Comparative study of the segregation behavior and crystallographic orientation in a nickel-based single-crystal superalloy

Accepted Manuscript Comparative study of the segregation behavior and crystallographic orientation in a nickel-based single-crystal superalloy Fu Wang, Dexin Ma, Samuel Bogner, Andreas Bührig-Polaczek PII:

S0925-8388(15)01455-3

DOI:

10.1016/j.jallcom.2015.04.237

Reference:

JALCOM 34267

To appear in:

Journal of Alloys and Compounds

Received Date: 13 March 2015 Revised Date:

19 April 2015

Accepted Date: 30 April 2015

Please cite this article as: F. Wang, D. Ma, S. Bogner, A. Bührig-Polaczek, Comparative study of the segregation behavior and crystallographic orientation in a nickel-based single-crystal superalloy, Journal of Alloys and Compounds (2015), doi: 10.1016/j.jallcom.2015.04.237. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Comparative study of the segregation behavior and crystallographic orientation in a nickel-based single-crystal superalloy Fu Wang, Dexin Ma*, Samuel Bogner, Andreas Bührig-Polaczek

RI PT

Foundry Institute, RWTH University, 52072 Aachen, Germany

Abstract:

The segregation behavior and crystallographic orientation of a nickel-based single-crystal CMSX-6 superalloy produced by the downward directional solidification process have been investigated. The results were compared

SC

with those in the Bridgman process. In comparison to the Bridgman process, the fluctuations in the concentrations of Al, Ti, Ta, Co, Mo and Cr became smaller along the path from the dendrite lobe through the interdendritic region to the adjacent dendrite lobe. This suggests that the degree of the segregation of these

M AN U

alloying elements was reduced by using the DWDS process. In addition to this, the angle between the dendrites’ preferred growth direction and the heat flow direction in the DWDS cast samples, and the splaying region were smaller than those in the Bridgman solidified samples. This indicates that better crystal orientation can be obtained in nickel-based single-crystal superalloy components by using the downward directional solidification process.

1. Introduction

TE D

Keywords: Alloys; Crystal growth; Solidification; Electron probe

Nickel-based single-crystal superalloys are widely used in turbine blades where high

EP

temperature strength and creep resistance are required [1]. The performance of these single-

AC C

crystal superalloy blades can be improved by either optimizing their composition [2-7] or their solidification process. Nowadays, the major directional solidification processes (DS) are the Bridgman process [8], the liquid metal cooling process (LMC) [9-11] and the gas cooling casting process (GCC) [12]. However, these processes exhibited some disadvantages when they are industrially used to produce highly-efficient single-crystal turbine blades, especially for large industrial gas turbine blades (IGB) [13]. In the Bridgman process, the thermal

*

Corresponding author. Tel.: +49 (0)2418095883; fax: +49 (0)2418092276. E-mail address: [email protected] (D. Ma). 1

ACCEPTED MANUSCRIPT gradient (GL) ahead of the liquid/solid (L/S) interface decreases with increasing distance from the chilling plate [14]. This decrease could produce an inhomogeneous microstructure [15]. In addition to this, the mold used in this process is thick and non-uniform [16] which can generate a nonhomogeneous thermal field leading to an occurrence of stray grains. The mold

RI PT

used in the LMC process is, on the one hand, also thick and nonuniform, and on the other hand the coolants may contaminate the casting. This is deleterious to the mechanical properties of the nickel-based superalloys [17]. Owing to its open structure, the gas used in

SC

the gas cooling casting process can cool the furnace and lead to the occurrence of defects such as stray grains.

M AN U

To offset these disadvantages in conventional DS processes and to meet the demands of highly efficient turbines, the downward directional solidification process (DWDS) was presented by Ma et al. [16, 18]. Since, in this process, there is no impact force or static pressure exerted by the melt during the casting process, a 1 mm thick ceramic mold can be

TE D

used which is thinner than that used in the conventional processes (7–8 mm). In addition to this, the occurrence of freckles can be effectively reduced due to the density inverse [19, 20]: an unstable density gradient, which is a major factor leading to the emergence of freckles, is

EP

restrained during the downward growth.

