Oxidation of Metals, VoL 38, Nos. 3/4, 1992

High-Temperature Oxidation Resistance of Sputtered Micro-Grain Superalloy K38G Lou Hanyi,* Wang Fuhui,* Xia Bangjie,* and Zhang Lixin* Received March 5, 1992; revisedMay 20, 1992

The oxidation of sputtered and cast superalloy K38G specimens was studied. The sputtered alloy was microcrystalline, with an average grain size <0.I 12m. The mass gains of the sputtered alloy were much less than those of the cast alloy at 800, 900, and IO00~ up to 500 hr, and were even less than those of pack aluminide on the cast alloy. K38G is a chromia-forming cast nickel-base superalloy, so the oxide scale formed on it is composed of Cr2Os, Ti02, Al203, and a spinel. The oxide scale formed on the sputtered alloy was Al2Os. This scale is thin, compact, and adherent. This result implied that microcrystallization reduced the critical aluminum content necessary to form alumina on the surface of this superalloy. No oxide spallation, as typically observed for cast of aluminized alloys, occurred on the sputtered superalloy. The reduction of the critical aluminum content for the formation of alumina and the improvement of the spallation resistance may be attributed to the microcrystalline structure formed during sputtering. The numerous grain boundaries favor outward aluminum grain-boundary diffusion, provide increased nucleation sites, and reduced stresses in the oxide scales.

KEY WORDS: microcrystal; sputtering; oxidation; superalloy. INTRODUCTION Besides chemical composition, the structure of superalloys can affect their resistance to high-temperature oxidation. Giggins and Pettit 1'2 reported that reduced grain size improved oxidation resistance of a N i - C r alloy. Merz 3 *Corrosion Science Laboratory, Institute of Corrosion and Protection of Metals, Academia Sinica, Shenyang, China. 299 0030-770X/92/1000-0299506.50/0 9 1992 Plenum Publishing Corporation

300

LouetaL

indicated that fine-grain (diameter about 3-4pm) sputter-deposited 304 stainless steel oxidized significantly slower than coarse-grain (diameter about 40 pro) wrought stainless steel at high temperature. Ledjeff e t al. 4 revealed that increasing grain size increased the oxidation rate of an Fe-25Cr-20Ni stainless steel. Wang and Lou 5 reported that the oxidation rate of sputtered Co-30Cr-5A1 microcrystalline coating is much lower than that of the cast alloy in air at 1000~ The structure and homogeneity of superalloys can affect their oxidation resistance. For example, carbides which are generally present in cast superalloys are readily attacked during high-temperature oxidation. It is difficult to change the homogeneity or grain size of cast superalloys by means of hot- or cold-working, and it is also very difficult to change their grain size by heat treatment. Therefore, the oxidation resistance of superalloys is usually improved not by the techniques mentioned above, but by the addition of reactive elements, such as yttrium6 or by various protective coatings. 7 A common problem with coatings is the interdiffusion between the coating and the substrate. This can lead to the possible formation of some brittle phases in the interface between the coating and the substrate. These brittle phases usually reduce the strength and/or the thermal-fatigue resistance of the superalloy components with the aluminide coatings. A homogeneous and micro-grain film of a superaUoy can be prepared by sputter-ion plating (SIP). One of the advantages of SIP is that the-sputtered film has the same chemical composition as the target, because the film is built up from a flux of atoms and ions from solid-source plates. This sputtered film, as a type of coating on the superalloy, may not only enhance the oxidation resistance but also eliminate the detrimental phases formed between an extraneous coating and the substrate during long exposure at high temperatures. The oxidation behavior and the oxidation mechanism of superalloys with normal grain size in the Ni-Cr-A1 system has been investigated. However, oxidation studies of this alloy system in the microcrystalline state is yet to be performed. This work investigated the oxidation resistance of the sputtered microcrystalline coating of a cast superalloy at 800-1000~ in an effort to define the oxidation mechanisms. EXPERIMENTAL PROCEDURES The cast superalloy used in the present study was K38G. Its chemical composition is as follows (weight %): C 0.16

Cr 16.3

Co 8.4

W 2.7

Mo 1.8

Nb 0.76

Ta A1 Ti B 1 . 7 5 4 . 0 1 3 . 8 1 0.01

Ni bal.

