Evolution of Ni-based superalloys for single crystal gas turbine blade applications

Evolution of Ni-based superalloys for single crystal gas turbine blade applications

Aerosp. Sci. Technol. 3 (1999) 513–523  1999 Éditions scientifiques et médicales Elsevier SAS. All rights reserved S1270-9638(99)00108-X/FLA Evoluti...

2MB Sizes 0 Downloads 13 Views

Aerosp. Sci. Technol. 3 (1999) 513–523  1999 Éditions scientifiques et médicales Elsevier SAS. All rights reserved S1270-9638(99)00108-X/FLA

Evolution of Ni-based superalloys for single crystal gas turbine blade applications Pierre Caron a, *, Tasadduq Khan b a ONERA, Metallic Materials and Processes Department, BP72, 92322 Châtillon cedex, France b ONERA, General Scientific Directorate BP72, 92322 Châtillon cedex, France

Received 19 July 1999; accepted 4 October 1999

Abstract

The chemistry of the Ni-based superalloys designed for single crystal gas turbine blades has significantly evolved since the development of the first generation of alloys derived from columnar grained materials. The overall performance of the second and third generations has been significantly improved by the addition of increasing amounts of rhenium. However, the problems of increased density, grain defects and microstructural stability have also become more and more acute and render necessary to carefully control the level of the various alloying elements in order to effectively benefit from the high potential of the most recently developed third generation alloys.  1999 Éditions scientifiques et médicales Elsevier SAS superalloy / single crystal / rhenium

Résumé

Evolution des superalliages à base de nickel destinés aux aubes monocristallines de turbine à gaz. La chimie des superalliages à base de nickel destinés aux aubes monocristallines de turbine à gaz a évolué de manière significative depuis le développement des alliages de première génération dérivés des matériaux à grains colonnaires. La performance d’ensemble des seconde et troisième générations d’alliages a été grandement améliorée par l’addition de quantités croissantes de rhénium. Cependant, les problèmes de masse volumique accrue, de grains parasites et de stabilité microstructurale sont aussi devenus de plus en plus aigus et rendent nécessaire le contrôle soigné des concentrations des différents éléments d’alliage afin de pouvoir effectivement bénéficier du potentiel élevé des alliages de troisième génération les plus récents.  1999 Éditions scientifiques et médicales Elsevier SAS superalliage / monocristal / rhénium

1. Introduction Turbine blades are critical components in both aeronautical and stationary gas turbines. The engine performance is closely related to the capability of materials to withstand higher and higher temperatures. Over the past 25 years, the turbine blade temperature capability has increased significantly. Some of the advances have been * Correspondence and reprints

achieved through improved alloy compositions, others have been accomplished by major innovation in processing, such as directional solidification or single crystal technology of nickel-based superalloys. Although the nickel based superalloys are certainly approaching their temperature asymptote, there is still room left for some further development in single crystal superalloys, the socalled third generation alloys.

514

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

mainly as solid solution strengthening elements. Cr also plays an essential role in the hot corrosion resistance since it promotes the formation of a protective Cr2 O3 oxide scale. These alloys contain a high volume fraction of strengthening ordered Ni3 Al-based γ 0 phase particles homogeneously distributed in the γ matrix as nearcubical precipitates (figure 1). The elements Ti, Ta, Nb and V strengthen the γ 0 precipitates by substituting to Al in Ni3 Al. Al also plays a fundamental role in promoting the formation of a stable Al2 O3 alumina surface scale which protects the alloy against further oxidation. The superalloys for single crystal blades do not generally contain voluntary additions of minor elements such as C, B, Zr or Hf used as grain boundary strengthening elements in conventionally cast or columnar grained superalloys. As an example, the alloy NASAIR 100 was

About twenty years ago, the chemistries of the first superalloys suited for single crystal casting derived from that of well-known nickel-based superalloys initially designed for conventional casting, such as Mar-M200. The major chemistry modification as compared to the polycrystalline materials was the suppression of the grain boundary strengthening elements C, B, Zr and Hf. The alloys’ designers then succeeded in optimising the mechanical properties of these alloys, especially the creep resistance, by introducing large amounts of refractory alloying elements such as W, Ta, Mo. A major step was the introduction of Re at a level up to 6 wt.%, but at the expense of the density, castability, cost and microstructural stability. The most recent developments included the addition of more exotic elements such as ruthenium and iridium. In some alloys, minor elements such as carbon and boron were recently re-introduced in order to render the material more tolerant to problems related to large angle sub-grain boundaries with the objective of reducing the number of rejected blades. It is the purpose of this paper to describe the evolution of the chemistry of this class of alloys by focusing on the advantages and drawbacks resulting from these chemistry changes. 2. Evolution of the chemistry of the single crystal superalloys 2.1. First generation alloys As detailed in table I the alloying elements of the first generation Ni-based single crystal superalloys are mainly Cr, Co, Mo, W, Al, Ti, Ta and sometimes Nb or V. The effects of these alloying elements on the properties of superalloys have been extensively studied during the last forty years and are currently well-known [32]. Cr, Co and Mo partition preferentially to the austenitic facecentred cubic nickel-based γ matrix where they act

Figure 1. Two-phase γ -γ 0 microstructure in fully heat-treated AM3 first generation single crystal superalloy.

Table I. Chemical compositions (wt.%) of first generation Ni-based superalloys for single crystal blades. Alloy

Cr

Co

Mo

W

Al

Ti

Ta

Nb

V

Hf Density (g.cm−3 ) Country

Ref.

