Effect of B on microstructure and properties of low thermal expansion ...

3 downloads 0 Views 1011KB Size Report
Effect of B on microstructure and properties of low thermal expansion superalloy. L. X. Yu. 1. , Y. R. Sun. 2. , W. R. Sun*. 1. , W. H. Zhang. 1. , F. Liu. 1. , X. Xin. 1.
Effect of B on microstructure and properties of low thermal expansion superalloy L. X. Yu1, Y. R. Sun2, W. R. Sun*1, W. H. Zhang1, F. Liu1, X. Xin1, F. Qi1, D. Jia1, X. F. Sun1, S. R. Guo1 and Z. Q. Hu1 The effect of B on microstructure and various properties including coefficient of thermal expansion (CTE), HV hardness, and both smooth and notch stress rupture properties of modified ThermoSpan alloy was studied. The results show that B hardly dissolves in matrix. Increasing B content constrains the formation of Laves phase and grain boundary (GB) precipitation of Laves and G phases, but promotes the formation of M(Co, Fe)NbB boride. In low B doped alloy, its intrinsic high susceptibility to intergranular cracks leads to reduced rupture life and notch sensitivity. Increasing B improves grain boundary cohesion, tying up vacancies and reducing GB diffusion, which constrains the nucleation and propagation of intergranular microcracks, prolongs the rupture life and eliminates the notch sensitivity in the new alloy. Compared with conventional Thermo-Span alloy, the B doped modified alloy shows lower CTE and improved notch sensitivity. Keywords: Thermo-Span, Coefficient of thermal expansion, Boron, Notch sensitivity, Grain boundary

Introduction Low CTE superalloys are attractive for aerospace and land based gas turbine engine applications, because they can solve the high temperature problems of high thermal stress and clearance, thus improving the efficiency of the engine and reduce CO2 emission.1–4 Most low CTE superalloys are based with Fe–Ni–Co system with compositions more complex than high strength stainless steels and maraging steels, and investigations are much fewer and less elaborate than steels.5 Compared with Cr free low CTE GH903, GH907 and GH909 (GH90X series) alloys, Thermo-Span alloy shows relatively high strength and much longer rupture life. Especially, due to the addition of 5?5%Cr, the oxidation resistance of Thermo-Span alloy is far better than the above mentioned GH90X series alloys, so that it can be used without additional oxidation resistant coatings, which will reduce the cost.2,3 However, the CTE is a bit higher than GH90X series alloys due to the addition of Cr. Aimed at balancing the CTE and oxidation resistance of the low CTE superalloy, on the basis of Thermo-Span alloy composition specification, Sun et al. decreased Cr from 5?5 to 3?0 wt-% and increased Al from 0?45 to 1?5 wt-%, and the results showed that the CTE was lowered noticeably. Both tensile strength and smooth stress rupture life were also enhanced.6,7 However, such Al–Cr modified alloy suffered serious notch sensitivity which was deduced to be related with the brittleness of GBs caused by the stress 1

Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110016, China Shenyang University of Technology, 111 Shenliao West Road, Shenyang 110178, China

2

*Corresponding author, email [email protected];[email protected]

1470

ß 2013 Institute of Materials, Minerals and Mining Published by Maney on behalf of the Institute Received 9 November 2012; accepted 13 May 2013 DOI 10.1179/1743284713Y.0000000299

accelerated grain boundary oxidation (SAGBO) resulting from the addition of too much Al.4,8 The brittleness of GBs might be improved through addition of GB strengthening elements. An investigation on the mutual effects of Al, Zr and B in Ni3Al intermetallic compound found that, increasing Al lead to the intergranular fracture, Zr segregation caused Al depletion at GBs, and B segregation markedly improved the ductility.9 Additions of B and Zr have been found with beneficial effects in improving hot workability, creep rupture life and elongation, and notch sensitivity in numerous alloys.10–16 Therefore, doping B and Zr in the Al–Cr modified Thermo-Span alloy might be able to eliminate the negative effect of Al and improve notch sensitivity. The present work investigates the microstructure and properties of the Zr containing Al–Cr modified ThermoSpan alloy doped with different B contents.

