Full Text

Design and Development of Hot Corrosion-Resistant Nickel-Base Single-Crystal Superalloys by the d-Electrons Alloy Design Theory: Part I1. Effects of Refractory Metals Ti, Ta, and Nb on Microstructures and Properties

J.S. ZHANG, Z.Q. HU, Y. MURATA, M. MORINAGA, and N. YUKAWA

A systematic study of the effects of refractory metals Ti, Ta, and Nb on the microstructures and properties was conducted with a hot corrosion-resistant alloy system Ni-16Cr-gA1-4Co-2W- l Mo-(0-4)Ti-(0-4)Ta-(0-4)Nb (in atomic percent) which was selected based on the d-electrons alloy design theory and some basic considerations in alloying features of single- crystal nickel-base superalloys. The contour lines of solidification reaction temperatures and eutectic (Y + Y') volume fraction in the Ti-Ta-Nb compositional triangle were determined by differential thermal analysis (DTA) and imaging analyzer. Compared with the reference alloy IN738LC, in most of the compositional ranges studied, the designed alloys show very low amounts of eutectic (y + y') (-<0.4 vol pct), narrow solidification ranges (-<65 ~ and wide "heat-treatment windows" (> 100 ~ This indicates that the alloys should have the promising microstructural stability, single-crystal castability, and be easier for complete solution treatment. In a wide compositional range, the designed alloys showed good hot corrosion resistance (weight loss less than 20 mg/cm 2 after 24 hours kept in molten salt at 900 ~ By summarizing the results, the promising alloy compositional ranges of the alloys with balanced properties were determined for the final step of the alloy design, i . e . , to grow single crystal and characterize mechanical properties of the alloys selected from the previously mentioned regions.

I. INTRODUCTION

THE critical boundary conditions (Mdt -< 0.991 and Mdy -< 0.93) for the alloy systems with high Cr contents have been determined in the first part of this study, t~ Those Md values are the electronic parameters calcu- lated from alloy or phase compositions. They are the most important reference criteria for the alloy design and development process with the d-electrons alloy design theory.

The new single-crystal alloys were required to have the desired balance between hot corrosion resistance and high-temperature mechanical property. The traditional trial and error method usually requires repeating many experiments, is expensive in time and money, and achieves the desired optimum results only with great difficulty. To overcome these problems and improve the efficiency for alloy development, some alloy design methods have been proposed, such as the Nv-PHACOMP 121 and Mongeau-Wallace methods, t-~j These methods have been applied to the alloy design of single-crystal superalloys.t4,5'61 The results of those stud- ies were not completely satisfactory because of the poor predictions of the microstructures and properties in com- plex alloy systems. On the contrary, the newly devel- oped d-electrons alloy design theory (or the New

J.S. ZHANG, Associate Professor, formerly with the Institute of Metal Research, Academia Sinica, Shenyang 110015, China, is with Beijing University of Science and Technology, Beijing 100083, China. Z.Q. HU, Professor, is with Institute of Metal Research, Academia Sinica, Shenyang 110015, China. Y. MURATA, Associate Professor, and M. MORINAGA and N. YUKAWA, Professors, are with the Department of Production Systems Engineering, Toyohashi University of Technology, Toyohashi 441, Japan.

Manuscript submitted June 5, 1992.

PHACOMP method) based on the D V - X a cluster cal- culations of the electronic structures has been success- fully applied in the design and development of various advanced materials such as superalloys, t7 1o1 titanium al- loys, till construction materials used in fusion reactors, ~2J intermetallic compounds, t~31 e t c . For example, the high- performance nickel-base single-crystal alloy TUT 92 de- veloped by the d-electrons alloy design concept t8'91 showed the promising balanced comprehensive proper- ties in high-temperature strength, ductility, hot corrosion resistance, density, and cost. This alloy is markedly su- perior to the first generation single-crystal superalloys t ~41 PWA1480, NASAIR 100, CMSX-2, and MXON. The comprehensive properties of the recently developed single-crystal alloys with the new alloy design concept such as TUT95 t~~ have exceeded those of the second generation single-crystal alloys such as PWA1484 and CMSX-4 and are among the most advanced single- crystal superalloys around the world.

The second part of the study is to develop the new hot corrosion-resistant single-crystal superalloys under the guidance of the d-electrons alloy design theory. The ef- fects of refractory elements Ti, Ta, and Nb in a hot corrosion-resistant alloy system with high Cr content were determined to define the promising compositional regions for the final examination.

II. ALLOY DESIGN CONSIDERATIONS

To design and develop the hot corrosion-resistant nickel-base single-crystal superalloys, the alloy system was first selected. The system Ni-Cr-A1-Co-W-Mo-Ti- Ta-Nb is usually the basic alloy system for the hot

METALLURGICAL TRANSACTIONS A VOLUME 24A, NOVEMBER 1993 - - 245 !

corrosion-resistant superalloys, with additions of the grain-boundary strengthening elements C, B, and Zr. According to the strengthening mechanisms of the nickel-base superalloys uS~ and the strengthening features of the single-crystal alloys (Figure 1), a high amount of A1 and (W + Mo) contents should be maintained. At the same time, the Co content should be taken at a relatively low level. 116'171 Therefore, the levels of A1, (W + Mo), and Co in the designed alloy system were fixed at 9.0, 3.0, and 4.0 at. pct, respectively. To maintain the hot corrosion resistance in the designed alloys, Cr content was kept high at 16 at. pct. According to the results of the first part of this study, lu in a similar hot corrosion- resistant alloy system, the elements AI, Ti, Ta, and Nb were the main y ' forming elements. They are important to hot corrosion resistance as well as mechanical prop- erties. The severe segregation of Ti and Nb during so- lidification processing of the nickel-base superalloys makes the alloys prone to TCP phase precipitation, ll81 Up to now, many studies have been conducted on the effects of Ti and Ta in superalloys, but the effects of Nb and its interaction with other elements have not been well understood. Therefore, the effects of refractory metals Ti, Ta, and Nb on the microstructures and prop- erties of the selected alloy system were systematically studied. The contents of the three elements were changed within a compositional range between 0 and 4 at. pct, which is often seen in currently used nickel- base superalloys.

