THE STRUCTURE OF RENE’ 88 DT

S.T. Wlodek*, M. Kelly** and D. A. Alden*** This study was performed at the

Engineering Materials Technology Laboratory GE Aircraft Engines

Cincinnati, OH 45215.

Abstract

As heat treated, Rene’ 88 DT was found to contain some 42.5 % y ’ , with both the cooling ( 24.5% ) and aging ( 18.0% ) forms exhibiting a very low positive ( 0.05% ) mismatch to the matrix. Small amounts of gram boundary MsB2 TiNblCJ, and traces of MsBs,were also present. The rate of coohng from the solutioning temperature could be related to the resultant cooling y ’ diameter by a regression equation, as could the aging conditions. The rate of cooling, as well as the grain size, were found to effect the gram boundary sermtions that were formed on cooling from the super - solvus annealing temperature. Both of these features, and thus the cooling rate, mav effect mechanical oronerties. On nrolonaed hiah tempemture’ exposure, the y ’ phases 1 grow, and more M&z precioitates above 650°C ( 1200°F 1. followed bv MT&L above 705 ‘“C ( i300°F ), and an inbagranul~ p phase at ?6O”C ( 1400°F ). Simple regression equations were fitted to the kinetics of growth of both the cooling and aging y ’ by the use of a Larsen-Miller parameter to normalize the-time and temperature of exposure. This abowed the mediction of the growth of the Y ’ Dhases and the possible iden&ication of the average service tempe&ure that a part experienced. The growth of y ’ during high temperature exposure, rather then any precipitation of deleterious phases, was found to limit the service range of this structurally stable alloy. Indeed, the formation of M&e and n start only after the aging y’ had comnletelv re - solutioned. Re& 8SDT is an extremely stable alloy for prolonged 650°C (1200°F ) service. Its sunerior orooerties reflect a hieh Y’ content and a y “whose low positive mismatch allows a highly coherent and finer y’ precipitation. These advantages are optimized by a super solvus annealing practice that eliminates large, sub-solvus y’ phases, which promote crack nucleation in fatigue, and interfere with the formation of serrated grain boundaries.

Introduction

Rene’ 88 DT was developed(t) to be more damage tolerant than Rene’ 95. hence the DT desienation. while offering imoroved creeo

Y ,

strength and fatigue crack growth resistance@). The-nickel base chemistry can be given as: 13% Co, 16% Cr, 4% MO, 4% W, 2.1% Al, 3.7 % Ti, 0.7% Nb, 0.03% C and 0.015% B. The production allov is alwavs orocessed through the powder metallurgy route:The Stan&d heat treatment con&ts of: a-super - solvus solution of 1.0 hr at 1150°C ( 2100“F ). followed bv a delayed oil quench, and aging for 8.0 hr at 760°C ( 14OO’F ).-Its main structural characteristics are thus a fine grain size, achieved through PM consolidation, and a duplex distribution of y ‘, the coarser forming on cooling from the super solvus solution, the finer predominantly on aging. Rene’ 88 DT is used in disk applications in advanced General Electric Co. engines.

Due to the super-solvus nature of me solunon anneal, the microstructure of Rene’ 88 DT contains no sub-solvus y ‘, which is usually termed prima#3). Normal convention refers to the cooling y’ as secondary, and the y’ that forms predominantly on aging, as tertiary. That terminology will be used here. This study completely characterized the structure of Rene’ 88, including the effect of heat treatment and prolonged exposure on the structure of this alloy. Due to the well documented effect of y ’ size(b) , and the possible effect of grain boundary morphologyc5) on properties, particular attention was paid to characterizing the effect of heat treatment on these features.