Nasair 100

9



1

10.5

5.75

1.2

3.3







8.54

USA

[37]

CMSX-2

8

4.6

0.6

8

5.6

1

6







8.60

USA

[21]

CMSX-3

8

4.6

0.6

8

5.6

1

6



0.1

8.60

USA

[21]

CMSX-6

9.8

5

3



4.8

4.7

2



0.1

7.98

USA

[41]

PWA 1480

10

5



4

5

1.5

12







8.70

USA

[20]

SRR 99

8

5



10

5.5

2.2

3







8.56

GB

[18]

RR 2000

10

15

3



5.5

4





1



7.87

GB

[18]

René N4

9

8

2

6

3.7

4.2

4

0.5





8.56

USA

[33,42]

AM1

7.8

6.5

2

5.7

5.2

1.1

7.9







8.60

F

[12]

AM3

8

5.5

2.25

5

6

2

3.5







8.25

F

[24]

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

derived from Mar-M247 essentially by suppressing these grain boundary strengthening elements [37]. The resulting increase of the incipient melting temperature from 1240 to 1330◦ C allowed a complete solutioning of the secondary γ 0 precipitates and an almost complete elimination of coarse primary interdendritic γ 0 particles by using a super-solvus high temperature solution heat treatment. The resulting increase in creep strength was shown to be equivalent to a 58◦ C temperature advantage at high temperatures and low stresses. Several alloy compositions were then developed by engine manufacturers, alloy makers or research institutes in order to improve the performance of the single crystal turbine blades (table I). The main goal was to improve the creep strength which is an important damage process for these components, together with the thermo-mechanical fatigue. Improvement in mechanical properties was therefore obtained by a careful balance between the various alloying elements. The main differences between the alloys concern the respective levels of refractory elements W, Mo and Ta which are known to reduce the bulk diffusion rate that slows down the coarsening kinetics of the strengthening γ 0 precipitates and the diffusion controlled creep mechanisms such as climb or cross-slip of dislocations. These elements also induced strong solid solution hardening effects due to their large atomic radii compared to that of Ni or Al. The CMSX-2 and CMSX-3 single crystal (SC) alloys were derived from Mar-M247 [21], whereas the low density SC superalloy CMSX6 was derived from the alloy IN6212 [41]. The SC alloys PWA1480 [20] and René N4 [33,42] were developed respectively by Pratt & Whitney Aircraft and General Electric; the alloys SRR99

(a)

515

and RR2000 were developed by Rolls Royce in United Kingdom [18]. The SC alloy AM1, jointly developed by ONERA, SNECMA, ARMINES and TECPHY [12], was selected by SNECMA as a blade and vane material for its M88 engine for the RAFALE fighter. The alloy AM3 [24], developed by ONERA, is now used as a SC blade material in the ARRIEL 2 and ARRIUS 2 TURBOMECA engines which equip a number of helicopters. As demonstrated with NASAIR 100, one great advantage of the first generation SC superalloys compared to equiaxed conventionally cast (CC) or columnar grained directionally solidified (DS) alloys was the possibility to achieve a homogeneous microstructure using a single high temperature heat treatment which allowed the elimination of almost all the γ /γ 0 eutectic interdendritic nodules and solutioning of all the secondary γ 0 precipitates. This solution heat treatment has to be performed within the so-called “heat treatment window” (interval between the incipient melting temperature and the solvus temperature of the coarse interdendritic secondary γ 0 precipitates). Such an example is provided by the alloy AM3 in which the γ 0 phase can be totally solutioned by using a simple treatment for 3 hours at 1300◦C (figure 2). Except for PWA1480 [20] which contains a very high level of Ta and consequently large amounts of γ /γ 0 eutectic phase, the first generation superalloys are therefore easy to homogenise, that allows us to obtain an optimised distribution of fine γ 0 precipitates by applying adequate subsequent ageing heat treatments. The ageing heat treatments performed on single crystal superalloy blades correspond generally to the diffusion coating treatment cycles. The pioneering work on CMSX-2 has shown that the creep behaviour of high γ 0 volume fraction alloys can be optimised by choosing a precipitation heat treatment leading

(b)

Figure 2. Transverse dendritic microstructure of AM3 single crystals; (a) as-cast; (b) heat-treated for 3 hours at 1300◦ C, air cooled.

516

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

to cubical γ 0 precipitates with a mean cube edge of approximately 0.45 µm [5]. Since that time alloy designers have attempted to balance these diffusion treatments in order to achieve optimised γ 0 size for creep strength.

may sometimes induce deleterious effects on the mechanical properties, such as loss of ductility, decrease of the impact strength or sometimes decrease of the creep strength as demonstrated by Dreshfield and Ashbrook in IN100 [13]. It is generally recognised that superalloys which contain high levels of refractory elements such as Re, Mo or W are more or less prone to the precipitation of TCP phases. Re-rich TCP phases have been thus reported to precipitate in CMSX-4 [23], whereas W-rich rhombohedral µ phase precipitates within the MC2 alloy under certain conditions [30]. A study performed on a series of experimental Re-bearing alloy showed that three types of TCP phases can coexist in the same alloy, rhombohedral µ, tetragonal σ and orthorhombic P, with similar compositions, all rich in Re and Cr [11]. It was however demonstrated that a small amount of µ phase in MC2 superalloy does not affect its mechanical behaviour [30], and it has been reported that the Re-rich TCP precipitates present in CMSX-4 did not detrimentally affect creep rupture properties [23]. On the other hand, a study performed on a series of experimental SC superalloys with various amounts of Re showed a drastic reduction in rupture life when the amount of Re-rich σ phase precipitating in these alloys increased [14]. This deleterious effect was attributed mostly to the depletion of the γ matrix of refractory strengthening elements. As Re partitions preferentially to the γ matrix, the levels of other γ -former elements Cr, Co, Mo and W must be reduced and carefully balanced in order to avoid the supersaturation of the solid solution, which, in turn, will lead to the precipitation of Re-rich TCP phase particles, while maintaining a good balance between the mechanical and the environmental properties. Thus, as the typical level of Cr encountered within the first generation SC alloys was around 8 wt.%, it decreased to 5–7 wt.% in the second generation alloys. It was shown that this Cr content is sufficient to ensure an acceptable level of corrosion resistance and the matrix can keep into solution high amounts of Re, W and Mo, necessary for an optimised creep strength. The role of the γ 0 -forming elements Al, Ti, Ta, Nb on the phase stability is also important as their total amount determines the quantity of γ 0 phase which is

