Oxidation behaviour of TiAl3 coatings and alloys

Oxidation behaviour of TiAl3 coatings and alloys

Corrosion Science,Vol. 35, Nos 5-8, pp. 1199-1208, 1993 Printed in Great Britain. 0010-938X/93 $6.00 + 0.00 Pergamon Press Ltd O X I D A T I O N B E...

508KB Sizes 11 Downloads 187 Views

Corrosion Science,Vol. 35, Nos 5-8, pp. 1199-1208, 1993 Printed in Great Britain.

0010-938X/93 $6.00 + 0.00 Pergamon Press Ltd

O X I D A T I O N B E H A V I O U R OF TiA13 COATINGS A N D ALLOYS J. L. SMIALEK NASA Lewis Research Center, 2100 Brookpark Rd, Cleveland, OH 44135, U.S.A.

Abstract--Various thicknesses of TiAI 3 coatings were produced on Ti3AI by pack aluminizing at various processing conditions. Thick coatings affected oxidation by increasing the amount of cracking and oxidation within the coating. Bulk TiA13 and AI67XsTi25 tau-phase alloys were oxidized in cyclic and isothermal oxidation from 800 to 1200°C. Anomalously high oxidation rates for TiA13 were related to second phase aluminum oxidation and were eliminated by alloying. Chromium had an overall beneficial effect, while manganese additions could be catastrophic. Cyclic oxidation below 1000°C proved to be excellent for all alloys, while at 1200°C, spalling to bare metal, aluminum depletion and excessive TiO 2 formation were related to more excessive degradation.

INTRODUCTION

TITANIUM base alloys are being developed for high temperature aircraft engine applications because of lower density compared to nickel superalloys. However, after exposures to temperatures above about 800°C, these alloys suffer oxygen embrittlement or rapid oxidation. There is therefore a clear nead for oxidation resistant coatings on titanium alloys. Published coating studies have generally focussed on pack aluminizing to produce a TiA13 layer. This and, in some cases, TiA1, represent the only prospects for an oxidation resistant phase in the Ti-Al binary system. The purpose of this report is to review key aspects of the oxidation of TiAla-based compounds with potential as either coatings or structural alloys. EXPERIMENTAL METHOD Ti3AI + Nb (Ti-19AI-11Nb at%) plates were produced by hot pressing powder. 1 They were sectioned into coupons 1.3 m m x 1.3 c m x 1.3 cm. The coupons were pack aluminized at 800 or 1038°C, from 4 to 25 h, with pure AI, AI-12Si or AI-25Cr sources and NaCI or NaF activators. The coatings were evaluated in 982°C, 200 h cyclic oxidation tests. 1 Bulk TiAI3, TiAI 3 +5Cr-1.6Si~).05Y, and the latter alloy with 12Nb substituted for Ti were produced by arc melting and drop casting in a copper hearth. Hot, isostatically pressed AI67X8Ti25 (X = Cr,Mn,V) tau-phase alloys were obtained from D. Mikkola of Michigan Technological University (MTU). Forged Cr and Mn modified tau alloys were obtained from S. Kumar of Martin Marietta Labs. Coupons were sectioned as before, polished to 600 grit, and isothermically or cyclically oxidized in the range 800-1200°C. The oxidation behaviour was characterized by weight change, X-ray diffraction (XRD) of surface phases, scanning electron microscopy (SEM) and metallography. EXPERIMENTAL

RESULTS AND DISCUSSION

Pack aluminide coatings

All the packs produced the TiAl 3 phase. The amount of pickup ranged from 2.5 to 35 mg cm -e. Thicker coatings resulted primarily from the NaF activator and longer times and secondarily from the amount of A1, Al activity and temperature. Thicker coatings exhibited greater oxidation weight gains after 200 1-h cycles at 982°C in air (Fig. 1). A broad minimum in oxidation weight gain occurred between 1199

1200

J.L. SMIALEK

~E

Bulk TiAI 3 baseline

2

"~

B

'~ 1 bCL

:|1 |1

0

r,

I 10

_

N

7

R,~I~

I 20

~

800Cpacks

I 30

I 4O

Coatingthickness,mg/cm2 Fro. 1,

Final weight change at 200 h for pack aluminized Ti3AI + Nb in 982°C cyclic oxidation.

