Phase equilibria of the Al-Li binary system

Phase Equilibria of the AI-Li Binary System SINN-WEN CHEN, CHIA-HONG JAN, JEN-CHWEN LIN, and Y. AUSTIN CHANG The solid + liquid phase equilibria between c~-A1 and fl-A1Li were determined using differential thermal analysis (DTA), metallography, and chemical analysis. Boron nitride (BN), which was found to be inert to these alloys, was used as the container. These measurements were carried out in order to resolve the discrepancies reported in the literature. The a-A1 + fl-A1Li eutectic temperature and composition were determined to be 600 ~ --- 1 ~ and 25.8 --- 0.5 at. pct Li. Using these data and data reported in the literature concerning the phase equilibria and thermodynamic properties, thermodynamic models for all the phases were obtained by optimization. The thermodynamic values obtained for the fl-A1Li phase describe not only the phase equilibria, but also yield structural defect data in agreement with measured values. The assessed enthalpies of formation, excess entropies of formation, and entropies of melting for all the intermetallic phases obtained are compared with empirical correlations when experimental data are not available. In addition to the stable diagram, a metastable diagram involving the 6'-A13Li is also calculated from the thermodynamic models. The calculated diagram is in good agreement with the experimental data. I. INTRODUCTION T H E aluminum-lithium alloys are emerging as a family of structural materials for applications in the aerospace, automotive, and marine industries. They are lighter and stiffer than conventional alloys and serve as the basis for developing multicomponent alloys with even better engineering properties. In order to define the processing conditions for making these alloys and subsequent treatments to obtain the optimum engineering properties, a knowledge of the phase diagram and thermodynamic properties of these alloys is essential. Although phase equilibria of the A1-Li binary were studied by many investigators, [1.3~ the data reported in the literature are not in agreement. For instance, the a-A1 + fl-A1Li eutectic composition reported by the following five groups of investigators varies from 23.4 to 30.8 at. pet Li: Muller, [2J 23.4 at. pct; Grube et al., I31 30.0 at. pet; Shamray and Saldau] qj 26.3 at. pet; Myles et al.,ts] 30.8 at. pct; and Schfirmann and Voss, ~6] 25.2 at. pct. In view of the intrinsic experimental difficulties in working with these alloys, such as vaporization and oxidation of Li and interaction of either Li or A1 with the container materials, it is not surprising that these discrepancies exist in the literature. These discrepancies did not disappear with subsequent thermodynamic assessments of the phase equilibria tl-3~ and thermodynamic I31-361 data. Saboungi and Hsu [37] modeled the thermodynamics of the system, assuming fl-A1Li to be a line compound. They obtained an c~-A1 + fl-A1Li eutectic composition of 27 at. pct Li. A subsequent evaluation by McAlister, I38~ using a Wagner-Schottky type of model to account for the nonstoichiometry of the fl-A1Li phase, yielded a value of 24 at. pct Li. The most recent assessment by Sigli and Sanchez, I391 using the SINN-WEN CHEN, Research Assistant, C H I A - H O N G JAN, Research Assistant, J E N - C H W E N LIN, Research Associate, and Y. A U S T I N C H A N G , Wisconsin Distinguished Professor and Chairman, are with the Department of Materials Science and Engineering, University of Wisconsin-Madison, 1509 University Avenue, Madison, WI 53706. Manuscript submitted November 28, 1988. METALLURGICAL TRANSACTIONS A cluster variation method, yielded a value of 23 at. pct Li. Again, this is not surprising, since the parameter values for the various thermodynamic models of the phases were obtained by optimizing the existing thermodynamic and phase equilibria data, which are not in agreement. Different compromises with regard to the data were made by the three groups of investigators. In order to resolve these discrepancies, conclusive experiments need to be carried out. Khachaturyan e t a / . [411 carried out a theoretical investigation of the precipitation of 6'-A13Li from a-A1 alloys. They examined the metastable equilibrium between a and 6', using the following model for the 6' phase. The enthalpy of formation was obtained by assuming the atoms interact in pairs and the entropy of mixing was evaluated using the mean field approximation. However, they did not examine the solid + liquid equilibria of this system. The objectives of the present study are (1) to determine the solid + liquid equilibria between a-A1 and /3-A1Li, using well-defined experimental methods and (2) to reassess the A1-Li binary system thermodynamically, using our data as well as those in the literature. The experiments are restricted to the compositions between a-Al and fl-A1Li because these equilibria are of primary importance in the development of aluminum alloys. II. EXPERIMENTAL METHOD A review of the literature shows that many investigators working with either A1-Li or A1-Cu-Li alloys experienced the loss of Li during the course of their experiments, as summarized in Table 1. [42-48] Several groups of investigators t42,44,481 quantified their losses of Li, while others only reported Li losses qualitatively. Mikheeva et al. [421 analyzed only three of their many samples. However, they did not correct for the losses of Li in any of their samples. In fact, it would be difficult to account for the loss of Li quantitatively, since Li loss depends strongly on the time the sample has been cycled through the liquid state. In the following, we will VOLUME 20A, NOVEMBER 1989--2247 Table I. Authors Mikheeva et al. [42[ Shamrai et al.[43] Ueda et al. I441 Smith [4sl Ashton et al.