Basic research into the effect of the solidification process parameters on the microstructural

AC C

development during the casting of superalloys is imperative since the as-solidified material dictates the subsequent microstructural development and the ultimate engineering performance. To date, however, the benefits of the DWDS process have not been directly measured in terms of micro-segregation behavior and crystallographic orientation. The primary aims of this research are to provide a direct comparison between the Bridgman and the DWDS process with respect to these aspects. By considering the results of this experimental program, the benefits of the DWDS process compared to the Bridgman process 2

ACCEPTED MANUSCRIPT are discussed. 2. Experiments In a previous paper [20], details of the DWDS experimental furnace were presented. Here,

RI PT

we only give a brief description of the procedure. Fig. 1 schematically depicts the procedure of the DWDS process for casting nickel-based single-crystal superalloy components. During the

experiment,

CMSX-6

(Ni-10Cr-5.0Co-3.0Mo-4.85Al-4.75Ti-2.0Ta-0.02C,

wt%)

SC

superalloy was overheated to 1773 K in a crucible and covered with hollow Al2O3 particles (1–3 mm in diameter) which act as a dynamic baffle. A ceramic tube (1 mm thick, 200 mm

M AN U

high, and 9 mm inside diameter), having a single-crystal seed of CMSX-6 superalloy with <001> crystal orientation in the longitudinal direction, was inserted into the melt. The opposite end of the tube was wrapped and sealed using a nickel foil which acts as plug to prevent the ingress of the dynamic baffle. When the foil melts, the alloy melt flows into the

TE D

mold and makes contact with the seed. After the seed has partially melted, the tube is elevated at a withdrawal rate (V) of 50 µm s-1 and cooled by gas (argon), resulting in the solidification of a single-crystal bar.

EP

Conventional Bridgman bars were solidified in an ALD Vacuum Technologies, Inc. furnace. The parameters, such as melting temperature, the geometry of the mold, the sampling

AC C

location and the withdrawal rate, were identical to those employed in the DWDS process. Here, however, the wall-thickness of the mold was 7 mm and was preheated to a temperature of 1773 K.

After solidification, the bars were sectioned transversely (perpendicular to the growth direction), and the samples were mounted and polished for microstructural analyses. Fig. 2 shows a single-crystal cylindrical bar and the section position of the samples. The segregation behavior of different alloying elements in the samples was determined using a JAX-8100 3

ACCEPTED MANUSCRIPT electron probe microanalysis (EPMA) device. EPMA line scanning was conducted to investigate the concentration distribution of the alloying elements between two adjacent dendrite lobes (A→B shown in Fig. 3 (a) and (b)). Three scanning lines on the transverse section of the examined samples were chosen for the analysis. According to the sizes of the

RI PT

dendrites in the samples produced by the two investigated processes, two different scanning spacing were chosen. These spacings were approximately 180 µm (L1) and 120 µm (L2) for the Bridgman process and the DWDS process, respectively. The individual data acquisition

SC

points along the analysis line were 2 µm apart. To comparatively plot the results from both samples onto one graph for the same alloying element, point A was set to zero, and the

M AN U

distances X1 and X2 for each examined point for the Bridgman and DWDS processes, respectively, were then normalized with respect to L1 and L2. These dimensionless numbers were used as the plot’s abscissa. The average values of the examined data at the same position along the scanning lines were used as the plot’s ordinate. The variation in the concentration of

TE D

the alloying elements was indicated by the standard deviations. The crystallographic orientations of the dendrites on the transverse samples were examined by using HKLNordlys