about 8 and 15 mg cm -2 of aluminum pickup (corresponding to about 40-70/~m of TiA13). XRD identified a-A120 3 as the major oxide, with only minor amounts of TiOz in almost all cases. Only the thinnest coating (Fig. 1, curve B) produced a substantial amount of TiOz. SEM studies of the scale revealed a surface of fine nodules or needles of AI203 decorated with a network of ridges composed of large TiO2 crystals over cracks in the coating. Metallographic cross-sections showed that the cracks extended through to the substrate in all cases and were in turn filled with oxide. For the thicker coatings, a wedge of oxide (bulb feature) also formed beneath the coating in the substrate (Fig. 2). In this particular case (coating D), the thickest coating for some reason did not have TiO2 crystals above the cracks. Thus the minor amount of TiO2 in ridges at the scale surface could not account for all the weight increases shown in Fig. 1. An attempt was made to account for the oxide inside and beneath the cracks. The crack and bulb dimensions were measured for each open crack in the cross section. The data in Fig. 3 show a definite trend in crack and bulb size with coating thickness. The total amount of oxide contained in the cracks was then determined, based on simple geometric shapes and assuming a randomly distributed angle of crack intersection with the cross-section. The weight of oxygen calculated to be in the ct-A1203 in the cracks, summed with that in the external scale, is shown as a function of oxidation weight gain in Fig. 4. The calculated amount (solid line) corresponds closely with the expected amount (dashed line). The increase in oxidation weight with coating thickness (Fig. 1) is therefore a result of the deeper, wider cracks in the thicker coatings. Although these cracks were present before oxidation, cyclic oxidation produced a widening effect. Oxidation o f bulk TiA13 alloys

Parabolic plots of the isothermal weight change vs t 1/2of cast TiA13 oxidized in dry 1 atm 02 are shown in Fig. 5. Typical parabolic behavior and temperature effects are seen above 1000°C. However, below 1000°C, two branches are present in the curves. Regression fits were made to these branches and initial, kp,i, and final, kp,f, rates were determined, z An Arrhenius plot of kp,i and kp,f is shown in Fig. 6. At high

FIG. 2.

Metallographic cross-section of thick TiAI3 coating (D) showing wide crack filled with oxide and substrate oxidation beneath crack.

120~

Oxidation of TiAI3

1203

fA

-

Crack depth ~

~150 -

Bulb diameter

F

--

~. ~ c

0

c r ~ c > . /. . . . . o - - - - - o r~.O " ~ ' ~ / ~ Crack width . ~ " 10

I 30

20

-

I 40

Aluminum pickup, mg/cm 2

F~. 3. Variation in crack and oxide dimensions with coating thickness.

temperature a close correspondence to NiA1 oxidation is seen. However, at low temperatures both branches are considerably higher than classical alumina growth kinetics and actually approach TiO2 kinetics. However, all surface scales were identified as a-Al203 from 600 to 1300°C. Here again the explanation was found in the microstructure as a second phase of aluminium metal was present in the casting. This was the result of a peritectic decomposition upon solidification and is a common observation for chill-cast TiA13. Attempts to remove this phase by annealing at 1100°C, hipping at 1200°C or directional solidification resulted in interconnected Kirkendall porosity and an anomalously high oxidation rate due to the extra surface area of the pores. Thus the determination of accurate oxidation kinetics of TiA13 at lower temperatures remains a problem. The normal behavior at high temperatures results from more rapid oxidation and sealing over the second phase aluminium regions. The following discussion deals with a preliminary screening study for 8Cr, 8Mn and 8Mn-3V modified TiA13 alloys. These modifications result in the L12 cubic structure (tau-phase) rather than the tetragonal DO22found for the binary. The DO22 aluminides are typically very brittle, due in part to limited symmetry and

•~>¢

3 ~ --

O

Prediction

E

N

Js ~ "

~

L

~ R C.s

ti> O

Q ~ S

S'~

/ ~

Outer scale only

~ / / / / / / / / / / / / / / / / / / ~ ~B I I 1

2

I 3

Actual oxygen uptake (weight gain), mg/¢m 2

Flo. 4. Oxygen weight pickup calculated from filled crack volumes compared to measured weight gain.

