99,

Transactions of The Indian Institute of Metals              

Design considerations for a high temperature creep resistant nickel base alloy M. Sundararaman and J.B. Singh Structural Metallurgy Section, Mechanical Metallurgy Section, Materials Group, Bhabha Atomic Research Centre, Mumbai 400085, India E-mail: [email protected] Received 19 December 2008 Revised 14 October 2009 Accepted 14 October 2009 Online at www.springerlink.com © 2010 TIIM, India

Keywords: nickel base superalloy; creep resistance; microstructure

Abstract Life of all high temperature materials is decided by their creep properties. The understanding of factors that control their high temperature properties is important in designing creep resistant alloys. Since dislocation movement is primarily responsible for creep, all microstructural parameters that increase resistance to dislocation motion like, low stacking fault energy, association of jogs and vacancies with dislocation and distribution of second phase particles influence the creep resistance of alloys. In this paper, detailed investigations carried out on the evolution of two phase microstructural parameters in nickel base binary and some commercial high temperature alloys are presented. The role of some of the observed microstructures in enhancing creep resistance will be discussed.

1.

Introduction

For high temperature applications, designers would ideally like to have materials with infinite life or zero creep deformation. However, in all real situations, the life of components is limited by creep deformation. In order to design a creep resistant alloy, one should understand all the mechanisms that occur in the material/microstructural modifications that occur in it under thermal as well as thermal/ stress conditions. The design of creep resistant alloys is thus governed by the microstructure, stress and temperature conditions. Stress and temperature are decided by the operating conditions, which are being stretched for a better productivity. This is stretching the limits of the material application, which can be achieved by intelligent manipulation of the microstructure so as to reduce the aforementioned effects. The basic mechanisms responsible for creep deformation are dislocation glide, dislocation creep, diffusional creep and grain boundary sliding [1]. The simplest mechanism of creep is diffusional creep, which occurs by the transfer of matter by the migration of vacant lattice sites. Creep involving motion of dislocations may occur either by the climb of edge dislocations (Harper-Dorn creep) or by the glide of dislocations that overcome obstacles with the aid of thermal activation (Power-law creep). The plastic deformation associated with glide/climb may be small, but once the obstacles have been overcome or annihilated, dislocations can glide over large areas and produce relatively large plastic strains. Grain boundaries (GBs) in a polycrystalline material also have a strong influence on the creep. GBs are regions where the structure of the crystal is perturbed. They are the sources as well as sinks for the vacancies which produce diffusional creep. GBs also act as barriers to dislocation motion, increasing the rate of work hardening and reducing

the creep rate. If a shear stress acts across a GB, one grain would slide over the other. However, the two grains would remain intact only if there is also deformation within the grains. If the adjacent grains fail to keep contact, as is quite common at junctions of three or four grains, voids are formed. Creep at the microscopic level is controlled by the migration of defects and the resistance different structural inhomogenities in the material offer to it. Large volume of research carried out in the last few decades has shown that creep strength of the matrix can be increased by solute addition which participates in one of the processes- (i) locking of dislocations by solute atoms (Cottrell locking), (ii) extended dislocations by solute atoms (Suzuki locking), (iii) elastic interaction of solute atoms with moving dislocations (friction hardening), (iv) lowering of stacking fault energy (climb/ cross slip is made difficult), (v) interaction of vacancies with dislocation jogs (climb is difficult) and migration energy of vacancies. The uniform distribution of second phase particles (either through precipitation or dispersion of oxide particles through mechanical alloying) was also found to drastically increase the creep resistance of the alloy. In addition to it, generation of interfaces within grains (through ordering of the lattice) and / lamellar microstructure which increase the obstacle density to dislocation motion within the grain has been reported to enhance the creep resistance considerably. Two general approaches have been reported in the literature to produce creep resistant materials. In cases where the operating temperature is very high (close to the melting point of alloy) and grain boundary creep contributes significantly to creep, the methodology adopted is to produce material with preferred orientation of grain boundary like directionally solidified material or material without any grain boundaries like single crystals [2]. When the operating temperature is close to 0.5 to 0.7 Tm where dislocation creep



 

  !" # $

%           

is the predominant mechanism, the major effort is towards producing microstructure with as much impediments to dislocation motion as possible like precipitates [2-6], different types of interfaces within the grains etc. [7]. Study on lamellar intermetallics in γ-TiAl alloys is a typical example of the latter case [7]. In the present paper a brief overview of enhancing creep resistance by distribution of second phase particles is covered including some work carried out in precipitation hardened nickel base superalloy Nimonic PE16. Some results on the evolution of interphase and variant interfaces in two phase ordered intermetallics is presented including the thermal stability of this microstructure and the efficacy of this structure in enhancing the creep resistance.