[46] Papazian et al. [471 Anyalebechi et al. [48[ Loss of Li Reported by Various Investigators Method Used Sample Size Li Loss Alloy alumina crucible alumina crucible vacuum, argon gas sealed air argon air, hydrogen vacuum, He in alumina crucible --0.05 to 0.24 g ---100 g 10 to 23 pet Li loss found 1.5 to 99.65 pet Li loss found Li loss found Li loss found 12 to 23.8 pet A1-Li-Cu A1-Li-Cu A1-Li A1-Li-Cu (Mg) A1-Li-Cu (Mg) A1-Li-Cu (Mg) A1-Li describe the experimental methods used to assure that no Li loss occurs in any of the samples. A. Design o f Sample Containers Graphite, graphite coated with Y203, quartz, and BN were tried as container materials for A1-Li and A1-Li-Cu alloys. As shown in Table II, BN is inert to these alloys. The YzO3-coated graphite is probably also inert if a perfect coating is made. Since we found BN to be compatible, we did not attempt to develop a perfect process for coating graphite with Y203. Figure 1 shows the design of our sample container system. B. Experimental Procedure and Characterization of the Samples The samples were prepared in a dry box, manufactured by Vacuum/Atmosphere Company, that provides an inert atmosphere with oxygen concentration less than 5 ppm. Since Li is easily oxidizable, the oxidized surface was removed in the dry box with a knife. After removal of the oxide layer, the Li samples remained shiny after several hours in the dry box, indicating the inertness of the atmosphere. Weighted Li and A1 samples of the desired proportions were loaded in a BN crucible, covered with a BN lid, and placed in a quartz capsule (Figure 1). The capsule was removed from the dry box, immediately evacuated, backf'dled with argon to 1/3 atm, and sealed. This sealed quartz capsule was held at 700 ~ for 20 minutes before quenching into ice water. The composition of the sample was analyzed using an inductively coupled plasma (ICP) method. C. Metallography Metallographic preparation of A1-Li alloys is extremely difficult because of the reactivity of Li. Kerosene instead of water was used as the lubricant. Silicon Table lI. D. DTA Experiments A PERKIN-ELMER* DTA 1700 system was used to *PERKIN-ELMER is a trademark of Perkin-Elmer Physical Electronics, Eden Prairie, MN. determine the eutectic, solidus, and liquidus temperatures of the A1-Li alloys. The procedure for preparing the samples was the same as that given in Section I I - B . Since the DTA sample container is rather small (the outside diameter is 4 mm), special attention was paid to the fabrication of the BN containers. In addition to preparing samples using the techniques described, we also obtained several large samples in rod form with a 0.25-in. diameter from Granger and C h u I491 of ALCOA. These samples were surface-cleaned by us and analyzed chemically, using the ICP method before any DTA experiments were carried out. Because the DTA sample containers are double-wailed (Figure 1), there is significant thermal lag between the sample and the measuring thermocouple. As shown in the calibration in Figure 2, there is a difference of - 2 5 ~ between the sample and measuring temperatures. All DTA experiments were carried out in the heating mode to avoid supercooling. The runs were made at 4, 2, and 1 ~ min, and the data were extrapolated to 0 ~ to obtain the true reaction temperatures. The calibration shown in Figure 2 is for 0 ~ In addition to utilizing the DTA system to carry out solidification experiments, we have also carried out independent solidification work. This was done primarily for the sake of convenience. For instance, we are not Li Losses for Various Container Materials Found in this Study Alloy Number Crucible Used 1 2 3 4 5 6 graphite graphite graphite graphite coated with Y203 quartz BN 2248--VOLUME 20A, NOVEMBER 1989 carbide sandpaper with 600 grits was used for initial polishing, and 1-/zm diamond paste was used for final polishing. Samples were cleaned by trichloroethylene after final polishing and were examined immediately by optical microscopy. Sample Size (g) At. Pet Li (Before) 0.51 20.4 0.47 48.2 0.56 33.1 0.49 20.4 reaction detected visually 0.43 20.4 At. Pet Li (After) Li Loss (Pct) 17.5 45.0 29.8 19.5 14 6.5 9.9 4 no detectable composition change METALLURGICAL TRANSACTIONS A To Vacuum Pump t Back Quartz Filled With Rod Argon "---------- S e a l e d Here < Tube Quartz Lid Saml Crucible 9 .j Fig. I - - A schematic diagram of the sample capsule used. limited b y the s a m p l e size i m p o s e d by the P E R K I N E L M E R D T A apparatus. A sealed sample capsule, as shown in Figure 1, was p l a c e d in a furnace at a temperature about 10 ~ a b o v e the liquidus, lowered to a d e s i r e d temperature, such as the ~ + /3 eutectic t e m perature, and then quenched into ice water. The s a m p l e was then characterized by m e t a l l o g r a p h y and c h e m i c a l analysis. III. EXPERIMENTAL RESULTS The D T A and m e t a l l o g r a p h i c results are s u m m a r i z e d in T a b l e III. Chemical c o m p o s i t i o n s for all the solidification samples characterized m e t a l l o g r a p h i c a l l y were a n a l y z e d c h e m i c a l l y after the experiment. No detectable change, i . e . , < 0 . 3 pet, in the Li contents was obtained. F o r the small D T A s a m p l e s , only m e t a l l o g r a p h i c examinations could be carried out b e c a u s e insufficient material was left for c h e m i c a l analysis. H o w e v e r , selected D T A s a m p l e s were run through the liquid phase several times, and the same thermal arrests were obtained in subsequent heating curves. This was not true when a graphite crucible was used. F o r instance, when a 30 at. Table III. pct Li sample was heated up the first time in the D T A apparatus, two thermal arrests were obtained, corresponding to the eutectic and liquidus temperatures. When the sample was heated up a second time, only the eutectic isotherm was observed. This clearly indicated loss o f Li. F u r t h e r m o r e , before we d e c i d e d to use BN containers, m a n y e x p e r i m e n t s were carried out to check the inertness o f BN to Li. Thus, although we d i d not analyze the p o s t - D T A samples c h e m i c a l l y , we are certain that the sample c o m p o s i t i o n s r e m a i n e d unchanged. The D T A d a t a are presented in F i g u r e 3, together with the literature data. t2-6] The a + /3 eutectic is placed at 873 +- 1 K and 25.8 • 0.5 at. pct Li. T h e uncertainty o f • 1 K is b a s e d on the calibration o f the D T A setup vs the melting points o f A1 and Zn. The uncertainty o f + 0 . 