Finally, the alloy system for this investigation was determined to be Ni- 16Cr-9AI-2W- 1Mo-4Co-(0~4)Ti- (0~4)Ta- (0~4)Nb (in atomic percent). For ease of com- parison, the volume fraction of Y' phase in the alloys was maintained near a fixed value, for which the total amount of the Y' forming elements was fixed as 13 at. pct, as in IN738LC alloy, i .e. , A1 + Ti + Ta + Nb = 13 at. pct. The contour lines of three important param- eters Mdt, Bot, and p (density) of designed alloys are given in a Ti-Ta-Nb compositional triangle (Figure 2). The values of Mdt, Bot, and p varied systematically in the ranges of 0.977 to 0.983, 0.705 to 0.728, and 8.3 to 8.9 (g/cm3), respectively.

# 03

g b~

/

i ~ o o 4 - /~,

""x i/s 0 \ Ol O~ 0 . \ \

\ / 3 ,, i 0

/ / \ \ 0 0

/ \ \ 0 / \ 0 ,.~-~ 2 / \

I 0 \\ �9 0 N /

I �9 \\0 0 / x

1 / ~ . ~ ~ ~ " ~ ' ,

/ / �9 �9 O ~ �9 "

0 1 2 3 4

W4Vb, a t -%

O single crys ta l alloy

�9 hot corrosion r e s i s t an t alloy

o 0 o 0 0 0@0

0 0 0 0 0

0 CO ~ -

Fig. l I C a t e g o r i e s of 67 nickel-base single crystal and 17 hot corrosion-resistant superalloys reported in the literature in the Ta (W + Mo) graph.

The values of Mdt and Bot are calculated in the fol- lowing equations: tm

M d t = 0.717Ni + 1.142Cr + 1.90AI + 1.655W

+ 1.55Mo + 0.777Co + 2.271Ti

+ 2.224Ta + 2.117Nb

Bot = 0.514Ni + 1.278Cr + 0.533A1 + 1.73W

+ 1.611Mo + 0.679Co + 1.098Ti

+ 1.67Ta + 1.594Nb

in which the compositions are in atomic percent. The densities are given as, t2u

/9 = ( P t + 0.1437 -- 0.00137Cr - 0.00139Ni

- 0.00142Co - 0.00125W - 0.00113Ta

+ 0.00040Ti - 0.00113Hf + 0.000187(Mo) 2

+ 0.000175(Nb) 2 - 0.0000506(Co) (Ti))

x 27.68 (g / cm 3)

with p~ = 100/p2 and

P2 = Ni/0 .322 + A1/0.0975 + Cr /0 .26 + Co/0 .322

+ Ti /0 .163 + Re/0 .76 + Ta /0 .6 + W/0 .697

+ Mo/0 .369 + Nb/0 .30 + Hf /0 .48

in which the alloy compositions are given in weight percent.

It should be pointed out that the Mdt values were var- ied within the limit of the phase stability boundary con- dition (Mdt -< 0.991), Ul which would guarantee stable microstructures for the alloys studied. The densities of the alloys selected were within the range for other cur- rently used superalloys. Based on the previously men- tioned analyses, the compositions of the alloys for this study were selected as given in Table I, including the

T i

de n~s~ t y g/cm3

2 L

T~ ~ 0 1 2 3 4

Nb, at%

Fig. 2 - - C o n t o u r lines of the important parameters Mdt, Bot, and density (p) in the Ti-Ta-Nb compositional triangle.

2452--VOLUME 24A, NOVEMBER 1993 METALLURGICAL TRANSACTIONS A

Table I. Compositions of the Alloys Studied (Atomic Percent/Weight Percent)

Alloy Ti Ta Nb A1 Cr W Mo Co Ni Mdt Bo----t Density (g/cm 3)

T1 0 0 4.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9770 0.7252 8.38 0 0 6.30 4.11 14.09 6.23 1.63 3.99 63.65

T2 0 1.00 3.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9780 0.7260 8.51 0 3.02 4.65 4.05 13.89 6.14 1.60 3.93 62.71

T3 0 2.00 2.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9791 0.7267 8.64 0 5.95 3.06 3.99 13.69 6.05 1.58 3.88 61.81

T4 0 3.00 1.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9802 0.7275 8.77 0 8.80 1.51 3.94 13.49 5.96 1.56 3.82 60.93

T5 0 4.00 0 9.00 16.00 2.00 1.00 4.00 64.00 0.9812 0.7282 8.90 0 11.57 0 3.88 13.30 5.88 1.53 3.77 60.07

T6 1.00 0 3.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9785 0.7202 8.35 0.82 0 4.76 4.15 14.20 6.28 1.64 4.02 64.14

T7 1.00 1.00 2.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9796 0.7210 8.48 0.81 3.04 3.13 4.08 13.99 6.19 1.61 3.96 63.19

T8 1.00 2.00 1.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9806 0.7218 8.61 0.79 6.00 1.54 4.02 13.79 6.10 1.59 3.91 62.27

T9 1.00 3.00 0 9.00 16.00 2.00 1.00 4.00 64.00 0.9817 0.7225 8.74 0.78 8.86 0 3.97 13.59 6.01 1.57 3.85 61.37

T10 2.00 0 2.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9800 0.7152 8.32 1.65 0 3.20 4.18 14.31 6.33 1.65 4.06 64.64

TI 1 2.00 1.00 1.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9811 0.7160 8.45 1.62 3.07 1.57 4.16 14.10 6.23 1.63 3.99 63.67

TI2 2.00 2.00 0 9.00 16.00 2.00 1.00 4.00 64.00 0.9822 0.7168 8.58 1.60 6.04 0 4.05 13.89 6.14 1.60 3.94 62.74

T13 3.00 0 1.00 9.00 16.00 2.00 1.00 4.00 64.00 0.9816 0.7103 8.29 2.49 0 1.61 4.21 14.42 6.38 1.66 4.09 65.14

T14 3.00 1.00 0 9.00 16.00 2.00 1.00 4.00 64.00 0.9827 0.7111 8.42 2.45 3.09 0 4.15 14.21 6.28 1.64 4.03 64.16

T15 4.00 0 0 9.00 16.00 2.00 1.00 4.00 64.00 0.983t 0.7054 8.26 3.35 0 0 4.24 14.54 6.43 1.68 4.12 65.65

IN738 4.05 0.54 0.55 7.19 17.56 0.81 1.01 8.23 60.06 0.9764 0.7229 8.22 3.40 1.70 0.90 3.40 16.00 2.60 1.70 8.50 61.80

reference al loy IN738. These des igned al loys were homogeneous ly dis t r ibuted in a T i -Ta-Nb compos i t iona l tr iangle (Figure 3). The composi t iona l t r iangle is s imilar to the c o m m o n ternary-phase d iagram. Each line paral lel to one of the bo t tom lines represents a contour line of

the e lement on the corresponding top. A crosspoint of two contour lines will de termine the composi t ion of an al loy, because any point in Figure 3 keeps Ti + Ta + Nb = 4 at. pct. It makes the select ion of al loy compo- sitions s impler .

