Effect of Sulfur on Creep Strength of Ni-Base Single-Crystal Superalloy, TMS-1700

Yuichiro Joh1,*1, Satoshi Utada1,*2, Makoto Osawa2, Toshiharu Kobayashi2, Tadaharu Yokokawa2, Kyoko Kawagishi2, Shinsuke Suzuki1 and Hiroshi Harada2

1Faculty of Science and Engineering, Waseda University, Tokyo 169–8555, Japan2National Institute for Materials Science (NIMS), Tsukuba 305–0047, Japan

We studied the effect of sulfur addition on a Ni-base single-crystal superalloy, TMS-1700, by performing creep tests at 1100°C/137 MPa using specimens doped with 0, 10, 20, and 100 ppm sulfur. The creep rupture life was found to decrease with increasing sulfur concentration. The scanning electron microscopy (SEM) observation of creep-ruptured specimens revealed the coarsening of the raft structure with increasing sulfur concentration. Their transmission electron microscopy (TEM) observation showed the formation of coarser γ/γ′ interfacial dislocation networks with increasing sulfur concentration. These microstructural differences may be the cause of the shorter creep rupture life observed in alloys with higher sulfur addition.　[doi:10.2320/matertrans.M2016032]

(Received January 28, 2016; Accepted May 31, 2016; Published July 1, 2016)

Keywords:　 nickel-base single-crystal superalloy, creep strength, sulfur, dislocation network

1.　 Introduction

Ni-base single-crystal superalloys have been used for the high-pressure turbine (HPT) components of jet engines and gas turbines. The improvement of the high-temperature prop-erties of Ni-base single-crystal superalloys is desired for im-proving the thermal ef�ciency of the turbine systems. For this purpose, rare metals such as Re and Ru are added to advanced Ni-base single-crystal superalloys.1) However, adding such rare metals increases the life-time cost. For the widespread use of turbine blades made with advanced single-crystal su-peralloys, it is necessary to reduce their life-time cost.

We previously proposed a direct recycling method for used turbine blades.2–4) This method is composed of three steps: collecting scrapped HPT components, remelting and re�ning them, and recasting new components. This process enables us to reproduce genuine HPT components without adding new materials and drastically reduce life-time cost.

However, a few problems must be solved to establish this recycling method. One of them is sulfur contamination origi-nating from the operating environment and/or repair history. In particular, sulfur is con�rmed to be a typical contaminant of scrapped HPT components, and it is important to deter-mine the allowable value of residual sulfur. Therefore, the effects of sulfur on creep strength, which is one of the high temperature properties of Ni-base single-crystal superalloys, must be clari�ed. It is generally accepted that sulfur reduces the high-temperature properties of Ni-base polycrystalline superalloys. Dong et al. intentionally added sulfur to a Ni-base polycrystalline superalloy (Alloy 718) and analyzed its effects on creep strength.5) It was clari�ed that the creep strength decreases with increasing sulfur content. In addition, the segregation of sulfur to grain boundaries causes grain-boundary embrittlement and decreases creep strength.5) While the effect of sulfur on polycrystalline superalloys is clear, its effect on single-crystal superalloys had not been clari�ed until the studies of our group. We intentionally add-

ed sulfur to the Ni-base single-crystal superalloy, TMS-17006), and clari�ed that sulfur reduces the creep strength of single-crystals as well. However, the mechanism of creep-strength decrease remains unknown.

The purpose of the present study is to investigate the mech-anism of the creep-strength decrease due to sulfur addition. We performed creep tests using specimens in which sulfur was intentionally added. The creep-ruptured specimens were observed using scanning electron microscopy (SEM) and transmission electron microscopy (TEM) to investigate the effect of sulfur on their microstructure.

2.　 Experimental Procedure

A �rst-generation Ni-base single-crystal superalloy, TMS-1700, was used as a model alloy. A master ingot of TMS-1700 was supplied by IHI Master Metal Co. Ltd.7) Table 1 lists the nominal chemical compositions of TMS-1700. NiS was sup-plied by Kojundo Chemical Laboratory Co. Ltd.

Approximately 2 kg of TMS-1700 ingots were cut from the master ingot and placed in an Al2O3 crucible. For sulfur-add-ed specimens, NiS was also placed in the crucible to achieve the desired sulfur contents: 10, 20, and 100 ppm by weight. For specimens with 0 ppm added sulfur, melting was per-formed without NiS addition. After this process, the crucible was heated in a directional solidi�cation vacuum furnace evacuated up to 10−2 Pa. The temperature of the molten metal

*1 Graduate Student, Waseda University*2 Corresponding author, E-mail: sat.utada-1992@fuji.waseda.jp

Table 1　Nominal composition of TMS-1700.