5 at. pet Li is our best estimate. Both the solidus and liquidus b o u n d a r i e s for the a-A1 and /3-AILi portion o f the binary were determined by D T A , as shown in Table IV and Figures 4 and 5. Typical heating curves are presented in Figures 4(a) through (e) for five alloys; two of the five involve only a + 1 alloys, and the other three involve a + /3 + l alloys. Three p h o t o m i c r o g r a p h s are shown in Figures 5(a) through (c). The heating curve for the 25.8 at. pet Li alloy shows o n l y one peak, as disp l a y e d in F i g u r e 4(d). S l o w e r c o o l i n g rates cannot resolve any m o r e peaks, which suggests this alloy is the eutectic alloy. A p h o t o m i c r o g r a p h (Figure 5(b)) for this alloy confirms the D T A results. As s u m m a r i z e d in Table III, alloys containing less than 25.8 at. pet Li show a-A1 as the p r i m a r y phase, and those containing more than 25.8 at. pct Li s h o w / 3 - A I L i as the p r i m a r y phase. T w o typical microstructures are shown in Figures 5(a) and (c) for 25 and 26.6 at. pet Li, respectively. As shown in F i g u r e 3 and T a b l e IV, the eutectic temperature obtained in this study is in a g r e e m e n t with the literature values, t3-6] if the older data o f M u l l e d -'] is discarded. This is understandable, since the eutectic temperature is i n d e p e n d e n t o f alloy c o m p o s i t i o n , as long as the c o m p o s i t i o n s lie between or-A1 and/3-A1Li. The eutectic c o m p o s i t i o n o f 25.8 at. pct Li falls between the data o f Shamray and Saldau t4] and Schiirmann and Woss. 161 The latter investigators also investigated the reactivity o f Fe as a c o n t a i n e r material for A1-Li alloys and found Experimental Results Metallographic Results DTA Results (K) At. Pet Li Te Ts TI Primary Phase 8.0* 9.2* 15.0" 20.7* 22.5 25.0 25.8 26.6 27.0 28.5 30.0* 33.7* 40.8 ---872 --873 ---873 872 873 915 913 883 ----------- 924 921 912 893 ------901 922.5 953.5 a a a a a a -- Eutectic i,J iI t,J 1,I pure t,J tJ 11 *Alloys were prepared by ALCOA but were chemically analyzed before use in the DTA and solidification experiments. METALLURGICAL TRANSACTIONS A VOLUME 20A, NOVEMBER 1989--2249 in the present study 9 The agreement among the data of Grube et al.,t3J Shamray and Saldau, t41 and ourselves is somewhat better. Lin tSq modeled the DTA peaks of the solid + liquid two-phase alloys in terms of heat transfer from the reference and working DTA cells to the surroundings. According to his study, the DTA peaks shown in Figures 4(a) and (b) are consistent with the shapes of the liquidus + solidus shown in Figure 3 but not with those constructed from the data of Schtirmann and Voss. [6] The solid lines in Figure 3 are calculated from our thermodynamic analysis, and they are also not consistent with Schiirmann and Voss's data. The calculated solidus slope decreases continuously with XLi, whereas the data of Schiirmann and Voss t6J indicate an initial decrease of the slope and then an increase with XLi. Earlier thermodynamic evaluations by Saboungi and Hsu, [37] McAlister, t381 and Sigli and Sanchez t391 yielded solidus and liquidus shapes which were also in disagreement with those of Schiirmann and Voss. t6j It is difficult to ascertain the reason for the lower temperatures reported by Schiirmann and Voss t6~ for the hypoeutectic alloys. Clearly, their data are lower in temperatures than ours and those of Grube et al. [31 and Shamray and Saldau. [4] 30-- AT O [c) 25-- 20-- I I 400 l 500 O I 600 700 C (Reading from DTA) Fig. 2--Temperature calibration for DTA measurements. AL Li BIHARY SYSTEM 1000. C • 26 N u l 2 980. O 35 6 r u 3 A 960, X ~' 37 Sha4 76 Nyl 5 81 $ch6 ~:) L IV. 920. (~ BOO,o ~ O X~ O V', 880.0 ~= _ Ar ~ ~ _ ~ A ~ -- • • ~ In view of the new experimental data obtained in the present study, a thermodynamic evaluation of all relevant data is carried out. In the following, we will present the models used, data assessment, and calculation of the phase diagram. Xr / 960. o ~ X / x 840.0 A. Thermodynamic Models 820.0 BOO. 0 THERMODYNAMIC EVALUATION ~ O. O0 0.10 02O L:L 0.30 ATOMIC tRACT 04O 0.5O IOM Fig. 3 - - T h e solid-liquid equilibria between a-Al and/3-A1Li: comparison between the data of the present study, those reported in the literature, and the model-calculated values. little interaction. This is consistent with the solubility of Li in Fe, as given by Kubaschewski.[5~ Tantalum is not inert to Li. This may explain why the data of Myles et al. ISJ for the hypereutectic alloys are too high in Li contents when compared with other data. The liquidus data for the hypereutectic alloys obtained in the present study fall in between those of Shamray and Saldau ~41and Schtirmann and Voss, 16J and agree with those of Grube et al. I31 For the hypoeutectic alloys, the agreement is less satisfactory. Both the solidus and liquidus temperatures reported by Schiirmann and Voss ~61are lower than those obtained T a b l e IV. Comparison Investigators 26Muller t21 35Grube et al. t31 37Shamrag and Saldau 141 76Myles et al.t51 81Schtirmann and Voss t61 This study 2250--VOLUME 20A, NOVEMBER 1989 To specify the thermodynamic expressions, the following superscripts, l, a, 6', /3, y, and e are used for the liquid, fcc, AI3Li (L12), A1Li (B32), A12Li3, and A14Li9 phases. The subscripts 1 and 2 are used to designate A1 and Li, respectively. 