T i

Ts ~ 0 1 2 3 4

N b , a t %

Fig. 3--Distribution of the experimental alloys in a Ti-Ta-Nb com- positional triangle.

I I I . E X P E R I M E N T A L

First , the sol idi f icat ion behavior of the des igned al loys l isted in Table I was studied by differential thermal anal- ysis (DTA). The specimens used were prepared by a tri- arc furnace in a purif ied argon atmosphere . The details of the sample preparat ion and D T A exper iments can be found in the first part of this study, m

Based on the D T A results, the al loys were d iv ided into three groups for heat treatment:

(a) 1260 ~ h / A C + 1050 ~ h / A C + 850 ~ 48 h / A C (for T3, T4, T5, T7, T8, T9, T11, T12, and T14 al loys); (b) 1230 ~ h / A C + 1050 ~ h / A C + 850 ~ 48 h / A C (for T2, T6, T10, T13, and T15 al loys); and (c) 1120 ~ h / A C + 850 ~ h / A C (for T1 and IN738LC al loys) .

It should be pointed out that the temperature of heat t reatment for a l loys T1 and IN738LC was lower than the y ' solvus temperature as de termined by DTA, because

METALLURGICAL TRANSACTIONS A VOLUME 24A, NOVEMBER 1993--2453

the severe solidification segregation lowered the incipi- ent temperatures of these alloys to a level lower than the solvus temperatures, t~8] In fact, here, the standard heat treatment for IN738LC was simply adopted.

Some specimens were long-term aged at 900 ~ for 500 hours and water quenched for characterization of phase stability and 3" coarsening behavior.

Heat-treated specimens were employed for hot cor- rosion experiments with the crucible method. The test pieces were polished down to a no. 1000 with dry abra- sive papers. The composition of mixed salts was 75 pct Na2SO4 + 25 pct NaC1. The selection of such a com- position of molten salt was based on the consideration that it is strongly aggressive to the Cr203 type scale. [ml In the alloys studied, the contents of Cr were high and the Cr203 scale would be formed during hot corrosion tests. Therefore, a possibly severe test medium was se- lected in this study. Figure 4 shows the hot corrosion experimental arrangements. The specimens were held in the molten salt at 900 ~ for 24 hours and then the weight loss (mg/cm 2) was determined after descaling. The corrosion products on the surfaces of the specimens were eliminated in two steps: (a) boiling in a solution of KMnO4 (5 g) + NaOH (8 g) + distilled water (77 g) for 30 minutes and washed in water with ultrasonic; and (b) boiling in a solution of (NH4)2HC6HsO7 (5 g) + dis- tilled water (45 g) for 20 minutes and washed in water with ultrasonic.

After fully drying, each specimen was weighed and the weight loss was calculated. The constituents of the corrosion products on the surface scale were determined by X-ray diffraction.

The microstructures of the alloys were studied by op- tical and scanning electron microscopy (SEM), electron probe microanalyzer (EPMA), and imaging analyzer. Before the observation, the specimens were polished and etched in a solution of CuSO4 (20 g) + HC1 (100 mL) + H20 (80 mL) + H 2 S O 4 (5 mL).

IV. E X P E R I M E N T A L R E S U L T S

Figure 5 shows the DTA results of the experimental alloys, in which the contour lines of solidification re- action temperatures P0 (3' + 3" ---> L), P~ (L --~ 3'), and

Specimen

Alumina crucible Pt wit

Q Heater

\ Salt (Na2SO4-25%NaCl)

Fig. 4 - - S c h e m a t i c show of the hot corrosion experiment with cru- cible method.

P3 (3 / "-+ 3'); solidification range (AT = P~ - P0); and eutectic (y + 3/) volume fraction are drawn. Figure 6 gives the microstructures of the alloys after DTA measurements. It is interesting that in a wide composi- tional range, the solidification range of the alloys is very low, -<65 ~ (Figure 5(d)), and the eutectic (3' + 3") almost disappeared (Figures 5(e) and 6).

Table II shows the results of the dendritic segregation features determined by EPMA. During solidification, the elements Nb, Ti, Ta, Mo, and AI segregated to inter- dendritic regions, while the other elements Cr, Co, W, and Ni segregated to dendritic arms. Among these ele- ments, Nb and Ti segregated most severely, which was similar to their behavior in other superalloys, t~SI The se- vere segregation of Ti and Nb resulted in a great deal of eutectic (y + y') precipitation in the final solidification stage in alloys around the Nb and Ti-rich corners (Figure 6). In the alloy with the highest Nb content (T1), a large amount of Nb-rich eutectic (y + y') formed (Figure 7), which resulted in the lowest solidification re- action temperatures around the Nb-rich corner (Figure 5).

Figure 8 shows the typical microstructures in the spec- imens after heat treatment. Except T1 and IN738LC al- loys, the coarse solidification structures in the designed alloys could be eliminated completely without incipient melting. This means that the alloys could be properly and easily heat treated, because of the lower eutectic (y + 3") amount and the wider heat-treatment window (P0 - P3 ~ 100 ~

Figure 9 shows the typical 3" morphologies in the specimens after heat treatment. The cubic 3" is aligned regularly in the alloys, which would be beneficial to the mechanical properties. 12~

After prolonged aging at 900 ~ for 500 hours, no TCP precipitation was observed in any of the experi- mental alloys. This means that the microstructures of the alloys were stable as predicted by the criterion (Mdt -< 0.991) of the alloy design concept. It should be noted that the 3" phase morphologies of the designed al- loys were also stable after the prolonged aging. On the contrary, the 3" phase in the reference IN738LC alloy had already coarsened and agglomerated (Figure 10(c)). This result indicates that the mechanical properties of the alloys would not be lowered rapidly by coarsening of the strengthening 3' phase during high-temperature long- term service. This long-term stability is one of the major requirements for marine and industrial gas turbine applications.