Superalloy K38G

301

The alloy was melted in a vacuum induction furnace and cast into a plate 4 mm thick and was then machined into specimens 20 x 10 x 3 (ram). The sputtering target was fabricated in the same manner. The sputtered materials were deposited on the specimens to a thickness of about 30 pm. The sputtering parameters were: sputtering voltage V= 700-750 V, power density 6W/cm 2, argon pressure PAr = 0.13 Pa, substrate temperature T= 300~ In order to compare the oxidation resistance of micro-grain material, cast-alloy specimens were aluminized by pack cementation. Diffusion coatings having a thickness of about 40 pm were obtained at 900~ after 5 hr in an Fe-A1 alloy powder with 1 wt.% NH4C1 as an activator. The specimens for oxidation tests in static air were placed in alumina crucibles in furnace and were kept at 800, 900, and 1000~ for 500 hr. The masses of the specimens were determined together with the crucibles with an accuracy of 0.01 mg. The specimens were analyzed using optical microscopy, X-ray diffraction (XRD), scanning and transmission electron microscopy (SEM, TEM), energy dispersive X-ray analysis (EDX) and electron probe microanalysis (EPMA). RESULTS

Characterization of the AHoy The microstructure of the cast superalloy is rather different from its sputtered film. The cast alloy K38G consists mainly of the matrix 7 phase (nickel solid solution, 52.6wt.%), ~" phase (Ni3(A1, Ti), 46.1 wt.%), and carbides (1.3 wt.%). The carbides include MC distributed evenly throughout and M23C6 located predominantly in the grain boundaries. The grain size of the cast alloy ranges from hundreds to thousands of/~m (see Fig. la). Because the grain size of the sputtered film was very small, TEM was used to reveal the microstructure of sputtered films that were thinned by electropolishing. The average grain size of the sputtered material is less than 0.1 pm.

Fig. 1. Microstructure of (a) cast alloy K38G, (b) TEM bright field image of sputtered film showing micro-grains and microtwins, and (c) SEM fractography of sputtered film showing columnar structure.

302

Lou

et

al.

It may be called nanocrystalline. As shown in Fig. lb, there are a number of defects of microtwins and stacking faults within the grains. Electron diffraction analysis revealed that the sputtered film consisted of only g phase, and no 7' phase and carbides were observed. The ~ had a columnar structure corresponding to the zone-II structure suggested by Movchan and Demchishin. s The structure consists of columnar grains separated by distinct, dense, intercrystalline boundaries.

Oxidation Kinetics It can be seen that the mass gains of sputtered micro-grain alloy are much less than those of the cast alloy at all test temperatures up to 500 hr (Fig. 2). It is interesting to note that the mass gain of the micro-grain alloy was even lower than that of the aluminized specimens on the alloy K38G. The appearance of the specimens after a long period of oxidation indicated that the oxide scales on the cast alloy cracked to some extent and spalled a little at 900~ and severe spallation occurred at 1000~ However, no spallation was seen on the sputtered alloy. These results suggest that the oxide scales on the sputtered microcrystalline alloy are more adherent than those on the cast one.

Scale Morphology and EDX Analysis On the cast alloy, the noncontinuous oxide scale formed after 1 hr at 1000~ and contained a mixture of Cr2Os and TiO2 (Fig. 3a). It can be seen that the Cr203 and TiO2 scales coexist on the specimen oxidized for 20 hr (Fig. 3b). For the sputtered micro-grain alloy, the aluminum peak in the EDX spectrum on the specimen after 20 hr of oxidation is much higher than that after l hr of oxidation, but the nickel and chromium peaks on the former specimen are much lower than those on the latter. This implied that the alumina scales grew rapidly during the early stage of oxidation, but the alumina thickness remained quite thin. Therefore, the nickel and chromium peaks in the EDX spectrum are from the matrix, especially in the case of 1hr oxidation (Figs. 3c and d).

(a}

~

8oo c

5 (b)

," ............ ~~176Z-'"

9oo c

9

' , / . .....~2. . . .