2.2. Second generation alloys Several alloy designers showed that a significant improvement of the creep strength of the single crystal superalloys may be obtained by the addition of rhenium at the expense of other refractory elements such as Mo or W. Thus, a study carried out on modified Mar-M200 SC alloys showed that additions of rhenium substantially lower the γ 0 coarsening kinetics and result in large negative γ -γ 0 misfits [19]. Atom-probe studies performed on CMSX-2 and PWA-1480 alloys modified by additions of rhenium also showed the existence of Re atom clusters within the γ matrix of these alloys, which is a more potent source of strengthening than the conventional solid solution effect [3,4]. The introduction of 3 wt.% Re to SC nickel based superalloys resulted in a temperature capability improvement of about 30◦ C [15]. The first rhenium-containing alloys were labelled as second generation SC superalloys. Typical examples of this class of alloys are PWA1484 [10], René N5 [43], SC180 [44], CMSX-4 [22] and SMP14 [40]. The chemical compositions of these alloys are reported in table II. However, a rational criterion for the definition of these second generation alloys should be based on the high temperature creep strength. In these conditions, the MC2 superalloy developed by ONERA is now considered as an integral part of this group, as it shows a high temperature creep strength comparable to that of second generation Re-bearing alloys, in spite of the fact that it does not contain rhenium [6]. It is noteworthy that the absence of rhenium makes this alloy much cheaper and lighter than the other second generation SC alloys. A major issue of concern in developing second generation alloys, and more specifically Re-bearing alloys, was their microstructural stability, i.e. the propensity to form undesirable topologically close-packed (TCP) brittle phases as σ , µ or P phases during exposure at high temperature. The presence of TCP phases in superalloys

Table II. Chemical compositions (wt.%) of second generation Ni-based superalloys for single crystal blades. Alloy

Cr

Co

Mo

Re

W

Al

Ti

Ta

Nb

Hf

Density (g.cm−3 )

Country

Ref.

CMSX-4

6.5

9

0.6

3

6

5.6

1

6.5



0.1

8.70

USA

[22]

PWA 1484

5

10

2

3

6

5.6



8.7



0.1

8.95

USA

[10]

René N5

7

8

2

3

5

6.2



7



0.2

8.70

USA

[43]

SC180

5

10

2

3

5

5.2

1

8.5



0.1

8.84

USA

[44]

SMP14

4.8

8.1

1

3.9

7.6

5.4



7.2

1.4



9.02

RSA

[40]

8

5

2



8

5

1.5

6





8.63

F

[6]

MC2

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

close to 70 vol.% in the SC superalloys. An excessive amount of γ 0 phase renders the matrix more prone to the precipitation of TCP phases due to the higher concentration of elements which partition preferentially to it. Methods have been developed to predict the propensity of superalloys to the precipitation of TCP phases. The oldest one is the electron vacancy (Nv ) or PHACOMP method [35]. The principle is to calculate an average electron vacancy number Nv for the alloy matrix, using specific Nv values for the transition elements, and to compare this value to a critical one above which the TCP phase is known to precipitate. It should be noted that the critical Nv value varies with the nature of TCP phase and the cast alloys, due to the dendritic segregations, have lower critical Nv values than the wrought alloys. This is of prime importance in Re-bearing alloys because this element segregates strongly to the dendrite cores and is difficult to homogenise due to its low bulk diffusivity. A more recent predictive method, known as the New Phacomp method, has been developed by Morinaga et al. [27]. A parameter Md is calculated which is an average energy level of d orbitals of alloying transition elements in the γ matrix. As in the case of PHACOMP method, an alloy will be prone to the precipitation of TCP phases when the Md value becomes larger than a critical value which depends on the phase type. The level of alloying elements must also be balanced considering other properties which can be of prime importance for the application. The environmental properties therefore must not be neglected even if the turbine blades are systematically coated to protect them against oxidising and corrosive combustion gas. Indeed, these protective coatings, aluminides or MCrAlY types, may crack under thermo-mechanical stresses or could be eliminated by erosion. The remaining life of the component will then depend on the intrinsic resistance of the alloy to high temperature oxidation and corrosion attack. Thus, Mo is generally kept at a low level, because this element is known to have a negative influence on the corrosion resistance of the nickel based superalloys, while alloy designers prefer to increase the level of Ta which is more beneficial for environmental properties. It is also generally considered that Re contributes positively to the hot corrosion and oxidation resistance partly due to its effect on the diffusion of reactive elements. The castability of single crystal components made of Re-containing superalloys must be also assessed carefully because it was demonstrated that it can be influenced negatively by excessive additions of Re and W. Thus, the occurrence of small chains of equiaxed grains referred to as freckles was observed in experimental SC superalloys containing high levels of Re and W [31]. These freckles form due to convective instabilities resulting from solute partitioning that lowers the density of the liquid in the mushy zone. The dendrite cores are enriched in Re and W and the freckles form from the low den-