Figure 1 l summarizes the results of hot corrosion ex- periments with a crucible method measured by weight loss (mg/cm 2) after immersion for 24 hours at 900 ~ in a molten salt of 75 pct Na2SO 4 + 25 pct NaC1. The experimental alloys showed good hot corrosion resis- tance around the Ti and Nb comers, but poor hot cor- rosion resistance around the Ta comer. In the Nb-rich corner, alloy T6 showed the best hot corrosion resistance (weight loss: 7.08 mg/cm 2) among the experimental al- loys. In the Ta-rich corner, alloy T9 showed the poorest behavior (fluxed during test). Figure 12 shows the de- pendence of hot corrosion resistance of a typical ex- perimental alloy T7 and the reference alloy IN738LC upon corrosion time up to 300 hours at 900 ~ Similar

2454--VOLUME 24A, NOVEMBER 1993 METALLURGICAL TRANSACTIONS A

(a)

Ti

T a ~ ~

N b , a t %

T i

Ts i 2 3 4 ONb

Nb, at% (b)

(c)

Ti

0 1 2 3 4

Nb, at%

2

T i

2 ~ N b , at% (d)

(e)

T i

/ 7 ,o.2 t

v /I o i 2

N b , a t %

%

~/ \ ~ 3 4

Fig. 5 - - T h e results of DTA analyses of the solidification behavior of the designed alloys. Contour lines of (a) P0 (Y + 7' ~ L), (b) Pt (L ~ 7), (c) P3 (Y ~ 7') , (d) AT (P, - Po), and (e) eutectic (7 + 7') volume fraction.

METALLURGICAL TRANSACTIONS A VOLUME 24A. NOVEMBER 1993--2455

E

Fig. 6--Microstructures of the alloys after DTA.

�9 i~i �84

N738I

Table II. Solidification Segregation Behavior for the Alloy System Studied (Segregation Ratio*)

Alloy A1 Ti Cr Co Nb M o Ta W Ni

T1 6.7 - - - 6 6 . 2 - 3 1 . 4 179.2 30.3 - - - 6 4 . 7 - 0 . 3 T2 6.5 - - - 3 . 5 - 5 . 0 146.9 63.6 50.0 - 4 7 . 6 - 5 . 4 T3 1.0 - - - 1.8 - 2 . 6 100.0 25.0 37.5 - 15.8 - 3 . 2 T4 4 .9 - - - 2 . 9 - 9 . 8 127.2 33.3 83.3 - 4 5 . 0 - 5 . 6 T5 5.2 - - - 3 . 5 - 7 . 7 - - 25.0 44.7 - 3 0 . 0 - 3 . 2 T6 6.3 54.5 - 9 . 0 - 7 . 5 109.4 20.0 - - - 4 0 . 0 - 3 . 4 T7 7.6 47.1 - 9 . 4 0 315.8 63.6 10.0 - 3 6 . 4 - 7 . 3 T8 1.0 91.7 - 1.1 - 11.9 218.2 20.0 100.0 - 4 3 . 5 - 6 . 7 T9 9.6 120.0 - 3 . 0 - 17,5 - - 36.4 104.3 - 6 5 . 0 - 5 . 5 T I 0 2.3 57.9 - 5 . 4 - 2 , 5 120.0 18.2 - - - 2 0 . 0 - 3 . 2 T I 1 4 .2 65 .0 - 1.8 - 12,2 70.0 9.1 100.0 - 3 4 . 8 - 1.9 T I 2 2.1 75 .0 - 2 . 8 - 9 , 5 - - 18.2 61.5 - 3 0 . 0 - 2 . 9 T I 3 7.5 87.5 - 6 . 0 - 16,3 163.6 16.7 0 - 6 8 . 4 - 4 . 3 T14 11.2 78.6 - 3 . 6 - 1 7 , 5 0 9.1 120.0 �9 - 5 5 . 0 - 2 . 7 T15 8.2 51.2 - 2 . 9 - 14,3 - - 0 .0 - - - 5 7 . 1 - 1.6

*Segregation ratio = [(DI-DC)/average composition] x 100 pct; DI = element composition in interdendritic region; and DC = element com- position in dendrite core region.

2456--VOLUME 24A, NOVEMBER 1993 METALLURGICAL TRANSACTIONS A

(a)

i66T~

C k l �84

(b) (c)

, , ~ 1 7 6 1 7 6

118908 20KV Xl,ggK 30um , , , , ~ 1 7 6

112904 20KV • 3gum

(d) (e) ( f )

112907 20K\ ,~ Xl.00k SOura

(g) (h) Fig. 7--Dis t r ibut ion of elements in the Nb-rich eutectic ('y + 7') in alloy T1 determined by EPMA in (a) morphology, (b) through (h) element distributions of AI, Nb, W, Cr, Mo, Co, and Ni, respectively.

weight change tendencies for the two alloys were ob- tained. It should be pointed out that in the present step of the study, polycrystalline specimens were used in the hot-corrosion tests. Such experiments usually give lower results compared to the experiments with single-crystal specimens. If single-crystal specimens were used, the

hot corrosion resistance of the designed alloys would be anticipated to be at the same or higher level compared to the polycrystalline IN738LC, which was shown in a latter investigation. 14~ This result shows that the newly developed single-crystal alloys will have hot corrosion resistance required by most of the practical applications.

METALLURGICAL TRANSACTIONS A VOLUME 24A, NOVEMBER 1993--2457

(a)

(b)

Fig. 8 - -Typica l microstructures after heat treatment in (a) T1 and (b) T l 1.

The microstructures and constituents of the corrosion scales were studied in detail. Figure 13 shows the micro- structures of the corrosion scales on the surfaces of the typical alloys with high and low hot corrosion resistance (alloys T6 and T5), respectively. In the Ti and Nb-rich corners, a thin layer (about 15/zm) of surface scale was formed during the hot corrosion process, which was composed of two parts, i .e. , the continuous and dense protective outside part and a somewhat porous sub- surface layer, while in the Ta-rich corner, a thick layer (about 150 /zm) of porous and nonprotective surface scale was formed. Figure 14 shows the results of ele- mental distributions in the surface scale in three typical alloys determined by EPMA. Table III gives the results

of the major corrosion products identified by X-ray dif- fraction with powders taken from the surface layers.

