You are on page 1of 9

Materials Science and Engineering, A 152 (1992) 9-17

Thermodynamic calculation of the ternary Ti-A1-Nb system


U. R. K a t t n e r * a n d W. J. B o e t t i n g e r Metallurgy Division, National Institute of Standards and Technology, Gaithersburg, MD 20899 (USA)

Abstract
Phase equilibria of the ternary T i - A I - N b system are dominated by the large range of homogeneity of (fl-Ti,Nb), the binary intermetallic compounds of the Nb-AI and Ti-AI systems and the formation of two ternary compounds. The available ternary experimental data, together with a thermodynamic extrapolation of the ternary system from the binary systems, have been used to calculate the ternary phase diagram. The model descriptions of the Gibbs energies of most of these compounds are given by the existing calculations of the binary systems. In order to model a phase which is present in only one binary system, but has a ternary homogeneity range, a hypothetical phase with the same structure was analytically described for each binary system. Such a phase would, of course, be metastable in the other binary systems. Constraints on the Gibbs energies of formation were derived from the crystal structures of the corresponding ordered compounds. These same constraints were employed for the corresponding phases in the ternary system. In a final optimization step, ternary parameters were introduced and adjusted to the available experimental data. The as-derived description of the ternary T i - A I - N b system can be used to estimate single or multiphase fields and thermodynamic quantities where no experimental data are yet available. It is also useful as an indicator of problem areas for which additional experimental data are required.

1. Introduction The Ti-A1-Nb system is of interest for the development of high temperature/low density intermetallic materials. Knowledge of stable and metastable phase relations in multicomponent materials provides valuable information for the development of processing strategies for these materials. Once derived, this thermodynamic description can be used to provide information which is not easily accessible through experimental techniques, such as metastable extensions or the temperatures where two phases have the same Gibbs energies at a given composition ( T, curves). Despite their importance, the phase relationships of this system are only partially known for certain temperature regimes [1-8]. From the available experimental data the following points can be determined. (1) The intermetallic compounds of the binary Nb-A1 and Ti-A1 systems have relatively wide ranges of homogeneity in the ternary system. (2) The compounds NbA13 and T i A I 3 form a continuous (D022)solid solution. (3) At least two ternary compounds exist near the compositions Ti2A1Nb and Ti4A13Nb. (4) The (fl-Ti, Nb) phase with b.c.c, structure orders to the (fl0-Ti, Nb) with B2 (CsCI) structure.
*Also at: Department of Materials Science and Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA. 0921-5093/92/$5.00

(5) A miscibility gap may exist in the ordered (fl,Ti,Nb) phase. The available ternary experimental data, together with a thermodynamic extrapolation of the ternary system from the binary systems, have been used to calculate the ternary phase diagram. This attempt at modeling the Ti-AI-Nb system must be viewed as preliminary. However, it provides a basis for the combination of data from several sources in a thermodynamically consistent manner and indicates composition and temperature regimes where further experimentation is essential. A continued interchange between calculation and experimentation is the quickest route to the true diagram.

2. History of calculations of the Ti-AI-Nb system Previous calculations of the Ti-A1-Nb system were based on the then accepted calculations of the binary systems and available experimental data. In a first calculation of the Ti-A1-Nb system [9], the thermodynamic descriptions of the binary systems of Nb-AI and Ti-A1 by Murray [10, 11] and of Ti-Nb by Kaufman and Bernstein [12] were used. Since this calculation, the experimental data and thermodynamic descriptions of the binary Ti-A1 and Nb-A1 systems have been subject to revision. For the calculation of the Nb-AI system, Murray [10] used phase diagram as well
1992--Elsevier Sequoia. All rights reserved

l0

U.R. Kattner, 14/.J. Boettinger

/
o

Terna~ T i - A I - N b system

as thermodynamic data to derive the thermodynamic description for this system. A re-evaluation of the experimental data for this system suggested that some of the thermodynamic data are inconsistent with the phase diagram data. The calculation based on this reevaluation yielded a simplified thermodynamic description and a more accurate fit between the calculated and experimental phase boundaries [13]. For the Ti-AI system, new experimental results showed that (a-Ti) is in equilibrium with the liquid phase and higher liquidus temperatures were established for the composition range of 35-80 at.% Al. These changes in the experimental phase diagram are reflected by a series of calculations in the literature (see Kattner et al. [14] for a summary). Since the accuracy of the thermodynamic descriptions of the binary systems is crucial for the calculation of a ternary phase diagram, the most recent calculations of the binary Nb-AI and Ti-A1 systems were used for a new approximation of the Ti-AI-Nb system. These three calculated binaries are shown in Fig. 1.

r...