Experimental

Most of the structuml studies reported here were performed on a full size production disc forging produced through the conventional PM route, with an average grain diameter of 60 pm. In order to ascertain the role of grain size on cooling rate effects, some limited work was also performed on a cast version of the alloy. This was produced by extruding a 43 cm. VAR ingot to a 15 cm diameter billet. The VAR material exhibited an average gram diameter of 130pm. The chemistries of both materials are given in Table 1. Both chemistries were similar, and well within specification limits. Conventional metallographic, x-ray, and analytical electron microscopy procedures, as used in similar studies(3). were employed-Ali the image analyses and grain boundary morphology studies were performed on SEM images obtained bv examining _I ~-a unmounted, electropolished, specimens in a field emission, scanning, electron microscope. As such morphological studies required very close control of specimen preparation t6), they will be described in greater detail. Measurements of y ’ were made on surfaces electropolished in cryogenic ( -40 to -50 “C ) 80% CHsOH - 20% HC104 at 25 volts, for about 25 seconds, followed by immediate etching performed at 5 volts for about 3 - 5 seconds, in the same electrolyte and at the same temperature. This procedure produced almost a mirror level of flatness, with just enough relief to allow easy phase discrimination at the large magnifications that must be used to resolve the sub - micron y ‘. Nevertheless, as the tertiary y ’ is as small as 0.01 nm, a truly planar surface cannot be achieved on that scale. The observed y’ phases thus would not always lie in exactly the same plane. This is exacerbated bv the excellent denth of field of a field-emission SEM. The result i’s that, although the size and shape of the minute y ’ phases are faithfully revealed, permitting the accurate measurement of the size of these features by image analysis, their volume is not. Attempts to determine the volume fraction of sub-micron precipitates by image analysis thus lead to an overstatement of the volume present. The amounts of phases were therefore determined by weighing electrolytic extractions, which is regarded as the preferred technique.

*Gamma Prime Consultants, San Diego, CA 92127-1272. **Shepherd Color Co., Cincinnati, OH 45246.

***RITSS, Cincinnati, OH 45237-0383. Superalloys 1996

Edited by R. D. Kissinger, D. J. Deye, D. L. Anton, A. D. Cetel, M. V. Nathal, T. M. Pollock, and D. A. Woodford

The Minerals, Metals & Materials Society, 1996 129

Due to the possibility that grain boundary morphology has some relation to mechanical properties, a simple procedure was also utilized to measure the straightness of g&in boundaries. The grain boundaries were best revealed by electropolishing in 90% CHsOH - 10% HCL, at -40 to -SO “C, for about 25 seconds at 25 volts, and etching for 3 to 5 seconds at 5 volts, in the same cycle, electrolyte and temperature. This etches out they ‘, which can obscure the true grain boundary shape, but leaves the carbides, borides and oxides unaffected. It was then possible to record the grain boundary morphology at a triple point, at a magnification of 25,000. This allowed a characterization of the grain boundary curvature. That characterization consisted of measuring the true length of the three grain boundaries, meeting at the triple point, and dividing it by the straight line distance, measured for the same points of reference. Averagmg six to ten such measurements yielded a grain boundary curvature ratio (GBCR) that served as a quantification of grain boundary morphology.

Analvsis of Fullv Heat Treated Structures

The microstructre of an as fully heat treated Rene’ 88DT forging is shown in Figures 1 and 2, as revealed by both preparation procedures.

Figure i. \ltrr[)!~~lii~~;\ o! 1 iri a:, hc;ii trentcd Rene’ 88 DT, perchloric - metholtc preparation. Note 0.14 brn cooling y’, and 0.017 urn, aging y’.

The y’ phases were extracted in the standard 1% ammonium persulfate + 1% citric acid solution. Using a combination of differential settling and centrifuging, the twosize fractions of y ’ were separated, to allow their chemical analysis and structural characterization bv x-rav diffraction. The lattice parameter of the cooling, secondary y’ which, in the as heat treated, condition exhibited an average diameter of 0.14 urn. was 0.35917 nm. The 0.017 y ’ aging, t&&y, y ’ lattice parameter was found to be 0.35923 nm. Compared to the 0.3590 nm parameter of the y matrix, both y’ phases reflected a positive mismatch of 0.05%. The chemical analysis of the cooling y ’ is included in Table I, as is the analysis of the y. The latter was calculated from the analysis of the liquor resulting from the phase extraction. The amount of the aging, tertiary, y ’ that was recovered was insufficient to permit chemtcal analysis. Weights of extracted residue indicated that the alloy contained a total of 42.5 wt. % y’. distributed amongst the cooling ( 60% ), and aging ( 40 % ) forms. X-my diffraction analyses, of the metholic-HCl+tartaric extraction residue, identified the presence of a tetragonal, high Cr plus MO and W containing M3B2 ( D5,, a = b = 0.5787 nm, c = 0.31228 nm ), and a high Ti, lower Nb, MC ( Bl, a = 0.43514 nm ) . SAED procedures identified, as shown m Figure 3, a rare, tetragonal, high MO and W plus Cr containing, MsBs ( D81, a = b = 0.572 nm, c = 1.08 nm ). Some trace amounts of acicular eta ( NisTi ) were probably also present. All of these minor constituents amounted to about 0.2% of the total weight. All of these boride and carbide phases were similar in their microscopic appearance, and can be observed in Figure 2 as small, ( 0.1 to 0.5 urn) equiaxed, grain boundary phases. Despite their low concentration, their amount was sufficient to partially impair grain growth.