Figure 15 summarizes the results of this investigation. Two regions with promising comprehensive properties (except mechanical properties) were determined in the Ti-Ta-Nb compositional triangle. These regions were outlined by the contour lines of some very important fac- tors considered in the alloy design process, 19'l~ including solidification range (AT -< 65 ~ eutectic (3' + 3") vol- ume fraction (-<0.4 vol pct), density (p -< 8.7 g/cm3), and weight loss after the 24 hour crucible hot corrosion tests at 900 ~ (-<20 mg/cm2). The mechanical behavior is the subject of the third part of the study. 14~

V. DISCUSSION

It is very interesting to notice that in a wide compo- sitional range, the experimental alloys showed outstand- ing solidification features with a very narrow solidification range (AT -< 65 ~ Figure 5(d)) and al- most no eutectic (3' + 3") precipitation in the last stage of solidification (Figures 5(e) and 6), which indicates that the single-crystal castability and phase stability of the alloys studied would be fairly good. These obser- vations are the first such results in alloys with such high Cr and AI contents. This may be directly related to the elimination of the grain boundary strengthening minor elements C, B, and Zr. The severe segregation of these minor elements during solidification processing will widen the liquid/solid two-phase region and cause the precipitation of a great quantity of nonequilibrium eu- tectic (3' + 3',)t~81 (please compare the as-cast structures of these alloys (Figure 6) with those of the alloys with C, B, and Zr in Figure 3 in the previous articlem). Ti and Nb are the most severely segregated elements as de- termined by EPMA measurements (Table II). Higher amounts of the eutectic (3' + 3") solidified in the alloys at the Ti- and Nb-rich corners (Figure 6). As for the segregation of Ta, it seems to be overestimated by EPMA measurements because of the great error in de- termining the Ta content. 1221 Different effects of Ta on solidified eutectic (3' + 3") were reported in the litera- ture. In B - 1900 + Hf and MAR--M247 alloys, c22j Ta showed no effect on the total amount of the eutectic (3' + 7'), but changed the morphology of the eutectic (3' + 7') in B-1900 + Hf. In B-1900 without the Hf addition, t231 the amount of eutectic (3' + 3") was shown to increase with Ta addition. Therefore, the effect of Ta on solidification structures strongly depends on alloy compositions.

According to the observations, the 3" volume fractions seemed to be a little bit low in the experimental alloys. This could be improved by adding more alloying ele- ments A1, Ti, Ta, or Nb, because in the present alloying status, the Mdt values of the alloys were still lower than the phase stability boundary condition (Mdt -< 0.991). We can further increase the alloying elements to increase the Mdt values to the critical value. For example, the Mdt value for alloy T7 is 0.980. Therefore, up to 0.9 at. pct A1 can be added to the alloy to increase the Mdt to the critical value of 0.991, which will increase the 3" volume fraction and the high-temperature mechanical properties.

2458--VOLUME 24A, NOVEMBER 1993 METALLURGICAL TRANSACTIONS A

Ca) (b)

(c)

Fig. 9 - - T h e y' morphologies in the specimens after full heat treatment in (a) IN738LC, (b) T7, and (c) TI3.

,

%oGa" .. , '6 . 'w e,~o.s" i + . , * r dp'.m B.4~ +e' .+.*_S �9 +~'~.q'.lP~ O . .e.oo'_4 . _ . . ~ w - . . , ~ + , o - e e ~ -i,,w-~-.

~ + ~ % % + 4 1 ' % + �9 �9 + . ~ _ % % - 4~+, l . + . ~ ~r" o o - ~ , . 4 + %. ~--++++

~. ~ +, �9 ~, +.'+~_~++~i,+41 d+... �9 , , . ~ ql, 41k ~ - I ~ ~ 1 7 6 ~ t ~ --'JsO+

�9 +% +, ~ ~+ .~ko~.~++.e~ +,, r . , + + '+"--~L~7. . ,"~+--"+" o,e ~

(a) (b) (c)

Fig. 10--Typical y' morphologies in the specimens after prolonged aging at 900 ~ for 500 h in (a) T7, (b) TS, and (c) IN738LC.

METALLURGICAL TRANSACTIONS A VOLUME 24A, NOVEMBER 1993--2459

T i

Taro I ~ N b , a t %

Fig. I 1 - -Summar ie s of hot corrosion experiments with crucible tests at 900 ~ for 24 h in a molten solution of 75 pct Na_,SO4 + 25 pct NaCI.

15

%

d o

4J t.-

, r - ~) 5

0 IN738LC �9 T7

0 ! I I

0 I00 200 300

time, h

Fig. 12--Dependence of hot corrosion resistance upon an exposure time of a typical experimental alloy T7 and the reference alloy IN738LC.

Microstructural analyses showed that the designed al- loys had promising aligned t" phase structures which were stable after long-term aging compared to the IN738LC. This indicates that the designed alloys will be more suitable for the applications in high-performance, long-term marine, and industrial gas turbine engines.

It is not difficult to identify the differences of hot cor- rosion products and microstructures of the scales in the typical designed alloys, which would provide useful in- formation for understanding the corrosion mechanisms.

In the Ti-rich comer, a dense outside protective scale mainly composed of A1203 and small amounts of Cr203 and TiO2 was formed on the surface of the alloy, which delayed the penetration of oxygen, while in the sub- surface, Cr reacted with S, forming Cr3S4, which de- layed the penetration of S to the inner metal matrix. Therefore, the rate of corrosion reaction would be low- ered after the formation of the above protective scale, because the contact of O and S with the matrix metals had to depend on the diffusion of the elements through the scale. Similarly, in the Nb-rich comer, a dense out- side protective scale mainly composed of A1203 and Cr203 was formed. But in this case, the role of Cr was replaced by Nb. A layer of NbS2 forms on the sub- surface. NbS2 is a stable sulfide with a very negative Gibbs energy of formation, t4" It seems that Nb was more effective than Cr in gettering S. In a very narrow band, S was restrained by Nb (Figure 14). This may be one of the major reasons for the improved hot corrosion resis- tance of the alloys with Nb additions. On the contrary, in the Ta-rich corner, O and S could easily penetrate the porous and nonprotective scale to contact directly with the inner metal matrix and continue the corrosion pro- cess, and finally resulted in the catastrophic corrosion of the alloys.

