THE SYSTEM ALUMINUM-ANTIMONY-BISMUTH A. N. CAMPBELL AND J . W I N K L E R

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/20/17 For personal use only.

Department of Chemistry, University of Manitoba, Winnipeg, Manitoba Received September 12, 1962

ABSTRACT -4 complete study, by the methods of thermal and isothermal analysis and of X-ray powder photography, has been made of the equilibrium diagram of the system aluminum-antimonybismuth. The critical solution temperature and composition of the congruent liquids in the system aluminum-bismuth have been determined as have also the compositions of congruent liquids in the ternary system. The only compound occurring in this system is AlSb and the solid model is largely occupied by the equilibrium surface of this compound. An explanation of the anomalous form of the solidus in the antimony-bismuth system is gi>en.

The ternary alloy described by the title is based on the three binary systems aluminumantimony, aluminum-bismuth, and antimony-bismuth. These systems are described briefly in what follows: complete bibliographies are given under References. In the system aluminum-antimony, the compound AlSb, melting congruently a t 1060' C, is formed. This compound decomposes rapidly in moist air, forming Al(OH)3 and hydrogen. The two eutectics lie very close to pure aluminum and pure antimony respectively, and the solid solubilities of aluminum in antimony and of antimony in aluminum are exceedingly small. The crystal structure of AlSb is face-centered cubic. The literature on the system is extensive (1). The system aluminum-bismuth forms two liquid layers a t all temperatures hitherto investigated (2). Kempf and van Horn give the monotectic temperature (equilibrium of liquid 1, liquid 2, and solid aluminum) as 657.1'. The solubility of bismuth in solid aluminum does not exceed 0.2%. In view of the close chemical similarity of bismuth and antimony it is perhaps surprising that aluminum and bismuth do not form a compound analogous to AlSb, but there is no evidence for the existence of such a compound. The equilibrium diagram of the system antimony-bismuth has presented difficulties (3). Alternative diagrams are shown in Fig. 1. There isno doubt that the liquidus is continuous

1

-

200

1

Bi

20

40

60

wt0/0 A n t i m o n y

FIG. I. Canadian Journal of Chemistry. Volume 41 (1963)

743

80

Sb

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/20/17 For personal use only.

744

CANADIAN JOURSAL O F CHEMISTRY. VOL. 41, 1963

and therefore one would expect that the solidus would also be so. Thermal analysis, however, as applied for instance by Cook, results in a solidus as shown in Fig. 1, that is, unbroken but horizontal up to 60Yo antimony. Otani, however, using the method of electrical resistance, obtains a solidus which may be described as normal, i.e. temperature rises continuously with increasing antimony content. X-Ray analysis shows that the (solid) system antimony-bismuth consists of one phase only, corresponding to a complete series of solid solutions. An explanation is offered in the present work of the apparently abnormal form of the solidus, which results from thermal analysis on cooling. The ternary system aluminum-antimony-bismuth has received little attention (4). Some preliminary work by Wright yielded nothing of importance. Some experiments were done by Kasten and these were offered, in an idealized form, as a practical illustration in a paper by Ritzau on the representation of quaternary systems. EXPERIMENTAL Aluminum was obtained from the Department of Mines and Technical Surveys, Mines Branch, Ottawa, and was stated to have the following composition: -41, 99.9946; Fe, 0.0019; Si, 0.0018; Cu, 0.00187,. The antimony used was Fisher Certified Reagent, the analysis being: Sb, 99.8; As, 0.0004; Cu, 0 001; Pb, 0.04; Fe, 0.002570. The bismuth was also Fisher Certified Reagent with impurities given as: .As, 0.0001; Cu, 0.002; Fe, 0.002; Pb, 0.005; Sb, 0.005y0. For the lower-temperature work, iron-constantan thern~ocoupleswere used, standardized by the freezing points of pure tin and pure cadmium. For the high-temperature work, thermocouples of platinum - platinum rhodium (10% Rh) were used. They were standardized from the freezing points of tin, cadmium, zinc, antimony, and silver, the last metal being allowed to freeze under a stream of hydrogen. For thermal analysis, a Honeywell Brown single-pen recorder was used. The thermocouple was connected in parallel with the recorder and a vernier precision potentiometer, by means of a six-point switch, so that the temperatures corresponding to points of interest on the r e ~ o r d e rdiagram could be determined. The potentiometer had a sensitivity of &0.001 mv, corresponding to a temperature uncertainty of k0.05" C with the iron-constantan thermocouple and of f0.1" with the less sensitive platinum - platinum rhodium thermocouple. The accuracy of reading does not, of course, represent the accuracy of measurement, since all thermal analysis measurements suffer from supercooling effects, the amount of which it is difficult to estimate. Our results were, a t least, all reproducible and we feel t h a t they are accurate to within &lo. A Hoskins combustion furnace (resistance type) was used for thermal and isothermal analysis but the alloys for investigation were prepared by melting appropriate weights of the pure metals, either under vacuum or in a current of argon, in a n Ajax-Northrup high-frequency induction furnace. For isothermal work, the resistance furnace was kept a t constant temperature by means of a photocell, a s described by Campbell and Kartzmark (5). Thermal analysis was used only for the determination of the surface of primary crystallization of the compound AlSb. The alundum crucibles, in which the melts were contained, invariably ruptured on solidification of the alloy, because of the well-known expansion on solidification of bismuth and antimony. I t was, therefore, not possible to repeat determinations with the identical alloy. As the compound AlSb forms from its components only very slowly a t temperatures below the melting point of the compound, the procedure adopted was to add the compound itself to a mixture of aluminum and bismuth or of bismuth and antimony. For example, the composition of alloy No. 3, on the AlSb surface, was 20.0% Al, 50.0y0 Sb, and 30.0% Bi: the 80-g specimen therefore contained 16.0 g Al, 40.0 g Sb, and 24.0 g Bi. This composition was attained by using 48.9 g ALSb, 7.1 g Al, and 24.0 g Bi. The compound itself was readily prepared by melting stoichiometric quantities of aluminum and antimony in the induction furnace under argon atmosphere, a t a temperature of 1200" C. I t was necessary to keep the preparation in a vacuum desiccator, out of contact with moist air. The determination of the eutectic temperatures, namely those of the eutectic A1-Bi, of the pseudobinary system AISb-Bi, and of the ternary eutectic, was carried o u t by means of a differential thermocouple, similar to that described by Desch (6, p. 95). T h e reference body was pure bismuth, of the same weight as the alloy. Because of the existence of (liquid) miscibility gaps in both ternary and in one of the binary systems, most of the work was isothermal. The alloys were kept a t constant temperatures until equilibrium was attained, when both liquid layers were sampled for analysis. For work on the binary miscibility gap, equal volumes (78.4 g Bi, 21.6 g Al) were melted in the induction furnace: the stirring action of the eddy currents in this type of furnace is advantageous in attaining equilibrium. The melt was then transferred to the Hoskins furnace, already under control a t the desired equilibrium temperature, and, after 1 hour, both layers were sampled by withdrawing through narrow (1.7-mm i.d.) alundum tubing. The variation of