Table I Chemical AnaIvses

Weight PM VAR Secondary Matrix Partitioned to % Disk Bar Y’ Y Secondary y ’ Cr 15.70 16.0X 2.1 22.0 0.095 : 1 co 12.90 13.05 5.0 17.0 0.29 : 1 MO 3.98 4.08 .I.2 6.0 0.20 : 1 W 3.98 4.05 2.8 3.3 0.85 : 1 Ti 3.84 3.66 9.0 0.7 1 : 0.078 Al 2.21 2.33 4.0 0.7 1 : 0.175 Nb 0.72 0.69 1.6 0.3 I : 0.19 Zr 0.04 0.05 * *

C 0.041 0.040 * * B 0.015 0.015 * *

P <O.Ol 0.01 * *

0 0.014 0.006 * *

N 0.029 0.006 * *

Nv3 2.26 2.30 2.17

* Not analyzed. Partitioning ratio expresses concentration in y ’ compared to y. AU y’ analyses refer only to secondary form.

Figure 2. 1 r~ple pomt in fully heat treated Rene’ 88 DT, hydrochlonc - metholtc preparatton, revealing equiaxed borides and convoluted. or serrated, grain boundaries ( GBCR = I. 16 ).

The Rene’ 88 that was consolidated by VAR, exhibited the same structural features and appreciable amounts of eta phase. Examination of the VAR ingot revealed center line segregation of eta and borides, that could not be homogenized. An ingot route could not thus be applied to this alloy. Differential thermal analysis (DTA), backed up by microstructural examination, was used to determine the primary transitions in Rene’ 88 DT. The more important were: Liquidus, 1355°C ( 2471 “F ), Solidus, 1250 “C ( 2283 OF ), y ’ solvus, 1130 “C (2065 OF ), Etasolvus, l18S°C(2167”F).

130

0.M 5.12 10.24

keV

15.36

Figure 3. Identification of a spheroidal, intragranular, phase as the tebagonal MsB3 (D&J. X-ray, energy dispersive, analysis indicated a high W and MO chemistry with an appreciable Cr level.

Structural Kesponse to Heat Treatment

Effective heat treatment of highly alloyed disk compositions required a controlled cooling rate from the solution annealing temperature, followed by an aging treatment. During cooling from the super - solvus solution anneal of 1 150°C, two structural changes occurred, the extent of both being a function of the cooling rate. The frst is the precipitation of secondary y’ , whose particle size varied inversely with the cooling rate. The second was the formation of convoluted grain boundaries, a process that partially depended on the as annealed grain diameter and became most pronounced within a specific range of cooling rates. Both ettects were studied by cooling 0.6 cm x 1.2 cm x 1.2 cm coupons, each spot welded to a thermocouple that recorded the cooling rate on a high speed recorder. In each case, cooling was initiated after a 1.0 hr. anneal at 1120°C and, for slow cooling rates, terminated at about 870°C by water quenching the specimen. The wide variation in cooling y’ morphology, that can occur, is illustrated by the microstructures that were observed on cooling at 30 “C /minute ( Figure 4 ) and 2800 “C I minute ( Figure 5 ). Note that copious y ’ formed at even the fastest rates and, at very slow cooling rates, some tertiary ” aging” y ’ was observed . The quantification of these factors is shown in Figure 6 for the effect of cooling rate on gamma prime size, and in Figure 7 for the effect of grain diameter, at two cooling rates, on grain boundary curvature. In Figure 6, each point reflects the average of some 100 y’ size measurements. It illustrates that at cooling rates faster then about 300 “C I min only one size of y ’ formed. At slower cooling rates, two sizes were observed. The larger is believed to form

Figure J. (‘ooled at 3OY / rnmute. Large 0.2 pm, secondary y’ formed on cooling approached almost a dendritic morphology, with trace amounts of tertiary ( 0.016 pm ) y’ also present at curved grain boundaries.

Figure 5. ( oolcd nt 2X0() Y / minute. Only ri very fine 0.020 urn coohng y ’ is present, and the grain boundaries are completely straight.