Experimental procedure The modified Thermo-Span master alloy was vacuum induction melted (VIM) using high purity raw materials. The master alloy was remelted to make two 10 kg ingots with constant 0?044 wt-%Zr, and the content of B was respectively 0?0046 wt-% in alloy 1 and 0?024 wt-% in alloy 2. Conventional Thermo-Span alloy was also prepared via VIM. Compositions of the above mentioned three alloys are shown in Table 1. The ingots were homogenized, forged, and then hot rolled into W18 mm bars. The alloys were solution treated at 1095uC, 1 h, air cooling (AC) and aged at 721uC, 8 h, 55uC h21 cooling to 621uC, 8 h, AC. Optical microscopy, scanning electron microscopy (SEM) and electron probe microanalysis wavelength dispersive spectrometry (EPMA-WDS) were carried out

Materials Science and Technology

2013

VOL

29

NO

12

Yu et al.

Effect of B on low thermal expansion superalloy

1 Microstructure of as treated a alloy 1, b alloy 2 and c Thermo-Span

to analyze the microstructure. Transmission electron microscopy (TEM) equipped with energy dispersive spectrometry (EDS) was used to characterise the fine grain boundary phases and c9 phase by analyzing the carbon extraction replicas. The crystal lattice constants were determined by TEM selected area diffraction (SAD) analysis. X-ray diffraction (XRD) was carried out on the bulk sample of the as treated alloys to characterise the phases. The CTE was determined from 100 up to 800uC using Unitherm-1252 ultra high temperature dilatometer. HV hardness of the grain matrix was also investigated. Samples of as-treated alloys were also statically oxidized at 650uC for 200 h and the cross-sectional microstructures were investigated. Both smooth and notch stress rupture properties were measured at 650uC under a stress of 600 MPa. Both smooth and notch stress rupture specimens were machined in accordance with Chinese Standard GB/T 2039-1997. The longitudinal microstructure of the ruptured test piece was observed to investigate the initiation and propagation of cracks.

2 Images (SEM) of as treated a alloy 1, b alloy 2 and c Thermo-Span

Results Microstructure The microstructure of the as treated alloys is shown in Fig. 1. The grain size of alloy 1 reaches about ASTM 8, while both alloy 2 and Thermo-Span alloy reach about ASTM 6–7. SEM images revealing the precipitation of micrometre scale particles in the as-treated alloys are shown in Fig. 2. Besides several large blocky particles, widely distributed small particles are also present in low B doped alloy 1 (Fig. 2a). However, the precipitation of small particles is reduced in high B doped alloy 2, which is

Table 1 Chemical composition of alloys/wt-% Alloy

Fe

Co

Ni

Nb

Ti

Si

Cr

Al

Zr

B

1 2 Thermo-Span

Bal. Bal. Bal.

29.4 29.4 29.4

24.5 24.5 24.5

4.8 4.8 4.8

0.85 0.85 0.88

0.27 0.27 0.27

3.0 3.0 5.5

1.45 1.45 0.46

0.044 0.044 …

0.0046 0.024 0.0059

Materials Science and Technology

2013

VOL

29

NO

12

1471

Yu et al.

Effect of B on low thermal expansion superalloy

3 Backscattered electron image of a precipitates and b distribution of Nb, c Zr and d B in as treated new alloy

comparable to standard Thermo-Span alloy (Fig. 2b and c). These particles control the grain growth during hot working and solution annealing process, and the mean grain size is relatively small in alloy 1 with denser precipitates. In order to characterise the precipitates in high B doped alloy 2, EPMA mapping examination is carried out to reveal the distribution of Nb, Zr and B as shown in Fig. 3 and the WDS analysis results of the precipitates are shown in Table 2. From the high magnification backscattered electron (BSE) image and the distribution map of Nb (Fig. 3a and b), nearly all these particles are Nb enriched. Only some larger particles are enriched in Zr (Fig. 3c) and some other particles are enriched in B as demonstrated in Fig. 3d. The composition of the marked particles (Table 2) reveals that with the exception of matrix (P-1), all these white precipitates contain a large amount of Nb. The solubility of B and Zr is quite low in matrix (P-1), while Al content is as high as 3?2 at-%, which might promote the precipitation of c9 phase. The

