the same general trend that the creep rupture strains decrease with specimen thickness. Coated 0.2 mm thick René N5 16◦ samples had remarkably shorter rupture lifetimes than their uncoated equivalents. It is straightforward to account the reduction of the load-bearing crosssection in coated samples for this observation. However, under the assumption that only the unaffected, bulk René N5 interiors are load-bearing, the loads would have been in fact higher by a factor of 3.0 for nominal 0.2 mm thick coated samples. Thus, the coated 0.2 mm thick René N5 16◦ sample loaded by a nominal stress of 270 MPa (see Fig. 6) would have seen an actual load of 810 MPa. On the other hand, the actual creep time to rupture of 1.3 h is much too long if one extrapolates the rupture times for uncoated 0.2 mm thick René N5 16◦ samples to a creep test under a load of 810 MPa (see Fig. 4). Apparently, the actual loads on nominal 0.2 mm thick specimens would be overestimated if one assumes only the bulk René N5 interiors as load-bearing. With a less restricted assumption that the bulk René N5 interiors together with the diffusion zones are load-bearing, the tension loads would be higher by a factor of 1.3 for 0.2 mm thick coated samples. The coated 0.2 mm thick René N5 16◦ sample loaded with a nominal stress of 270 MPa, which ruptured after 1.3 h, would have seen an actual load of 350 MPa (see Fig. 6). An extrapolation of the creep rupture times of uncoated 0.2 mm thick René N5 16◦ samples to a load of 350 MPa (see Fig. 4) would fall close to 1.3 h. Hence, there is evidence that in addition to the bulk interiors the diffusion zones and the aluminide scales of coated samples carry some load, too. The creep rupture times at 1253 K of nominal equally loaded uncoated and coated 1.0 mm thick René N5 16◦ samples (compare Figs. 5 and 7) are in the same order of magnitude. The poor statistics of the limited number of tests does not allow to deduce an effect due to the reduction of the load-bearing cross-section through the coating. 5. Summary Creep testing has been performed at 1253 K on thin and thick specimens of the single-crystal nickel-base superalloy René N5 with two orientations. One specimen set had a misorientation to [0 0 1] of less than 5◦ and the other of up to 16◦ . (i) Uncoated and coated René N5 with a misorientation of up to 16◦ had shorter creep live times and higher minimum creep rates at all investigated stresses when the specimens were nominally 0.2 mm thick as compared to specimens with 1 mm in thickness. (ii) The thickness effect was less pronounced for René N5 with a misorientation of less than 5◦ . Uncoated René N5 specimens only showed a decrease in lifetime at a low stress level of 230 MPa. (iii) The crystal orientation seems to influence the thickness effect. (iv) Noticeable short creep lifetimes of thin coated specimens can be explained only partly by the reduction of the load-bearing cross-section.

R. Hüttner et al. / Materials Science and Engineering A 510–511 (2009) 307–311

Acknowledgements This work was performed as part of the Graduate School 1229. The author wants to thank the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) for their financial support and the MTU Munich for the testing material. References [1] D.N. Duhl, in: C.T. Sims, N.S. Stoloff, W.C. Hagel (Eds.), Superalloys II, John Wiley, New York, 1987, pp. 189–214.

311

[2] M. Doner, A. Heckler, SAE Technical Paper 851785, Society of Automotive Engineers, Inc., Warrendale, 1985. [3] Seetharaman, in: K.A. Green, H. Harada, T.W. Howson, T.M. Pollock, R.C. Reed, J.J. Schirra, S. Walston (Eds.), Superalloys 2004, Tenth International Symposium, The Minerals, Metals, and Materials Society, Warrendale, PA, 2004, pp. 207– 214. [4] Y. Koizumi, T. Yokokawa, H. Harada, T. Kobayashi, J. Japan Inst. Metals 70 (2006) 176–179. [5] R. Völkl, B. Fischer, Experimental Mechanics 44 (2004) 121–127. [6] P. J Henderson, in: J. Lecomte-Beckers, F. Schubert, P.J. Ennis (Eds.), Schriften des Forschungszentrums Juelich, Energy Technology, 5 (3, Materials for Advanced Power Engineering 1998, Pt. 3), Juelich, 1998, pp. 1559–1568. [7] V. Sass, U. Glatzel, M. Feller-Kniepmeier, Acta Mater. 44 (1996) 1967–1977.