1

.l

.Ol

001

3 TERTIARY y’ t I I

FULL DATA POINTS PM DISK, OTHER VAR. . I ” ““I . *.--

1 10 100 1000 10000

COOLING RATE, “CIMIN

Figure 6. Effect of cooling rate on the y ’ size. Data from both the production PM disk ( full points ), and VAR consolidated bar, show no effect of consolidation or grain diameter.

131

strictly .on cooling and is termed secondary. It is believed that the very small y ’ that forms at cooling rates slower then about 300” C / mm, is precipitated in the latter part of the cooling cycle, through essentially an aging reaction. It is thus termed as tertiary, or aging y ’ . The particle size of the tertiary y ’ is not a strong function of cooling rate. It is always spherical. The secondary, cooling y’ is spherical at fast cooling rates, becoming cuboidal at about 100°C I mm and tending to “dendritic” in the octahedral directions, almost cloverleaf in section, at even slower cooling rates. In all cases a curved, and probably coherent, interface is retained between the y ’ precipitate and y matrix. The diameter D (urn ) of the secondary, cooling y ’ , can be related to the cooling rate d”C / dt ( “C per a minute ) by the equation :

logDy’2= 0.178-0.551 (logd“C/dt), R2 = 0.93, (Eqn. 1)

The extent of grain boundary curving was measured on specimens electropolished and etched in metholic HCI, so as to etch out the y ’ , and clearly delineate the grain boundaries. The measurements were quantified by comparing the actual length of the three grain boundaries at a triple point, as measured at X25,000, to the straight line distance, to determine a quotient that was termed the Grain Boundary Curvature Ratio ( GBCR ). As shown in Figure 7, the degree of grain boundary curvature was found to be a function of both the cooling rate and the initial grain diameter.

1.4"" : ." i *',I

1.0; *n 8 i -- - i n - *! 0 100 200 300

GRAIN DIAMETER, MICRONS

Figure 7. Variation of the ratio, actual grain boundary length divided by the straight line distance ( GBCR ), as a function of grain diameter, for two cooling rates. Slower cooling rates, and grain diameters in the 200 micron range, promote the formation of convoluted grain boundaries, indicated by higher GBCR ratios.

The effect of variations in aging conditions was studied by oil quenching coupons ( 0.6 cm x 1.2 cm x 1.2 cm ) and aging them for 1 to 16 hr at 650” to 1040°C. Figure 8 shows the resultant relationship between the measured Rc hardness and the Larsen- Miller ( L-M ) parameter ( P = OK 125 + Log t 1 x 10-s ), where the aging temperature is expressed in OK and t is the aging time in hours. The variation in the diameter of the tertiary y ’ that formed on aging, and the Larsen-Miller parameter P, is given in Figure 9 . When the aging was performed at L-M parameters much above 29, the tertiary y’ was no longer observed. This suggested that a solvus for tertiary of y ’ exists at this combination of time and temperature. The data given in Figures 8 and 9 can be fitted to the following regression equations, which relate the Rc hardness and the diameter ( & I3 ) of the tertiary y ‘, in r.lrn to the parameter of the aging exposure, P = OK ( 25 + log t ) x 10 -3.

Rc = -253.92+ 21.371 P - 0.38722 Pa > R2 = 0.94 (Eqn. 2)

Dy13= 1.0286 - 0.085216 P + 0.0017691 P2 R2=0.99 @In. 3)

Since, at higher exposure conditions no tertiary y’ was observed, the validity of equation 3 is limited to Pc29. During aging, slight further growth occurred of the secondary, y ’ . The amount of this growth increased inversely with the cooling rate. The main structural process that occurred during aging was, however, the precipitation of the tertiary y ’ and it’s growth. The precipitation was most concentrated along grain boundaries and continued until this form amounted to about 40% of the total y ’ .

24 26 28 30 32 34

AGING, P = “K ( 25 + Log t ) x IO- 3

Figure 8. Relationship between hardness and the L - M parameter of the aging exposure. Note that at P>28 the hardness begins to decrease.

.061

.05;

.04:

.03;

.02-~

.Ol:

.00-I 25 26 27 28 29 30 31

-3 AGING, I’ = “K ( 25 + Log t )x10

Figure 9. Relationship between tertiary y ’ size and the L-M parameter of the aging exposure. The tertiary y ’ was not observed in specimens subjected to aging conditions, whose time and temperature exposure > P = 29.