4 X-ray diffraction spectrum of bulk sample of as treated alloys

Table 2 EPMA-WDS results (at-%) of phases marked in Fig. 2 Phase

Fe

Co

Ni

Nb

Ti

Si

Cr

Al

Zr

B

Matrix (P-1) Laves (P-2) Laves (P-3) M(Co, Fe)NbB (P-4)

37.5 22.4 24.9 16.7

28.8 28.2 30.1 19.2

23.6 14.9 16.3 7.5

2.0 24.9 19 21.7

1.0 1.4 1.4 1.4

0.5 4.9 4.8 0.1

3.3 1.8 2.1 1.3

3.2 0.7 0.9 0.4

ND* 0.8 0.5 0.1

ND* ND* ND* 31.7

*ND represents content of the element is too low to be detected.

1472

Materials Science and Technology

2013

VOL

29

NO

12

Yu et al.

Effect of B on low thermal expansion superalloy

shown in Fig. 6. The relatively larger phase enriched in Nb is determined to be Laves phase (hcp: a50?478 nm, c50?7744 nm) (Fig. 6a, c and e), which is the same type as identified in the bulk sample (Fig. 4). The fine high Sicontaining phase is determined to be G phase (face centred cubic (fcc): a51?120 nm). Moreover, small smounts of (Nb,Ti)C, TiN, and (Ti, Zr)2(S,C) precipitates are also present in alloy 1 and alloy 2. As has been demonstrated, the contents of B and Zr are quite low in the matrix. Therefore, the precipitation of c9 phase of alloy is supposed to be hardly affected. The precipitation of c9 phase in the as-treated alloys is shown in Fig. 7. From the thin film TEM micrographs (Fig. 7a and b), the c9 phase in alloy 2 is a bit denser than ThermoSpan alloy. From the carbon replica TEM images (Fig. 7c and d), it can be determined that in both alloys, the size of c9 phase is nearly the same and reaches about 12 nm.

Properties

5 High magnification SEM micrographs showing precipitation behaviour of as treated a alloy 1, b alloy 2 and c Thermo-Span

large blocky particles (P-2) and the small particles (P-3) contain not only comparable Nb, the variations of other elements are also slight. It is believed that they belong to the Laves phase.1 For the B–Nb containing particles (P4), both B and Nb contents are quite high, which are more likely to be M(Co, Fe)NbB borides. The XRD spectrum of the bulk sample of as treated alloys is shown in Fig. 4. The spectrum demonstrates that in all the three tested alloys, besides c phase (a50?360 nm), Laves phase (MgZn2 type hexagonal close packed (hcp): a50?478 nm, c50?7744 nm) is also present. High magnification SEM images of the as treated alloys are shown in Fig. 5. Besides the randomly distributed micrometre scale particles, there are also fine GB precipitates. The GB precipitation in alloy 1 is comparable with Thermo-Span alloy, while, there are not as many precipitates in alloy 2. Fine GB precipitates are analyzed by the TEM carbon extraction replicas as

