Grain Boundary Embrittlement of the Iron-Base Superalloy IN903A R. H. B R I C K N E L L A N D D. A. W O O D F O R D It is shown that a low coefficient of expansion, iron-base superalloy, IN903A, suffers severe tensile embrittlement following high temperature air exposure at 1000 ~ This embrittlement involves a transition to intergranular failure at low strains, with no reduction in yield strength, and is manifested in the room temperature to 800 ~ range. In parallel with earlier observations on nickel-base superalloys, ductility is regained at 1000 ~ However, in contrast to these earlier results, air exposure enhances rather than hinders grain growth in the near surface regions, and, in addition, suppresses the occurrence of the jerky flow seen in vacuum-exposed material. Oxygen is demonstrated to be the damaging species, and it is shoran that boundaries are embrittled far ahead of any matrix internal oxidation. Small additions of boron are successful in eliminating the embrittlement, as they were in nickel-base alloys. The results of stress rupture tests are then reviewed, and it is concluded that the rapid failures which occur on air testing are a consequence of embrittled grain boundaries f~/iling in tension, rather than the stress accelerated grain boundary oxidation mechanism previously proposed.

1. I N T R O D U C T I O N L o w coefficient of thermal expansion alloys based on the system Fe-Ni-Co are attractive for certain aircraft engine applications, as they can reduce the spacing between stator and rotor. The first two such alloys developed were IN903 (Huntington Alloys, International Nickel Co.) and CTX 1 (Carpenter Technology), and have had their basic properties outlined in publications by Carpenter ~and Muzyka e t a l , 2 respectively. Their chemistries are very similar and based on an approximate composition of Fe-38Ni- 15Co with Nb, AI, and Ti additions to provide strengthening precipitates. An early indication of the extreme sensitivity of the mechanical properties of these alloys to the environment was given by the tensile tests of Carpenter.~ In these tests, a dramatic decrease in elongation was noted at elevated temperatures when testing was conducted in air as opposed to vacuum. It was subsequently discovered 3,4 that these alloys were also greatly sensitive to the environment on stress rupture, and particularly notched stress rupture testing, and that very limited lives were obtained on air testing in the range of operational interest, 540 to 650 ~ As the failures occurred intergranularly, and as the parted grain boundaries contained oxide wedges, it was concluded that these failures occurred by the penetration of oxide down grain boundaries, aided by the applied stress. Hence, this phenomenon was dubbed "Stress Accelerated Grain Boundary Oxidation" (SAGBO). Certain approaches had shown themselves to be capable of alleviating this problem. Firstly, additions of 1 pct Hf to CTX1 greatly improved the air notched R. H. BRICKNELLand D. A. WOODFORD are Staff Members, Metallurgy Laboratory, General Electric Corporate Research and DevelopmentCenter, Schenectady,NY 12301. Manuscript submitted December 30, 1980.

stress-rupture properties, 3 and secondly, removal of AI from IN903 produced the same effect. To compensate for the loss in strengthening precipitate, this new alloy (IN903A) has its N b + Ta level augmented to approximately 5 pct. 4 In addition, the thermomechanical history of the alloy may also play an important role. In the past decade, studies have been carried out on the embrittlement of nickel-base superalloys 5-~~and pure nickel t~ following high-temperature air exposure. This embrittlement has been shown 9,t2 to be due to oxygen diffusing down grain boundaries and is not connected with any continuous oxide formation. In addition, it has been noted in both tensile and creep rupture studies. This form of embrittlement and the poor stress rupture properties seen in the low CoE alloys have several points in common: 1) both require exposure in air; 2) both are manifested by a loss in mechanical properties in the mid-temperature range; 3) both are typified by a transition from transgranular to intergranular failure; and 4) both can be greatly ameliorated by Hf additions. These similarities suggest that the two embrittling mechanisms may be closely linked, and the current study seeks to explore these links by investigating whether IN903A suffers a loss in tensile ductility following high temperature air exposure. In addition, this study should also provide useful clues as to the connection between any embrittlement seen in this iron-base alloy, and that seen in nickel and the nickel-base superalloys. 2. E X P E R I M E N T A L Experiments were conducted on material machined from a forging of IN903A supplied by Huntington Alloys, and of nominal composition given in Table I. As prior embrittling treatments on the nickel-base superalloys had been carried out at 1000 ~ it was decided to explore the oxidation characteristics of IN903A in an