L
--~

?8 u,I

o 0)

b-

ro

(iTi,) (AJ
10 20 30 40 , 50 , , 60 , f 70 , 80 , f 90 , 100

T"

At.% AI

AI

(.3
o

~3 0

3. Available experimental data


The available experimental data for the Ti-AI-Nb system are sparse when compared with the complexity of this system. Most of the experimental information is available for the isothermal section at 1200 C. The tieline data of refs. 6 and 15 are in good agreement and establish the phase boundaries with sufficient accuracy. The data of ref. 2 for the (fl-Ti, Nb)-Nb3A1-Nb2AI and Nb2AI-TiAI-(Ti, Nb)A13 three-phase equilibria are in general agreement with the phase boundaries indicated by refs. 6 and 15. It is noteworthy to mention that the tie-line data of ref. 6 indicate an isolated single-phase area near the composition Nb~0A145Ti45. This single phase is believed to have the B2 (CsC1) structure. It is also known that the (fl-Ti, Nb) phase undergoes an ordering transition from the disordered b.c.c, structure to the ordered B2 structure (fl0-Ti, Nb) in the ternary system [1, 3, 6]. The occurrence of two single-phase fields with the same crystal structure is suggestive of a miscibility gap. Most of the experimental data which are available for the 1100 C section determine the (fl-Ti,Nb)-Ti3A1 boundary. The data of Perepezko [7] show a relatively wide two-phase area, while the data of Muraleedharan and Banerjee [4] show a relatively narrow one. Further experimental data report a tie-line for (/30Ti, Nb)-Nb2AI [1], as well as data for two tie-triangles (/30-Ti,Nb)-Nb2A1-TiAI [16] and (fl0-TiNb)-Ti3AlTiA1 [5]. The compositions of the (fl0-Ti,Nb) phase reported for these two three-phase equilibria are contradictory to the extrapolation of the phase bound-

iI I
b o

b)

o 1
o
I I I I I I

~NbA

Is

W)
100

1'0

2'0

30

4'0

5'0

6'0

7'0

80

90

Nb

At.% AI

AI

~o
2
o

p--

8-

10

2'0

'

3'0

'

4'0

'

5'0

'

6'0

'

7'0

'

8'0

'

g'0

'

100

Ti

At.% Nb

Nb

Fig. 1. The calculated binary phase diagrams. (a) Ti-AI, (b) Nb-AI and (c) Ti-Nb.

ary data given by Perepezko [7]. This conflict can be resolved if the assumption is made that two singlephase areas of the (fl0-Ti,Nb) phase also exist at this temperature. In this case one (fl0-Ti, Nb) would be

U. R. Kattner, W. J. Boettinger

Ternary T i - A I - N b system

11

observed in the three-phase equilibria [8], while the other (/30-Ti,Nb) was observed by Perepezko [7]. For temperatures other than 1200 and 1100 C, only few tie-line data are available. For 700 C, Bendersky et al. [8] reported tie-line data between (/3-Ti,Nb) and Ti2AINb, and Bendersky et al. [5] reported data between TiaAI3Nb and TiA1 in equilibrium with a trace of Ti3AI. Experimental data involving the liquid phase are available only for the TiAl3-rich part of the quasibinary NbA13-TiAi~ section [6]. Based on the observation of primary fields of crystallization and preliminary differential thermal analysis (DTA) results, Perepezko et al. [6] proposed an estimated liquidus projection. Experimental results have established the existence of two ternary compounds: Ti2A1Nb [17] and Ti4AI3Nb [5]. The Ti~A1Nb compound has an orthorhombic ordered structure. It is based on ternary ordering of the D0z9 structure of Ti3AI which is an ordered derivative of h.c.p. The Ti4A13Nb compound has the B82 (Ni2In) structure and is believed to have a relatively small range of homogeneity. Neither ternary compound was observed in alloys that were annealed at 1100 C [5, 8] or higher temperatures. Bendersky et al. [8] reported the occurrence of the Ti2AINb compound in samples that were annealed at and below 970C, and the Ti~AI~Nb compound was observed in samples that were annealed at 700 C [5].

described with a sublattice model, considering two or more sublattices and allowing substitution to occur on at least one of them. Phases for which few data were available on their homogeneity ranges were assumed to be stoichiometric. For the present calculation, the program PMLFKT by Lukas et al. [18] was used. In this program the different model descriptions can all be treated with an equation of the form G ' = E a E Yk ( G k + R T l n ( y k ) )
/ k /

+ ~ Alypy,lvz,
I
ill It 0

v,i"

+ Z Boyp Y,I Y,
jj

with
v v = yp + ( 1 - yp - yq)

and
t',t = Yq + ( 1 - yp - yq )

4. Analytical description of the phases

The phases considered in the calculation of the Ti-AI-Nb system are the disordered solution phases (liquid, (/3-Ti,Nb), (a-Ti) and (AI)), the ordered intermetallic compounds of the Nb-AI and Ti-A1 binaries (Nb3AI, Nb2A1, (Ti, Nb)AI3, Ti3A1, TiA1, TiA12 and Ti2AIs) and the ordered ternary compounds (Ti2A1Nb and Ti4AI3Nb). Since the transition between disordered (fl-Ti,Nb) and ordered (flo-Ti,Nb) is believed to be a second-order transition, and in order to keep the analytical description as simple as possible, these two structures will not be distinguished in the present calculation; both are represented as the (fl-Ti, Nb) phase. For the calculation of phase equilibria, the Gibbs energies (G) of the phases present must be expressed as analytical functions of the variables of interest, i.e. composition and temperature. For the calculation of the ternary system, the type of equation (model) for the phases existing in the binaries has been predetermined by the thermodynamic description used in the calculation of the binary systems. In all three binary systems the disordered solution phases were described with quasi-subregular solution models. The ranges of homogeneity of the ordered intermetallic compounds were