1. L i q u i d p h a s e The following expressions for the excess Gibbs energies are used for the liquid phase: AXSGt/RT= (~)XlX2[(wt12 + w~,) + (wt,2-- wlzl) of the a + /~ E u t e c t i c T e m p e r a t u r e s Container Materials -Fe porcelain, Fe Ta Fe BN 9(X2 -- XO -- 8VtXlX2] [la] x2[(wl2 + w21) + (w'~2 - w21) ' (1 - 4 x l ) + 8x1(-2 + 3x])v l] [lb] and Compositions T e (~ 590 600 602 600 602 600 --- 1 At. Pet Li 23.4 30.0 26.3 30.8 25.2 25.8 + 0.5 METALLURGICAL TRANSACTIONS A 1.00 2.00. 2.2 WT, 73 5g. 0 0 m9 1. O0 doS/mAn SCAN RATEL ATMOSPHEREs ARGON WT, 54. O0 m9 ATMOSPHEREI 0 cc/mln SCAN RATSI ARGON 2.00 d e O / m l n O cr 8 o.00 o 0 B00.00 610.00 a20.O0 ~aD.m E~O.00 ~00 TEMPERATURE (E) (a): 5~O.00 ~m SlO.00 6~I. 00 A 1 - 8 . 0 a t % Lt 630.00 840.00 DTA (b) : A1- 15.0 a t % LI BT 3 1 . 0 0 m9 SCAN RATE, ATMOSPHERE, ARGON ~00 ~.00 TEMPERATURE (r') 74 WT, 000.00 OTA ~00 2.00 ST, des/mln sTo. 00 ~oo ~00 ~00 6ltt 00 ~00 ~00 TEMPERATURE (r') (c) : I S . 00 ms ATMOSPHERE, O cc/mln AIR SCAN RATE, 1.00 dmg/mln 0 cc/mtn ~m 64O.OO TEMPERATURE DTA (C) DTA ( d ) : A1- 2 5 . 8 a t % LI AI- 20.7 at% L| 76 u 14. SO .g ATNOSPHEREJ ARGON 5511-00 ~0.00 SCAN RATE, 2.00 deg/mtn 0 cc/mln 5~. DO 610.00 530.00 TEMPERATURE (C) 650.00 ~&00 DTA (e) : i'd- 3 3 . 7 at% L! F i g . 4 - - H e a t i n g c u r v e s o f D T A : the t e m p e r a t u r e axis is not c o r r e c t e d to the a c t u a l t e m p e r a t u r e . (a) A I - 8 . 0 at. p c t Li a l l o y , T s = 6 4 2 ~ ( 9 1 5 -+ 2 K ) , a n d T ~ = 651 ~ ( 9 2 4 -+ 2 K); (b) A I - 1 5 . 0 at. p c t Li a l l o y , T ' = 6 1 0 ~ ( 8 8 3 -+ 2 K ) , a n d T ~ = 6 3 9 ~ ( 9 1 2 -+ 2 K); (c) A12 0 . 7 at. p c t Li alloy, T e = 5 9 9 ~ ( 8 7 2 -+ 1 K), a n d T t = 6 2 0 ~ ( 8 9 3 -+ 2 K); (d) A 1 - 2 5 . 8 at. pct Li a l l o y a n d T e = 6 0 0 ~ ( 8 7 3 -+ 1 K); a n d (e) A I - 3 3 . 7 a t . p c t L i , T e = 5 9 9 ~ ( 8 7 2 -+ 1 K ) , a n d T ~ = 6 5 2 ~ ( 9 2 5 +- 2 K ) . METALLURGICAL TRANSACTIONS A VOLUME 20A, NOVEMBER 1 9 8 9 - - 2 2 5 1 AXSGt2/RT= (~)x~([wt12 + w~l) -~- (W/12 -- W~I) (4x2 - 1) + 8x2 ( - 2 + 3x2) v; ] [lc] AXSG = excess Gibbs energy; excess partial Gibbs energy o f component i; x; = atom fraction of component i; T = absolute temperature; R = gas constant; and w;12, w~l, and v; = parameters of the model. where mxsG i = (a) : A1- 25.0 at% Li 2. F c c p h a s e The equations used to describe the fcc phase are the same as Eqs. [la] through [lc], except that the superscript 1 is replaced by 0/. 3. B32 p h a s e The assumptions used to construct the thermodynamic expressions for the B32 phase are as follows: (1) vacancies are formed only on the Li sublattice; (2) antistructure defects occur only on the AI sublattice; (3) only first and second nearest-neighbor interactions need to be considered; and (4) the mixing o f vacancies and Li atoms and the mixing o f A1 and Li atoms on the A1 and Li sublattices, respectively, are random. The final expressions obtained are identical to those for the triple-defect B2 phases. I52,53) The basic parameter o f the model is given below: 0/ ~ (b): AI- 25.8 at% Li where a N~ N~ N X = = = = = () N N ~ x=0 = 2 - N / x 0 [2] intrinsic disorder parameter; number o f vacancies on the 0/ sublattice; number o f atom 1 on the/3 sublattice; total number o f atoms; and deviation from stoichiometry = x2 - 1/2. The disorder parameter, z = ( N ~ / N ) x, is the sum of thermal and constitutional defects, and its compositi.on dependence is related to 0/and X by the following equation: (1 + 2)2(1 - Z) (z - 2X) (4z 2) (1 + 0/)2(1 -- 0/) = 40/3 [3] The expressions for the partial quantities are (Adi - a(,;,o)/RT = In (ai/ai,o) 2 = In (c) : A1- 26.6 at% Li Fig. 5--Microstructures of samples from solidification experiments. (a) AI-25.0 at. pcI Li alloy; the sample was solidified at 1 ~ with a (white) being the primary phase. (b) AI-25.8 at. pct Li alloy; the sample was solidified at 1 ~ exhibiting eutectic structure. (c) A1-26.6 at. pct Li alloy; the sample was solidified at 1 ~ with/3 (dark) being the primary phase. 2252--VOLUME 20A, NOVEMBER 1989 (1 + 0/)2(1 -- Z) (1 + 2X) (1 + z)2(1 -- 0/) + 2ei In (z - 2X) 0/(1 + 2X) [4] where the subscript o indicates X = 0 (the equiatomic composition), and e; = - 1 / 2 when i = 1 (A1) and ei = + 1 / 2 when i = 2 (Li). The model also yields an approximate relationship METALLURGICAL TRANSACTIONS A b e t w e e n the intrinsic d i s o r d e r p a r a m e t e r and the enthalpy o f formation: 0 RCTIVITY OF LI FOR THE LISUIB PHflSE -I AH//RT=(~)ln2+(~)Ina g57K - Temp-957K [5] 4. ~', 7, and e phases T h e s e three phases are a s s u m e d to be line c o m p o u n d s , and their integral G i b b s energies o f formation are represented by a t w o - t e r m equation o f the t y p e AG/= A + BT [6] . - - ..... z /, -4 t f/~ . . . . Sobouncj i o n d Hsu B. Evaluation of Thermodynamic Data ......... McRI i s t s r -5 T h e values o f the solution parameters for the liquid and intermetallic phases were obtained using the available thermodynamic and phase equilibrium data, as listed in T a b l e V. The G i b b s e n e r g y differences b e t w e e n the various forms o f A1 and Li are taken from the literature and are also given in T a b l e V. A detailed description for each o f the phases is given b e l o w , using the 1968 International Practical T e m p e r a t u r e Scale (IPTS-68). _~ 0.0 i i i J i i i i i 0.1 0.2 0.3 0.4 0.5 0.0 0.7 0.8 0.9 fll 1.0 X Li Li (a) tCT[VITY OF LI FOR THE LISUIB PHRSE - 987K Temp=g87K 1. Pure components The G i b b s energies o f m e l t i n g o f fcc A1 and bcc Li are f r o m Hultgren et a l . , [54] while those for bcc A1 and fcc Li are from K a u f m a n and Bernstein. I55} -1 d~ ..... 2. Liquid phase V a l u e s o f Li activities were d e t e r m i n e d b y H i c t e r et al. }31] at 957 and 987 K using the K n u d s e n m e t h o d and b y Yatsenko and Saltykova/321 at 1023 K using an e m f method. Y a o and F r a y [331 d e t e r m i n e d Li activities in m o l t e n A1 with 10 -4 to 10 -6 at. pct Li at 985 and 1050 K and found that H e n r y ' s l a w is o b e y e d in the dilute concentration region. F i g u r e s 6(a) through (c) s h o w Table V. 0 Z-3 J / ~ O=''" 9/ t , , / ? ~ /: -5 _~ Thermodynamic Values Used 0.0 10,795 - 1 1 . 