Figure 8 shows the mean CTE of the as treated alloys. It demonstrates that the CTE of alloy 1 and alloy 2 is similar and much lower than that of Thermo-Span alloy. The oxidation kinetic curve of the as treated alloys tested at 650uC is shown in Fig. 9. It shows that the oxidation kinetic curves are nearly next to each other, which means that the oxidation resistance of alloy 1 and alloy 2 is comparable with Thermo-Span alloy. The mean HV hardness of the interior grain of the as treated alloys is respectively 426 in alloy 1, 419 in alloy 2, and 399 in conventional Thermo-Span alloy. The high HV hardness is supposed to be related with the denser precipitation of c9 phase in new alloy due to addition of Al. Stress rupture properties of the smooth specimens and notch specimens tested at 650uC and 600 MPa are shown in Table 3. The smooth rupture life of the three alloys shows no significant difference. More appreciable differences are presented for notch stress rupture test. For high B doped alloy 2, the notch rupture life is nearly 3 times of alloy 1. Moreover, the notch rupture life of alloy 1 is much shorter than the smooth rupture life. Thus, alloy 1 shows notch sensitivity under the test condition. Apparently, increasing B can markedly improve the notch effects and eliminate the notch sensitivity of the modified alloy. Compared with Thermo-Span alloy, the new B–Zr containing alloy 2 shows lower CTE, comparable oxidation resistance and stress rupture property. Longitudinal microstructure of the ruptured specimens of low B doped alloy 1 (ruptured at 397 h) and high B doped alloy 2 (ruptured at 530 h) is shown in Fig. 10. In low B doped alloy 1 as shown in Fig. 10a, many intergranular cracks are observed both within the test piece and at surface, and the cracks at surface are longer and broader than cracks within the test piece which indicates cracks at surface are formed prior to these within the test piece. However, in high B doped alloy 2, visible cracks are scarce within the test piece and only a few cracks are found at surface. Evidently, the intergranular cracks are significantly reduced due to the addition of B. Figure 11 shows the microcracks within the test piece of alloy 1 and alloy 2. In low B doped alloy 1, besides the large wedge type intergranular cracks, many microcracks and micro-voids are shown at interfaces of small intergranular precipitates (Fig. 11a). However, there are few large microcracks in high B doped alloy 2 and the

Materials Science and Technology

2013

VOL

29

NO

12

1473

Yu et al.

Effect of B on low thermal expansion superalloy

6 TEM extraction replicas showing precipitation of grain boundary a Laves phase and b G phase, EDS spectra of c Laves phase and d G phase and SAD pattern of e Laves phase [001] zone axes and f G phase [012] zone axes

micro-voids around the interface of precipitates are also scarce (Fig. 11b). Figure 12 shows intergranular cracks at the surface of the ruptured test piece. In low B doped alloy 1, besides many microcracks as marked by arrows, a large intergranular oxidation site (O-11) is also shown in Fig. 12a. In high B doped alloy 2 shown in Fig. 12b, several oxides (as marked by number O-21 and O-22) intruding into the test piece along grain boundaries are found. There are not cracks in the advancing front of the oxidized grain boundaries. In a larger oxide marked by O-23, cracks are also identified within the intergrnaular oxidation site. The thickness of the oxide layer varies a lot from the initial oxidation site at the surface of the test piece to the ending point. Observing the tips of O-23 oxide as shown in Fig. 12c, though cracks are identified within the body, no cracks are found at the tip of this oxide.

1474

Materials Science and Technology

2013

VOL

29

NO

12

Discussion The solubility of B in Laves phase is quite limited as can be derived from Table 2. Moreover, there is strong interaction between B and Nb, and they always segregate together and form the B–Nb complexes,18 and even forming high Nb containing boride as shown in Fig. 2 and Table 2. The B–Nb complexes and Nb containing borides take up much Nb, which decreases the precipitation of another Nb enriched but B free Laves phase. The Laves phase in low CTE superalloys can be employed to control grain size during hot working.1,3 Owing to the abundant precipitation of Laves particles, the grain size in alloy 1 is the finest, while the grain size in alloy 2 and conventional Thermo-Span alloy is relatively larger. Many researchers supposed that the addition of B and Zr can block the oxygen penetration along the grain

Yu et al.

Effect of B on low thermal expansion superalloy

7 Images (TEM) showing precipitation of c9 phase in a and c as treated new alloy and b and d Thermo-Span alloy: a, b TEM images of thin film; c, d TEM images of carbon replicas

boundaries during service.14,15 However, intergranular oxidation sites are found in both alloy 1 and alloy 2 isothermally oxidised at 650uC for 200 h as shown in Fig. 13. Comparing the depth of intergranular oxidation sites, it is demonstrated that both alloys show comparable interganular oxidation behavior, that is to say, B has no obvious advantages in preventing the GB oxidation under static oxidation condition at 650uC. Owing to the surface oxidation, intergranular cracks at surface are usually formed firstly which can be deduced from the fact that the cracks are broader and longer than those within the test piece as shown in Fig. 10. The formation of cracks at surface induces larger net stresses imposing on the rest material of that diametral plane. If the interfacial or intergranular cohesion is unable to withstand the enhanced stress, microcracks will be formed at interfaces and GBs. In low