where i is the phase index; l is the sublattice index; k, p, q, r are the species indices; j is the polynomial index (binary terms); 1) is the polynomial index (ternary terms); G i is the molar Gibbs energy of phase i; a zis the fraction of sublattice l in phase i; Yk, Y~,,Y,~,Yr are the concentration of species, referred to 1 mole of sublattice sites; Gk is the Gibbs energy of species k; Aj is the Gibbs energy coefficients of binary polynomial terms; Bjj is the Gibbs energy coefficients of ternary polynomial terms; and m , n , o are the exponents of polynomial terms. The Gk, Aj and Bjj parameters are either constant or linear functions of the temperature. The exponents m, n and o as well as the indices p, q and r are given individually for each j or jj. The first term on the right-hand side of eqn. ( 1 ) represents the Gibbs energies of the species and their contribution to the configurational entropy, the second and third term are binary and ternary polynomial interaction energies, respectively. The polynomial interaction terms consist of two types (as seen in Table 1). Terms involving products of concentrations on different sublattices correspond to the energy of formation of the phase with only one species present on each sublattice. Terms involving products of concentrations on the same sublattice correspond to the change in energy owing to the mixing of the species on this sublattice. The above equation reduces to the binary quasisubregular solution model if only one sublattice (l = 1 and a ~= 1.0) is considered. In this case the species concentration, Yk, is identical to the elemental concentration, xk, and therefore, the variable xk is used instead of Yk, and with t' l = x I +(1 - x, - x 2 ) = x I

12

U. R. Kattner, W. J. Boettinger

Ternary T i - A I - N b system

T A B L E 1. T h e r m o d y n a m i c d e s c r i p t i o n o f t h e T i - A I - N b s y s t e m (all quantities are given in J m o l - ~) Multiplier Liquid XNb XAj P a r a m e t e r G~, Aj or B H Multiplier Nb2AI (three sublattices) 0.00 0.00 N i o b i u m sublattice: a yb = 0 . 5 3 3 3 3 3 4 0 yNbNb -- 2 2 6 6 6 . 7 0 + 1 0 . 1 7 5 6 7 T yAiNl' -- 3 9 8 9 . 0 0 + 8 . 2 6 4 1 3 T YTiNb -- 1 3 5 3 6 . 0 0 + 8 . 4 5 8 1 3 T YAiNbyTi Nb 8046.15
Y T i N b y A ) Nb -- 10000.00

P a r a m e t e r Gk, A / o r Bjj

Xmi XNt,XAIt'NI,'XNbXAII~NblIAI
XNbXAll'AI 2

0.00
-- 1 0 3 7 3 7 . 1 0 + 2 9 . 4 5 9 0 4 -- 2 8 9 7 1 6 . 2 0 + 8 7 . 9 7 0 0 6 -- 1 2 7 7 1 3 . 1 0 + 5 8 . 5 1 1 0 2 -- 1 2 0 5 2 1 . 0 0 + 4 1 . 1 1 3 7 8 -- 1 0 4 6 1 9 . 4 0 + 4 1 . 1 1 3 7 8 13058.26 -- 1 0 0 0 0 0 . 0 0 -30000.00+ -- 6 2 8 . 0 0 - 14146.00 -- 1 3 3 9 8 2 . 5 0 -- 1 2 5 4 0 1 . 5 0 -- 1 2 9 3 9 6 . 7 0 13075.00 -- 8 0 0 0 0 . 0 0 -- 1 8 0 0 0 0 . 0 0 -- 7 0 0 0 0 . 0 0 T T T T T

X-I.iXAIUTi XTiXAII'AI
XTiXNb
XNbXAIXTi

(fl-Ti,Nb) XNb XAI Xfi

XNbXAWNI,
XNbXAll'AI

XTiXAI XTiXNb
XNb2XA~X.I,i

+ + + + +

10.90910 T 6.65980 T 7.28800 T 41.69422 T 41.69422 T 40.06310 T

A l u m i n u m sublattice: a a~ = 0 . 3 3 3 3 3 3 3 0 YNb AI -- 2 2 6 6 6 . 7 0 + 1 0 . 1 7 5 6 7 T yAiAI -- 3 9 8 9 . 0 0 + 8 . 2 6 4 1 3 T yTi AI - 13536.00+8.45813 T yAtalyNb AI -- 5 4 9 9 . 8 2 + 1 . 6 6 8 2 5 T