5 6 4 . T AGA~~--~l = 627.6 - 6 . 6 9 4 . T AGLfC,~ l = 1786.6 - 7.155 * T AG [cc-W = 3001.6 - 6.61 * T = -- et al. T h I s Stud~l .......... N ~ l igter i I i i i i i I i 0.1 0.2 0.3 0.4 0.5 0.0 0.7 0.8 8.9 fll The Gibbs energy difference of the elements A G ~ ~t O Hitter 1.0 X Li J / g atom Li (b) J / g atom J / g atom J / g atom o :ICTIVITY OF LI FOR THE LISUID PHflSE-1023K Temp= 1023K The solution parameters for the liquid and fcc phases (l, a) w~12 = W~l = vI = w72 = w~] = Va ~ -0.073443 -3.833834 -1.347969 -16.47515 0.1091068 + + + 2093.594/T 2515.617/T 1770.301/T 11,287.59/T - 1518.836/T -;-3 The parameters for the B32 phase (fl) AG7 = - 1 6 , 6 6 0 + 6 . 8 3 . T [Standard states: AI(I), Li(1)] a = [exp (-Q/RT)]/[22/3] Q = 22,960 Z [Standard states: AI(I), Li(l)] METALLURGICAL TRANSACTIONS A A Yotegnko / &/ -- - ,y J / g atom o n d Sol tl=lkovo -4 Th I m S t u d y . . . . Soboung I o n d H s u ......... I~RI l e t e r -5 --.-- S I g I i m d J / g atom The Gibbs energies of formation of tS', y, and e AG?' = - 11,290 + 7.33 * T AGf = - 14,950 + 6 . 4 2 . T AG~ = - 1 2 , 5 0 0 + 6 . 2 9 . T ,'"'~" O 0 J / g atom J / g atom J / g atom -8 i 0.0 RI 8.1 Smchez i i i i i i i l 0.2 0.3 0.4 0.5 0.8 0.7 0.8 0.9 X Li 1.0 Li (c) Fig. 6 - - A c t i v i t y of Li for the liquid phase: comparison between the model-calculated values and experimental data. (a) 957, (b) 987, and (c) 1023 K. VOLUME 20A, NOVEMBER 1989 2253 comparisons between the experimental data and our model-calculated values. The values calculated from Saboungi and Hsu, [37] McAlister, ~381 and Sigli and Sanchez ~39J are also shown for comparison. (The values according to Sigli and Sanchez I391 were obtained by us numerically from their integral values.) Our calculated values are closest to the experimental activity data. Bushmanov and Yatsenko ~4~ measured the heat of mixing at 1023 K. Our calculations agree with their data for the composition around 80 at. pct Li. For A1-30 at. pct Li to A1-60 at. pct Li alloys, the difference between their measurements and our calculations is about 1 kJ. 3. Fcc a-AI p h a s e Figure 7 shows comparisons between the experimental data of Wen et al. [34] and the values calculated by us and the other three groups of investigators. All the calculated values are more positive than the experimental data. However, it is noteworthy that the available thermodynamic and phase equilibrium data are not mutually consistent. We rely on our own c~ + /3 eutectic data. A compromise is made among all the other data. 4. B32 /3-AILi phase Figure 8 shows a comparison between our calculated Li activity values at 688, 733, and 778 K with experimental data. Wen et al. {281 are the only investigators to determine Li activities as a function of composition within the homogeneity range of/3. As shown in this figure, the calculated values are much higher than their measured values. Many attempts were made to adjust the model parameter values for the/3-A1Li phase to reproduce the experimental data. The difficulty is that these data are not consistent with the thermodynamic data of the liquid phase and the phase equilibrium data. Considering the thermodynamic data of the liquid and the a-A1 and/3-A1Li phases and the phase diagram data, it is our conclusion that the data of Wen et al. t:BI are in error. However, if the data of Wen et a1.{281 are shifted to lower Li concentration by - 2 at. pct, there would be reasonable accord between the model-calculated values and their experimental data. As will be discussed in Sec- -2 --This .........McFII ister study o Wen o t o l . -4 ---- S ~ o u n g i -.-Sigli ond Hsu ~ m z _j 700K~-,~ "4 ~ -5 n 9 - o :~n-12 -14 -16 AI ' 0.01 ' 0.07 ' 0.03 J 0.04 ' 0.05 X Li ' O.OO ' 0.07 ' 0.08 ' 0.09 0.10 Li Fig. 7 - - A c t i v i t y o f Li f o r the a p h a s e at 6 9 6 K: c o m p a r i s o n b e t w e e n the m o d e l - c a l c u l a t e d v a l u e s a n d e x p e r i m e n t a l d a t a . 2 2 5 4 - - V O L U M E 20A, NOVEMBER 1989 O O08~< <Wen lit ctl .> (OSEK <Vallol:d<lm> ol .> A "1- /~ 700K <VIr et QI .> + ~1< <Yamat QI .> 0 077B~ <Vrm'~et al .> 0'66a4 <Yoo ot o i . > 5tUdlJ 0.44 i i i 0.48 0.48 0.50 ~700K <~elm~l e t i 0.52 i 0.54 0.50 XLi F i g . 8 - - A c t i v i t y o f the Li f o r the /3-A1Li p h a s e ( B 3 2 ) at 7 7 8 , 7 3 3 , a n d 6 8 8 K: c o m p a r i s o n b e t w e e n the m o d e l - c a l c u l a t e d v a l u e s a n d experimental data. tion V, there is evidence to support the supposition that the Li concentration of their alloys should be lower. Veleckis t291 determined the Li activities in the a + /3 two-phase field using the hydrogen titration method (HTM). Their activity data and the /3/a + /3 phase boundary obtained in the present study are shown in Figure 8. The calculated Li activity values are more positive than their data. On the other hand, the modelcalculated values are more negative than the data of Selman et a/. t361Our calculation is also more positive than the data of Yao et al. t351 5. ~', y, and e phases Values of AG i for the 6', % and e phases are given in Table V. V. PHASE DIAGRAM CALCULATION G~ = G~ [7a] or 3-10 rl -1E 0.(30 9 730K <V. IIClkIIk> 0 --Thlm -0 ~ /%~ A ~ 9 o l d Sonchez -8 -8 -1 The phase diagram is calculated using the equality of chemical potentials of the component elements in a t w o phase field, i e , tCTIVITY OF L[ FOR THE ALPHA PHASE - 696K 0 ACTIVITY OF LI FOR THE BETA PHASE + RT In a7 = ~ + R T In a~ [7b] Using the Gibbs energies for any two phases, Eq. [7] may be used to solve for the compositions of the coexisting phases as a function of temperature. The calculated phase diagram of A1-Li is compared with the experimental data in Figure 9. For the hightemperature s + l equilibria involving a-A1 and/3-A1Li, the comparison is presented in Figure 3. As shown in Figure 9, the calculated solvus for the a-A1 phase is in agreement with the data of Jones and Das, I91 with those of Costas and Marshall []~ at high temperatures, and with those of SchiJrmann and Geissler [3m at lower temperatures. The calculated/3/(c~ + /3) phase boundary agrees with the data of Schtirmann and Geissler. ~3~ The data of METALLURGICAL TRANSACTIONS A BINARY AL--Li 1000 O. L9?O ' SYSTEM ' 970 ' = ?1 N o b 14 L 900. 