ISSN 0360-2133/81/0911-1673500.75/0 METALLURGICALTRANSACTIONSA 9 1981 AMERICAN SOCIETY FOR METALS AND VOLUME 12A, SEPTEMBER 1981--1673 THE M E T A L L U R G I C A L SOCIETY OF AIME

Table I. Nominal Composition of the IN903A Forging Studied (wt%) Fe Ni Co Nb + Ta Ti A1 C

bal. 37.5 14.0 5.0 1.6 0.02 0.1

unstressed condition at various temperatures to determine a suitable preexposure. Exposures were made in flowing air for 100 h. The microstructures were subsequently examined metallographically and by microprobe analysis. For the tensile embrittlement study, specimens of 1.14 cm effective gage length, and 0.25 cm gage diam were used. After degreasing, these were exposed in air for 100 h at 1000 ~ and then furnace cooled. This treatment was selected on the basis of the oxidation study. To remove any effects of thermal aging on the subsequent tensile properties, further samples were given identical vacuum exposures. These were encapsulated in quartz tubes, baked out at 350 ~ in a dynamic vacuum of 10 -5 torr, and then sealed off. Vacuum and air exposures were carried out simultaneously in the same furnace. Tensile tests were conducted at a crosshead speed of 0.05 c m / m i n (~ ~ 7 • 10 -4 S - l ) in vacuum at a variety of temperatures from room temperature to 1000 ~ The friable external oxide was removed from the gage sections of air-exposed samples before measurement and testing. Following testing, the fracture surfaces were examined by scanning electron microscopy, and samples prepared for optical metallography.

3. RESULTS 3.1 Oxidation Behavior Figure 1 shows the alloy after oxidation for 100 h at 1000 ~ Only small amounts of the friable external oxide are retained, but extensive internal oxidation can be seen extending to a depth of approximately 140/~m. There is preferential oxide penetration at grain boundaries to a further 50 tzm. X-ray analysis showed the external oxide to consist chiefly of iron, but that the internal oxides were niobium rich. From recent work on the internal oxidation of Ni-Nb alloys at 850 ~ it is assumed that these oxides are Nb205. This result established that it was feasible to expose samples of IN903A for 100 h at 1000 ~ without losing too much specimen material to external or internal oxide regions. It also demonstrated that no protective oxide layer formed which would hinder the inward diffusion of gaseous species involved in any embrittling reaction. Hence, as this was the temperature chosen for prior exposures in the nickel-base superalloys 5-10and pure nickel, 1~ it was again selected for the air and vacuum 1 6 7 4 - - V O L U M E 12A, S E P T E M B E R 1981

Fig. I - - T h e internal oxide formed in IN903A after 100 h exposure in air at 1000 ~ Most of the external oxide has been lost.

exposures of this iron-base superalloy, before tensile testing at a variety of temperatures. 3.2 Grain Growth During High-Temperature Exposure It has been noted previously that oxygen penetration during high temperature air exposure of pure metals pins the grain boundaries in near surface regions and prevents thermal grain growth. This effect has been reported in copper, ~4a5silver,~6,~vand nickel," and has been linked to the concurrent embrittlement." To investigate any grain pinning effects, samples were exposed in air and vacuum at 1000 ~ for 100 h. The starting grain size of about 50 # m increased to around 125 t~m after the vacuum exposure, but this grain growth was actually enhanced by the air exposure, and here a grain size of 300/~m was seen throughout the sample. To further explore this effect, a coupon was exposed for 10 h at 1000 ~ see Fig. 2. The dramatic effect of the air exposure on grain growth can again be seen, with 300/~m grains within two grain diameters of the surface, and then 50 # m grains beyond. Solute pinning of grain boundaries is well known, and the pinning effect of oxygen on grain boundaries of copper, silver, and nickel previously noted. Here, it would appear that we are dealing with a reverse phenomenon, that of the loss of a pinning element by a surface reaction. One possibility would be carbon, reacting with the oxygen and being lost as carbon dioxide or monoxide. The loss of carbon from this alloy is also indicated by some mechanical property measurements described below. 3.3 Effect of High Temperature Air Exposure on Tensile Properties The total elongation and percentage reduction of area of the vacuum and air-exposed samples over the range of temperatures tested are shown in Fig. 3(a) and (b), respectively. The degradation in ductility caused by air exposure at all testing temperatures below 1000 ~ is immediately apparent, with both elongation and pct RA showing similar trends. In parallel to the situation in METALLURGICAL TRANSACTIONS A