YNbAlyTiAI
YTiAlyalAI

5028.85 -- 6000.00

N i o b i u m sublattice: a Nb = 0 . 1 3 3 3 3 3 3 0
yNb Nb --

22666.70 + 10.17567 T

XNbXAI2XTi
XNbXAIXTi 2

+ 60.00000 T + 60.00000 T + 60.00000 T

(a-Ti) XNh XAj x-li

XNhXAI XTiXalt,.ri XTiXAIUAI XTiXNI~ XNbXAIXTi


(Al) Xy/~ xA~ XTi XNbXAI X,riXAi XTiXN~ Nb3AI (two sublattices) N i o b i u m sublattice: ysb Nb
YAl Nb y T i Nb a Nb = 0 . 7 5

-- 1 3 0 0 0 . 0 0 + 9 . 7 0 9 1 0 T --5151.40+9.61780 T - 18316.00+ 10.89840 T -- 1 2 0 0 0 0 . 0 0 + 4 0 . 0 0 0 0 0 T -- 1 3 9 8 2 3 . 4 0 + 4 5 . 3 9 1 7 4 T -- 1 0 7 7 5 3 . 6 0 + 2 1 . 0 2 6 3 0 T 13075.00 -- 1 1 5 0 0 0 . 0 0 - 8000.00 + - 10711.00 + 12316.00+ -- 1 8 5 7 8 3 . 3 0 + -- 1 2 4 2 6 9 . 6 0 + 13075.00


-

Polynomial t e r m s b e t w e e n s p e c i e s o n different sublattices YNbNbyalAlyNbNb -- 4 6 8 9 5 . 3 0 + 1 0 . 4 6 5 6 9 T YAiNbyNbAlyNbNb 19747.45 yalYbyalalyNb Nb -- 2 7 1 4 8 . 6 9 + 1 0 . 4 6 5 6 9 T YTiNbyAiAlYNbNb -- 3 5 2 0 0 . 0 0 + 6 . 2 0 0 0 0 T YAiNbyyiAlyNbNb -- 3 5 2 0 0 . 0 0 + 6 . 2 0 0 0 0 T (Ti,Nb)AI~ (two sublattices) (Ti,Nb) sublattice: a T M = 0.25

yNbTM
yaj T M yl, iT M yNuTNyAI TM yq-iTNyAi TM

yNbfNyTi TM
8,70910 T 11.47280 T 10.79840 T 92.57490 T 43.89675 T A l u m i n u m sublattice: a AI = 0.75
YNb al

- 8000.00 + 8.70910 T - 10711.00 + 11.47280 T - 12316.00+ 10.79840 T 12500.00 12500.00 3268.75


-- 8 0 0 0 . 0 0 + -- 1 0 7 1 1 . 0 0 + -- 1 2 3 1 6 . 0 0 + -- 5 6 5 9 2 . 5 6 + -- 6 1 8 4 9 . 4 6 + 9806.25 -- 1 0 0 0 0 0 . 0 0

yAW AI

YTiAI
Y A i A l Y N b AI

yAiAlYTiAI YNbAlyTiAI

8.70910 T 11.47280 T 10.79840 T 33.80652 T 36.53370 T

YNbAlyAiAlYTiAI
-- 2 4 5 0 0 . 0 0 + 1 0 . 3 5 9 1 0 T -- 3 1 4 8 . 8 0 + 7 . 8 6 3 0 5 T -- 1 3 6 8 8 . 5 0 + 8 . 1 6 5 6 0 T 9806.25 + + + + 10.35910 T 7.86305 T 8.16560 T 2.47012 T

yNbYbYTiNb.

A l u m i n u m sublattice: a A~= 0.25 T yyh a~ - 24500.00 YAi AI -- 3 1 4 8 . 8 0 YTi AI -- 1 3 6 8 8 . 5 0 YAiglyNbgl -- 8675.01 yybAlyTi Ar 3268.75 YTiAlyAIAI -- 4 0 0 0 . 0 0

Polynomial t e r m s b e t w e e n s p e c i e s o n different sublattices yNbTNyA~ AI -- 5 0 1 3 1 . 9 0 + 1 2 . 1 8 2 8 7 T YAITNyNbAr 50131.90-- 12.18287 T YTi'rNyAiAI -- 4 0 3 4 9 . 6 0 + 1 0 . 3 6 5 2 5 T y A I T N y T i AI 40349.60- 10.36525 T YNbTNyTiTNyAIAI -- 1 7 0 0 0 . 0 0 + 6 . 0 0 0 0 0 T TizAl5 ( s t o i c h i o m e t r i c c o m p o u n d ) 1 -361654.70+ 145.66830 T T

TiAI 2 (stoichiometric compound) 1 - 165313.20+66.80632 T i A I (two sublattices) T i t a n i u m sublattice: a v~ = 0.5 yNbTi yA~ T~
YTi T i
YTiTiyAiTiiuTiTi

Polynomial t e r m s b e t w e e n species o n different sublattices YNbNbyAi al -- 3 5 5 1 8 . 5 0 + 6 . 9 3 6 1 7 T Y A I N b y N b AI 35518.50-6.93617 T yTiYbyAiAI -- 2 6 0 0 0 . 0 0 + 7 . 0 0 0 0 0 T yAiNbywiAI 26000.00 - 7.00000 T YNbNbyTiNbyAiAI -- 6 0 0 0 . 0 0