EP

electron back-scatter diffraction (EBSD) detector. 3. Results and discussions

AC C

3.1 Solute segregation behavior

Fig. 3 shows the average intensity distributions of the primary alloying elements Al, Ti, Ta, Mo, Co, and Cr along the scanning spacing in both of the processes’ cast samples. It can been seen from Fig. 3 (e) that the intensity of Ti in the interdendritic region is stronger than that in the dendrite lobe. In addition to this, the difference in the intensity between the two regions in the DWDS solidified samples is smaller than that in the Bridgman process. Since the intensity distribution of the alloying elements is proportional to the concentration distribution, this 4

ACCEPTED MANUSCRIPT finding suggests that Ti segregated to the interdendritic region during the solidification and the degree of the segregation was slighter in the DWDS solidified samples than that in the Bridgman process. In the case of Al and Mo (shown in Fig. 3 (c) and (d)), the concentration of Al in the interdendritic area of the Bridgman solidified sample was smaller than that in the

RI PT

dendrite lobe while, for this sample, a larger concentration of Mo was observed in the interdendritic region than in the dendrite lobe. These results contradict some investigations [21, 22] which reported that Al is enriched in the interdendritic region, whilst Mo is enriched

SC

in the dendrite lobe during the solidification. However, other researchers [23] suggest that there is a reversal in the segregating direction with increasing sample withdrawal rate.

M AN U

Compared to the Bridgman process, the difference in the concentrations of the two elements between the two positions was smaller for the DWDS solidified samples. In contrast to this, a sharp increase or decrease in the concentration of Cr, Ta and Co between the interdendritic region and the dendrite lobe was not observed in either the DWDS or Bridgman solidified

TE D

samples presented in Fig. 3 (f), (g) and (h). In addition to this, a significant reduction in the intensity fluctuations of Al, Ti, Mo, and Cr along the scanning spacing was found in the DWDS solidified samples. The fluctuations of the elements of Ta and Co were also reduced,

EP

but these reductions were not as large as those of Al, Ti, Mo, and Cr. Based on the above observation, a nickel-based single-crystal component having more homogenious distrubution

AC C

of the alloying elements can be obtained by using DWDS process. It is especially favorable for nickel-based single-crystal superalloys having an enhanced content of refractory elements (i.e., Re and Ru) because the sluggish diffusivity of these elements in the solid-state leads to significant segregation [24]. Liu et al. [11] reported that these chemical inhomogeneities are not removed, even by employing a subsequent high-temperature homogenizing process. Thirumalai et al. [25] and Harold and Merton [26] found that the influence of elevated thermal gradient on the segregation behavior for the constituent elements can be largely 5

ACCEPTED MANUSCRIPT attributed to the homogeneous back-diffusion in the solidified solid. A critical factor for solid back-diffusion is the diffusion distance (approximately equal to λ1/2, where λ1 is the primary dendrite arm spacing). Under higher thermal gradients, the diffusion distance reduces significantly due to the greatly refined dendritic structures. Our previous investigation [27]

RI PT

showed that the thermal gradient ahead of the L/S interface in the DWDS process and the conventional Bridgman process was 23.6 K mm-1 and 2–4 K mm-1 [9], respectively. This suggests that the thermal gradient in the DWDS process is almost 10–12 times larger than that

SC

in the Bridgman process. Subject to this high thermal gradient, the microstructures were significantly refined. Therefore, the samples cast by using the DWDS process had a smaller λ1

M AN U

which resulted in a shorter back diffusion distance. For this reason, the primary alloying elements Al, Ti, Ta, Mo, Co, and Cr distributed more homogeneously in the DWDS cast samples, and the fluctuations of these elements along the scanning spacing were reduced. In the analysis of the concentration of the alloying elements, a larger (120 µm) and

TE D

smaller scanning spacing (180 µm) between two adjacent dendrite lobes was chosen in the DWDS and Bridgman samples, respectively. It seems that the spacing in the DWD samples is not in accordance with the high thermal gradient of the process reported in our previous study