900 O g ~ O. 16Z * 0.162 , (0, Z58. 873) ", ( 0 Z 5 8 , - 8 7 3 ) a ~ 59 J ~ 63 L e v 9 A § llO Sch 3 0 80 Wen 34 81 Sch 6 84 3en 21 88 l i u 2 5 ~ 3? S h . 4 56 Now 8 + 83 Bau 23 ....\ , "'", C> 35 G r u O 77 Cer 2 o :] 0 800 0 700.0 ~ . 700.0 E/"i "",, ~ 88s 600 Or V 500.0 4 0 0 . O' O. O0 5ooo[. T ~ l i l Study V 0 20 0 30 0 40 0.50 0 60 0.70 0 80 0 90 I 00 400.0 0.00 ~ 0 i0 ~ /'/ ,, / , ' 0 ?0 0 30 ' 0 40 0.50 0.6O 0.70 0 B0 0 9O ~ 00 Fig. 9--AI-Li phase diagram: comparison between the calculated and experimental data. See Section V for detailed discussion. Fig. ]0--AI-Li phase diagram including the metastable 6' phase. Wen et al., 128j as discussed in Section I V - B 4 , are believed to be in error. In fact, if their data are shifted by - ( - 2 ) at. pct Li, they would then be in agreement with the phase boundary data of SchSrmann and Geissled 3~ (Figure 9). Moreover, this shift of composition by - ( - 2 ) at. pet Li would bring better agreement between the Li activity data of Wen e t a / . 1281 and the calculated values (Figure 8). Also shown in Figure 9 are the [3/I + /3 phase boundary data of Schiirmann and Voss, I61 which deviate from the calculated values. If we accept the [3/a + /3 data of SchSrmann and Geissler 43~ to be correct and the a + [3 eutectic temperature to be 873 K, it would be difficult to accept the [3/l + [3 values of Schiirmann and Voss. 161 This suggests that the calculated phase boundaries and the thermodynamic models used provide the most consistent description of this system between a-Al and/3-AILi. The calculated peritectic temperature for the formation of y from the melt is in agreement with the data of Myles et al. ISl and Grube et al, I3I The calculated peritectic temperature for the formation of e and the e + Li eutectic temperature are in agreement with the measured values. The calculated range of homogeneity of [3 at low temperatures is not verified by experimental data. However, it is reasonable to expect that the range of homogeneity would decrease with decreasing temperature. We believe the calculated phase equilibria between [3-AILi and Li are quite reasonable in view of the fact that the AH and AS of 3' and e used are consistent with well-known thermochemical behavior of alloys and intermediate phases. Once we have the solution models for the various phases, we can calculate the various metastable equilibria. Figure 10 shows such a diagram, including the metastable 6' phase. In view of the limited data available, the 6' phase is taken to be a line compound. As shown in Figure 10, the calculated a / a + 6' phase boundary is in agreement with the data of Noble and Thompson, I~4j Williams and Edington, 1~7j Ceresara et al., I2~ Baumann and Williams, 12~1 Jensrud and Ryum, 12~j and Liu and Williams. pSI At compositions higher than 25 at. pet Li, the metastable fcc phase encounters phase separations resulting in a metastable eutectoid decomposition of at to 6' and a2. The extension of the a~/a + 6' boundary through the miscibility gap is also shown. Figure 11 shows high-temperature AI-Li phase equilibria with the To(l/at) and To(l~[3) curves. The equilibria below 600 K are not shown, since the model used for the /3 phase becomes physically unrealistic a large deviations from stoichiometry, and the calculated To(l~[3) curves at lower temperatures would be in serious error. METALLURGICAl. TRANSACTIONS A VI. DISCUSSION A. [3 P h a s e Kishio and Brittain r561 determined the defect concentrations in [3-A1Li at 773 K. Figure 12 shows the experimental data, with values calculated using Eq. [3] and the value of a = 0.0177 at 773 K obtained from the equation given in Table V. Good agreement is obtained between the experimental and calculated values, further AL- 1ooo o r - / Li ) i gTO ~ / / / ' I / / .... / , / 0 oo SYSTEM - P 8000t BINARY 0.10 ' " , O. 4 5 Z \ / , 0 20 0 30 L~L 0.40 ATOMIC 0.50 0 60 0 70 0.00 0.90 1.00 F R A C T I O N Fig. 11--AI-Li phase diagram including the curves. To(I/cO and To(I~[3) VOLUME 20A. NOVEMBER 1989--2255 6, DEFECTS CONCENTRATIONS IN BETA PHASE A N U35 O i--i ~-4 Z -~nti-structure Kishio ond Brittoin CRLCULRTED / vo..n.? / ~ I--z the thermodynamic properties of the /3-A1Li phase. McAlister assumed the formation of vacancies on the A1 sublattice and antistructure defects on the Li sublattice. This assumption is contrary to the data of Kishio and Brittain, t56[ as shown in Figure 12. B. Integral Properties of the Various Phases uJ L~3 O Z o'J 2 u_ 0 44 0 I I I 46 48 5(] I 57 54 X Rl(Z) Fig. 1 2 - - D e f e c t concentrations of the/3 phase: comparison between the model-calculated values and experimental data. supporting the validity of the model used. Chang and Neumann I53~ found a relationship between ( - A H / / R T ) and ( - I n a) for triple-defect B2 phases, as shown in Figure 13. The dashed line is obtained from Eq. [5], while the solid line represents the experimental data for several triple-defect B2 phases. Even though/3-A1Li has a B32 structure with similar defects, i.e., two vacancies and one antistructure defect at the stoichiometric composition, the derived thermodynamic equations are identical to those for the triple-defect B2 phases] TM This suggests that the cohesive energy for these two types of phases with closely related structures may be similar. This is supported by the evidence presented in Figure 13, which provides an additional evidence to support the validity of the model. It is noteworthy to point out that, in assessing the thermodynamic and phase equilibrium data of A1-Li, McAlister ml also used a Wagner-Schottky type of model to account for the compositional dependence of Since thermodynamic data for the A1-Li binary are not as well established as those for many ferrous or noble metal alloys, it is useful to discuss the optimized ~r ASI, and AS m values in terms of empirical correlations. For instance, Chart I571reported an empirical relationship between Ax~ss and ~rar/for many ( - 1 0 0 ) compounds and alloys with ionic, covalent, and metallic bonding, as shown in Figure 14. This correlation was based on the earlier findings of Kubaschewski et al. [581and Slough. 