1204

J.L. SMIALEK oc

1.8 1.6 ~E 1.4

~,

~.o .a

-

aoo

-

900

.6

0

1

2

3

4 5 6 Time1/2, hrl/2

7

8

9

10

FIG. 5. Effect of temperature on the isothermal oxidation kinetics of cast TiAI3 with time 1/2.

I

I

\ \

I

\\ \

I

I

I

/._ Ti3AI

I I

i-2

/ - - Ti

I I

1 NIAI (Zr) .~

, \ _ TiAI31

--4

(~r)

~-- TUU3f -5 / - - TiAI

-6 6

Fl~. 6.

I

I

7

8

I

I

9 10 1/T 10 -4 °K-1

I 11

12

Arrhenius plot of cast TiAI3 initial and final kp values compared to NiAI and other Ti-A1 alloys.

Oxidation of TiAI3

1205

10 59.2

.00c

D

1000C

• ,2oocI ~6 E

== ¢o

~

4

2

0 TIAI 3 +CrSiY

+Nb

MTU-AI

-Cr

-Mn

-Mn,V

MM-Cr MM-Mn

TiAIa alloys

FIG. 7.

Comparison of final 100-h weight change for isothermal oxidation of various TiAI 3 and tau-alloys from 800 to 1200°C.

independent slip systems. However, there has been considerable effort to develop ductile, strong AI67XsTi25 tau alloys that possess the required symmetry. Gravimetric results were obtained in 100-h isothermal tests at 800, 1000 and 1200°C and in 200-h cyclic tests at 982 and 1200°C. The isothermal behavior often exhibited large initial rates or non-parabolic behavior, making kp comparisons difficult. The results are summarized, therefore, only by final weight change in Fig. 7. The first alloy grouping shows the TiA13 baseline data and two in-house alloys designed to match the successful 5Cr-I.6Si-0.05Y alloying approach taken for the DO22 NbA13 alloy. 3 Results for a similar alloy but with Nb substituted for half the Ti are also presented. One of the major effects of this alloying is to form DO22 + L12 with no aluminum and eliminate the interior oxides. This produced a lower weight gain at 800°C for the CrSiY alloy. All the tests produced primarily a-Al20 3 scales with a small amount of 3Y203-SA120 3 for the Cr-Si-Y alloy at 1200°C and TiO2 for the Nb-Cr-Si-Y alloy at 800 and 1000°C. The binary alloy exhibited a Widmanstatten array of TiA12 platelets after 1200°C isothermal oxidation, as revealed when the scale spalled to bare metal. This is a nonuniform means of accommodating aluminum depletion in a line alloy. The hipped materials obtained from MTU are shown in the second grouping. The binary alloy exhibited rapid oxidation due to porosity, the Cr-modified alloy was extremely oxidation resistant at the lower temperatures, and the Mn-modified alloys exhibited catastrophic rates at 1000°C. The forgings from Martin Marietta are shown in the third grouping and have the best overall behavior due to the lack of porosity, second phases or chemical inhomogeneities. A small amount of TiO2 formed on the Mn-modified alloy. The 982°C cyclic oxidation data in Fig. 8 show much simpler overall behavior. All

1206

J.L. SM1ALEK

t~

~2

2.01

d

1.51

.C

0.32

TiAI 3 +CrSIY

F1G. 8.

+Nb

MTU-AI -Cr -Mn TiAI 3 alloys

-Mn,V

MM-Cr MM-Mn

Comparison of final weight change after 200 h of cyclic oxidation of various TiAI3 and tau-alloys at 982°C.