Chemical composition, mass%

Ni Bal.

Cr 9.0

Mo 0.6

W 7.6

Al 5.4

Ta 10.0

Hf 0.1

Si 0.04

Materials Transactions, Vol. 57, No. 8 (2016) pp. 1305 to 1308 ©2016 The Japan Institute of Metals and Materials

was measured using a B-type thermocouple. The melt was held at 1600°C for 10 s so that sulfur was homogeneously distributed. Subsequently, the molten metal was poured into a mold with cylindrical cavities having a selector for each, and the mold was pulled out from the heated zone at a speed of 200 mm/h. After the unidirectional casting process, single- crystal bars (11 mm diameter) were obtained. The cast bars were solution heat treated at 1320°C for 5 h, following which it was primary aged at 1150°C for 4 h and secondary aged at 870°C for 20 h, where each heat treatment was followed by air-cooling. The quantity of sulfur was similar to the added amount of sulfur in all specimens.6) In this paper, sulfur con-centrations of the specimens represent the added amount in-stead of the analyzed value.

Creep-test specimens (4-mm diameter and 22-mm gauge length) were fabricated from the cast bars with a longitudinal axis within 6° from the crystal growth orientation of <001> obtained using the X-ray back-re�ection Laue method (Riga-ku DXG-2).

The creep tests were performed under the condition of 1100°C/137 MPa. The temperature in the chamber contain-ing the creep testing machine was measured using an R-type thermocouple �xed on the longitudinal center of the creep specimens. A load was applied on the creep testing machine after the temperature reached 1100 ±  1°C. The strain of the creep specimens was obtained by measuring the displace-ment of gauge length by using CCD cameras during the creep tests.

The microstructures before and after creep tests were ob-served using SEM (JEOL JSM-6060). Furthermore, γ/γ′ in-terfacial dislocation networks were observed using TEM (JEOL TEM-4010). After the creep tests, the specimens were cut into discs perpendicular to the <001> orientation at a po-sition more than 3 mm away from the fracture surface and mechanically thinned to 80 μm using a disc grinder. Finally, the samples were electrochemically thinned by using the twin jet method with a solution of HClO4 (50 mL) and CH3COOH (500 mL) at 12°C and an electrolytic polishing voltage of 30 V.

The dislocation spacing was evaluated using the point counting method from the obtained TEM images. In this method, lines parallel to the [100] or [010] orientation were drawn on each TEM image, and the number of points where these lines cross dislocation networks was counted. The value of dislocation spacing was de�ned as the quotient of the line length and number of interaction points. This operation was performed 30 times for each sulfur-added specimen. Then, the means and standard deviations of the measured values were calculated.

3.　 Results and Discussion

3.1　 Initial microstructureFigure 1 shows the microstructures after heat treatment ob-

tained from the SEM observation. The observation position was the center of the face perpendicular to the <001> orien-tation. All specimens had the γ matrix phase (white part) and γ′ precipitated phase (black part). There was no difference in the size of the precipitated γ′ phase between specimens with and without sulfur addition. However, the specimen with

100-ppm sulfur addition generated plate-like topologically closed pack (TCP) phases with γ′ envelopes. TCP phases are known to precipitate on the {111} plane, which is a slip plane of Ni-base superalloys. Since dislocations easily move through the γ′ envelopes, TCP phase precipitates could cause the creep-strength decrease.8,9)

3.2　 Results of creep testsFigure 2 shows the results of creep tests under the condi-

tion of 1100°C/137 MPa. The creep rupture time tended to decrease with increasing sulfur addition. The rupture time of the specimen with 100-ppm sulfur addition is approximately 50% of that of the specimen without sulfur addition. We �tted the obtained creep curves with the creep constitutive eqs. (1)–(3):

ε = I(A1 + S 2 + S 3), (1)

I = exp[−exp{(−t + µ)/c1}], (2)

S i = Aiexp{(t − λ)/ci} (i = 2, 3), (3)

where ε denotes the creep strain. Ai, ci (i =  1, 2, 3), µ, and λ are all parameters corresponding to the �rst, second, and third creep regions.10) The minimum creep rate for each creep test was calculated by differentiating the �tting equation. The minimum creep rate also tended to increase with increasing

Fig. 1　SEM images of the microstructures of sulfur-added specimens.