Effect of Lone Time Exnosure

The effect of long time exposure was studied for just the conventional PM version of Rene’ 88 DT. As heat treated material was exposed for up to 1000 hours at 650” to 790°C with shorter time exposures to 1040 “C. In order to evaluate the effect of stress, three unfailed creep specimens, that had been tested for up to 6400 hours in the temperature range of 680” to 760°C and at sbess levels of 276 to 689 MPa, were also examined. The growth of y ’ on long time exposure is portrayed in Figure 10. All the data, whether obtained in the presence or absence of stress, could be fitted to a Larsen-Miller relationship P = OK ( 25 + Log t ) x 10 -3, where me temperature of exposure is in OK, and t is the time, in hours.

132

TERTIARY Y ’

P = “K ( 25 + Log t ) x 10 - 3

Figure 10. The change in diameter of both the secondary, cooling, r’ and the tertiary, aging, y’ can be correlated to the exposure conditions through a L - M parameter with C = 25. Note that the y’ that grew under stress, designated by full data points, falls in the same population as that which grew in the absence of stress.

This allowed the development of the following regression equations, that gave the diameter D, in urn, of both the secondary Y’, that formed on cooling, and with less accuracy, the tertiary y ’ , that formed largely on aging, as a function of P.

For secondary y ‘, D,‘z = 0.02833 P - 0.57557, R2 = 0.93 ( Eqn. 4 )

For tertiary y ’ , Dy’3 = 0.01311 P - 0.32435, R2 = 0.82 ( Eqn. 5 1

It should be noted that the growth of the aging, tertiary y ‘, stopped when this precipitate reached a diameter of some 0.05 urn. Exposure at L - M parameters of about 29, or larger, caused the tertiary form of y’ to completely dissolve. In addition to the above changes in y’ morphology, grain boundary films of y’ began to form at exposure conditions with a LM parameter of 25.8, or above. This is equivalent to exposures above 650°C and 1000 hr. Although the total amount of secondary and tertiary y’ does not change ( < 5% ) appreciably during high temperature exposure, changes do occur in the relative amounts of these two forms. At 650 “C the amount of tertiary y ’ increases with exposure, reaching about 50% of the total y’ content after 1000 hr. At higher temperatures the growth of the tertiary y’ was accompanied by a decrease in the amount of this form. After 1000 hr at 705°C and 760°C the tertiary y’ content was, respectively, 30% and 16% of the total. These are estimates, based on image analysis, and urobablv over-state the amount of tertian, y’. Because of these , complex changes, the variation of hardness with exposure is nenheible. until a L-M oarameter of 25 is exceeded. The scatter in h~dn”e&data was large, and a statistically significant equation could not be used to describe the change in hardness with high temperature exposure. X-ray diffraction analysis of the metholic-HCI-tartaric electrolytic extraction residue, combined with analytical electron microscopy procedures, served to identify the phases that formed on exposure. Prolonged exposure ( > 2,000 hr ) at 650 “C, or for shorter times at higher temperatures, produced grain boundary precipitation of a high Cr and MO + W, M3B2 ( DSa, a = 0.579 nm, b = c = 0.311 nm ). For exposure conditions longer than 1000 hr. and at 705°C or above, possibly associated with the solution of the tertiary y’, a high Cr, M&6 ( D84, a = 1.0706 nm ) phase formed. Both the MsB2 and Mz3C6 precipitated largely at grain boundaries, often within the previouslv noted y’ films. In addition, an intragranular precipitation of small amount of MC ( Bl a = 0.434 nm )-which, being high in Zr, rather then Nb, differed in chemistry from that found in the as heat treated condition. It must be stressed that these precipitation reactions were observed above the expected temperature range for the long term utilization of Rene’ 88 DT, and

involved very small amounts of precipitation. For instance, the total amount of all borides and carbides recovered by extraction from a specimen that had been aged for 1000 hr at 760 “C, well above any expected service temperature, was only 0.7wt %. In addition to these largely inter-granular precipitation reactions, the formation of an intragranular, acicular, precipitate was observed in a specimen that had been aged for 6,300 hr at 760°C and 276 MPa. This phase, as shown in Figure 1 I, was identified using AEM procedures as a high Cr, MO and W containing n. All of the results on alloy stability were obtained on production disk material with the chemistry given in Table I and NV3 = 2.26.

2

Figure I 1. Al:M idcntilicatior 01 rhombohetlral u phase after 6300 hr. at 76O“C and 276 MPa. Note high Cr and Mo+W chemistry.