Fig. 2. Curves of mass gain vs. time of cast alloy K38G and its sputtered films oxidized in air at (a) 800~ (b) 900~ (c) 1000~

Superalloy K38G

303

Fig. 3. SEM micrographs with corresponding EDX analyses showing the morphology and average composition of the oxide scales formed on (a, b) cast coarse-grain alloy K38G, and (c, d) its sputtered micro-grain film after oxidation at 1000~ for 1 or 20 hr. Figure 4 shows the morphologies of the oxide scales formed on the specimens oxidized for 100 hr at I000~ It can be seen that the thick and porous scales of Cr203 on the cast alloy show h e a w spallation, and a scale of A1203 was formed at the position where Cr203 has spalled off. On the sputtered micro-grain alloy, the adherent scale A1203 was retained and had very little change as compared with the early stage.

Mierostructure of Oxidized Specimens Obvious differences were observed between the microstructure beneath the oxide scales for the cast alloy and that of the sputtered alloy. For the cast alloy, the external surface scale is thick, continuous, and nonadherent.

304

Loll et aL

Fig. 4. SEM micrographs of morphologies of oxide scales formed on (a, b) cast alloy K38G, and (c, d) its sputtered film after 100 hr oxidation in air at 1000~

Numerous internal oxides were present beneath this scale in the alloy. The EPMA results indicate that the Cr and Ti oxides coexist in the outer layer of the external scale and that A1 oxide predominates in the inner layer. Rod-shaped particles were observed beneath the internal oxides, and were identified as TiN 9 (Fig. 5a). Occasionally, the inner A1203 scale linked up and became a continuous layer (Fig. 5b). In the sputtered alloy a thin,

Fig. 5. Photomicrographs of the oxide scales on the cross-section of (a, b) cast alloy K38G, and (c) its sputtered micro-grain film after oxidation in air at 1000~ for (a, b) 200 hr, and (c) 500 hr.

Superalloy K38G

305

continuous, and compact layer of A1203 formed on the surface, and no internal oxides were observed in the alloy (Fig. 5c). These results suggest that the oxide scales on sputtered alloy are very dense and may inhibit the transportation of oxygen and metal through it. X-ray diffraction analysis on the surface of the specimens after oxidation for 200 hr at 900 and 1000~ showed that the oxides formed on the cast alloy are mainly Cr203 with some TiO2 and little A1203; however, only A1203 formed on the sputtered alloy. These results are in agreement with those obtained by EPMA. DISCUSSION Nickel superalloys may be divided into two groups: "chromia former" and "alumina former. ''1~ The group to which an alloy belongs is determined by its aluminum and chromium contents. Felix H reported that the alloy with a chromium/aluminum ratio (weight %) greater than 4 is a chromia former, and alloys with a ratio less than 4 are alumina formers. The Cr/A1 ratio in alloy K38G is a little higher than 4, so, according to Felix, 1~ it should be a chromia former, although its composition is very near the alumina/chromia boundary. In fact, at 1000~ many regions on the surface showed only internally oxidized aluminum (Fig. 5a), whereas in certain locations, a com, plete A1203 layer formed at the alloy/oxide interface (Fig. 5b). For the micrograin superalloy produced by magnetron sputtering, improvement of the oxidation resistance is due to the formation of alumina instead of chromia and to increased adhesion of the a-A1203 scale. This may be the result of microcrystallization of the sputtered alloy which favors grain-boundary diffusion. In binary Ni-Cr alloys 1'2 and stainless steels, 3 decreases in grain size reduced the oxidation rate. This was attributed to selective oxidation of chromium at grain boundaries. If the grain sizes were less than 10 pm, the lateral diffusion of chromium from grain-boundary regions to the center of grain surfaces may facilitate the formation of a protective Cr203 scale) Grain-boundary diffusion becomes dominant with the microcrystallization of the chromium- and aluminum-containing alloy K38G. The diffusion of the active species along the numerous grain boundaries may dominate the oxidation process. Which of the reactive species, aluminum or chromium, has priority in forming an oxide scale becomes an important consideration. It may be dependent upon the relative concentration of aluminum and chromium and the Gibbs free energy AG ~ of alumina or chromia formation. Although K38G contains more chromium than aluminum, and a Cr203 layer forms on the outer surface of the coarse-grain cast alloy, A1203 forms on the outer surface of the sputtered micrograin alloy. The reasons for this may