B doped alloy 1, much more cracks are found within the test piece. On the contrary, in high B doped alloy 2, although the rupture life is much longer, the intergranular cracks within the specimen are scarce, which indicates its intrinsic GB is resistant to creep cracking due to the addition of B. Generally, large interganular cracks are grown from the microvoids. Comparing Fig. 11a with b, the large

9 Oxidation kinetics curves of as treated alloys tested at 650uC Table 3 Stress rupture life tested at 650uC/600 MPa of as treated alloys

8 Mean CTE of as treated alloys

Alloy

Smooth/h

Notched/h

1 2 Thermo-Span

405¡15 485¡65 345¡40

175¡25 600¡115 …

Materials Science and Technology

2013

VOL

29

NO

12

1475

Yu et al.

Effect of B on low thermal expansion superalloy

10 Longitudinal microstructures of ruptured test piece of stress rupture specimen of a alloy 1 and b alloy 2

11 Cracks within test piece of a alloy 1 and b alloy 2: cracks are marked by black arrows

12 Microstructures at surface of ruptured specimen of a low B doped alloy 1 and b high B doped alloy 2, and c magnified image revealing tip of oxide O-23: rolling direction is indicated by white arrow

1476

Materials Science and Technology

2013

VOL

29

NO

12

wedge type intergranular cracks and microvoids around the smaller particles are markedly reduced in high B doped alloy 2. Apparently, both nucleation and growth of microvoids are inhibited in high B doped alloy 2. The intergranular precipitates might play a role in inhibiting cracking. However, due to the insufficient precipitatation, many interfaces and GBs are free of precipitates as shown in Fig. 5. Therefore, the intrinsic cohesion of GB in high B doped alloy 2 is expected to be much crucial. As has been widely accepted, B segregation at GBs can increase the GB cohesion, tie up vacancies and reduce GB diffusion reaction, and thus prevents a premature formation of cavities and micro-cracks at the GBs.13–15 In the case of cavities formed, it has been found that B has segregated on the nucleated cavity surface during creep test,16 which decreased the diffusivity along the cavity surface and suppressed creep cavitation. It is reasonable to conclude that both the nucleation and growth of the microcracks are suppressed and constrained in high B doped alloy 2. For failure at surface, the issue is more complicated due to environment enhanced creep fracture.14 SAGBO and oxygen embrittlement are two commonly used mechanisms and a schematic diagram comparing SAGBO and oxygen embrittlement mechanisms is also provided by Ref. 14. As no micro-crack is identified in the advancing front of the intergranular oxide in high B doped alloy 2 as shown in Fig. 12b and c, the primary failure mechanism is believed to be the SAGBO. During notch rupture test, the stress and strain concentration at the notch throat determines a high risk of damage and failure there.20,21 In low B doped alloy 1, the intergranular cohesion is inadequate and the formation of microvoid is much easier. As the service time prolongs, the oxidation deteriorates the intergranular

Yu et al.

Effect of B on low thermal expansion superalloy

13 Cross-sectional microstructure of a low B doped alloy 1 and b high B doped alloy 2 oxidised at 650uC for 200 h

14 Microstructures of notch ruptured specimen of a alloy 1 and b alloy 2: rolling direction is indicated by a white arrow and cracks are marked by black arrows

cohesion in the vicinity near the notch root, and the enhanced stress concentration leads to the formation of intergranular cracks at the notch root. The formation of intergranular cracks further increases the net stress concentration at the diametral plane at the notch throat, which leads to the quick failure of the notched specimen of low B doped alloy 1 and few cracks can be found in other planes near the fracture surface as shown in Fig. 14a. Although SAGBO may be enhanced at the notch root surface due to the stress concentration, increasing B markedly increases the intergranular cohesion, which retards both the formation and the propagation of intergranular microcracks in case of stress concentration at the notch throat. Therefore, the specimen of high B doped alloy 2 is able to hold without cracks for a longer time. Logically, holding for such a long time under a high stress will lead to creep. Consequently, the stress concentration is redistributed at planes other than the diametral plane near the notch throat, so that many cracks are also found in planes near the fracture surface in high B doped alloy 2 (Fig. 14b). It can be concluded that the improvement in GB cohesion and inhibition of the formation of microcracks due to GB segregation of B substantially contribute to the elimination of notch sensitivity in high B doped alloy 2.