(a) Fig. 2 - - E n h a n c e d grain growth in the near surface region in a sample exposed for 10 h in air at 1000 ~

air-exposed nickel l~ and a range of nickel-base superalloys, 5-1~a tensile ductility minimum is observed at intermediate temperatures, here at 600 ~ However, in contrast to the earlier results, only a slight recovery in properties occurs at the lower temperatures. The prior exposures have had little effect on the yield stresses (Fig. 4), or the ultimate tensile strengths. The air samples appear slightly stronger, presumably due to internal oxidation, but have slightly lower ultimate tensile strengths, presumably due to lower total elongations. For both exposures, the dramatic loss in strength above 600 ~ reflects the resolution of the ~//y" strengthening precipitate. Examination of the load-time curves revealed that serrated yielding, the Portevin-LeChatelier Effect, occurs over a certain temperature range for the vacuumexposed samples. This effect is just noticeable at 400 ~ very prominent at 600 ~ and still observable at 800 ~ Examples of the load-time curves for air and vacuum-exposed samples at 600 ~ are shown in Fig. 5. The air-exposed samples show conventional load-time curves at all temperatures except this, where slight serrated yielding is seen just before final failure.

(b) Fig. 3 - - T h e tensile ductility of IN903A as a function of temperature after prior exposures for 100 h at 1000 ~ in air and vacuum as measured by: (a) total elongation; (b) percentage reduction in area.

3.4 Optical Metallography The samples tested after vacuum exposure were essentially similar at all test temperatures, but showed greater ductility and tendency to voiding as the test temperature increased. No surface or internal wedgeshaped cracks were observed. An example of the microstructures observed is given in Fig. 6(a) for the 800 ~ test. The large cavities formed on testing can clearly be noted; the smaller features are carbide precipitates. The microstructure of the air-exposed sample after testing at 800 ~ is shown in Fig. 6(b), and can be directly compared to the vacuum-exposed sample in Fig. 6(a), their pct RA being 19 and 68, respectively. All the air-exposed samples showed extensive surface cracks emanating from grain boundaries in the internally oxidized region, and penetrating far beyond the META LLURGICAL TRANSACTIONS A

Fig.4--The 0.2 pct yield strengths as a function of temperature for the air and vacuum-exposed samples. V O L U M E 12A, SEPTEMBER 1981-- 1675

Fig. 5--Load-time curves for the air and vacuum-exposed samples at 600 ~ showing the pronounced serrated yidding observed after vacuum exposure.

(a)

continuous grain boundary oxide. However, in the sample tested at 1000 ~ the underlying material displayed sufficient ductility to blunt these cracks, which propagated rapidly to failure in the other sampies. In addition, these samples also showed internal grain boundary cracking. This is particularly evident in Fig. 6(b) where wedge-shaped cracks can be seen at many triple points. 3.5 Fractography The fracture surface of the vacuum-exposed sample tested at room temperature is shown in Fig. 7, and can be seen to involve a transgranular failure at approximately 45 ~ to the tensile axis. The fracture of the vacuum-exposed samples remained transgranular at higher temperatures, but became more ductile in nature and displayed failures in regions of heavy void formation. The air-exposed samples showed a brittle, intergranular failure at room temperature, and this fracture mode persisted with increasing temperature. At 800 ~ the failure was still totally intergranular, Fig. 8. However, at 1000 ~ the air-exposed sample became ductile and necked down to failure at a point.