Fig. 2　Results of creep tests of specimens with various amounts of added sulfur at 1100°C/137 MPa.

1306 Y. Joh, et al.

sulfur addition (Fig. 3). It was concluded that sulfur addition caused the decrease of creep rupture time and increase of minimum creep rate. This result suggests that sulfur addition causes the abnormality of microstructure.

3.3　 Microstructure after creep testsFigure 4 shows the SEM observation results of microstruc-

tures of the specimens with 0- and 100-ppm sulfur additions after creep tests. The observation position was the center of the parallel face to the <001> orientation 3 mm away from the fracture surface. Both microstructures had a raft structure after the creep tests regardless of the sulfur addition. Howev-er, the morphology of the raft structure is different between the microstructures. Whereas the specimen with 0-ppm sulfur addition had a �ne raft structure with narrower γ′ spacing, the specimen with 100-ppm sulfur addition had a coarse raft structure with wider γ′ spacing. It is generally accepted that a raft structure formed perpendicular to the stress axis prevents the dislocation movement at the γ/γ′ interface in Ni-base sin-gle-crystal superalloys.11) When the raft structure is �ner, the dislocation movement becomes slower. Since the creep defor-mation of TMS-1700 at 1100°C/137 MPa depends on the dis-location movement, this prevention of dislocation movement results in higher creep strength. Therefore, the coarse raft structure caused by sulfur addition is considered as one of the factors decreasing the creep strength.

3.4　 γ/γ′ interfacial dislocation networksFigure 5 shows the TEM observation results of γ/γ′ inter-

facial dislocation networks of all the specimens after creep

tests. These �gures are dark-�eld images at the (001) plane. The black lines show dislocations. All specimens formed γ/γ′ interfacial dislocation networks consisting of octagon units. By using these �gures, we quantitatively measured the dislo-cation spacing to investigate the effect of sulfur addition. The dislocation spacing became wider with increasing sulfur ad-dition (Fig. 6). The difference of the dislocation-spacing val-ue between the specimens with 0- and 100-ppm sulfur addi-tion was approximately 20 nm. It is clear that sulfur addition increases the spacing of the γ/γ′ interfacial dislocation net-work. In addition, the difference of dislocation-spacing value showed a linear relation with the minimum creep rate (Fig. 7). This tendency agrees with the results of a previous study by Zhang on sulfur-free superalloys.11) This seems to be the prin-cipal reason for the shorter creep rupture life of sulfur-added specimens.

3.5　 Mechanism of creep-strength decrease caused by sulfur addition

This chapter considers the reason why sulfur addition coarsens the raft structure and increases the γ/γ′ interfacial dislocation network spacing.

The creep strength of Ni-base superalloy is fundamentally

Fig. 3　Relationship between sulfur addition and minimum creep rate.

Fig. 4　SEM images of the raft structures of sulfur-added specimens after creep tests: (a) 0 ppm; (b) 100 ppm.

Fig. 5　TEM images of the dislocation networks of each sulfur-added spec-imen after creep tests.

Fig. 6　Change of dislocation spacing caused by sulfur addition. The dislo-cation spacing increases with increasing sulfur addition. The line is sim-ply a guide to clarify the tendency of the data. The error bars show stan-dard deviations of the measured values.

1307Effect of Sulfur on Creep Strength of Ni-Base Single-Crystal Superalloy, TMS-1700

enhanced by the following mechanism:The lattice mis�t δ is the lattice-parameter difference be-

tween the γ and γ′ phases and is de�ned by eq. (4):

δ = (aγ − aγ)/aγ, (4)

where a denotes a lattice parameter.The lattice mis�t and elastic mis�t cause an elastic strain

energy at the γ/γ′ interface. A driving force is obtained by the elastic strain energy generated to form a raft structure with the dislocation network. The γ/γ′ interfacial dislocation net-works prevent the movement of dislocations during the creep. When the raft structure becomes �ner, the creep rupture life increases. Therefore, the lattice mis�t is a key factor deter-mining the higher creep strength in Ni-base single-crystal su-peralloys.