In general, Ta can improve the high-temperature me- chanical properties and hot corrosion resistance of alloys with low Cr contents, t23'24,25j However, in the alloys with high Cr contents, the effect of Ta on hot corrosion re- sistance was reversed as proved by this study. The de- teriorated hot corrosion resistance of the alloys and the most severe hot corrosion attack were observed around the Ta-rich corner (Figure 11). As for the effect of Ti on hot corrosion resistance of the designed alloys, it was similar to the effect reported in the literature, i .e . , it im- proved the hot corrosion resistance of the nickel-base superalloy, t26'27j At the present time, no unanimous con- clusions have been drawn for the effect of Nb on hot corrosion resistance of superalloys, t:8~ Sims et al. 1291 showed the unfavorable effect of Nb on hot corrosion resistance, while alloy S-816, with up to 4 pct Nb, showed a satisfactory hot corrosion resistance at 8 7 0 ~176 Another example that should be noted is the well-known hot corrosion resistant superalloy IN738 with about 1 pct Nb added.

Therefore, the effects of alloying elements on hot cor- rosion resistance can change greatly. According to the alloy system and the actual experimental condition, re- versed behavior could result. An analysis of the results should be done according to the actual situations.

The element Cr is unanimously believed to be bene- ficial to the hot corrosion resistance of superalloys. It is the most effective and important alloying element for this p u r p o s e . 13u321 The element AI is also generally ben- eficial to the hot corrosion resistance of superalloys. By comparing the compositions of the experimental alloys (Table I), it can be seen that the Cr and A1 contents (weight percent) are almost the same. Therefore, the ef- fects of other elements should be taken into consider- ation. According to the test results, the effect of Ti and Nb may be important. Lewis and Smith t33j proposed an "effective chromium content" (EC) concept based on hot

2460-- VOLUME 24A, NOVEMBER 1993 METALLURGICAL TRANSACTIONS A

(a)

(b)

Fig. 13--Micros t ruc tures of the hot corrosion products formed on the surfaces in typical alloys with low and high hot corrosion resis- tance in (a) T5 and (b) T6.

corrosion tests with the crucible method of Nimonic alloy series, as

EC = Cr + 1.1 (Ti + Nb) + 0.7AI 11]

All components are in weight percent. The distribution of the contour lines of the EC values

in the Ti-Ta-Nb compositional triangle is shown in Figure 16. It is interesting that EC and hot corrosion re- sistance of the experimental alloys showed similar dis- tribution. In the Nb- and Ti-rich corners, EC values are also higher (23.9 pct and 21.8 pct for alloys T1 and T15, respectively). While in the Ta-rich corner, EC values are the lowest (EC = 16.03 pct for alloy T5). The alloy with the highest Nb content (alloy T1) has the highest EC value. But the alloy showing the best hot corrosion re- sistance is alloy T6 with an EC value (23.33 pct) lower than the alloy T1. This may have resulted from the se- vere segregation of Nb during solidification of the alloy T1, which caused a great amount of Nb-rich eutectic (3' + Y') precipitation in the interdendritic regions (Figure 7), Such nonequilibrium precipitates cannot be eliminated through post heat treatment (Figure 8). They consumed a certain amount of Nb and actually lowered

the effective EC value as well as the hot corrosion re- sistance of the alloy. It should be noted that the typical hot corrosion-resistant superalloy IN738 also has a high EC value (23.11 pct). Therefore, the hot corrosion re- sistance of the alloy systems with high Cr contents could be predicted to a certain extent with Eq. [1].

The hot corrosion process in an alloy is very compli- cated, and up to now the study of the mechanisms of the process is limited and not satisfactory in most of the cases. But the constituents and microstructures of the corrosion products in the surface layer of an alloy will doubtlessly provide important information about the mechanisms of the hot corrosion process of the alloy. According to the results of this study, the authors tried to explain the abnormal effect of Ta on the hot corrosion resistance of the alloys.

The most commonly used theory to explain the hot corrosion process is the acidic-basic fluxing mechanism proposed by Goebel eta/ . 134'35'36] The basis of this model is the loss of the protective function of the surface oxide layer because of the fluxing of the layer in the molten salt. In this process, the behavior of the 02- plays an important role. When fluxing of the oxide layer is caused by the process in which the oxides combine with 0 2- to form anions, it is called basic fluxing. When fluxing is caused by the decomposition of oxides into correspond- ing ions and O 2-, it is called acidic fluxing, which takes place when the O:- activity is markedly lowered in the molten salt. Since this model was proposed, it has been widely applied to explain the hot corrosion behavior of superalloys. In general, hot corrosion of the alloys with high AI and Cr contents occurs according to the basic fluxing mechanism. For the alloys with high W, Mo, or V contents, hot corrosion takes place according to the acidic fluxing mechanism. Although basic fluxing is still an accelerated oxidation process, it is not so severe as the catastrophic oxidation caused by the acidic fluxing mechanism. The alloys used in this study were based on the hot corrosion-resistant IN738LC alloy which con- tains high A1 and particularly high Cr contents. The hot corrosion of IN738LC occurred according to the basic fluxing mechanism. 1371 From the results of this study, the hot corrosion of the alloys around the Ti- and Nb-rich corners may also occur according to the basic fluxing model. However, for the alloys around the Ta-rich cor- ner, catastrophic hot corrosion controlled by acidic flux- ing took place. It seems hard to explain this result with the above fluxing model, because basic fluxing should occur in such alloys with the A1 and Cr contents so high. According to the results of analyses of constituents and the microstructures of the oxidation scales, this phenom- enon may be explained by an alternative basic-acidic fluxing model. I38.39I

During hot corrosion of the Ta-rich alloys, the basic- acidic fluxing processes took place alternately and promoted each other, finally causing the catastrophic corrosion. The results of the X-ray diffraction phase identification showed the existence of NaA102, NaNiO2, N a 2 S O 4 , A I W O 4 , and CrWO2 as the corrosion products in alloy T5 (Table III). These were also the reaction products as predicted by the alternating basic and acidic fluxing models. In the present study, the identity of the

METALLURGICAL TRANSACTIONS A VOLUME 24A, NOVEMBER 1993--2461

0 0 S o AL ~, , . , . - . , .~ 1

cr, Cr

Nb

S W ~ S

Co ~ C l ~

& 3 _l_ A i _I , , ,a | r ix I scale [ ma t r ix l sca le t n, atr ixl sca le ]

(a) (b) (c) Fig. 14--Element distribution (EPMA line analyses) in the corrosion scale in Nb-, Ta-, and Ti-rich alloys, (a) T6, (b) T5, and (c) T15.