(b) Fig. 6--The microstructure of the samples tested at 800 ~ uum exposed; (b) air exposed.

(a) vac-

4. DISCUSSION 4.1 Jerky Flow As reported above, jerky flow (the Portevin-LeChatelier Effect) was observed in IN903A on testing in the mid-temperature range. The full results, for both air and furnace-cooled samples after both air and vacuum exposures, are available elsewhere. ~8However, one fact of pertinence to the current work emerges, and that is the almost complete suppression of the phenomenon in air-exposed samples. Jerky flow may be the result of either substitutional or interstitial solute, but in nickel alloys jerky flow is not normally observed in the absence of interstitials such as C or N. 19The temperature range for jerky flow in the present study is higher than that for pure Ni or Fe, but Co is known to increase the temperature range in Ni-Co alloys,2~and here, there 1676--VOLUME 12A, SEPTEMBER 1981

Fig. 7----Transgranular failure in vacuum-exposed material tested at room temperature. METALLURGICAL TRANSACTIONS A

Fig. 8--Intergranular failure in air-exposed material at 800 ~

are several alloying additions which could increase this temperature range, Hence, if carbon is considered to be the controlling species, then the absence of jerky flow in air-exposed samples can be explained by the loss of carbon from solution at the metal surface as carbon monoxide. This loss of carbon could also explain the enhanced grain growth seen in air-exposed samples if the boundaries were pinned either by carbon in solution, or by boundary carbides which have redissolved. Chemical analysis of vacuum-exposed IN903A gave an average carbon content of 96 ppm, compared with a value of 83 ppm in air-exposed material. Whether or not this represents a significant change in the carbon in solution in the matrix, as opposed to that combined in carbides, is not clear. Interpretation is further hindered by the uneven carbide distribution observed in the samples.

800 ~ and gave nearly identical elongations and pct RA as did the full-sized samples, see Fig. 9. An indication that this might be the case came from the copious wedge cracking previously observed in the centers of air-exposed samples, and unrelated to any surface cracks. Hence, the air exposure has embrittled the samples to a depth far beyond that evinced by visible oxidation. This embrittlement, in contrast, say, to liquid metal embrittlement, has led to no loss in yield strength, which, in fact, is increased slightly, presumably due to the internal oxidation. The ultimate tensile strengths were slightly less after air exposure, but this is a consequence of the earlier failures allowing less work hardening. This form of embrittlement after prior air exposure, involving no loss of strength but a transition to intergranular failure, has been observed in a number of nickel-base alloys as was described above. One further characteristic of the embrittlement in nickel-base alloys is the recovery of tensile ductility at high temperatures, and its partial recovery at low temperatures. The current alloy parallels this behavior at high temperatures, but shows little recovery at low temperatures, and thus, the results are difficult to reconcile with the theory advanced 1L~2to explain the form of embrittlement previously seen. This theory rested on grain boundary pinning by oxygen penetration, and led to severe embrittlement when grain boundary sliding was a prominent deformation mode. Obviously, this is not expected to be the case in IN903A at the low tempertures. A further complication arises from the fact that air exposure of the current alloy enhances grain boundary mobility and grain growth, rather than suppressing them as had previously been observed. Hence, the theory advanced to explain the form of embrittlement seen in the nickel-base alloys does not appear apposite in this case. The damaging species identified in the nickel-base alloys was oxygen, and this too is the responsible element in the current study. This was confirmed by exposing a sample at 1000 ~ in dried, pure oxygen and

4.2 Embrittlement The immediate and striking result of this investigation is the severe degradation of tensile properties caused by prior air exposure of IN903A. This exposure also results in a change in fracture mode from transgranular to intergranular. As the intergranular failure is initiated in the internally oxidized region, where the boundaries contain a large fraction of niobium oxide, the question arises as to whether this forces the rest of the sample similarly to fail in this brittle, intergranular manner. To explore this, a further series of specimens was air exposed as before, but these were then machined down to half their starting gage diameter prior to testing. This involved the removal of 0.65 mm of material, compared to a combined depth of internal and external oxide of about 0.4 mm, and left material with no visible damage from the air exposure. However, upon testing, the samples still failed in a brittle, intergranular manner from room temperature to METALLURGICAL TRANSACTIONS A