2. Creep in precipitation hardened / oxide dispersion strengthened material As mentioned in section 1, dispersion of second phase particles acts as a physical obstacle to dislocation motion. How effective these dispersions are, is determined by the characteristics of the dispersion, namely, the lattice parameter misfit of precipitate with the matrix, their size, volume fraction and uniformity of dispersion. The smaller the misfit at operating temperature, the higher is the thermal stability of microstructure at operating temperature. In fact, in alloys containing large volume fraction of γ′ precipitates, rafting of precipitates (growth of precipitates in a preferred direction depending on the direction of applied stress) was noticed leading to change in microstructure and modified creep behaviour. The thermodynamic stability of the second phase particles decides the maximum upper temperature limit for service. In order to enhance the use of these materials to higher temperatures (close to melting point), dispersion of fine oxide particles is employed. The creep behaviour of nickel base superalloys containing γ′ particles has been well studied and reported extensively in the literature [3-6]. The presence of γ′ particles has been found to raise the value of exponent in the relation ε ∝ σn to about 8 to 23 from a value of 3 to 5 in the case of solution treated materials [4]. When ( ε )1/4 is plotted against applied stress, with n = 4 one could obtain linear plots with different slopes. The intercept of each plot with stress axis gives a threshold stress which has been attributed to a specific creep mechanism. In the case of Nimonic PE16, two slope behaviour has been reported [6]. The threshold stress determined for different slopes were found to correspond to local climb and general climb mechanisms theoretically predicted by Arzt and Ashby [8]. The CRSS for precipitate shearing and Orowan bypassing for different γ′ particle sizes determined following the equations derived by Nembach and Neite [9] was employed to calculate theoretical threshold stress for general and local climb mechanisms. Using these values, the creep rate was theoretically determined for different operating mechanisms for a test temperature of 650°C and its plot is shown in Fig. 1. The details of the calculations are beyond the scope of the paper and it will be described elsewhere. In this plot, the regimes corresponding to different mechanism are clearly demarcated. The dislocation structure corresponding to different applied stress and particle sizes have been characterized in detail. Illustrative examples of microstructures corresponding to different operating mechanisms are shown in Fig. 2. For 550 MPa stress, dislocations are confined mainly to planar arrays suggesting that mainly shearing of particles has occurred lowering of stress resulted in more uniform distribution of dislocation and loops could also be noticed. When the stress was

Fig. 1 : The plot of theoretically estimated creep strain rate versus γ′ particle size for different loads at test temperature of 650°C. The regimes corresponding to different dislocation precipitate interaction mechanisms are marked in the figure. Actual data points corresponding to different particle size and loads are also shown.

Fig. 2 : The distribution of dislocations after creep deformation corresponding to different applied stresses (mentioned at the top of each micrograph) for γ′ particle size of 21nm is shown in figure. See text for details.

 

  !" # $

%           

reduced to 250 MPa, mostly straight dislocations typical of general climb behaviour have been observed in the alloy. The oxide dispersion in the matrix is employed to increase the operating temperature. Rossler and Arzt observed attractive interaction between dislocation and dispersiod during TEM examination of crept specimens [10]. From the morphology of dislocation-precipitate configuration, it was concluded that higher stress is required to detach dislocation from the particle. They theoretically derived creep rate equation taking into account the detachment stress as an important factor [10]. On the basis of that expression, strong stress and temperature dependence of creep rate has been explained.

3.