-- 8 0 0 0 . 0 0 + - 10711.00+ -- 1 2 3 1 6 . 0 0 + -- 1 0 2 9 7 8 . 4 0 +

8.70910 T 11.47280 T 10.79840 T 7.79282 T

U. R. Kattner, W. J. Boettinger
T A B L E 1 (continued) Multiplier P a r a m e t e r G k, A / o r Bi,

7~'rnao' T i - A I - N b system

13

Multiplier

P a r a m e t e r G k, Aj or Bij

Titanium sublattice: a T i = 0 . 5 (continued) YTiTiyAiTil'Ai Ti - - 24001.71 + 7.79282 T YNbTiyTi AI 6537.50 YNbTiyAiTiyTi I'i -- 4 0 0 0 0 . 0 0
A l u m i n u m sublattice: a a~ = 0.5 yNt,al
YAi AI

--

yli al
YNbAlyAi AI

YAiAlyTi AI
YNbAlyTiAI

- 8000.00 + 10711.00+ - 12316.00+ -- 6 0 0 0 0 . 0 0 + -- 2 8 3 1 1 . 6 3 + 6537.50

8.70910 T 11.47280 T 10.79840 T 11.00000 T 10.85167 T

Polynomial terms between species on different sublattices Y N b T i y A i AI - - 3 8 0 0 0 . 0 0 + 7.00000 T YAiTiyNb AI 3 8 0 0 0 . 0 0 - 7.00000 T YTiTiyAiAI - - 2 9 6 3 3 . 6 0 + 6.70801 T yAiriyfi AI 2 9 6 3 3 . 6 0 -- 6.70801 T YNbTiy.l i'l iy Ai AI -- 9400.00

Ti2AINb (three sublattices)


T i t a n i u m sublattice: a f~ = 0.5

YNbTi
yl] i
Yni,Tiy Niobium I i Ii sublattice: a Nb = 0.25

13000.00 + 9.70910 T - 1 8 3 1 6 . 0 0 + 10.89840 T - - 10000.00


--

Polynomial terms between species on different sublattices YNbTiYAi AI - - 4 2 0 0 0 . 0 0 + 8.00000 T YAiTiyNb AI 4 2 0 0 0 . 0 0 - 8.00000 T yTiTiyAi AI - - 3 7 4 4 5 . 1 0 + 16.79376 T yAiTiyTi AA 3 7 4 4 5 . 1 0 - - 16.79376 T yybTiyTiTiyAIAI -- 2 0 0 0 0 . 0 0 Ti3AI (two sublattices) T i t a n i u m sublattice: a f~ = 0.75
yyb Ti
YAi li

Ynbnh y-riNB

-- 13000.00 + 9.70910 T - 18316.00 + 10.89840 T

A l u m i n u m sublattice: a a~ = (t.25 ya( al -- 5151.40 + 9.61780 T Polynomial terms between species on different sublattices YNbTiyNbNbyAi AI - - 3 8 0 0 0 . 0 0 + 7.00000 T YTiTiyTiNbyAi AI - - 2 9 6 3 3 . 6 0 + 6.70801 T yTiTiYNbNbyAi AI - - 3 4 0 0 0 . 0 0 + 7.00000 T YNbTiyTinbyAi AI 3 4 0 0 0 . 0 0 - 7.00000 T Ti4A13Nb (stoichiometric compound) 1 - 4 8 0 0 0 0 . 0 0 + 180.00000 T

.YTiri

yybTiyAi Ti yTifiyAi Ti yNbTiyri "ri


A l u m i n u m sublattice: a AI = 0.25
YNb AI YAi AI

13000.00 + 9.70910 T -5151.40+9.61780 T - 1 8 3 1 6 . 0 0 + 10.89840 T -- 3 0 0 0 0 . 0 0 - 71277.87 + 25.46998 T 9806.25


--

--

--

Y T i AI

--

YNbAlyl iAI

13000.00 + 9 . 7 0 9 1 0 T 5151.40 + 9.61780 T 1 8 3 1 6 . 0 0 + 10.89840 T 3268.75

and v , = x 2 +(1 - x I - x 2 ) = x 2 the equation can be rewritten as


G i= x l ( G 1 + R T ln(x I )) + x2(G 2 + R T ln(x2) )
nl tl

+ ~ AjXl x,
J

In order to provide a thermodynamic description of a binary phase whose compositional stabilities extend into the ternary region, the Gibbs energies of the counter phases in the other binaries need to be estimated. These complimentary phases are not stable in the respective binaries under ordinary conditions. For the Nb-AI and Ti-AI systems, the Gibbs energies of formation of the stable phases were used to estimate the Gibbs energy of formation of the hypothetical metastable compounds. The phases in the Ti-Nb system exhibit no ordering tendency. Therefore, the Gibbs energy of formation of the ordered compounds was set to zero for the Ti-Nb system and the total

excess Gibbs energy was set equal to that of the disordered compounds. In the analytical descriptions of the ordered intermetallic compounds of the Nb-AI and Ti-A1 systems, the number of parameters in the description was minimized by using constraints for the different Gibbs energies of formation. If, in an ordered compound, all the elements are assumed to occur on all sublattices of this compound, this compound will exist over the entire composition range. The structures of most of these compounds are ordered derivatives of the structures of the pure elements. For complete substitution, i.e. at the composition of one of the pure elements, the degree of ordering in these compounds must be zero and the Gibbs energy of these compounds must then be equal to the Gibbs energy of the pure elements of the corresponding crystal structure. This implies constraints for the Gibbs energy of formation of the substitutional atoms. The same constraints were applied to the thermodynamic quantities of the hypothetical metastable compounds in the respective binaries.