Fig. 9--Ductility of the specimens machined down after air exposure compared to the full-sized samples. VOLUME 12A, SEPTEMBER 1981--1677

subsequently testing at 800 ~ Similar embrittlement as obtained after air exposure was seen. The manner in which oxygen prompts grain boundary embrittlement in the room temperature to 800 ~ range is not clear, but this embrittlement to low temperatures parallels that seen in a cobalt-base alloy, 2~ which will be reported subsequently. The samples in the current study were tested after furnace cooling, which produced a relatively high strength level. In order to evaluate the effect of alloy strength on the observed embrittlement, further samples were air and vacuum exposed for 100 h at 1000 ~ and then air quenched. This reduced the yield strengths by more than a factor of two when compared to the furnace-cooled samples, except for specimens tested at 800 ~ and above, where precipitate resolutioning occurred in all samples. Although the ductilities obtained were greater for both air and vacuum-exposed samples to 800 ~ where the furnace-cooled samples had softened, the air-exposed samples still showed a significant loss in ductility and a transition to intergranualr fracture. Hence, the intrinsic ductility of the material appears to play little part in the embrittling process. An example for the 600 ~ tests is shown in Fig. 10 where the tensile elongation of furnace-cooled and airquenched materials are compared. The relative yield strengths of the vacuum-exposed material are 558 and 234 MPa. As studies in the nickel-base superalloys had shown that boron additions could prevent subsequent air embrittlement, s-l~ and as studies in a model system 22 have shown that boron can block the diffusion of oxygen along nickel grain boundaries, it was decided to explore whether a similar effect could be obtained with this iron-base alloy. Hence, a casting of INg03A with the addition of 0.1 wt pct B was made, and given the standard heat treatment. The composition of this alloy and details of its heat treatment are given in Table II. Tensile samples were machined from this casting, and exposed in air or vacuum and tested as before. The air-exposed samples showed similar external and internal oxidation as did the boron-free samples. The ductilities obtained from the air and vacuum-exposed samples are shown in Fig. 11, and indicate little effect of exposure environment. Hence, as in the nickel-base alloys, boron appears to prevent grain boundary embrittlement. This is believed to be the first reported example of an iron-base alloy embrittling by air exposure; whether this phenomenon will be as general as it appears to be in nickel-base alloys and whether boron will always be as effective remains to be determined. It is now instructive to apply these results to a consideration of the mechanism leading to the rapid failures seen in notched stress rupture testing in the 540 to 650 ~ range. Currently, these are attributed to the phenomenon of stress accelerated grain boundary oxidation, 2,3 whose mechanism is represented in the upper half of Fig. 12. Initially, an oxide "wedge" is preferentially formed down the grain boundary a) this oxide then cracks and parts in tension b) the oxide then further penetrates down the boundary c) and the process is repeated. However, on the basis of the above 1678--VOLUME 12A, SEPTEMBER 1981

INgO3A EXPOSEDIOOhrs IO00*C

% ELONG

600'C TEST

60 --

AiR QUENCHED 02YS:234MPo

50 --

,///

--

/ / / /

/ / / ,

40 --

". VAC/ , / / . ....

FURNACE COOLED

30 --

-

z / x /

02YS = 558MP0

///,

1 1 / ,

zo -

V//,

io

Y / // // "/

-

o

Fig. 10---Comparison of embrittlement of 600 ~ for furnace-cooled and air-quenched material, showing the effect of changing strength level. I

I00 9C

I

INgO3A +

I

I

I

0.t 8

o

I00 Hrs, LO00*C

8C 0

70 60 5C 4(3

za

~

IC O0

I 200

I t 400 600 TEST TEMPERATURE

I 800

I I000

Fig. 11--Tensile ductility v s . test temperature curve after vacuum and for the boron-modified material.