159[ Although there is a large scatter of as much as ---7 J / K g atom at AHy = - 2 5 k J / g atom, a definite trend exists. Values of Axss; decrease rapidly at first, with decreasing i H I values near the origin. They tend to approach a constant value, certainly decreasing much less with decreasing AH/ at large negative values. Values of Axss1 and AHI, obtained in the present study for (~'-A13Li, fl-A1Li, y-A12Li3, and e-AlaLi9, are shown in Figure 14 and summarized in Table VI. These values are in excellent agreement with Chart's (571 correction. For /3, y, and e (Figure 14 and Table VI), values of Ax~s and AHs obtained by McAlister I3sj for/3 and y also fall within the spread of the correlation. But, for e, his assessed values fall outside the spread of the correlation. The values of Axss and AHI obtained by Saboungi and H s u , [371 a s well as by Sigli and Sanchez, ~39~fall outside the spread for all phases. Although this does not prove that the A~S and All/values used by the other investigators are wrong, it does indicate that their values are questionable. In fact, it is difficult to rationalize the origin for such large negative excess entropy values for alloys with enthalpies of formation of - 2 5 k J / g atom. Kubaschewski et al. t581 gave an empirical rule for estimating the entropy of melting of intermetallic phases: I0 Pdln -5 o ~ PdRI 8 CoAl o ~ / o.,%....... ~8 t /" ,-~ -10 I S ~d ,~ -15 o This Studu m //./// NIRI o / CoGo/ <1 /..I..I 1I Sigli -L~5 i -35 ~ I L l.CcCeFlI As "/0 HcRlister ~0 [] Soboungi and Hsq 9 -35 X.0 0 c~d Sonchez i , ? 4 -ln (x , , 8 8 -200 10 Fig. 1 3 - - T h e relationship between (AHs/RT) and ( - I n a) for several triple-defect B2 phases and the/3-AILi (B32) phase. 2256--VOLUME 20A, NOVEMBER 1989 -150 -lOg Hf (kJ/gatom) -50 s i 0 Fig. 1 4 - - A n empirical relationship between A~'S and AH[. [571Values for 6', /3, y, and e used in this study and those in the literature are included for comparison. METALLURGICAL TRANSACTIONS A T a b l e VI. C o m p a r i s o n s o f V a l u e s o f AxsS a n d A l l y A13Li ( 6 ' ) This study A~'S AHs - 1.64 -2400 McAlister i3sl Ax'S --- AH s Saboungi and Hsu 1371 Sigli and Sanchez 1391 A1Li (/3) A12Li3 (Y) AI4Li 9 (e) -3.46 -9770 -3.35 -8830 -3.23 -7190 - 8.6 - 14,985 -9.3 - 14,745 -23.7 -21,303 AXSS -- --18.3 -29.8 AHI -- -23,030 - 30,896 -26.4 - 23,809 A:'ss -- -- A / -b -- __ -20.5 -24,863 -34.2 -30,289 Ax's ( J / K g atom) and M-/+. ( J / g atom) for disordered phases, the entropy of melting is close to the arithmetic average of those of the two elements; for ordered intermetallic phases, an entropy of mixing term must be included. This approximation would yield a maximum value for the entropy of melting of an ordered phase. Table VII summarizes the values of A S m for 6', /3, y, and e obtained in the present study and those calculated from the empirical relationship. For 6', /3, and y, the values obtained by us agree well with those calculated, including the ideal entropy of mixing term. For the e phase, the value of A S m = 15 J / K g atom s e e m s somewhat high in comparison to a value of 14 J / K g atom obtained from this empirical rule. However, if we adjust the A S m value to be less than 14 J / K g atom, we would obtain an e + Li eutectic temperature of 435 K (162 ~ 5 K lower than the measured value (Figure 9). Since the affinity of Li for oxygen is high and the presence of a small amount of oxygen would raise the liquidus temperature, there is considerable uncertainty associated with this measurement. Additional experimental determination is needed before a definitive statement can be made concerning the eutectic temperature and the thermodynamic properties of e. C. 6' Phase The 6' phase is metastable and is formed readily from supersaturated a-A1 phase. Its Gibbs energy is obtained from the metastable solvus curve and the thermodynamic properties of the a-A1 phase. Figure 15 shows schematically the Gibbs energy curves of 6', as compared to those of a and I. According to this diagram, if formation of the stable phases is suppressed, 6' would form from a supercooled fiquid at T ~'~t. Alternatively, the a-A1 phase T a b l e VII. C o m p a r i s o n s o f A S m V a l u e s O b t a i n e d in the P r e s e n t S t u d y w i t h those C a l c u l a t e d f r o m the E m p i r i c a l R e l a t i o n s h i p I45] AS m AS m AS m (This Study) (No Mixing) (Completely Mixing) 10 9.1 8.6 8.5 15 15 14 14 AI3Li (6') A1Li (13) AlzLi3 (y) AI4Li9 (e) ASm(J/K g atom) 15 14 13 15 METALLURGICAL TRANSACTIONS A $i~s Eae~ of 40-25 ~X L~ 0 0 \ ,?, ~J -4 I- c_ 6 rl n ~a ~-8 T ot-L 6' -I0 400 i i i i 500 600 700 800 i 000 1000 Temp( K ) Fig. 15--Gibbs energies of 6', a, and 1 at 25 at. pct Li. formed at T ~--'t (the To curve shown in Figure 1 1) would transform to 6' at T ~'--'". In fact, experiments have shown that 6' forms readily when a-A1 alloys are annealed in the solid state. D. e Phase Myles e t a l . m found a thermal arrest at 548 K for several alloys between y and Li. They attributed this thermal arrest to a solid-state transformation of e. This isotherm is added as a dashed line in Figure 10. ACKNOWLEDGMENTS The authors wish to acknowledge the assistance of Dr. Y.-Y. Chuang (now deceased), who participated in the initial assessment of this system, H. Beumler for reviewing the manuscript, and D. Granger and Men G. Chu of A L C O A Laboratories for supplying many of the A1-Li samples used in this study. We are also grateful to the National Science Foundation (Grant No. NSFDMR-88-19758) and A L C O A Laboratories for financial support. VOLUME 20A, NOVEMBER 1989--2257 REFERENCES 1. R. Assmann: Z. Metallkd., 1926, vol. 18, pp. 51-54. 