14

U. R. Kattner, W. J. Boettinger

Ternary Ti-AI-Nb system

Because the homogeneity range of the ternary Ti4A13Nb compound is believed to be small, this phase is modeled as stoichiometric. The structure of the ordered Ti2AINb phase can be derived through additional ordering of titanium and niobium on the titanium sublattice of Ti3AI. It was therefore modeled as a solution phase with three sublattices (the titanium, niobium and aluminum sublattices). This phase reveals a significant range of homogeneity with respect to titanium and niobium, but the range of homogeneity

T~ Colculo~ed Liquidus

o
~T[2AI 5

, 0 10 20

, 30 40 50 60 70 80 90

O 100

Nb

At.% AI
Ti

AI

b Exper imer;tol Liquidus Surfoce

~ QV T ~ ~o / ~: 7 kv~

q
(#Ti ,Nb

,g

with respect to aluminum is smaller. It was, therefore, assumed that niobium and titanium atoms mix only on the titanium and niobium sublattices and no mixing occurs on the aluminum sublattice. Thus, the TizAINb compound composition is constrained to (Ti, Nb)3AI. For complete substitution on either the titanium or niobium sublattice, the degree of ordering must again be zero and the Gibbs energy of formation must then be identical to the corresponding binary Ti3A1 and Nb3A1 phases with the D0j9 structure. No attempt was made to develop a miscibility gap in the (fl0-Ti,Nb) phase in order to fit the data of Perepezko et al. [6]. For the present calculation, the simplification of treating the (fl-Ti, Nb) and (fl0-Ti,Nb) as one disordered phase has the consequence that this miscibility gap would have to occur in the (fl-Ti, Nb) phase. In the three binary systems, the (fl-Ti,Nb) phase only forms a miscibility gap in the Ti-Nb system at a relatively low temperature. However, the experimental data suggest that the ternary miscibility gap occurs on the (Ti,Al)-rich side of the system at relatively high temperatures. In order to obtain the specific curvature of the Gibbs energy needed for the formation of a miscibility gap, the number of ternary polynomial terms would have to be increased. Considering the simplifications already introduced for the analytical description of the (fl/flo-Ti,Nb) phase, increasing the number of polynomial terms is not justified. At the beginning of the calculation, attempts were made to use a least squares method in order to adjust the Gibbs energy coefficients of the polynomial terms. Since the amount of available data with respect to the number of adjustable terms is relatively small and most of these data cover only the temperature interval between 1100 and 1200C, the results of the least squares fitting could only be used as initial values for trial and error calculations. In the trial and error method the values of the adjustable parameters are estimated. The change of the phase boundaries between two steps of the calculation is used to estimate new values in order to adjust the calculated phase boundaries to the experimental results. The reported values of the adjustable parameters were estimated by such a trial and error approach.

i2AI 5

5. Results and discussion


2
0 10

,
20

,/'
30 40 50 60 70 80

'
90

.
100

Nb

At.% AI

AI

Fig. 2. The liquidus projection (circles, maximum; squares, minimum). (a) Calculated. (b) Initially, after Perepezko et al. [6], the composition of liquid of the binary three-phase equilibrium was taken from the binary calculations.

The thermodynamic description of the Ti-AI-Nb system is given in Table 1. This description was used to calculate the liquidus projection, several isothermal sections and an isopleth. The calculated liquidus projection is compared with a preliminary projection estimated by Perepezko et aL [61 in Fig. 2. The calculated 1400 C section is shown in Fig. 3. The calculated

U. R. Katmer, W. J. Boettinger T;

Ternary Ti-AI-Nb system

15

Ti

G
Calculated ~~ / ~ ~o Calculated
2oooc

~o
~.~o

7 Y

(~'T] ,Nb)

/x

,,o "7 o
0 1 0 20

% 2A 0

30

40

50

60

Nb

At.% At

70 I 80 (T[,NbbAI3

90

100

10

20

30

40

50

60

AI

Nb

At.% AI

70 80 (Ti 'Nb)A]3

90

100

AI

Fig. 3. Calculated isothermal section at 1400 C.