517

sity interdendritic liquid enriched in Al, Ta and Ti. Some isolated spurious grains can also form during directional solidification and the potential for nucleation and growth of these grain defects was also demonstrated to increase with the levels of Re and W. The nucleation of spurious grains is due to the fragmentation of secondary and tertiary dendrite arms under the action of the interdendritic fluid flow generated by density differences. This tendency can however be counterbalanced by increasing the level of Ta which is rejected to the interdendritic areas, therefore decreasing the density inversions. The alloys which were most resistant to freckling were therefore high in Ta and low in Re and W [31]. An other drawback due to Re addition is that it increases the density as compared to the first generation alloys, which is a drawback for the rotating parts of aeronautical engines. The density of these alloys are in the range 8.7–9 g.cm−3 whereas the density of MC2 is only 8.6 g.cm−3 . 2.3. Third generation alloys More recently, alloy designers tried to improve again the high temperature capability of the SC blade alloys by increasing the content of rhenium up to about 6 wt.%. The challenge was to achieve improved creep strength, without increasing the density and by keeping the alloy not too much prone to the precipitation TCP phases. Two typical third generation alloys are CMSX-10 developed by Cannon-Muskegon [17] and René N6 developed by General Electric [39]. More recent development work conducted by GE was devoted to third generation SC alloys containing also some additions of ruthenium [28]. A new generation of SC alloys, a typical example of which is the MC-NG alloy, is developed in France by ONERA [8] and the alloys TMS75 and TMS80 were developed in Japan [25]. The chemical compositions of some of the third generation alloys are reported in table III. The tendency to the precipitation of TCP phases in the third generation alloys is a still more important problem than in the second generation alloys since it is difficult to attain the right balance between the alloying elements promoting this instability. Once again, the content of Cr was decreased as compared to the second generation alloys in order to keep the alloys less prone to TCP phase precipitation. The typical level of Cr encountered within the third generation alloys is now between 2 to 4.2 wt.%. The role of Co on the precipitation of TCP phases is still very controversial. Whereas Erickson limited the Co level at 3 wt.% in CMSX-10 claiming that it reduces the tendency to form TCP phases [17], Walston et al. recommended a high level of Co, 12.5 wt.%, in René N6 in order to improve phase stability [39]. Re-rich TCP phase particles have however been reported to precipitate in CMSX-10 after high temperature exposure, the greatest propensity for phase instability oc-

518

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

Table III. Chemical compositions (wt.%) of third generation Ni-based superalloys for single crystal blades. Alloy

Cr

Co

Mo

Re

W

Al

Ti

Ta

Nb

Hf

Others

Density (g.cm−3 )

Country

Ref.

CMSX-10

2

3

0.4

6

5

5.7

0.2

8

0.1

0.03



9.05

USA

[17]

4.2

12.5

1.4

5.4

6

5.75



7.2



0.15

0.05C

8.97

USA

[39]

8.91

USA

[2]

René N6

0.004B 0.01Y Alloy 5A

4.5

12.5



6.25

5.75

6.25



7



0.15

0.05C 0.004B 0.01Y

TMS-75

3

12

2

5

6

6



6



0.1





J

[25]

TMS-80

2.9

11.6

1.9

4.9

5.8

5.8



5.8



0.1

3 Ir



J

[25]

8.75

F

[8]

MC-NG

Patent pending

Figure 3. Interdendritic carbides and dendrite core Re-rich TCP phase precipitates within a René N6 single crystal fully heattreated and aged for 200 hours at 1050◦ C.

curring within the temperature range 1090–1150◦C [17, 39]. A reduction of creep life associated with the precipitation of these Re-rich TCP phase particles was recognised in CMSX-10 for long exposures between 1100 and 1160◦ C [17]. Re-rich phase precipitation has also been reported in René N6 [39] but it did not detrimentally affect the creep rupture properties of this alloy. The precipitation of Re-rich TCP phases is illustrated in figure 3 showing the microstructure within a dendrite core of a René N6 SC sample melted and cast at ONERA and exposed for 200 hours at 1050◦C. However, a new type of instability was observed by Walston et al. [38] in superalloys containing high levels of refractory elements and evaluated during the development phase of René N6. A typical alloy used in this study was Alloy 5A [2] (see table III). This instability, termed secondary reaction zone (SRZ), was observed be-

neath the diffusion zone of aluminide coatings, in dendrite cores and along some low angle boundaries (LAB’s) away from the coating. These so-called LAB’s delimit sub-grains containing groups of dendrites and with relative misorientations up to 15◦ . These SRZs are described as areas with a γ 0 matrix containing γ and P phase needles and are referred to as cellular colonies when present in dendrite cores and sub-grain boundaries. The nucleation and growth of the SRZs is thought to be controlled by local chemical supersaturations resulting from the presence of a coating or from internal segregation (to dendrite core or LBS’s segregation) and by strain energy introduced by surface preparation prior coating, or by misfit strains along grain boundaries or between γ and γ 0 . Rhenium has been demonstrated to be the most potent element for determining the alloy propensity to form SRZ [38]. The main concern was thus to evaluate the eventual deleterious effects resulting from the presence of SRZ beneath the coating or away from the surface. Walston et al. demonstrated that SRZ beneath the coating can eventually affect the rupture strength due to the reduction of cross section [38]. The cellular colonies formed at the low angle grain boundaries may also induce some reduction of the rupture properties for large misorientations. However, the most detrimental effect was attributed to the cellular colonies in the dendrite cores. Some cracks forming at the SRZ interfaces indeed induce premature failure leading to a life reduction of up to 70%. The development of René N6 was thus based upon the work performed on the SRZ containing alloys, with the primary goal to obtain a microstructure stable with respect to sub-coating SRZ and dendritic cellular colonies. This goal was achieved essentially by decreasing the level of Re and by introducing Mo in order to keep a comparable total amount of strengthening refractory elements. Walston et al. pointed out the difference existing between these SRZs and the well known TCP phases [39]. They