EP

[18]. This report showed that a 53% reduction in average λ1 value was observed for the DWDS solidified CMSX-6 samples in comparison to the Bridgman process, and large

AC C

variations in λ1 values were found in both of the DWDS and Bridgman samples. The maximum λ1 value in the DWDS samples was a little smaller than the minimum λ1 value in the Bridgman samples. The selected scanning spacing in the DWDS samples was near the maximum λ1 value, whereas it was near the minimum λ1 value in the Bridgman samples. Therefore, a significant reduction was not found between the two spacings. Above experimental results indicate that the distribution of the alloying elements in the DWDS samples was homogeneous, even if the larger scanning spacing was selected. One can obtain 6

ACCEPTED MANUSCRIPT if a smaller scanning spacing in the DWDS samples is chosen, a much more homogeneous distribution of the alloying elements along the spacing can be observed in comparison to that in the Bridgman process. Consequently, a nickel-based single-crystal superalloy component possessing a more homogeneously-distributed alloying elements can be obtained by

RI PT

employing the DWDS process. 3.2 Crystallographic orientation

The preferred growth direction of the nickel-based superalloys is <001> [28]. This natural

SC

axial texture coincides with the direction of the minimum Young’s modulus, and leads to

M AN U

reduced thermal stresses and enhanced thermal fatigue resistance of the single-crystal components in service [29]. The deviation of the preferred growth direction from the heat flow direction (axial direction of the component) greatly influences the mechanical properties [30-32]. The angle between these two directions is defined as the deviation angle (θ). Fig. 4 presents the inverse pole figures (IPFs) of the cast samples in both processes. According to

TE D

this figure, the measured deviation angles of the primary dendrites showed that the deviation angle in the DWDS cast samples (depicted in Fig. 4(a)) was 0° ± 1°, and the splaying region

EP

of the dendrites was about 0°–5°. However, the angle in the samples produced by the Bridgman process (depicted in Fig. 4(b)), was 7.8° ± 1° and the splaying region was 2°–12°.

AC C

This suggests that by employing the DWDS process, it is possible to produce nickel-based single-crystal superalloy components possessing improved crystal orientation. A small thermal gradient ahead of the L/S interface has been known to bend the solidification front [33]. Curved solidification fronts typically contain higher horizontal thermal gradient components (i.e., perpendicular to the solidification direction) and lower vertical thermal gradient components (i.e., parallel to the solidification direction). Deviations from the vertical thermal gradient component cause the developing dendritic tip to splay from the desired growth orientation [34]. Similar report was also given by Ardakani et al. [35] who 7

ACCEPTED MANUSCRIPT found that under high thermal gradient the deviation angle was smaller than that under low thermal gradient, and a straight L/S interface guaranteed a precise <001> growth direction. As stated above, the thermal gradient ahead of the L/S interface in the DWDS process is much higher than that in the Bridgman process. The DWDS process is more capable of preserving

RI PT

finely straight solidification fronts than the Bridgman process. In addition to this, previous study [36] indicates that the convection situation ahead of the L/S interface of the DWDS sample was different from the Bridgman process, and the convection contributes to alleviate

SC

the accumulation of the solute and preserve a straighter solidification front. Thus, the higher horizontal thermal gradient components can be alleviated, the splay of the developing

M AN U

dendritic tip from the desired growth orientation was inhibited. Consequently, the deviation angle in the DWDS cast samples and the diverging region of the dendrites were smaller than that in the Bridgman solidified samples.

TE D

4. Conclusions

The downward directional solidification casting process was compared to the conventional Bridgman casting process with regard to the segregation behavior and crystallographic

EP

orientation. In comparison to the Bridgman process, the concentration distribution of the primary alloying elements along the path of the dendrite lobe through the interdendritic region

AC C

to the adjacent dendrite lobe became more homogeneous in the DWDS process. This indicates that the degree of the segregation of these elements was alleviated by using the DWDS process. In addition to this, the deviation angle in the DWDS cast samples and the diverging region of the dendrites were smaller than that in the Bridgman solidified samples. This indicates that nickel-based single-crystal superalloy components possessing better crystal orientation can be obtained by using the DWDS process.