Single phase intermetallic alloys

Intermetallic compounds with ordered atomic arrangements have shown a great potential for hightemperature materials. However, their application at high temperatures has not been realized in the single phase, whose creep strength is generally far inferior to that of advanced superalloys. Ni3Al system is one which has been studied extensively which has the same structure and composition as γ′ particles in nickel base superalloys. It was thought that this alloy will have better creep resistance because of inherently low diffusion coefficient. However, the thermal APBs produced within the matrix on ordering were found to have no effect on reducing creep as they are weak obstacle to dislocation motion. On the contrary, there are many intermetallic alloys with tetragonal distortion like Al3Ti and



Ni3 V based on D022 structure (body centred tetragonal) derived from fcc structure exists where many ordered variants form within each grain [11]. A typical microstructure of lamellar structure in stoichiometric Ni3V alloy is shown in figure 3. The thermal stability of these structures is governed by the magnitude of tetragonal distortion - the smaller the distortion, the higher is the stability. In these types of alloys, a group of perfect dislocations generated in one variant can not easily propagate into another variant without the formation of high energy faults within them [12]. Hence, these boundaries, like precipitates, can act as effective obstacles to dislocation motion within the grains. However, it should be borne in mind that many of these intermetallics are deformed by propagation of twins which can often cut across many ordered variants and the stress required to propagate them is very small though the stress necessary for their formation is very high [13]. More research is required to optimize the microstructure and chemistry of single phase intermetallics so that mechanism of deformation can be controlled to enhance creep resistance. In order to improve the creep resistance of single phase intermetallics, Arzt and co-workers introduced yttria dispersion in the matrix in ordered Ni3Al [14, 15] and NiAl [16] intermetallics. Experimentally, they showed that superpartials have to surmount the obstacles to propogate deformation across the matrix. They formulated a new creep equation based on sequential detachment of leading and trailing superpartials from the particles. Using this model, they could find out an optimum particle size for a given anti phase boundary energy of the alloy to maximize creep strength [17].

4. Two phase ordered intemetallics

Fig. 3 : Lamellar microstructure developed within grains due to tetragonal distortion associated with Ni3V phase.

Recently, lot of research work is being carried out on two-phase intermetallics which inherently have advantages over the single phase one for creep properties because of generation of more interfaces within the matrix. In this case, in addition to the variant interfaces, interphase interfaces also act as impediment to deformation propagation. γ - TiAl is an example of one such two phase intermetallic alloy where a lot of research activities is going on for the last two decades to optimise the microstructure to improve its creep properties [7]. In this context, the hyperstoichiometric Ni3V alloy with V in the range of 27 at% to 33 at% undergoes phase transformation by eutectoid decomposition reaction occurring below 908°C [18] to form two ordered phases (Ni3V

Fig. 4 : Distribution of different variants in Ni-29at% V alloy which has undergone eutectoid transformation at temperature below 908°C. (a) BF, (b) DF microsgraph corresponding to one variant of Ni2V and (c) schematic of the area showing distribution of different variants.



 

  !" # $

%           

phase with tetragonal structure and Ni2V with orthorhombic structure) with both phases derived from the disordered fcc lattice. Understanding the creep phenomenon in two phase materials depends upon (i) the generation of detailed knowledge on the evolution of microstructure and (ii) the interaction and coupling of the elementary deformation and damage processes with these interfaces. The characterization as well as identification of different types of interfaces that forms between the two-phases and their characteristics that is useful for creep resistance is described in the following sections 4.1 Microstructure of the Ni-29at%V Alloy The ordered Ni3V phase has a tetragonal structure (D022 - I4/mmm; a=0.3542 nm and c/a=2.036 [4]) and its c axis can align itself along any one of the three a-axes of the high temperature (fcc) structure resulting in the formation of three mutually orthogonal variants called transformation twins. The ordered Ni 2V phase has an orthorhombic structure (D 25 - Immm, Pt2Mo type; A=0.2559 nm, B=0.7640 nm, 2h C=0.3549 nm [19]) and forms six independent orientation variants of the Ni2V superlattice with respect to the fcc structure. The six variants of the ordered structure are twin related and form two types of twins, viz., parallel and perpendicular. Parallel twins share the same C axis, while their respective A and B axes are rotated by 90º from each