As explained above, this study clari�ed that sulfur addition coarsens the raft structure and increases the γ/γ′ interfacial dislocation network spacing. According to Brooks formula, the dislocation spacing is inversely proportional to the abso-lute value of lattice mis�t, as shown in eq. (5):

|δ| = |b|/d, (5)

where b denotes the Burgers vector and d denotes the disloca-tion spacing.12) According to this equation, the disloca-tion-spacing increase caused by sulfur addition implies a de-crease of the absolute value of lattice mis�t. Then, it can be assumed that the following mechanism causes this phenome-non. As mentioned above, the lattice parameter is different between the γ and γ′ phases. The γ phase has an elastic com-pression strain, whereas the γ′ phase has an elastic tensile strain parallel to the γ/γ′ interface. Thus, the γ/γ′ interface is unstable. Furthermore, a previous study reported that sulfur substitutes Ni in Ni-base alloys.13) Therefore, we can assume that the Ni at the γ/γ′ interface is substituted by sulfur to re-lieve interfacial strain because the atomic radius of sulfur is less than that of Ni. In this manner, the absolute value of lat-tice mis�t of sulfur-added alloys becomes less than that of alloys without sulfur addition. The elastic strain energy re-duces with lattice-mis�t relief at the γ/γ′ interface. This inter-

facial stress relaxation should result in the observation of the wider dislocation network spacing. Moreover, the reduction of absolute value of lattice mis�t should contribute to the de-crease in the driving force for forming the raft structure.

4.　 Conclusion

The effect of sulfur on the creep strength of a Ni-base sin-gle-crystal superalloy was clari�ed through creep tests. The following results were obtained:

(1) The creep-rupture life decreases with increasing sulfur addition.

(2) Sulfur addition makes the raft structure rougher and the γ/γ′ interfacial dislocation network larger.

The obtained results can be explained by possible γ/γ′ in-terfacial stress relaxation caused by the substitution of Ni by sulfur at the γ/γ′ interface.

Acknowledgements

The authors are grateful for support from Dr. T. Yuyama at NIMS for creep tests. We would also like to thank Mr. Y. Takebe at Waseda Univ. for technical support in TEM obser-vation. This research was supported by Japan Science and Technology (JST), under the Advanced Low Carbon Technol-ogy Research and Development Program (ALCA) project “Development of direct recycle method for single-crystal tur-bine parts”.

REFERENCES

1) H. Harada: Proc. Int. Gas Turbine Cong. 2003 Tokyo, (Gas Turbine Soc. Japan, 2003) pp. 1–6.

2) T. Kobayashi, T. Yokokawa, H. Harada, Y. Koizumi, M. Sakamoto and M. Osawa: Proc. 40th Annu. Conf. GTSJ, (Gas Turbine Soc. Japan, 2012) pp. 181–184.

3) H. Harada et al., Japan Patent Pending, application number 2015-160748 (2015, 8, 18).

4) S. Utada, Y. Joh, M. Osawa, T. Yokokawa, T. Kobayashi, K. Kawagishi, S. Suzuki and H. Harada: Proc. Int. Gas Turbine Cong. 2015 Tokyo, (Gas Turbine Soc. Japan, 2015) pp. 1039–1043.

5) J.X. Dong, X.S. Xie and R.G. Thompson: Met. Mater. Trans. 31A (2000) 2135–2144.

6) Y. Joh, T. Kobayashi, T. Yokokawa, K. Kawagishi, M. Osawa, S. Suzu-ki and H. Harada: Proc. 10th Liege Conf, Mat. Adv. Power Eng., (Univ. Liege, 2014) pp. 538–544.

7) K. Kawagishi et al., Japan Patent Pending, application number 2014-034720 (2014, 2, 24).

8) T. Yokokawa, Y. Koizumi, T. Kobayashi and H. Harada: J. Japan Inst. Metals 70 (2006) 670–673.

9) Y. Koizumi, H. Koguchi, T. Kobayashi, T. Yokokawa, M. Osawa, T. Kimura and H. Harada: J. Japan Inst. Metals 68 (2004) 54–57.

10) H. Izuno, T. Yokokawa, Y. Koizumi, S. Odaka and H. Harada: J. Japan Inst. Metals 68 (2004) 526–529.

11) J.X. Zhang, J.C. Wang, H. Harada and Y. Koizumi: Acta Mater. 53 (2005) 4623–4633.

12) A. Lasalmonie and J.L. Strudel: Philos. Mag. 32 (1975) 937–949. 13) L. Rivoaland, V. Maurice, P. Josso, M.P. Bacos and P. Marcus: Oxi.

Met. 60 (2003) 137–157.

Fig. 7　Relationship between the dislocation spacing and minimum creep rate of TMS-1700 with various sulfur additions.

1308 Y. Joh, et al.