Table III. Phases Identified by X-ray Diffraction of the Hot Corrosion Products Taken from the Surface Scales

of Some Typical Experimental Alloys

Phase T6 T5 T 15

Simple oxide NiO NiO AIzO3 AlzO3 C~O3 ChO3 (Cr205) (Cr205)

Complex oxide NaAIO2 NaA102 (NaNiO2) (NaNiO2) Na2CrO4 Na2CrO4

CrWO4 AIWO4

Sulfide NbS2 Cr3S4

NiO A[203 Cr203 (Cr205) TiO2 (Ti.~Os) NaAIO2 (NaNiO~) Na2CrO4

Cr384

T•

q, r

T~ ~b 0 I 2 3

Nb, at%

Fig. 15--Summaries of the comprehensive properties in the Ti-Ta-Nb compositional triangle, in which, two promising composi- tional areas (shadowed) were indicated for the selection of alloys for single-crystal growth and mechanical property characterizations.

2462-- VOLUME 24A, NOVEMBER 1993 METALLURGICAL TRANSACTIONS A

T i

T~O ~Nb 3 N b , at%

Fig. 16--Contour lines of EC values of the designed alloys in the Ti-Ta-Nb composihonal triangle.

exact role of Ta during the hot corrosion process is un- certain. This problem remains to be studied further. It is certain that the lower EC values for alloys at the Ta-rich corner are related to the lower hot-corrosion resistance of the alloys.

As summarized in Figure 15, the promising compo- sitional ranges were determined, in which most of the important factors considered in alloy development were included except for the mechanical properties. The al- loys in these promising ranges had the excellent com- prehensive properties including predicted single-crystal castability, microstructural stability, hot-corrosion resis- tance, and density. The only remaining goal for this alloy design and development study is to determine the alloy composition having the required mechanical prop- erties in the foregoing promising regions. The results will be given in another article./4~

VI. CONCLUSIONS

Through a systematic study of the effects of three re- fractory metals Ti, Ta, and Nb in the hot-corrosion resistant alloy system Ni-16Cr-9A1-4Co-2W-IMo-Ti- Ta-Nb (in atomic percent and with Ti + Ta + Nb = 4 at. pct), the following conclusions are drawn:

1. DTA analyses show that the solidification reaction temperatures and eutectic (3' + 3") volume fraction change systematically with compositions and can be expressed by the contour lines in a Ti-Ta-Nb com- positional triangle. Except for narrow regions around the Ti- and Nb-rich comers, the eutectic (3" + 3") volume fraction as well the solidification range are relatively low in the alloy system studied, which is beneficial to microstructural stability and single- crystal castability. At the same time, the alloys show a wide heat-treatment window (>100 ~ which makes for easier complete solution treatment.

2. After 900 ~ h long-term aging, no TCP phase precipitates in the experimental alloys. The alloys

can be further alloyed to improve the properties ac- cording to the phase stability critical condition (Mdt _< 0.991).

3. Microstructural observations show that the 3" mor- phologies in the alloys are fairly aligned cuboids. No obvious coarsening and agglomeration of the pani- cles are observed after long-term aging, which is ben- eficial to the improvement of the mechanical properties and the long-term property stability. Therefore, the alloys should be useful in applications where high temperature, long-term service is re- quired, such as in the marine and industrial gas turbines.

4. Solidification segregation analyses show that ele- ments Cr, Co, W, and Ni segregate to dendrite arms. Other elements Ti, Ta, Nb, Mo, and AI segregate to the interdendritic regions among which Nb and Ti segregate most severely. High amounts of Nb-rich eutectic (3' + 3") form at the Nb-rich comer.

5. In a certain compositional range, hot corrosion resis- tance of the alloys is fairly good. In the Nb and Ti- rich comers, hot corrosion resistance is improved because of the formation of dense protective surface scales. In the Nb-rich alloys, Nb combines with S, which effectively prevents the penetration of S into the inner part and enhances the hot corrosion resis- tance of the alloys. However, in the Ta-rich comer, a porous multiphase surface layer forms, which does not prevent catastrophic corrosion.

6. Through a comprehensive evaluation, the promising compositional regions with balanced properties have been determined, from which alloy compositions can be selected for further single-crystal growth and me- chanical property characterization. This finally can lead to hot corrosion-resistant single-crystal alloys with balanced comprehensive properties satisfying the alloy design goals.

A C K N O W L E D G M E N T

The authors are grateful to Drs. H. Ezaki and K. Matsugi for their assistance in the study and helpful discussions on the results.

REFERENCES

1. J.S. Zhang, Z.Q. Hu, Y. Murata, M. Morinaga, and N. Yukawa: Metall. Trans. A, vol. 24A, pp. 2443-50.

2. W.J. Boesch and J.S. Slaney: Met. Prog., 1964, vol 86, pp. 109-17.

3. D.E. Mongeau and W. Wallace: Scripta Metall., 1975, vol. 9, pp. 1185-90,

4. T. Yamagata, H. Harada, S. Nakazawa, and M. Yamazaki: Superalloys 1984, Seven Springs, Champion, PA, 1984, M. Oell, C.S. Kortovich, R.H. BricknetI, W.B. Kent, and J.F. Radavich, eds., TMS, Warrendale, PA, 1984, pp. 157-66.

5. H. Harada, K. Ohno, T. Yamagata, and M. Yamazaki: Superalloys 1988, Seven Springs, Champion, PA, 1988, DN. Duhl, G. Maurer, S. Antolovich, C. Lund, and S. Reichman, eds., TMS, Warrendale, PA, 1988, pp. 733-42.

6. T. Ohno, R. Watanabe, and K. Tanaka: Tetsu-to-Hagane, 1988, vol. 74, pp. 2193-200.

7. M. Morinaga, N. Yukawa, H. Adachi, and H. Ezaki: Superalloys 1984, Seven Springs, Champion, PA, 1984, M. Gell, C.S. Kortovich, R.H. Bricknell, W,B. Kent, and J.F. Radavich, eds., TMS, Warrendale, PA, 1984, pp. 523-32.