other. Perpendicular twins are those variants for which the C axes are mutually perpendicular. The orientation axis of six variants with respect to the fcc structure is given in Singh et al. [20]. The Ni2V phase precipitates in a plate shaped morphology within Ni3V lamellae with {120} type habits on ageing below Tc (Fig. 4). This is because the structures of the Ni 3 V and Ni 2V phases belongs to the {420} family of structures and due to the fact that 1/2<110> type superpartial dislocation are present within the Ni 3V phase. These superpartials create {210} type non-conservative APBs, which contain Ni2V nuclei. Precipitation of the Ni2V phase within the Ni 3V lamellae causes the interfaces between adjacent Ni3V domains to assume a zig-zag appearance. The zig-zag morphology of these interfaces is a consequence of the formation of different coherent interphase, and intraphase interfaces when any two domains meet. Such interfaces are excellent barriers against boundary sliding. A schematic illustration of the [001] projection of atoms across different types of interfaces encountered is shown in figure 5. Intravariant as well as inter-variant interfaces between different Ni3V and Ni2V variants are coherent. A HRTEM image of interfaces is shown in Fig. 6. Due to the presence of low energy coherent interfaces, the two-phase microstructure is quite stable against coarsening. However, re-nucleation of the two phases in the vicinity of the grain boundaries, in a manner similar to “recrystallization”, takes place upon prolonged heat treatment.

Fig. 5 : A schematic illustration of the [001] projection of atoms across different types of interfaces encountered in Ni-29at% V alloy on ordering.

 

&

  !" # $

%           

twelve modes generate true twins of either combined, type 1 or type 2 [22]. The nature of twins generated in different variants of Ni3V and Ni2V phases for different types of shear on octahedral planes is given in table 1 and it has been worked out on the basis of crystallography of twins. From this table it emerges out that, for a fixed shear, twins of different types would be created in different variants of the two phases. From this analysis, it can be inferred that as twins propagate across interfaces into another adjacent phase, they would usually undergo a change in their nature, which would be dependent upon the relative orientation of different variants of the adjoining phases. 4.3 Deformation microstructure of the two-phase alloy Fig. 6 : High resolution image of interfaces between Ni3V and Ni2V phases revealing its coherent nature. The FFT from different regions confirm the nature and identity of variants.

4.2 Crystallographic Features of Twins in Ni3V and Ni2V Phases Twins in ordered structures can form by either true or pseudo mode. True twins retain the ordered state of the structure whereas pseudo twins destroy it. The formation of true and pseudo twins in D022 and Pt2Mo type structures has been discussed by Christian and Laughlin [21]. Accordingly, upon ordering, in the Ni3V phase, all the {111} planes of the parent fcc phase become {112}Ni3V planes. The mixed parenthesis used throughout the text means that all permutations are possible only on first two indices while the third index is fixed. Of the three partials with Burgers vector 1/6<11-2> in the (111) plane of the disordered lattice, only one of them generate true twins while the propagation of others generate pseudo twins in any variant of Ni3V. It means out of the twelve twining modes only four of them generate true twins. In the case of Ni2V phase, ten out of the

Deformation in this two-phase alloy takes place predominantly by the propagation of SF and/or twins across coherent Ni2V/Ni 3V interfaces. This is possible without disturbing the long range order of the two superlattices only when the two phases bordering the interface correspond to specific orientations with respect to the parent fcc lattice (Table 1). This property is illustrated by the example shown – in Fig. 7, where twins are cutting across (120) interfaces in lamellar two-phase alloy. Figure 7(a) depicts schematically the situation where the Ni2V phase was bounded as plates within a Ni3V lamella. The two phases are separated by – coherent interfaces lying parallel to (120) plane. Figure 7b shows a BF image of the region and delineated Ni3V ([010] variant) and Ni2V (variant 1) phases are shown in Figs. 7c and 7d respectively. Twins that were cutting across Ni2V/ Ni3V interfaces are shown by arrow marks in Fig.7b. The – – habit plane of these twins was found to lie on the (111) plane. A DF image of twins taken with a fundamental twin reflection is shown in Fig. 7(e). The propagation of twins was apparently stopped by an interface between [010] and [100] variants of the Ni3V phase (see double-headed arrows). These twins were made up of overlapping geometric SF that does not disturb the long-range ordered state of either of the two

Table 1 : The nature of twins created in different variants of the Ni3V and Ni2V phases for different types of shears on the octahedral plane. (V: variant; C: combined twin; I: type I twin; II: type II twin; P: pseudo twin)

S.No.