Conclusions 1. Compared with conventional Thermo-Span alloy, the new B and Zr containing Al–Cr modified alloy shows much lower CTE, comparable oxidation resistance and improved notch sensitivity. 2. Increasing B content promotes the formation of B–Nb enriched boride phase, but constrains the formation of Laves phase and the GB precipitation of Laves and G phases. 3. Owing to the GB segregation of B, both the nucleation and propagation of intergranular cracks

have been controlled, and the enhancing intergranular cohesion leads to the improvement in notch sensitivity.

References 1. K. Sato and T. Ohno: in: ‘Superalloys 1992’, (ed. S. D. Antolovich et al.), 247–255; 1992, Pennsylvania, TMS. 2. B. Deng, G. W. Han and D. Feng: J. Aeronaut. Mater., 2003, 23, 244–249. 3. E. A. Wanner and D. A. Daniel: in: ‘Superalloys 1992’, (ed. S. D. Antolovich et al.), 237–246; 1992, Pennsylvania, TMS. 4. K. A. Heck, D. F. Smith, M. A. Holderby and J. S. Smith: in: ‘Superalloys 1992’, (ed. S. D. Antolovich et al.), 217–266; 1992, Pennsylvania, TMS. 5. W. Sha: ‘Steels: from materials science to structural engineering’, 1st edn, Chap. 6, 141–161; 2013, London, Springer. 6. Y. R. Sun: ‘Study of a modified new low expansion Thermo-Span superalloy’, PhD thesis, Shenyang, 2007. 7. Y. R. Sun, W. R. Sun, X. F. Sun, S. R. Guo, Z. Liu and Z. Q. Hu: Rare Metal Mater. Eng., 2007, 36, 1788–1792. 8. D. F. Smith, E. F. Clatworthy, D. G. Tipton and W. L. Mankins: in: ‘Superalloys 1980’, (ed. J. T. Tien et al.), 521–530; 1980, Ohio, TMS. 9. T. H. Chuang: Mater. Sci. Eng. A, 1992, A141, 169–178. 10. R. T. Holt and W. Wallace: Int. Met. Rev., 1976, 21, 1–24. 11. R. H. Bricknell and D. A. Woodford: Metall. Trans. A, 1981, 12A, 1673–1680. 12. J. Kameda: Acta Metall. Mater., 1993, 41, 517–525. 13. E. P. George, R. L. Kennedy and D. P. Pope: Phys. Stat. Sol. (A), 1998, 167, 313–333. 14. D. A. Woodford: Energ. Mater., 2006, 1, 59–79. 15. L. Xiao, D. L. Chen and M. C. Chaturvedi: Mater. Sci. Eng. A, 2006, A428, 1–11. 16. K. Laha, J. Kyono, S. Kishimoto and N. Shinya: Scr. Mater., 2005, 52, 675–678. 17. P. J. Zhou, J. J. Yu. X. F. Sun, H. R. Guan and Z. Q. Hu: Mater. Sci. Eng. A, 2008, A491, 159–163. 18. W*. Chen, M. C. Chaturvedi, N. L. Richards and G. McMahon: Metall. Mater. Trans. A, 1998, 29A, 1947–1954. 19. S. J. Sijbrandij, M. K. Miller, J. A. Horton and W. D. Cao: Mater. Sci. Eng. A, 1998, A250, 115–119. 20. D. R. Hayhurst, J. Lin and R. J. Hayhurst: Int. J. Solids. Struct., 2008, 45, 2233–2250. 21. J. B. Pu, G. Z. Wang, X. L. You and J. H. Chen: J. Lanzhou Univ. Technol., 2005, 31, 9–13.

Materials Science and Technology

2013

VOL

29

NO

12

1477