b

T~
Experimental 1200oc Isotherm ~ ~/ ~ ~ Ref.2 o Ref.6 0 Ref.15

isothermal sections at 1200, 1100 and 700 C are compared with the experimental tie-line and phase boundary data as shown in Figs. 4-6. The isopleth at 25 at.% AI with the experimental data is shown in Fig. 7. The calculated and the estimated experimental liquidus projections as given in Figs. 2(a) and 2(b) are in qualitative agreement. The calculation predicts correctly the phases occurring in the invariant fourphase equilibria. However, the liquidus temperatures predicted by the calculation for the primary TiAI region are lower than the temperatures from first experimental results [6]. This has as a consequence that the temperature of the calculated three-phase equilibrium L + ( a - T i ) + T i A ! decreases with increasing niobium concentration while the first experimental results indicate that the temperature increases. A ternary eutectic (L ~ Nb2AI + (Ti, Nb)A13 + TiAI) is predicted for the invariant equilibrium involving L+ Nb2A1 + (Ti,Nb)A13 + TiAl, but the experimental results indicate that it is a transition-type reaction (L+ (Ti,Nb)A13 ~ Nb2AI + TiAI). The calculation gives a maximum for the three-phase equilibrium L+Nb2AI+TiA1 close to the eutectic composition. Small changes of Gibbs energies of the compounds involved in this equilibrium could shift this maximum closer to or even beyond the eutectic composition. In the latter case, the invariant equilibrium would then be a transition type and the maximum would occur in the metastable three-phase equilibrium. The calculated 1400 C section reveals a large range of homogeneity for (#-Ti,Nb). This is in agreement with experimental results [1], which indicate that this

~%

q-

10

20

30

40

50

60

70

80

90

100

Nb

At.% AI

AI

Fig. 4. Isothermal section at 1200 C. (a) Calculated. (b) Experimental data and phase boundaries.

range of homogeneity is even larger. The calculated and experimental phase boundaries agree fairly well at 1200 and 1100C, while the calculated tie-line directions deviate from those experimentally determined. The increased range of homogeneity observed for the (Ti,Nb)AI 3 phase in the ternary system was well matched by the calculation at 1200 C. In the present calculation the temperature of the binary eutectoid (a-Ti) --*Ti3AI + TiA1 decreases initially with increasing niobium concentration. This results in an isolated ternary (a-Ti) phase field between Ti3AI and TiAI.

16

U. R. Kattner, W. J. Boettinger
Ti

Ternary Ti-AI-Nb system


Ti
0 c) 0

Calculated

00oc
Isotherm ~./

\
~ eo

Culculoted 700C Isotherm

~.

//~//////

TiA X~!TiA,

fiAI 2

:2

!o
80 90 100

10

20

30

40

50

60

'

7'0 I

8'0 '

9'0 ' 1(}0

10

20

30'

40

5'0

6'0

70

Nb

At.%
Ti

AI

(Ti ,Nb)AI~

AI

Nb

Nb2AI A t . %
Ti

AI

AI

A~ ~ o

Exper imentat

~ / ~

1100o0
Isotherm
/

7<
~.// \
0~/I.,

'

a t3 e ~u

Ref.1 Ref.4 Ref.5 Ref.7 R e f . 16

Experimental 7000 Isotherm ~/

~ / \~ ~o ~/~\ \\'//~aTi ) /~"'-'i,.ik~

o Ref.5 e Ref.8

dko

:/

,'~

/. . . . .

1~."

"

~.

Ti ~, ~,o
/" .'~.T i A I 2 '
K ~- i

//
2A I 5

~/

//
i~~
/'1

//
/ /-. i_~

~)' / ~'./ / _-. ./ i ~ / / !..~x .~./ , i , I,'/ ,


0

. . . .
40 50

. . . .

10

20

30

60

70

80

90

100

10

20

30

40

50

60

70

80

90

100

Nb

At.%

AI

AI

Nb

At.%

AI

AI

Fig. 5. Isothermal section at 1100 C. (a) Calculated. (b) Experimental data and phase boundaries.

Fig. 6. Isothermal section at 700 C. (a) Calculated. (b) Experimental data and phase boundaries.

The agreement between the calculated isothermal section at 700 C and the experimental data is poor. The single-phase region of the phase denoted by Ti2A1Nb exists in the present calculation at an excessively high niobium content and, in fact, does not include the stoichiometric composition of Ti~A1Nb. Thus the important (fl-Ti,Nb) + Ti3AI + Ti2AINb three-phase equilibrium also exists at an excessively high niobium content. With the current thermodynamic description of the (fl-Ti,Nb) phase, the phase boundaries of the (fl-Ti,Nb)-Ti3Al, (fl-Ti,Nb)-Ti2A1Nb and (fl-Ti,Nb)-NbaAl equilibria have very similar

slopes and small changes in the Gibbs energy of the Ti2AINb compound would result in different topologies for this section. Further experimental effort is needed to establish the phase boundaries for this isothermal section. The process of adjusting the Gibbs energy coefficients of the polynomial terms showed that the phase relationships obtained from the calculation are extremely sensitive to small changes in the Gibbs energies. Similar behavior had been previously observed during the calculation of the Ti-AI binary system [14, 19]. The fact that small changes in the Gibbs energies

U. R. Kattner, 14/.J. Boettinger

Terna O' Ti-AI-Nb system

17

g
r-~D o L oo ~

positional and temperature regions in order to elucidate the phase equilibria of this technologically important system.