1 2 3 4 5 6 7 8 9 10 11 12

Shear Shear Plane Direction K1 η1

(111) – (111) – (111) – (111) (111) – (111) – (111) – (111) (111) – (111) – (111) – (111)

T W I N Ni3V Phase

T Y P E

Ni2V Phase

[100]

[010]

[001]

V1

V2

V3

V4

V5

V6

P

P

C

P

C

II

I

II

I

P

P

C

C

P

I

II

II

I

P

P

C

C

P

II

I

I

II

P

C

P

C

I

II

I

II

C

P

I

II

C

P

II

I

C

P

II

I

P

C

II

I

P

C

P

II

I

C

P

I

II

P

C

P

I

II

P

C

I

II

C

P

P

I

II

II

I

C

P

C

P

P

II

I

I

II

C

P

C

P

P

II

I

II

I

P

C

C

P

P

I

II

I

II

P

C

– [112] – [112] – [112] [112] – [121] – [121]

P P P

[121] – [121] – [211] [211] – [211] – [211]



 

  !" # $

%           

Fig. 7 : Propagation of deformation twins across Ni3V and Ni2V interface. See text for details.

– – Fig. 8 : Schematic illustration of arrangement of atoms on the (111) close packed plane in Ni3V ([010] variant) and Ni2V (variant 1) – – – – phases. This plane correspondingly becomes the (1 1 2)Ni V and (101)Ni V planes of the two phases. Open and shaded circles 3 2 represent Ni and V atoms whereas a solid (black) circle in each drawing represents the position of a minority atom on the next plane.

phases [23]. Since the continuity of the glide plane is – maintained across coherent (120) interfaces SFs created in one phase would move in to the other if the moving Shockley does not disturb the long range order of the lattice. Figure 8 illustrates a typical arrangement of atoms – – – – –– on close packed (111) planes, which is (11 2)Ni V and (101 3 )Ni V planes of the Ni3V ([010] variant) and Ni2V (variant 1) 2 phases. Since the ordered arrangement of atoms on close packed planes in the two phases is different, the movement of an atom (such as shown by a solid black circle in each figure) on the next layer in a particular direction would disturb the neighbouring atoms differently in the two phases. A Shockley partial in the Ni3V lattice can move an – – – atom only in the [1 2 1 ] direction without disturbing the long-range order of the lattice whereas an equivalent – movement by other two Shockley partials (i.e., 1/6[211] and – 1/6[112]) in this plane would destroy the D022 order resulting in the formation of complex stacking faults (CSFs). In the Ni2V phase an equivalent movement of an atom by any one of the three Shockley partials (equivalently expressed as 1/ – – –– 6[111]Ni V, 1/3[101]Ni V, 1/6[111]Ni V respectively) would all 2 2 2 retain the ordered state of the Ni2V lattice. Therefore, in the – event of deformation propagation across (120) interface, if –– – a stacking fault is created by the movement of a 1/6[1 2 1 – – ] partial dislocation on the (111) plane in the [010] variant of the Ni3V phase, the fault would be able to propagate across Ni3V/Ni2V interfaces if the adjoining Ni2V domain has the orientation of variant 1. On the other hand, though a stacking fault can be created by any one of the three Shockley partials in the Ni2V phase, the fault would be able to cut across Ni3V/Ni2V interfaces into the adjoining Ni3V phase only when the Shockley partials has the Burgers – – – vector 1/6[1 2 1]. The propagation of stacking faults and