306

Lou et aL

be that the standard free energy of alumina formation is much lower than that of chromia, and the mobility (or diffusivity) of aluminum is more rapid than that of chromium at grain boundaries. Since the average grain size of the sputtered material was less than 0.1 pm, a large number of grain boundaries are beneficial to aluminum diffusion from the lattice to the surface and may act as sites for nucleation of alumina. Hence, a continuous, thin A1203 scale may form on the outer surface during the initial stage of oxidation (Fig. 3c). The aluminum concentration of the cast alloy was less than that required for an alumina-forming alloy,~l therefore chromia and TiO2 formed on the surface. It can be seen that microcrystallization, by SIP, decreases the critical aluminum content necessary to form protective alumina films on the surface of alloy K38G. The reasons for the superior adhesion of oxides A1203 on the sputtered alloy may be attributed also to microcrystallization. The grain size of a A1203 formed on the sputtered alloy is very small. Therefore, plastic deformation of the fine-grain oxide scales a-A1203 would be easier than for the coarse-grain ones formed on the aluminized coating. The micrograin superalloy films prepared by magnetron sputtering exhibit superior resistance to high-temperature oxidation compared to a conventional cast alloy with the same composition. This sputtered material is more resistant to oxidation than a conventional aluminized coating (Fig. 2). Therefore, the micrograin films may be applied as a protective coating on the blades of aero-gas turbines. One of the important advantages of such micrograin coatings over diffusion or overlay coatings is that they would have the same composition as the substrate material, thereby eliminating the usual problem of interdiffusion between the coating and the substrate. The formation of possible brittle phases, such as o- phase, beneath the coating is also eliminated.

CONCLUSIONS This work clearly demonstrates that significant improvement can be obtained in the oxidation and spallation resistance of a superalloy by microcrystallization. From the results presented above, we conclude the following: 1. Sputtered films of superalloy K38G consisted of only ),-phase. No 7/' (Ni3A1) phase and carbides were observed. The average grain size of the film alloy was less than 0.1/~m, and the sputtered alloy microstructure was microcrystalline. 2. The sputtered superalloy K38G film exhibited superior resistance to high-temperature oxidation in comparison to that of a conventional

Superalloy K38G

307

cast alloy of the same composition. Its oxidation resistance was even better than that of an aluminized coating on the cast alloy. 3. Chromia-rich external scale formed on cast alloy K38G, while sputtered alloy films formed a protective alumina scale. This implies that microcrystallization altered the oxide-scale-formation mechanism. 4. Oxide scales of Cr203 and TiO2 formed on cast alloy K38G may spall, but the a-A1203 scale formed on the sputtered film displayed excellent adhesion during oxidation tests. ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundation of China under Contract 59071025. The authors wish to thank Professor Wu Weitao for many helpful discussions. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.

C. S. Giggins and F. S. Pettit, Trans. TMS-AIME 245, 2495 (1969). C. S. Giggins and F. S. Pettit, Trans. TMS-AIME 245, 2509 (1969). M. D. Merz, Met. Trans. 10A, 71 (1979). K. Ledjeff, A. Rahmel, and M. Schore, Oxid. Met. 15, 485 (1981). F. Wang and H. Lou, Mater. Sei. Eng. A129, 279 (1990). D. P. Whittle and J. Stringer, Phil. Trans. Royal Soc., London Ser. A, 295, 309 (1980). G. W. Goward, Mater. Sei. Technol. 2, 194 (1986). J.A. Thornton, in Deposition Technologies for Films and Coatings, Developments and Applications, R. F. Bunshan, ed. (New Jersey, 1982), p. 170. 9. M. J. Bennett, A. T. Tuson, C. F. Knights, and C. F. Ayres, Mater. Sci. Technol. 5, 841 (1989). 10. D. P. Whittle, in High Temperature Alloys for Gas Turbines, D. Coutsouradis et al., eds. (Appl. Sci. Pub., London, 1978), p. 109. 11. P. Felix, in Deposition and Corrosion in Gas Turbines, A. B. Hart and A. J. B. Cutler, eds. (Appl. Sci. Publ, London, 1972), p. 331.