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

suggest that the precipitation of TCP phases assisted by excess concentrations in Cr and Mo will deplete the matrix from these and then reduce the chemical driving force for SRZ. They report the formation of SRZs in CMSX10 beneath a PtAl coating after exposure for 400 hours at 982◦ C, but only TCP phase precipitation in failed creep rupture bare specimen of CMSX-10 [39]. It is noteworthy that the third generation SC superalloys containing some additions of ruthenium [8,28] or Ir [25] seem to show a reduced tendency to form undesirable TCP phases. This beneficial effect is presumably due to the modification of the partitioning ratios of the elements between the γ and γ 0 phase, that will decrease the concentration of Re in the γ matrix and thereby rendering it less prone to the precipitation of Re-rich TCP phases. These alloys exhibit excellent creep strength, especially at high temperatures, but complementary investigations are needed to evaluate and understand the influence of Ru and Ir on other properties. A significant difference in the design of the third generation alloys concerns the levels of minor elements such as C, B, Hf and Y. Whereas C and B are not intentionally added within the alloys CMSX-10, TMS-75 and -80, Walston et al. [39] choose to reintroduce these elements in René N6 in order to improve the castability and the tolerance to low angle grain boundaries. The level of carbon was thus set at 0.05 wt.% in René N6 for a cleaner melting because it helps in reducing the oxides, thus improving the castability. In order to minimise the possible deleterious effects of LABs which are inevitably present in the “single crystal” components, low amounts of C, B and Hf are also added to René N6 in order to strengthen the LABs. These additions improve the yield of the components, because LABs with misorientations as large as 12◦ are accepted, instead of the value of 6◦ generally fixed for rejection of the castings when the grain boundary strengthening elements are omitted [2]. Yttrium is also added in N6 to improve the adherence of the Al2 O3 protective layer formed at high temperature whereas Hf is very often added in the SC superalloys in order to improve their coatability. Indeed, it was clearly demonstrated that small additions of reactive elements such as yttrium, hafnium, cerium and lanthanum improve the adherence of alumina scales on the superalloys [36]. One of the mechanisms proposed to explain this beneficial effect is that these reactive elements tie up sulphur in the alloy and prevent its segregation to the alloy/oxide interface [34]. Residual content of yttrium is however difficult to control within the single crystal component due to reactions between the molten alloy and foundry ceramics and yttrium volatilisation during melting and directional solidification [1]. Moreover, the segregation of yttrium during the directional solidification may promote the formation of low melting point phases in the interdendritic areas which will in turn reduce the temperature of incipient melting as reported for the Y-enriched

519

Figure 4. Cyclic corrosion behaviour at 850◦ C of CMSX-10, MC-NG and René N6 alloys.

version of CMSX-4 [26]. On the other hand, work performed at ONERA showed that simultaneous additions of Si and Hf dramatically improve the cyclic oxidation in air at 1100◦ C of the AM1, AM3 and MC2 single crystal alloys at a level comparable to that obtained with Y [7]. These additions can be a promising alternative to Y, because additions of Si and Hf are easy to control within the cast components. It must be pointed out that the hot corrosion resistance will be difficult to maintain for the low-chromium third generation SC superalloys because Cr plays a key role in promoting the formation of a protective chromium rich oxide scale at the temperatures where hot corrosion is active. On the basis of isothermal burner rig corrosion tests performed at 950◦ C with 2 ppm salt ingestion, Erickson reported comparable corrosion resistance for CMSX10 and CMSX-4 at least up to 100 hours [17]. However, these corrosion test conditions are not very severe as compared to the real situation in the turbine where the blades are subjected to cyclic thermal stresses and where hot corrosion damaging may occur at temperatures lower than 950◦ C. Some cyclic corrosion tests were thus performed at ONERA in air at 850◦ C with one-hour cycles and additions of Na2 SO4 renewed every 50 hours, on bare samples of René N6, CMSX-10 and MC-NG alloys. The salt level deposited on the samples corresponds to that typically found in a burner rig using kerosene fuel containing 0.15% sulphur and a 5 ppm NaCl contamination. The specific weight changes are plotted as a function of number of cycles in figure 4. The CMSX-10 alloy which contains only 2 wt.% Cr exhibits a very poor hot corrosion resistance as compared to René N6 which contains 4.2 wt.% Cr, the MC-NG alloy exhibiting an intermediate behaviour. The strong difference in corrosion resistance between CMSX-10 and René N6 is perhaps not entirely due to the Cr content variation, but these tests show that the low content of Cr in some third generation SC alloys may not be sufficient to obtain a satisfactory hot corrosion strength in severe environment. Hot corrosion dip tests using Na2 SO4 -25% salt mixture performed at 900◦ C demonstrated that alloys TMS-75 and -80, con-

520

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

taining only 3 wt.% Cr, exhibit a corrosion resistance slightly better than for CMSX-4 [25]. This result has however to be confirmed by testing under conditions prevailing in aeroengines. The problem of the formation of grain defects, freckles and spurious grains during directional solidification of single crystal components, mentioned in the case of secondary generation alloys, becomes more acute for the third generation alloys due to the higher content of Re. It renders in particular more difficult the growth of large blades for industrial gas turbines [29]. One of the characteristics of the third generation SC alloys is that their high total amount of alloying elements, in particular Ta and W, promotes the formation of a large volume fraction of γ /γ 0 eutectic. A typical example is shown in figure 5 illustrating the as-cast microstructure of a single crystal of the CMSX-10 alloy melted and cast at ONERA. Experience shows that it is impossible to eliminate these eutectic phases by using a single isothermal solution heat treatment as in the case of first generation SC alloys. It is necessary to apply a complex heat treatment cycle generally including a pre-homogenisation heat treatment aiming at increasing the incipient melting temperature, before performing a final solution treatment which eliminates the major part or all of the eutectic phases. Complete elimination of these γ /γ 0 nodules has been thus obtained in CMSX-10 by using the following heat treatment procedure: 1337◦C/3h, ramp up at 3◦ C.h−1 , 1367◦ C/3h/Air cooled (figure 5b). To attain this result, Erickson reported durations of solution heat treatments of 30–35 h, with a peak soak temperature of about 1366◦ C [15]. The choice to reintroduce minor elements such as C, B and Y which lower the incipient melting temperature in N6 render more difficult the elimination of γ /γ 0 eutectic