The structural data for all the phases found in this study, for both the as heat treated condition and after prolonged elevated temperature exposure, are compiled in Table II.

133

Table II. Phases Found in Rene ’ 88 DT These measurements gave a good approximation of the specification aging conditions, the resultant hardness and the optimum tertiary y ’ size, above which the hardness would drop. The absence of ternary y ’ after aging at L-M parameters >29, should be noted. Phase

Gamma

Secondary y’

Tertiary y’

M3Bz Cr, MO, W

MS& MO, W, Cr

Ti,Nb ( C )

Zr,Ti ( C )*

M23c6 +

Lattice Parameter nm Structure

Al a = 0.35890

Ll2 a = 0.35917

Ll2 a = 0.35923

m3 a = b = 0.5787, c = 0.3 1228

D81 a=b=0.572 c = 1.08

Bl a = 0.435 14

Bl a = 0.434

D84 a = 1.0706

* Found only after high temperature exposure. Eta in the as heat treated condition and, after prolonged exposure, a n phase, were

also identified but their exact lattice parameters were not determined.

Discussion

The processing of Rene’ 88 DT through a powder metallurgy route allows greater alloying levels and a finer grain size. The use of a super-solvus anneal further benefits properties. This alloy exhibits the lowest y’ misfit of any commercial alloy c7). Although the solute distribution between the y’ and y phases is essentially the same as has been documented for other commercial alloys @z9), the balance of elements is such, that effective alloying of both they and y’ phases has been achieved without introducing structural instabilities in it’s normal operating range. In addition, the boron and carbon are low enough, so that the precipitation of grain boundary borides and carbides is not embrittling. The most critical operation in the heat treatment of an advanced disk alloy is the achievement of an optimum cooling rate from the solution temperature. The faster the cooling rate, the smaller is the resultant secondary y’ size, and usually, the better the creep and tensile properties, The practical limit of this trend is that very fast cooling-rates can cause high residual stresses, leading to quench crackine. Fortunately. Rene’ 88 DT appears to be less sensitive to quench&acking then other disk alloy@). The relationship between the cooling rate dT/dt and the diameter of the secondary y ‘, can be calculated by equation 1. Previous studies(*), of Rene’ 88 DT, have shown that preferred properties are associated with a cooling rate of some 140°C / min, a rate that should result in a secondary y ’ size of some 0.1 pm. The availability of equation 1 permits the metallographic verification of the cooling rate that had been used in the production forging. Thus, the 0.13 pm y’ size of the disk which was used in this study, suggests that the cooling rate from the super - solvus anneal was about 85”C/min. The effect of aging on tertiary y ’ formation can be fully quantified, not only in terms of hardness ( equation 2 ), but also in terms of the tertiary y ’ size ( equation 3 ). Both the size of the tertiary y ’ and the Rc hardness can be related to the Larsen-Miller relationship that describes the time and temperature of aging. These equations may allow the verification of appropriate aging conditions in a Droduction hart through a metallomanhic examination. In addition, they can clarify someof the chara&&stics of the aging reactions in Rene’ 88 DT. When solved. eouations 2 and 3. predict a maximum hardness on aging at P = 27.4 or, for an 8 hr age, at 785”C, at which value the tertiary y’ diameter would be some 0.022 nm.