twins generated in one variant of a phase into another variant of second phase can be analysed on similar lines. The analysis detailed here is in agreement with that expected from crystallographic consideration as given in Table 1. From the above results, it is clear that twinning is the major mode of deformation in Ni-V system in the composition range under consideration. Some of the twins from one variant can easily propagate into adjacent ones and it is to be noted that twinning requires hardly any extra stress to propagate it. It is also known that in this ordered systems, superpartials generated in one variant on entering another variant requires extra energy to change the nature of anti phase boundaries and this requirement of increase in energy is responsible for confining superpartials to the same variant and imparting enhanced creep resistance. Hence in order to produce better creep resistant intermetallic alloys, the mode of propagation has to be confined to mainly by dislocation motion and this necessitates further research. Another important aspect of ordered intermetallics mentioned here is their stability under stress. The thermal stability of these ordered intermetallics has been well studied. In fact, it has been reported in the literature that the lamellar structure changes to equiaxed one during creep in γ-TiAl alloy [24]. The stability of microstructures under stress could be enhanced by controlling the tetragonal distortion.

5. Conclusions The various microstructural parameters that have to be controlled to improve the creep resistance in disordered and ordered alloys have been presented in this paper. The nature of interfaces in hyper-stoichiometric Ni3V alloys and their role

 

  !" # $

%           

in propagation of faults and twins across them have been elucidated. The stress and strain rate regimes under which different dislocation precipitate interaction mechanisms operate has been identified theoretically in the case of Nimonic PE16 for test temperature of 650°C and found to broadly confirm to experimental observations.

7.

Acknowledgements

12.

The authors would like to thank Dr. J. K. Chakravarthy, Head, MMS and Dr. A. K. Suri, Director, Materials Group for their keen interest and constant support in this work.

13.

References

15. 16. 17. 18.

1. 2. 3. 4.

5. 6.

Dieter G E, Mechanical Metallurgy, SI Metric Ed., McGraw –Hill Company,(1988) 432. Duhl D N, in The Superalloys II, (eds.) Sims C T, Stoloff N S and Hagel W C, John Wiley and Sons, New York, (1987) 189. Lagneborg R and Bergman B, Metal Science J., 10 (1976) 20. Wahi R P, Chen W, Kutumbarao V V and Skopp A, Proceedings of the 8th Risø International Symposium on Metallurgy and Materials Science, RISØ National Laboratory, Denmark, (1988), 537. Chatirvedi M C and Han Y, Mater. Sci. Eng., 89 (1987) l7. Mathew M D, Sundararaman M, Chen W and Wahi R P, Proceedings of the Second International Symposium on Plasticity and its Current Applications, Tsu, Japan, (1989).

8. 9. 10. 11.

14.

19.

20. 21. 22. 22. 24.

'

Appel F and Wagner R, Mater Sci. Eng., R22 (1998) 187. Arzt E and Ashby M F, Scripta Metall., 16 (1982) 1285. Nembach E and Neite G, Prog. Mater. Sci., 29 (1985) 177. Rossler J and Arzt E, Acta Metal. Mater., 38 (1990) 671. Yamaguchi M and Umakoshi Y, Prog. Mater. Sci., 34 (1990) 1. Singh J B, Sundararaman M and Mukhopadhyay P, Philos. Mag. A, 80 (2000) 1983. Sundararaman M, Mukhopadhyay P and Banerjee S, Acta Metall., 36 (1988) 847. Klotz U E, Mason R P, Goehring E and Arzt E, Mater. Sci Eng., A231 (1997)198 Arzt E and Goehring E, Acta Mater., 46 (1998) 6575. Arzt E and Grahle P, Acta Mater., 46 (1998) 2717. Arzt E and Goehring E, Scripta Metall., 8 (1993) 843. Massalski T B, Binary Alloys Phase Diagrams, ASM International, 3 (1992) Villars P and Calvert L D, Pearson’s Handbook of Crystallography Data for Intermetallic Phases, ASM, Metals Park (OH), 3, (1986), p.2907. Singh J B, Sundararaman M and Mukhopadhyay P, Metall Trans., 29A (1998) 1883 Christian J W and Laughlin D E, Acta Metall., 36 (1988) 1617. Singh J B, Ph.D. dissertation, IIT Bombay, 2002. Singh J B, Sundararaman M, Mukhopadhyay P and Prabhu N, Mater. Character., 49 (2003) 347. Kim H Y and Maruyama K, Acta Mater., 51 (2003) 2191.