air exposures of 100 h at 1000 ~

T a b l e II. A. C o m p o s i t i o n of the B o r o n - C o n t a i n i n g Alloy

(Weight Percent) Fe balance

Ni

Co

Nb

B

C

37.5

14.0

5.0

0.1

0.03

B. Heat T r e a t m e n t

843 ~ 1 h 718 ~ 8 h, F.C. 55~ to 621 ~ 8 h, air quench.

results, it is possible to offer an alternative mechanism based on oxygen embrittlement, which is shown in the lower portion of Fig. 12. Here the boundary is embrittled by oxygen penetration in the near surface region a) this embrittled boundary then fails in tension b) and the free surfaces thus produced are oxidized METALLURGICAL TRANSACTIONS A

( -r (a)

(b)

(c)

STRESS ACCELERATED GRAIN BOUNDARY OXIDATION

I~ (a)

I

(b)

(c)

OXYGEN EMBRITTLEMENT Fig. 12--A comparison of the SAGBO and oxygen embrittlement mechanisms of crack propagation. For description see text.

c) while oxygen diffuses down the boundary at the head of the crack embrittling the next segment and the process is repeated. Although the temperatures employed for rupture testing are considerably lower than that used to produce the bulk embrittlement, we are concerned only with the embrittlement of the area at the head of the crack, and this can proceed rapidly even at the lower temperatures. If one assumes that the distance, x, penetrated into the specimen along the grain boundary network is given by x = c o n s t . ~v/e-Q/Rr X where Q is the activation energy for grain boundary diffusion, and t is the time, then one can calculate the distance embrittled at any temperature from the knowledge that the center of a 0.25 cm diam bar was embrittled after 100 h at 1000 ~ If Q g b is taken as 37 kcal/mol (i.e., one-half the bulk value, 23) then at 650 ~ an additional area of grain boundary greater than 1/~m deep will be embrittled in one minute, and if this is assumed to be the interval between crack advancements, rapid rates of crack growth can be postulated at these operating temperatures. Furthermore, the bulk embrittlement was achieved by permeation through an oxide film, while in the incremental process where tensile stress is applied, it cali~zcur by rapid absorption at the bare metal surfaces cr~m~ed. Support for this mechanism.over that of SAGBO comes from the severe oxygen embrittlement of grain boundaries displayed in the temperature range of stress rupture testing and reported above; from the total lack of any preferential grain boundary oxide formation at these temperatures in the absence of stress; ~gand from some results on boron-containing alloys. It was reported above that IN903A containing 0.1 wt pct B was immune to grain boundary embrittlement in air, and this alloy also displays excellent notched stress rupture properties in air in the 540 to 650 ~ range. 18An alloy of similar composition, Udimet LX, 24which was developed to provide superior castability, was also shown to have benefited similarly in air rupture tests. As these small boron additions produced no noticeable change in the METALLURGICAL TRANSACTIONS A