(a)

and the solutioning of coarse γ 0 precipitates. This difficulty was circumvented by accepting some residual eutectic and incipient melting after solutioning the alloy. In these conditions, a 20◦ C temperature window is claimed for acceptable solution heat treatment, the optimum being at temperatures in the range 1315–1335◦C for approximately six hours [39]. The size and amount of such eutectic phases must however be carefully controlled because they can act as fracture initiation sites and thereby affect the tensile ductility or the endurance limit in fatigue. It is also noteworthy that the voluntary addition of carbon led to the formation of interdendritic blocky (Ti, Ta)C carbides as shown in figure 3 illustrating the microstructure of a René N6 single crystal cast at ONERA, fully heattreated and finally aged for 200 hours at 1050◦C. As for the residual eutectic nodules, these carbides may penalise the ductility and the fatigue properties by promoting early rupture. As Re and W segregate strongly to the dendrites during the directional solidification, the benefit of a long solution heat treatment is to improve the chemical homogeneity of the alloy and hence to prevent phase instabilities within the dendrite cores. This beneficial effect has been demonstrated in CMSX-10 and René N6 [15,39]. The duration of the solution treatment must however be optimised, by considering also the cost of this heat treatment. The high level of refractory elements such as W, Re and Ta can also be a drawback for the third generation alloys, as they induce an increase of the density up to values around 9 g.cm−3 (table III). Moreover, high levels of Re, Ta and eventually Ru and Ir, which are expensive metals, increase the cost of the alloy, which could be a drawback for cost effective applications in civil engines.

(b)

Figure 5. Transverse dendritic microstructure of a CMSX-10 single crystal; (a) as-cast; (b) solution heat-treated.

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

521

Figure 6. Typical temperature advantages over CC superalloys obtained with DS and SC superalloys estimated from stress rupture tests performed at 982◦ C and 248 MPa (from reference [16]).

3. Evolution of the mechanical properties of the single crystal superalloys The constant creep strength improvements resulting from the transition from CC superalloys, to DS materials, including second generation DS alloys with 3 wt.% Re, and then to the three generations of SC alloys are illustrated in figure 6 where the temperature advantages are reported for stress rupture tests performed at 982◦ C and 248 MPa. The major objective of the alloy designers in developing third generation SC superalloys was to improve significantly the stress rupture strength of this class of alloys compared to previous first and second generation alloys. Data on René N5, René N6 [39] and CMSX-10 [9, 25] alloys are compared to data produced at ONERA on MC-NG alloy in figure 7 where the specific stress-rupture strengths are reported in a Larson Miller diagram. The third generation SC alloys exhibit comparable stress rupture strengths significantly higher than for the second generation alloy René N5. Thus, Walston et al. [39] claim for a 30◦ C benefit obtained with René N6 as compared to René N5. It is noteworthy that the MC-NG alloy shows a very high creep resistance with a density significantly lower than those of CMSX-10 and René N6, which is a great advantage because it helps to reduce the stresses on the disk, and the weight of the rotor. The temperature advantage of CMSX-10 over CMSX4 appears to be equal to 36◦ C in the temperature range 913–1010◦C. However, the CMSX-10 creep strength benefit at 982◦ C as compared to CMSX-4 decreases with increasing exposure times [17], which suggests that there may be a point beyond which the third generation alloys would not really be better than the second generation alloys. This analysis was confirmed by Ross and O’Hara [33] who compared the stress rupture life at

Figure 7. Comparative Larson–Miller stress-rupture curves for second and third generation SC superalloys.

982◦ C of different generations of SC superalloys. It appears that the first generation SC René N4 surpasses the second generation alloys in the 7000–10000 hour life regime at this temperature and data extrapolation indicates that this alloy could attain the level of the third generation alloys for rupture life around 20000 hours. On the other hand, some comparative creep tests performed at ONERA on various single crystal alloys demonstrate that a Re-free superalloy such as MC2 exhibits stress rupture lives at 1050◦C comparable to that of superalloys containing Re contents close to 6 wt.%. A great advantage of the third generation SC superalloys is that they maintain a rather high creep resistance at temperature above 1100◦ C. Indeed, the stress rupture life at 1150◦C and 100 MPa of the MC-NG alloy is over 150 hours, whereas the first generation alloys show, typically, a rupture life less than 10 hours. Contrary to the creep strength, tensile properties of the single crystal superalloys are not reported to be