METALLURGICAL TRANSACTIONS A VOLUME 24A, NOVEMBER 1993--2463

8. N. Yukawa, M. Morinaga, H. Ezaki, and Y. Murata: High Temperature Alloys for Gas Turbines and Other Applications 1986, Leige, Belgium, Oct. 1986, D. Reidel Publishing Company, Dordrecht, Holland, 1986, pp. 935-44.

9. N. Yukawa, M. Morinaga, Y. Murata, H. Ezaki, and S. Inoue: Superalloys 1988, Seven Springs, Champion, PA, 1988, D.N. Duhl, G. Maurer, S. Antolovich, C. Lund, and S. Reichman, eds., TMS, Warrendale, PA, 1988, pp. 225-34.

10. K. Matsugi, R. Yokoyama, Y. Murata, M. Morinaga, and N. Yukawa: Proc. Int. Conf. on High Temperature Materials for Power Engineering 1990, C.R.M., Liege, Belgium, 1990, pp. 1251-60.

I 1. M. Morinaga, N. Yukawa, T. Maya, K. Sone, and H. Adachi: Proc. 6th Worm Conf. on Titanium, Society Francaise de Metallurgie, Paris, France, t988, pp. I601-06.

12. N. Yukawa, M. Morinaga, K. Nishigama, Y. Matsumoto, Y. Murata, and H. Ezaki: Reduced Activation Materials for Fusion Reactors, ASTM, Philadelphia, PA, 1988, pp. 30-46.

13. M. Morinaga, J. Saito, N. Yukawa, and H. Adachi: Acta Metall. 1990, vol. 38, pp. 25-29.

14. M. Gell, D.N. Duhl, D.K. Gupta, and K.D. Sheffler: J. Met., 1987, vol. 19, pp. 11-15.

15. N.S. Stoloff: The Superalloys, C.T. Sims and W.C. Hagel, eds., Wiley--Interscience, John Wiley & Sons, New York, 1972, p. 79.

16. M.V. Nathal and L.J. Ebert: Metall. Trans. A, 1985, vol. 16A, pp. 1849-62.

17. K. Matsngi, R. Yokoyama, Y. Kawamura, H. Ishihara, Y. Murata, M. Morinaga, and N. Yukawa: Report of the 23rd Committee on Heat-Resisting Metals and Alloys, Japan Society for the Promotion of Science, Tokyo, Japan, 1989, vol. 30, pp. 277-86.

18. Y.S. Zhu, S.N. Zhang, L.Y. Xu, J. Bi, Z.Q. Hu, and C.X. Shi: Superalloys 1988, Seven Springs, Champion, PA, 1988, D.N. Duhl, G. Maurer, S. Antolovich, C. Lund, and S. Reichman, eds., TMS, Warrendale, PA, 1988, pp. 703-12.

19. M. Kawakami, K. Goto, R. Raba, and M. Kajiyama: Tetsu-to- Hagane, 1979, vol. 65, pp. 811-19.

20. R.F. Decker and C.T. Sims: The Superalloys, C.T. Sims and W.C. Hagel, eds., Wiley-Interscience, John Wiley & Sons, New York, 1972, p. 33.

21. F.C. Hull: Met. Prog., 1969, vol. 96, pp. 139-40.

22. G.M. Janowski, R.W. Heckel, and B.J. Pletka: Metall. Trans. A, 1986, vol. 17A, pp. 1891-1905~

23. Z.Y. Meng, G.C. Sun, M.L. Li, and X.S. Xie: Superalloys 1984, Seven Springs, Champion, PA, 1984, M. Gell, C.S. Kortovich, R.H. Bricknell, W.B. Kent, and J.F. Radavich, eds., TMS, Warrendale, PA, 1984, pp. 563-72.

24. G.C. Fryburg, C.A. Stearns, and F.J. Kohl: J. Electrochem. Soc., 1977, vol. 124, pp. 1147-53.

25. H. Warlimont and L. Delaey: Progress in Materials" Science, Pergamon Press, Oxford, England, 1974, vol. 18, pp. 36-41.

26. J. Billingham and J. Lanridsen: Metallography, 1973, vol. 6, pp. 85-100.

27. N.S. Bornstein, M.A. DcCrescente, and H.A. Roth: Metall. Trans., 1973, vol. 4, pp. 1799-1810.

28. C.T. Sims: Niobium, TMS-A1ME, Warrendale, PA, 1984, pp. 1169-94.

29. C.T. Sims, P. Berbman, and A. Beltran: Naval Eng. J., 1970, April, pp. 39-45.

30. A.M. Beltran: Cobalt, 1970, vol. 46, pp. 3-11. 31. C.W. Wallwork: Rep. Prog. Phys., 1976, vol. 39, pp. 401-10. 32. M. Yoshiba, O. Miyagawa, and D. Fujishiro: Tetsu-to-Hagane,

1981, vol. 67, pp. 996-1005. 33. H. Lewis and R.A. Smith: Proc. 1st Int. Conf. on Metallic

Corrosion, Butterworth's, London, 1965, pp. 202-07. 34. J.A. Goebel and F.S. Pettit: Metall. Trans., 1970, vol. 1,

pp. 1943-54. 35. J.A. Goebel and F.S. Pettit: Metall. Trans., 1970, vol. 1,

pp. 3421-29. 36. J.A. Goebel, F.S. Pettit, and G.W. Goward: Metall. Trans.,

1973, vol. 4, pp. 261-78. 37. G.C. Fryburg, F.J. Kohl, and C.A. Stearns: Proc. Symp. on

Properties of High Temperature Alloys, Z.A. Foroulis and F.S. Pettit, eds., The Electrochemical Society, Pennington, NJ, 1976, pp. 585-94.

38. M. Kawakami, A. Baba, C. Ikeda, N. Yukawa, and M. Morinaga: Proc. 1988 MRS Int. Meeting on Advanced Materials, Tokyo, Materials Research Society, Pittsburgh, PA, 1988, vol. 4, pp. 431-36.

39. K. Matsugi, M. Kawakami, Y. Murata, M. Morinaga, and N. Yukawa, Tetsu-tu-Hagane, 1991, vol. 77, pp. 1503-11.

40. J.S. Zhang, Z.Q. Hu, Y. Murata, M. Morinaga, and N, Yukawa: paper in preparation.

41. S.R. Shatynski: Oxid. Met., 1977, vol. 11, pp. 307-22.

2464--VOLUME 24A, NOVEMBER 1993 METALLURGICAL TRANSACTIONS A