When solutioned and aged in the preferred processing range, Rene’ 88 DT exhibits a structure containing some 42.5 % Y’. distributed between the secondary ( 60 % ) and&tiary ( 40 % ‘) forms. Both types of y ’ exhibit only a small ( 0.05 % ), positive mismatch with the matrix. Because of this high level of coherency, the tertiary y ’ is always spherical, and when the secondary y ’ departs from spheroidicity at larger sizes, it tends to maintain a structure typified by a spheroidal morphology. When compared to other disk alloys, such as Rene’ 95(m) or N18(3), Rene’ 88 DT forms a finer secondary y ‘, for an equivalent cooling rate. This characteristic benefits properties and results directly from the low mismatch of the y ‘, which not only produces a highly coherent precipitate, but also reduces the free energy change that must be overcome on precipitation, increasing the nucleation rate of the y ’ . The low r ‘ly mismatch! that is so beneficial to properties in the temperature range in which disk alloys are used, is a direct result of the judicious use of tungstenUs) which, unlike MO (91, increases the lattice parameter of y , without appreciably effecting y ’ . The beneficial effects of a coherent y’ may also reflect in the relatively low rate of growth of this phase during service. The growth of y ’ during high temperature exposure can also be described by exploiting the L-M relationship in equations 4 and 5, to relate the time and temperature of exposure to both the secondary and tertiary y ’ sizes. The measurement of the y ’ size, after service, could thus allow the approximation of a parts average service temperature. Only a gross over temperature has any measurable effect on room temperature hardness, and usually no correlation of hardness and service temperature is possible for Rene’ 88 DT. Measurements of y ’ size, through a field metallographic procedure, and the use of equations 4 and 5, could prove particularly useful in any investigation where an estimate has to be made of the average service temperature. It should also be noted that Rene’ 88 DT achieves its property level at a much lower y’ content ( 42.5% ) then Rene’ 95 (to) ( 50% ), or N18 (3) ( 58% ). The relationship between grain boundary curving, or formation of serrations, and propeties, is not well understood. Grain boundary curving is effected by both the cooling rate and grain diameter, but the relationships are not simple. The serrations ate a result of grain boundary movement during sub- solvus cooling from the solution anneal. The serrations documented here are not due to the growth of y ’ into a grain boundary. They are produced by the dynamic movement of a grain boundary into freshly nucleated, secondary, y ’ precipitates. For a constant cooling rate, the force driving this movement appears to increase with the length of the unsupported grain boundary. For an ectuivalent grain size, such grain boundary movement is much more pronounced in alloys that are given an anneal above, rather then below. the Y’ solvus. Allovs such as Rene’ 95 and N18, which are annealed below the y’* solvus, and thus contain large, primary, y ’ during cooling from the annealing temperature, do not develop curved, or serrated, grain boundaries as readily as Rene’ 88 DT. In addition, annealing above the y’ solvus removes the large primary y ’ phases that occur in Rene’ 95 and N 18, and serve as a preferred point of fatigue crack nucleation c3). Annealing above the y’ solvus, however, introduces the possibility of grain growth. In the case of a sub-solvus annealed alloy, such grain growth is contained by the presence of large, primary, y ’ . During prolonged high temperature service the secondary and tertiary y’ grow and increase slightly in amount. Their growth, which, can be predicted by equations 4 and 5, does not vary greatly with the presence of stress during exposure. If the relationship between the mechanical properties and the y’ size could be established, and this has been achieved for other alloys c4, l l and 12), these equations should prove useful in predicting the approximate level of properties that could be expected after prolonged service exposure.

134

800

600

----- - eor_vug - - + AGlNGy’

YSOLVUS q AGING y +M,B,

q M3B2

A M3B2 + M23C6

0 M3B2+MPc6+~

2200

2100

2000

1900

1800

1700

1600

1500

1400

1300

1200

1100

1 10 100 1000 10000

TIME(HR)

Figure 12. Schematic summary of phase reactions in Rene’ 88 DT. Prolonged exposure produces the growth of both the secondary and the tertiary forms of y ’ and the formation of grain boundary y ’ films. Ultimately, at exposure conditions equivalent to LM = 29 the tertiary y ’ that formed on aging dissolves. Continued exposure produces grain boundary MsB2 and M&j and, after prolonged exposure at 76O”C, p phase appears. Full points were stressed exposures.

In the meantime, it is possible to point out that when the L-M

Parameter exceeds 29, the tertiary Y’ is completely dissolved and a very large reduction in load bearing properties should result. Some *OSS Of Properties should, of course, be also expected when the secondary and/or tertiary y ’ diameter exceeds some critical size. If the relationship between any specific mechanical property and he sizes of the Y’ phases were known, equations 4 and 5 could be used to calculate the L-M parameter associated with any minimum property level. The tertiary y ’ solvus, as well as the areas of stability for the various secondary phases found in Rene’ 88 DT are indicated in Figure 12. As heat treated, Rene’ 88 DT contains only some 0.2 wt. % borides and carbides, predominantly MsB2, with some Nb rich MC, both largely at gram boundaries. Small amounts of eta phase and a uses constituent were also observed, but in truly trace amounts. None of these phases are in any way unusual, but note should be taken of the fact that the major gram boundary constituents were borides and not carbides. At the NV3 = 2.26 chemistry used in these studies, the alloy is completely free of any deleterious precipitation reactions in its nominal service range. At 650 “C, a 2000 hr. exposure is required before even any additional MsB2 forms. Although, gram boundary films of ~~~(3~ and intragranular n appear at higher temperatures, this happens only after the tertiary y ’ solvus is exceeded, at which point the alloy is well outside its long time service temperature capability. Rene’ 95 and N - 18 are more prone to the formation of a topological close packed phase, than Rene’ 88 DT. Both form n. and N- 18 is also subject to the precipitation of u ( 3, to ).