522

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

very sensitive to their chemical compositions. The tensile behaviour of the third generation alloys is therefore comparable to that of first and second generation alloys with a peak strength at about 760◦ C and a good ductility within the temperature range of interest [17]. However, some significant improvements in low cycle fatigue (LCF) and high cycle fatigue performance have been reported both for CMSX-10 [16] and René N6 [39] alloys as compared to their predecessors containing 3 wt.% Re. The significant increase of LCF properties observed at temperatures around 950◦ C are of prime importance in the case of cooled blades or vanes. Indeed, the thin airfoil walls are subjected to huge thermal strains and stresses due to the thermal gradients occurring within the blade during the working cycles of the gas turbine engines, which lead to damaging crack initiation and growth processes. It is therefore thought that the fatigue strength benefit afforded by the third generation alloys could improve significantly the durability of the single crystal components. 4. Conclusions Significant progress in the improvement of the performance of the nickel based single crystal superalloys has been made with the optimisation of the chemical compositions. The addition of rhenium up to 6 wt.% in the third generation alloys was a key factor for the increase of the mechanical properties. The development of such alloys has however taken into account some specific features such as increase in cost and density, tendency to phase instability and grain defects, which must be carefully controlled in order to ensure that these alloys will be effectively exploitable as turbine blade materials. Recent development at ONERA of new generation single crystal superalloys with additions of both rhenium and ruthenium could pave the way for the development of improved third generation superalloys with reduced density and better phase stability compared to third generation alloys containing high levels of rhenium. Acknowledgements The authors acknowledge the French Ministry of Defence for the financial support of a part of the work presented in this paper and would like to thank J.-L. Raffestin for single crystal preparation. References [1] Aimone P.R., McCormick R.L., The effect of yttrium and sulfur on the oxidation resistance of an advanced single crystal nickel based superalloy, in: Antolovich S.D. et al. (Eds.), Superalloys 1992, The Minerals, Metals & Materials Society, Warrendale, PA, USA, 1992, pp. 817– 823.

[2] Austin C.M., Darolia R., O’Hara K.S., Ross E., General Electric Company, US Patent 5 151 249, 1992. [3] Blavette D., Caron P., Khan T., An atom probe investigation of the role of rhenium additions in improving creep resistance of Ni-base superalloys, Scripta Metall. Mater. 20 (1986) 1395–1400. [4] Blavette D., Caron P., Khan T., An atom-probe study of some fine-scale microstructural features in nickel-based single crystal superalloys, in: Reichman S. et al. (Eds.), Superalloys 1988, The Metallurgical Society, Warrendale, PA, USA, 1988, pp. 305–314. [5] Caron P., Khan T., Improvement of creep strength in a nickel-base single-crystal superalloy by heat-treatment, Mat. Sci. Eng. 61 (1983) 173–194. [6] Caron P., Khan T., Development of a new nickel based single crystal turbine blade alloy for very high temperatures, in: Exner H.E., Schumacher V. (Eds.), Advanced Materials and Processes, Vol. 1, DGM Informationsgesellschaft mbH, Oberursel, Germany, 1990, pp. 333–338. [7] Caron P., Navéos S., Khan T., Improvement of the cyclic oxidation behaviour of uncoated nickel based single crystal superalloys, in: Coutsouradis D. et al. (Eds.), Materials for Advanced Power Engineering 1994 – Part I, Kluwer Academic Publisher, Dordrecht, Holland,1994, pp. 1185–1194. [8] Caron P., Raffestin J.-L., Navéos S., ONERA, Patent pending. [9] Caron P., ONERA, unpublished work. [10] Cetel A.D., Duhl D.N., Second-generation nickel-base single crystal superalloy, in: Duhl D.N. et al. (Eds.), Superalloys 1988, The Metallurgical Society, Inc., Warrendale, PA, USA, 1988, pp. 235–244. [11] Darolia R., Lahrman D.F., Field R.D., Sisson R., Formation of topologically closed packed phases in nickel base single crystal superalloys, in: Duhl D.N. et al. (Eds.), Superalloys 1988, The Metallurgical Society, Inc., Warrendale, PA, USA, 1988, pp. 255–264. [12] Davidson J.H., Fredholm A., Khan T., Théret J.-M., French patent N◦ 2 557 598, 1983. [13] Dreshfield R.L., Ashbrook R.L., Effects of sigma-phase formation on some mechanical properties of wrought nickel-base superalloys (IN 100), NASA TN D-7654, 1974. [14] Duhl D.N., Alloy phase stability requirements in single crystal superalloys, in: Stocks G.M. (Ed.), Alloy Phase Stability and Design, Mat. Res. Soc. Symp. Proceedings, Vol. 186, Materials Research Society, 1991, pp. 389–399. [15] Erickson G.L., A new, third-generation, single-crystal, casting superalloy, J. Metals 47 (4) (1995) 36–39. [16] Erickson G.L., The development of CMSX-10, a third generation SX casting superalloy, in: Proceedings of the Second Pacific Rim International Conference on Advanced Materials and Processing (PRCIM-2), Kyongju, Korea, 18–22 June 1995. [17] Erickson G.L., The development and application of CMSX-10, in: Kissinger R.D. et al. (Eds.), Superalloys 1996, The Minerals, Metals & Materials Society, Warrendale, PA, USA, 1996, pp. 35–44. [18] Ford D.A., Arthey R.P., Development of single crystal alloys for specific engine applications, in: Gell M. et al.

P. Caron, T. Khan / Aerosp. Sci. Technol. 3 (1999) 513–523

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28] [29]

[30]