8

ACCEPTED MANUSCRIPT

Acknowledgements This work was supported by the Deutsche Forschungsgemeinschaft under Grant No. MA

RI PT

2505/3-1. One of the authors (FW) would like to acknowledge the China Scholarship Council

AC C

EP

TE D

M AN U

SC

for supporting his stays in Germany.

9

ACCEPTED MANUSCRIPT References [1] X.B. Zhao, L. Liu, W.G. Zhang, G. Liu, J. Zhang, H.Z. Fu, Mater. Lett. 63 (2009) 2635-2638. [2] D. Blavette, P. Caron, T. Khan, Scripta Metall. 20 (1986) 1395-1400. [3] J.R. Li, D.Z. Tang, R.L. Lao. Riling, S.Z. Liu, Z.T. Wu, J. Mater. Sci. Technol. 15 (1999) 53-57. [4] A.C. Yeh, S. Tin, Scripta Mater. 52 (2005) 519-524.

RI PT

[5] A. Heckl, S. Neumeier, M. Göken, R.F. Singer, Mater. Sci. Eng. A 528 (2011) 3435-3444. [6] A.F. Giamei, D.L. Anton, Metall. Trans. 16 (1985) 1997-2005.

[7] S. Neumeier, F. Pyczak, M. Göken, in: R.C. Reed et al. (Eds.), Superalloys 2008, TMS, Warrendale (PA),

SC

2008, pp. 109-119.

[8] P.W. Bridgman, Crystals and their manufacture, US Patent 1793672 (Field on February 16, 1926, Grand on

M AN U

February 24, 1931)

[9] A.F. Giamei, J.G. Tschinkel, Metall. Trans. A 7 (1976) 1427-1434.

[10] C.L. Brundidge, D. Vandrasek, B. Wang, T.M. Pollock, Metall. Mater. Trans. A 43 (2012) 965-76. [11] L. Liu L, T.W. Huang, M. Qu, G. Liu, J. Zhang, H.Z. Fu, J. Mater. Process. Technol. 210 (2010) 159-165. [12] M. Konter, E. Kats, N. Hofmann, in: T.M. Pollock et al. (Eds.), Superalloys 2000, TMS, Warrendale (PA), 2000, pp. 189-200.

(2004) 3221-3231.

TE D

[13] A.J. Elliott, S. Tin, W.T. King, S-C. Huang, M.F.X. Gigliotti, T.M. Pollock, Metall. Mater. Trans. A 35

[14] P.K. Rohatgi, K. Pasciak, C.S. Narendranath, J. Mater. Sci. 29 (1994) 5357-5366.

EP

[15] S.F. Gao, L. Liu, N. Wang, X.B. Zhao, J. Zhang, H.Z. Fu, Metall. Mater. Trans. A 43 (2012) 3767-3775. [16] D.X. Ma, H. Lu, A. Burig-polaczek, in: IOP Conf. Ser: Mater. Sci. Eng. 27 (2012) 012036.

AC C

[17] R.T. Holt, W. Wallace, Inter. Meter. Rev. 21 (1976) 1-24. [18] F. Wang, D.X. Ma, J. Zhang, S. Bogner, A. Bührig-Polaczek, J. Mater. Process. Technol. 214 (2014) 31123121.