Rene’ 88 DT IS, however, very highly alloyed and is an excellent example of the advantage of processing a highly alloyed composition through a powder metallurgy route. Due to the tendency to form segregation induced eta phase, in even small ingots, this alloy could never be processed through a conventional ingot practice. Studies of this type allowed an understanding of an alloy’s behavior, identification of the factors that must be controlled in it’s processing and heat treatment, a method of identifying occurrences of inappropriate heat treatment, identification of service induced over-temperature. When combined with appropriate mechanical data, such studies may allow the prediction of behavior in service, particularly the degradation of mechanical properties. This could allow a more effective design philosophy, that need not be based completely on virgin, as heat treated, mechanical property levels. In closing, one caution is required. The regression equations presented here may not reflect a level of great theoretical validity. Their use should not, therefore, be extrapolated past the range of the data on which they were based. Still, they are presented in the spirit of Lord Kelvin’s admonition:

“When you can measure what you are speaking about, and explain it in numbers, you know something about It.”

Acknowledgments

The authors are indebted to the management of the Engineering Materials Technology Laboratory, GE Aircraft Engines, for permission to publish this paper.

135

References

1. D.D. Krueger, R.D. Kissinger, R.D. Menzies and C. S. Wukusick, US Patent 4,957, 567 .

2. D.D. Krueger, R.D. Kissinger and R.D. Menzies, “Development and Introduction of a Damage Tolerant High Temperature Nickel-Base Disk Alloy, Rene’ 88 DT”, Superalloys 1992 S.D.Antolovich et al Editors, TMS-AIME, Warrendale PA, (199i), 277- 286.

3. S.T. Wlodek. M.Kellv. and D. Alden. “The Structure of N 18”. Su~erallovs 1992, S. 6.’ Antolovich et al Editors, TMS-AIMEI Warrendale PA, (1992), 467- 476.

4. P.R. Bhowal, E.F.Wright and E.L. Raymond, “Effects of Cooling Rate and Morphology on Creep and Stress Rupture of a Powder Metallurgy Superalloy”, Met.Trans.A, 21, (1990), 1709- 1717.

5. M. Zhiping, Y. Ruizeng, and G. Liang, ” Effect of Zigzag Grain Boundary on Creep and Fracture Behavior of Wrought y ’ Strengthened Superalloy”, Materials Science and Technology, 4, (1988), 540 - 547.

6. S.T. Wlodek, “SEM Metallography of Superalloys”, Microstructural Science T.A. Place et al Editors, ASM, Metals Park OH, 18, (l%@$?‘- 429.

7. B.A. Parker and D.R.F. West, ’ The influence of Particle/Matrix Interfacial Mismatch on the Size, Shape and Distribution of Coherent y ’ Particles in Nickel Based Alloys”, J. Australian Inst. of Metals”, 14 , (1969), 102-l 10.

8. O.H. Kriege and J.M. Baris, ” The Chemical Partitioning of Elements in Gamma Prime Sepemted from Precipitation - Hardened, High - Temperature Nickel Base Alloys”, Trans. ASM, 62, (1969), 195 - 200.

9. J.P. Collier, P.W. Keefe and J. K. Tien, “The Effects of Replacing the Refractory Elements W, Nb, and Ta with MO in Nickel - Base Superalloys on Microstructural, Microchemistry, and Mechanical Properties”, Met. Trans. A, 17A, (1986), 651 - 661.

10. S.T. Wlodek, Unpublished results.

11. J.J. Schirra and S.H.Goetschius, “Development of an Analytical Model Predicting Microstructure and Properties Resulting From the Thermal Processing of a Wrought Powder Nickel - Base Superalloy Component”, Superalloys 1992, S.D. Antolovich et al editors, TMS-AIME, Warrendale PA, (1992), 437-446.

12. S.T.Wlodek and R.D. Field, “The Effects of Long Time Exposure on Alloy 718”, Superallovs 718. 625. 706 and Various Derivatives”, E.A.Loria editor, TMS-AIME, Warrendale PA, (1994),659-670.

13. A. Havalda. “Investieation of the Influence of Tungsten on the Morphology of the y’ Precipitation in Nickel - Base Alloys”, Trans. ASM, 62, (1969), 477-480.

136