(Eds.), Superalloys 1984, The Metallurgical Society of AIME, Warrendale, PA, USA, 1984, pp. 115–124. Giamei A.F., Anton D.L., Rhenium additions to a Nibase superalloy: effects on microstructure, Metall. Trans. A 16A (1985) 1997. Gell M., Duhl D.N., Giamei A.F., The development of single crystal superalloy turbine blades, in: Tien J.K. et al. (Eds.), Superalloys 1980, American Society for Metals, Metals Park, OH, USA, 1980, pp. 205–214. Harris K., Erickson G.L., Schwer R.E., Mar M 247 derivations – CM 247 LC DS alloy, CMSX single crystal alloys, properties and performance, in: Gell M. et al. (Eds.), Superalloys 1984, The Metallurgical Society of AIME, Warrendale, PA, USA, 1984, pp. 221–230. Harris K., Erickson G.L., Schwer R.E., Frasier D.J., Whetstone J.R., Process and alloy optimization for CMSX-4 superalloy single crystal airfoils, in: Bachelet E. et al. (Eds.), High Temperature Materials for Power Engineering 1990, Part II, Kluwer Academic Publishers, Dordrecht, Holland, 1990, pp. 1281–1300. Harris K., Erickson G.L., Sikkenga, S.L., Brentnall W.D., Aurrecoechea J.M., Kubarych K.G., Develoment of the rhenium containing superalloys CMSX-4 and CM 186 LC for single crystal blade and directionally solidified blade vane applications in advanced turbine engines, in: Antolovich S.D. et al. (Eds.), Superalloys 1992, The Minerals, Metals & Materials Society, Warrendale, PA, USA, 1992, pp. 297–306. Khan T., Recent developments and potential of single crystal superalloys for advanced turbine blades, in: Betz W. et al. (Eds.), High Temperature Alloys for Gas Turbines and Other Applications 1986, D. Reidel Publishing Company, Dordrecht, Holland, 1986, pp. 21–50. Kobayashi T., Koizumi Y., Nakazawa S., Yamagata T., Harada H., Design of high rhenium containing single crystal superalloys with balanced intermediate and high temperature creep strengths, in: Strang A. et al. (Eds.), Advances in Turbine Materials, Design and Manufacturing, The Institute of Materials, London, UK, 1997, pp. 766–773. Marchionni M., Goldschmidt D., Maldini M., High temperature mechanical properties of CMSX-4 + Yttrium single-crystal nickel-base superalloy, in: Antolovich S.D. et al. (Eds.), Superalloys 1992, The Minerals, Metals & Materials Society, Warrendale, PA, USA, 1992, pp. 775– 784. Morinaga M., Yukawa N., Adachi H., Ezaki H., New phacomp and its applications to alloy design, in: Gell M. et al. (Eds.), Superalloys 1984, The Metallurgical Society of AIME, Warrendale, PA, USA, 1984, pp. 523–532. O’Hara K.S., Walston W.S., Ross E.W., Darolia R., General Electric Company, US Patent 5 482 789, 1996. Parks Jr. W.P., Hoffman P., Karnitz M.A., Wright I.G., The advanced turbine systems program in the USA, in: Lecomte-Beckers J. et al. (Eds.), Materials for Advanced Power Engineering 1998 – Part III, Forschungzentrum Jülich, Jülich, Germany, 1998, pp. 1789–1805. Pessah M., Caron P., Khan T., Effect of µ phase on the mechanical properties of a nickel-base single crystal su-

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42] [43] [44]

523

peralloy, in: Antolovich S.D. et al. (Eds.), Superalloys 1992, The Minerals, Metals & Materials Society, Warrendale, PA, USA, 1992, pp. 567–576. Pollock T.M., The growth and elevated temperature satbility of high refractory nickel-base single crystals, Mater. Sci. Eng. B32 (1995) 255–266. Ross E.W., Sims, C.T., Nickel-base superalloys, in: Sims C.T., Stoloff N., Hagel W.C. (Eds.), Superalloys II, John Wiley & Sons, New York, USA, 1987, pp. 97–133. Ross E.W., O’Hara K.S., René N4: a first generation single crystal turbine airfoil alloy with improved oxidation resistance, low angle boundary strength and superior long time rupture strength, in: Kissinger R.D. et al. (Eds.), Superalloys 1996, The Minerals, Metals & Materials Society, Warrendale, PA, USA, 1996, pp. 19–25. Sarioglu C., Stinner C., Blachere J.R., Birks N., Pettit F.S., Meier G.H., Smialek J.L., The control of sulfur content in nickel-base, single crystal superalloys and its effect on cyclic oxidation resistance, in: Kissinger R.D. et al. (Eds.), Superalloys 1996, The Minerals, Metals & Materials Society, Warrendale, PA, USA, 1996, pp. 71– 80. Sims, C.T., Prediction of phase composition, in: Sims C.T., Stoloff N., Hagel W.C. (Eds.), Superalloys II, John Wiley & Sons, New York, USA, 1987, pp. 217–240. Smialek J.L., Meier G.H., High-temperature oxidation, in: Sims C.T., Stoloff N., Hagel W.C. (Eds.), Superalloys II, John Wiley & Sons, New York, USA, 1987, pp. 293–326. Strangman T.E., Hoppin III G.S., Phipps C.M., Harris K., Schwer R.E., Development of exothermically cast single crystal Mar-M 247 and derivative alloys, in: Tien J.K. et al. (Eds.), Superalloys 1980, American Society for Metals, Metals Park, OH, USA, 1980, pp. 225–234. Walston W.S., Schaeffer J.C., Murphy W.H., A new type of microstructural instability in superalloys – SRZ, in: Kissinger R.D. et al. (Eds.), Superalloys 1996, The Minerals, Metals & Materials Society, Warrendale, PA, USA, 1996, pp. 9–18. Walston W.S., O’Hara K.S., Ross E.W., Pollock T.M., Murphy W.H., René N6: third generation single crystal superalloy, in: Kissinger R.D. et al. (Eds.), Superalloys 1996, The Minerals, Metals & Materials Society, Warrendale, PA, USA, 1996, pp. 27–34. Waudby P.E., Benson J.M., Stander C.M., Pennefather R., McColvin G., Advanced high strength single crystal superalloy – SMP14, in: Strang A. et al. (Eds.), Advances in Turbine Materials, Design and Manufacturing, The Institute of Materials, London, UK, 1997, pp. 322–338. Wortmann J., Wege R., Harris K., Erickson G.L., Schwer R.E., in: Proceedings 7th World Conference on Investment Casting, Münich, Germany, 29 June–2 July 1988. Wukusick C.S., Final Report NAVAIR/N62269–78-C0315, 25 August 1980. Wukusick C.S., Buchakjian, Jr. L., General Electric Company, Patent Application #GB 2 235 697 A, March 1991. Xuan Nguyen-Dinh, Allied-Signal, Inc., US patent 4 935 072, 1990.