[19] D.X. Ma, Q. Wu, A. Bührig-Polaczek, Metall. Mater. Trans. B 43 (2012) 344-353. [20] F. Wang, D.X. Ma, J. Zhang, A. Bührig-Polaczek, J. Alloys Compd. 620 (2015) 24-30. [21] R.M. Kearsey, J.C. Beddoes, K.M. Jaasalu, W.T. Thompson, P. Au, in: K.A. Geen et al. (Eds.), Superalloys 2004, TMS, Warrendale (PA), 2004, pp. 801-810. [22] D.X. Ma, U. Grafe, Mater. Sci. Eng. A 270 (1999) 339-342. [23] B.C. Wilson, E.R. Cutler, G.E. Fuchs, Mater. Sci. Eng. A 479 (2008)356-364. 10

ACCEPTED MANUSCRIPT [24] N.D. Souza, H.B. Dong, in: R.C. Reed et al. (Eds.), Superalloys 2008, TMS, Warrendale (PA), 2008, pp. 261-268. [25] A. Thirumalai, A. Akhtar, R.C. Reed, Mater. Sci. Technol. 22 (2006) 1-13. [26] D.B. Harold, C.F. Merton, Trans. Metall. Soc. AIME 236 (1966) 615-23. [27] F. Wang, D. Ma, J. Zhang, L. Liu, J. Hong, S. Bogner, A. Bührig-Polaczek, J. Cryst. Growth 389 (2014) 47-

RI PT

54.

[28] X.B. Zhao, L. Liu, Z.H. Yu, W.G. Zhang, J. Zhang, H.Z. Fu, J. Mater. Sci. 45 (2010) 6101-6107. [29] A. Wagner, B.A. Shollock, M. McLean, Mater. Sci. Eng. A 374 (2004) 270-279.

SC

[30] D.M. Shah, A. Cetel, in: T.M. Pollock et al. (Eds.), Superalloy 2000, TMS, Warrendale (PA), 2000, pp. 295-304.

[31] R.A. MacKay, R.D. Maier, Metall. Trans. A 13 (1982) 1747-1754.

M AN U

[32] J. Yu, J.R. Li, J.Q. Zhao, M. Han, Z.X. Shi, S.Z. Liu, H.L. Yuan, Mater. Sci. Eng. A 560 (2013) 47-53. [33] M.L. Clemens, A. Price, R.S. Bellows, JOM 55 (2003) 27-31.

[34] N.D. Souza, M.G. Ardakani, M. Mclean, B.A. Shollock, Metall. Mater. Trans. A 31 (2000) 2877-2886. [35] D.G. Ardakani, N. D’Souza, A. Wagner, B.A. Shollock, M. McLean, in: T.M. Pollock et al. (Eds.), Superalloy 2000, TMS, Warrendale (PA), 2000, pp. 219-228.

AC C

EP

TE D

[36] F. Wang, D. Ma, J. Zhang, L. Liu, J. Hong, S. Bogner, A. Bührig-Polaczek, Int. J. Mater. Res. 105 (2014) 2.

11

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Fig. 1 Schematic illustration of the DWDS procedure for casting single-crystal nickel-based superalloy components.

Fig. 2 Locations of sectioned samples in the single bar.

12

ACCEPTED MANUSCRIPT (a)

(b)

(c)

50µm

M AN U

SC

(d)

RI PT

50µm

(f)

(h)

AC C

(g)

EP

TE D

(e)

Fig. 3 Examined solute distribution between the adjacent dendrite lobes. (a) and (b) were scanning paths in the Bridgman process and DWDS process; (c), (e) and (g) were the intensity distribution of Al, Ta and Ti; (d), (f) and (h) were the intensity distribution of Mo, Co and Cr. 13

(b)

M AN U

SC

(a)

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Fig. 4 Inverse pole figures at three regions on the transverse of the samples produced by the Bridgeman and DWDS processes. (a) DWDS process; (b) Bridgman process.

14

ACCEPTED MANUSCRIPT Hightlights We examined the segregation behavior in the DWDS solidified components.

•

Compared to the Bridgman process, the degree of the segregation was alleviated.

•

The crystallographic orientation was comparatively investigated.

•

Better crystal orientation can be obtained by using the DWDS process.

AC C

EP

TE D

M AN U

SC

RI PT

•