2. A. Muller: Z. Metallkd., 1926, vol. 18, pp. 231-35. 3. G. Grube, L. Mohr, and W. Bruening: Z. Elektrochem., 1935, vol. 41, pp. 880-83. 4. F.I. Shamray and P.Ya. Saldau: lzv. Akad. Nauk SSSR, Otd., Khim., 1937, pp. 631-40. 5. K.M. Myles, F.C. Mrazek, J.A. Smaga, and J.L. Settle: Proc. Syrup. and Workshop on Adv. Battery Res. and Design, U.S. ERDA Report ANL-76-8, Mar. 1976, pp. B50-B73. 6. E. Schtirmann and H.V. Voss: Giessereiforschung, 1981, vol. 33, pp. 33-42. 7. H. Vosskuhler: Metallwirtsch., 1937, vol. 16, pp. 907-09. 8. S.K. Nowak: Trans. A1ME, 1956, vol. 206, pp. 553-56. 9. W.R.D. Jones and P.P. Das: J. Inst. Met., 1958-59, vol. 87, pp. 338-40. 10. L.P. Costas and R.P. Marshall: Trans. AIME, 1962, vol. 224, pp. 970-74. 11. E.D. Levine and E.J. Rapperport: Trans. A1ME, 1963, vol. 227, pp. 1204-08. 12. J.M. Silcock: J. Inst. Met., 1959-60, vol. 88, pp. 357-64. 13. T. Yoshiyama, K. Hasebe, and M. Mannami: J. Phys. Soc. Jpn., 1968, vol. 25, p. 908. 14. B. Noble and G.E. Thompson: Met. Sci. J., 1971, vol. 5, pp. 114-120. 15. I.N. Fridlyander, V.S. Sandier, and T.I. Nikol'skaya: Metalloved. Term. Obrab. Met., 1972, vol. 3, pp. 41-43. 16. D.B. Williams and J.W. Edington: Phil. Mag., 1974, vol. 30 (5), pp. 1147-53. 17. D.B. Williams and J.W. Edington: Met. Sci., 1975, vol. 9, pp. 529-32. 18. D.B. Williams and J.W. Edington: Acta Metall., 1976, vol. 24, pp. 323-32. 19. R. Nozato and G. Nakai: Trans. JIM, 1977, vol. 18, pp. 679-80. 20. S. Ceresara, G. Cocco, G. Fagherazzi, A. Giarda, and L. Schiffini: Phil. Mag., 1977, vol. 35 (2), pp. 373-78. 21. O. Jensrud and N. Ryum: Mater. Sci. Eng., 1984, vol. 64, pp. 229-36. 22. E.S. Balmuth: Scripta Metall., 1984, vol. 18 (4), pp. 301-04. 23. S.F. Baumann and D.B. Williams: AI-Li H, T.H. Sanders, Jr. and E.A. Starke, eds., TMS-AIME, Warrendale, PA, 1983, pp. 17-29. 24. S.F. Baumann and D.B. Williams: Acta Metall., 1985, vol. 33 (6), pp. 1069-78. 25. D.-R. Liu and D.B. Williams: Scripta Metall., 1988, vol. 22, pp. 1361-65. 26. J.M. Papazian, C. Sigli, and J.M. Sanchez: Scripta Metall., 1986, vol. 20, pp. 201-06. 27. S. Spooner, D.B. Williams, and C.M. Sung: Al-Li 111, C. Baker, P.J. Gregson, S.J. Harris, and C.J. Peel, eds., Institute of Metals, London, 1986, pp. 329-336. 28. C.J, Wen, B.A. Boukamp, R.A. Huggins, and W. Weppner: J. Electrochem. Soc., 1979, vol. 126, pp. 2258-66. 29. E. Veleckis: J. Less-Common Met., 1980, vol. 73, pp. 49-60. 30. E. Schtirmann and I.K. Geissler: Giessereiforschung, 1980, vol. 32, pp. 163-74. 31. J.M. Hicter, A. Vermande, I. Ansara, and P. Desre: Rev. Int. Hautes Temper. et R~fract., 1971, vol. 8, pp. 197-200. 32. S.P. Yatsenko and E.A. Saltykova: Russ. J. Phys. Chem., 1974, vol. 48, pp. 1402-03. 2258--VOLUME 20A, NOVEMBER 1989 33. P.C. Yao and D.J. Fray: Metall. Trans. B, 1985, vol. 16B, pp. 41-45. 34. C.J. Wen, W. Weppner, B.A. Boukamp, and R.A. Huggins: Metall. Trans. B, 1980, vol. lIB, pp. 131-37. 35. N.P. Yao, L.A. Heredy, and R.C. Saunders: J. Electrochem. Soc., 1971, vol. 118, pp. 1039-42. 36. J.R. Selman, D.K. DeNuccio, C.J. Sy, and R.K. Steunenberg: J. Electrochem. Soc., 1977, vol. 124 (8), pp. 1160-64. 37. M.L. Saboungi and C.C. Hsu: CALPHAD, 1977, vol. 1 (3), pp. 237-51. 38. A.J. McAlister: Bull. Alloy Phase Diagrams, 1982, vol. 3 (2), pp. 177-83. 39. C. Sigli and J.M. Sanchez: Acta Metall., 1986, vol. 34, pp. 1021-28. 40. V.D. Bushmanov and S.P. Yatsenko: Russ. J. Phys. Chem., 1981, vol. 55 (11), pp. 1680-81. 41. A.G. Khachaturyan, T.F. Lindsey, and J.W. Morris, Jr.: Memll. Trans. A, 1988, vol. 19A, pp. 249-58. 42. V.I. Mikheeva, Z.K. Steryladkina, and O.N. Kryukova: Russ. J. Inorg. Chem., 1960, vol. 5 (3), pp. 867-71. 43. V.F. Shamrai, I.N. Fridlyander, and I.N. Sokolov: Symposium: Research m Nonferrous Alloys, Publishing House of the USSR Academy of Sciences, Moscow, 1963, vol. 4, pp. 100-07. 44. H. Ueda, A. Matsui, M. Furukawa, Y. Miura, and M. Nemoto: J. Jpn. Inst. Met., 1985, vol. 49 (7), pp. 562-68. 45. A. Smith: Al-Li Alloys 111, C. Baker, P.J. Gregson, S.J. Harris, and C.J. Peel, eds., Institute of Metals, London, 1986, pp. 164-72. 46. R.F. Ashton, D.S. Thompson, E.A. Starke, Jr., and F.S. Lin: Al-Li Alloys III, C. Baker, P.J. Gregson, S.J. Harris, and C.J. Peel, eds., Institute of Metals, London, 1986, pp. 66-77. 47. J.M. Papazian, J.P.W. Wagner, and W.D. Rooney: AI-Li Alloys IV, G. Champier, B. Dubost, D. Miannay, and L. Sabetay, eds., Journal de Physique, Paris, 1987, pp. 513-19. 48. P.N. Anyalebechi, D.E.J. Talbot, and D.A. Granger: Metall. Trans. B, 1988, vol. 19B, pp. 227-32. 49. D. Granger and M.G. Chu: ALCOA Laboratories, Aluminum Company of America, Alcoa Center, PA, private communication, 1987. 50. O. Kubaschewski: Iron-Binary Phase Diagrams, Springer-Verlag, Berlin, 1982, p. 4. 51. J.-C. Lin: Ph.D. Thesis, University of Wisconsin-Madison, Madison, WI, 1986. 52. J.P. Neumann, Y.A. Chang, and C.M. Lee: Acta Metall., 1976, vol. 24, pp. 593-604. 53. Y.A. Chang and J.P. Neumann: Progress in Solid-State Chem., G.M. Rosenblatt and W.L. Worrell, eds., 1982, vol. 14, pp. 221-301. 54. R. Hultgren, P.D. Desai, D.T. Hawkins, M. Gleiser, K.K. Kelley, and D.D. Wagman: Selected Values of the Thermodynamic Properties of the Elements, ASM, Metals Park, OH, 1973. 55. L. Kaufman and H. Bernstein: Computer Calculation of Phase Diagrams, Academic Press, New York, NY, 1970. 56. K. Kishio and J.O. Brittain: J. Phys. Chem. Solids, 1979, vol. 40, pp. 933-40. 57. T. Chart: High Temp. High Pressures, 1973, vol. 5, pp. 241-52. 58. O. Kubaschewski, E.LL. Evans, and C.B. Alcock: Metallurgical Thermochemistry, 4th ed., Pergamon Press, London, 1967. 59. W. Slough: in Metallurgical Chemistry, O. Kubaschewski, ed., Her Majesty's Stationery Office, London, 1972, pp. 301-22. METALLURGICAL TRANSACTIONS A