the modified alloys exhibited significantly lower weight changes than the binary alloy. Also the binary alloy exhibited a significant amount of TiO2 compared to the other alloys, which formed primarily a-A1203 scales. The good cyclic behavior of the MTU Mn-modified alloys is at odds with the poor isothermal behavior at 1000°C and no explanation appears plausible at this time. These alloys were also tested for 1000 h at 815 ° and 982°C and the weight gains were generally under 0.4 and 0.7 mg cm -2, respectively. The situation is more complex for 1200°C cyclic oxidation (Fig. 9). The binary alloy exhibited breakaway oxidation due to excessive amounts of TiO2 formation. This resulted from the oxidation of TiA12 plates formed by aluminum depletion and exposure due to spalling. The Cr-Si-Y and Nb-Cr-Si-Y modifications were improvements, but well away from optimized alumina-formers. The higher Cr MTU and Martin Marietta alloys were clearly the best and exhibited only a small amount of TiO 2. Spalling to bare metal is still a concern here, although it is much improved over the case for the binary alloy. The MTU Mn-modified alloys were both worse than the binary alloy and exhibited a large amount of both TiO2 and A120 3 formation and minor amounts of Mn304 .4 Finally, the forged Martin Marietta Mn-modified alloy now shows higher weight gains than the forged Cr alloy, as opposed to the similar behavior in Figs 7 and 8. The high temperature cyclic behavior represents the largest difference between TiA13 alloys and other oxidation resistant alumina formers, which generally show minimal growth and spallation after 200 cycles to 1200°C. Even the tau alloys containing Cr exhibited a weight loss and bare metal spalling. It should also be noted that every bulk alloy tested spalled completely to bare metal after the ll0-h

Oxidation of TiAI3

1207

20

62.0

1

99.4

(;

(lo I~r)

15

10.3 E 10 6.8

E s

I

i,ii!iii

._~

-1.5

-2.0

-6.3 -10 f TiAI 3 +CrSIY +Nb

MTU-Cr -Mn

-Mn,V

MM-Cr MM-Mn

TiAI3 alloys FIG. 9.

Comparison of final weight change after 200 h of cyclicoxidation of various TiAI3 and tau-alloys at 1200°C.

isothermal exposure at 1200°C. Since the differential thermal expansion between tau- and a-A1203 (15 vs 8 x 10-6°C -1) is similar to that for nickel-base alloys, the stress on the scale is.not likely to be any larger. Furthermore, the presence of yttrium in two of the alloys did not prevent interfacial spalling. On the other hand T i - 1 5 C r 50AI alloys (which have a good coefficient of thermal expansion match, but no yttrium) appear to form quite adherent scales. There is a need for a greater understanding of alumina scale adhesion for TiA13 alloys. ASSESSMENT The TiA13 coating that forms quite easily upon pack aluminizing under a number of conditions is also quite oxidation resistant at intermediate temperatures. Its greatest detriment lies in its propensity for cracking on titanium aluminide substrates. The cracks account for all the anomalous weight gains and direct attack or embrittlement of the substrate. Successful application in the 1000°C regime could occur if the coatings are between 40 and 70 ~tm thick (1-3 mg cm-a) and if the cracks could be prevented from propagating into the substrate under stress. Modifications to increase ductility or improve the C T E match would alleviate the crack problem, but this appears unlikely or difficult. The bulk TiA13 or A167X8Ti25 alloys have a propensity to be alumina formers, but are compromised in a number of ways. The binary alloy suffers from processing artifacts when cast, but appears to oxidize with slow kinetics at the higher temperatures or when produced as a coating. Alloying eliminates this problem by producing a different phase that is not a peritectic line compound. However, all the compositions studied suffer from a tendency to bare metal spalling after high temperature

1208

J.L. SMIALEK

oxidation. In the binary alloy, an aluminum-poor phase is exposed by spalling, which then triggers TiO 2 growth. Mn-modified alloys are subject to deterioration by an unknown mechanism, especially for the castings. Cr modifications clearly offer the most promise. While these undesirable effects occurred in accelerated tests at high temperature, low temperature (815 and 982°C) cyclic tests of forged alloys indicate oxidation lifetimes in excess of thousands of hours. This is the most important engineering result, since the titanium aluminide class of alloys is intended for use under 1000°C. REFERENCES 1. J. L. SMIALEK,Scripta Metall. Mater. 24, 1291 (1990). 2. J. L. SM1ALEKand D. L. HUMPHREY,Scripta Metall. Mater. 26, 1763 (1992). 3. M. G. HEBSUR, J. R. STEPHENS, J. L. SMIALEK, C. A. BARRETTand D. S. Fox, Oxidation of HighTemperature Intermetallics (eds T. GROBSTEINand J. DOYCHAK), p. 171. The Minerals, Metals and Materials Society (1989). 4. L. J. PARFITT,J. L. SMIALEK,J. P. NIC and D. E. MIKKOLA, Scripta Metall. Mater. 25,727 (1991).