~g
o

Acknowledgments
J

cz o2

Re
Calculated f.6: o
o
~, i

(Ti,nb)AI
o

Liquidus
Solidus
i i i ~ I 1 1 1 1

, 60 70

~0

20

50

40

50

i 80

i 90

,
1 O0 5

NbAI3

Mole

TiAI

TiAI

Fig. 7. Calculated isopleth at 25 at.% AI and experimental data. Note that this section is quasibinary, except on the TiA13-rich side.

The authors wish to thank Dr. H. L. Lukas of the Max-Planck-Institut fiir Metallforschung (Stuttgart, FRG) for providing the software used in the present assessment, Professor J. H. Perepezko of the University of Wisconsin-Madison for helpful discussions and Professor Y. A. Chang of the University of WisconsinMadison for critically reading the manuscript. This research was partially supported (W.J.B.) under DARPA Order 7469.

result in different phase relationships may not only help to explain the experimental difficulties in determining the accurate phase diagram for the T i - A I - N b system, but also explain the difficulties of obtaining a satisfactory thermodynamic descripton for the phases in this system. A change in the thermodynamic description of one phase affects all the phase equilibria involving that phase. In the trial and error calculation it is advisable to modify only a few parameters between two steps of the calculation in order to clearly recognize the effects of the modifications on the locations of the calculated phase boundaries. To model the wide homogeneity ranges of the ordered intermetallic compound a large number of adjustable parameters had to be included. This large number of adjustable parameters makes the trial and error method for the adjustment of these parameters relatively inefficient. A straightforward method of adjusting all of the parameters simultaneously is the least squares method [20]. However, in order to apply this method successfully, more experimental data are needed. These data are not only needed to constrain the phase boundaries at certain temperatures, but even more importantly, to establish the change of the phase boundaries over a larger temperature interval.

References

I L.A. Bendersky and W. J. Boettinger, Mater, Res. Soc. Symp. Proc., II (1989)45. 2 K. Kaltenbach, S. Gama, D. G. Pinatti, K. Schulze and E.-Th. Henig, Z. Metallkd., 80(1989) 535. 3 H.T. Kester-Weykamp,C. H. Ward, T. F. Broderick and M. J. Kaufman, Scr. Metall., 32 (1989) 1697. 4 K. Muraleedharan and D. Banerjee, Metall. Trans. A, 20 (1989) 119. 5 L. A. Bendersky, W. J. Boettinger, B: P. Burton, E S. Biancaniello and C. B. Shoemaker, Acta Metall. Mater., 38(1990) 931. 6 J. H. Perepezko, Y. A. Chang, L. E. Seitzman, J. C. Lin, N. R. Bonda, T. A. Jewett and J. C. Mishurda, in Proc. Symp. High Temperature Aluminides and lntermetallics, Indianapolis, ASM/TMS-AIME, Metals Park, OH, 1990, p. 19. 7 J. H. Perepezko, private communication, 1990. 8 L. A. Bendersky, W. J. Boettinger and A. Roytburd, Acta Metall. Mater., 39 ( 1991 ) 1959. 9 U.R. Kattner, unpublished results, 1988. 10 J.L. Murray, Alcoa report, 1987. 11 J.L. Murray, Metall. Trans. A, 19(1988) 24. 12 L. Kaufman and H. Bernstein, Computer Cah'ulation of Phase Diagrams, Academic Press, New York, 1970. 13 U.R. Kattner and H. L. Lukas, unpublished results, 1990. 14 U. R. Kattner, J.-C. Lin and Y. A. Chang, Metall. Trans. A., Feb. (1992) accepted for publication. 15 D.T. Hoelzer and F. Ebrahimi, in E. D. Verink Jr., Processing
and Protection of High Temperature Structural Materials, Annual Report, University of Florida, Gainesville, F L 1990,

6. Conclusion

The calculated phase diagrams of the Ti-A1-Nb ternary systems are undoubtedly subjected to large uncertainties owing to the lack of sufficient thermodynamic and phase equilibria data to fix the model parameter values and the complicated phase relationships. However, they do provide valuable and useful information for alloy development and indeed suggest experiments to be carried out in critical corn-

p. 26. 16 L.A. Benderski, private communication, 1991. 17 D. Banerjee, A. K. Gogia, T. K. Nandi and V. A. Joshi, Acta Metall., 36 ( 1988 ) 871. B. Mozer, L. A. Bendersky, W. J. Boettinger and R. G. Rowe, Scr. Metall. Mater., 24(1990) 2363. 18 H. L. Lukas, J. Weiss and E.-Th. Henig, CALPHAD, 6 (1982) 229. 19 J. C. Lin, T. Jewett, J. C. Mishurda, Y. A. Chang and J. H. Perepezko, Second Annual Report, supported by DARPA through ONR contract (0014-86-K-075), 1988. 20 H. L. Lukas, E.-Th. Henig and B. Zimmermann, CALPHAD, 1 (1977) 225.