oxidation characteristics of the alloys, it can be concluded that the enhancement of the stress rupture properties is a direct result of the suppression of grain boundary embrittlement by the boron. One final point of interest is the possible relevance of this oxygen embrittlement mechanism to certain observations in fatigue tests. Oxide wedges along grain boundaries have been observed following high temperature air testing of a number of materials, and cracking of these oxides has been suggested to be the crack nucleation mechanism in both Udimet 50025 and U d i m e t 700. 26,27 It is well-known that many nickel base superalloys suffer grain boundary embrittlement in air: 10and it is the current authors' belief that the oxygen embrittlement process described above can be equally pertinent in fatigue. 5. CONCLUSIONS 1) The iron-base superalloy IN903A is shown to be embrittled by high-temperature air exposure at 1000 ~ This embrittlement is manifested in a loss in tensile ductility and a transition to intergranular failure. In parallel with previous observations on nickel-base superalloys, ductility is regained at the highest test temperatures. 2) Oxygen is the responsible species and damages grain boundaries to a depth far greater than that indicated by matrix internal oxidation. Again, in parallel with the studies on nickel-base alloys, small additions of boron prevent this embrittlement. 3) Rapid failures on stress rupture testing in air are re-interpreted in terms of this grain boundary embrittlement. Intergranular fracture results from embrittled boundaries failing in tension and then being oxidized, rather than from oxide wedging due to the SAGBO mechanism. ACKNOWLEDGMENTS The authors wish to thank P. Dupree, C. Palmer, J. Methe, and C. Robertson for valuable experimental assistance. REFERENCES 1. H. W. Carpenter, Met. Prog., 1976, vol. |10, pp. 25-29. 2. D. R. Muzyka, C. R. Whitney and D. K. Schlosser, J. Met., 1975, vol. 27, pp. 11-15. 3. MATE Report submitted to NASA, Lewis Research Center, vol. 1, pp. 128-68, General Electric Co., Aircraft Engine Group, Evendale, OH, 1977. 4. D. F. Smith, E. F. Chatworthy, D. G. Tipton, and W. L. Mankins, "'Superalloys 1980,'" Proc. 4th Int. Conf. on Superalloys, Seven Springs, Pa., pp. 521-30, ASM, Metals Park, OH, 1980. 5. E. Ross, General Electric Co., Evendale, OH, unpublished researck, 1968. 6. W. H. Chang, "'Superalloys-Processing, '" Proc. 2nd Int. Conf. o n Superalloys, Seven Springs, Pa., Section V, AIME, 1972. 7. J. H. Wood, General Electric Co., Schenectady, NY, unpublished research, 1973-80. 8. D.A. Woodford, Int. Conf. on Engineering Aspects of Creep, Univ. of Sheffield, England, 1980, I. Mech. E., London, 1981, Paper 55. 9. D.A. Woodford, Metall. Trans. A, 1981, vol. 12A, pp. 299-308. 10. D. A. Woodford and R. H. Bricknell, "'Superalloys 1980, "" Proc. 4th Int. Conf. on Superalloys, Seven Springs, Pa., pp. 633-41, ASM, Metals Park, OH, 1980. VOLUME 12A, SEPTEMBER 1981--1679

11. R. H. Bricknell and D. A. Woodford, Metall. Trans.A, 1981, vol. 12A, pp. 425-33. 12. R. H. Bricknell and D. A. Woodford, General Electric Co., Schenectady, NY, Report No. 80CRD164, 1980. 13. T. Yoneoka, M. Yamawaki, and M. Kanno, J. Jpn. Inst. Met., 1979, vol. 43, pp. 1144-50. 14. L. L. Wyman, Trans. AIME, 1934, vol. 104, pp. 205-17. 15. D. L. Wood, J. Met., April 1957, vol. 9, pp. 406-08. 16. D. L. Martin and E. R. Parker, Trans. AIME, 1943, vol. 152, pp. 269-77. 17. J. C. Chaston, J. Inst. Met, 1945, vol. 71, pp. 23-35. 18. R. H. Bricknell and D. A. Woodford, General Electric Co., Schenectady, NY, 1980, Report No. 80CRD268. 19. J. S. Blakemore, Metall. Trans., 1970, vol. 1, pp. 1281-85.

1680--VOLUME 12A, SEPTEMBER 1981

20. D. E. Sonon and G. V. Smith, Trans. TMS-AIME, 1968, vol. 242, pp. 1527-33. 21. D. A. Woodford and R. H. Bricknell, Metall. Trans. A , in press. 22. D. A. Woodford and R. H. Bricknell, Metall. Trans. A , 1981, vol. 12A, p. 1467. 23. R. Barlow and P. J. Grundy, J. Mater. Sci., 1969, vol. 4, pp. 797-801. 24. U.K. Patent Application, GB2009787A, 1978. 25. C. J. McMahon and L. F. Coffin, Metall. Trans., 1970, vol. 1, pp. 3443-50. 26. G. F. Paskiet, D. H. Boone and C. P. Sullivan, J. Inst. Met., 1972, vol. 100, pp. 58~62. 27. C. J. McMahon, Mater. Sci. Eng., 1974, vol. 13, pp. 295-97.

METALLURGICAL TRANSACTIONS A