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/20/17 For personal use only.

CAMPBELL AND WINKLER: Al-Sb-Bi

745

SYSTEM

temperature in the furnace was about fl o . The method was used successfully up to 1180' b u t above this temperature the alundum tubing always broke. This temperature then, 118O0,represents the upper limit of our worlr, so t h a t direct determinations of critical solution temperature and composition were not possible. The samples withdrawn as described above were weighed, dissolved in concentrated nitric acid, diluted to volume, and analyzed. Essentially the same method was used for the investigation of congruent solutions in the ternary system, except t h a t here more than one alloy had to be prepared, a t each temperature, to obtain suitable variation in the tie lines. Standard methods were used for the analysis of the alloys. In the bismuth-rich layer of the system aluminum-bismuth, the bismuth was determined by direct titration with E D T A (ethylenediaminetetracetic acid) using xylenol orange indicator (7). The aluminum mas determined with 8-hydroxyquinoline. In the aluminum-rich layer only, bismuth was determined as BiOCl (8). Alloys containing all three metals were analyzed as follows: The antimony and bismuth were precipitated with H2S and the antimony separated with ammonium sulphide. The bismuth was then titrated with E D T A and the antimony determined by titration with potassium bromate. Small quantities of aluminum were determined directly with 8-hydroxyquinoline but large amounts of aluminum were obtained b y difference. X-Ray powder photography was carried out in the usual way, using collodion rollings. E X P E R I M E X T A L RESULTS TABLE I Equilibrium composition of the two congruent liquids in the system aluminum-bismuth Temperature (" c)

W t % Bi Liquid 1 Liquid 2

Wtyo Bi ~lTemRfure Mean

Liquid 1

Liquid 2

Mean

When the data of Table I are plotted, it appears that the miscibility gap will close a t about 1300" C.-The critical composition is then obtained by application of the law of the rectilinear diameter, as about 6870 bismuth. Eutectic Points (a) Binary Eutectic in System Aluminum-Bismxth The maximum decrease in the freezing temperature of bismuth was found to be 1.8", produced by the addition of 0.30y0 Al. The effects of progressive additions of aluminum are shown in Table 11, sample Nos. Bl-B7. ( b ) The Eutectic in the Pseudobinary System Bi-AlSb The maximum decrease in the freezing temperature of bismuth was found to be 2.30" C, caused by the addition of 0.50% AlSb. The experimental results are given in Table 11, Nos. Cl-C4. ( c ) The Ternary Eutectic The addition of 0.20y0 A1 to an alloy of the composition of the pseudobinary eutectic alloy caused a further decrease of 0.20" C. Hence we conclude that the eutectic temperature in the ternary system Al-Sb-Bi is 268.8" C, a t a composition of 99.370 Bi, 0.40% Sb, and 0.30% Al. The experimental data are given in Table 11, Nos. TI-T4. The solid phases are All AlSb, and solid solution of antimony in bismuth. Since antimony and bismuth form a continuous series of solid solutions, there is only one ternary eutectic in this system. Thermal Analysis on Cooling Since the equilibrium solid model is largely occupied by the AlSb surface, it was

746

C.4NADIAN JOURNAL O F CHEMISTRY. VOL. 41, 1963

TABLE I1 Eutectic data: decrease in temperature and eutectic compositions

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/20/17 For personal use only.

Sample No. Addition wt%

A T (" C)

B1 B2 B3 B4

0 . 0 5 A1 0 . 1 0 A1 0.20 3 1 0 . 3 0 A1

1.2 1.8

B7 C1

1 . 5 A1 0 . 2 5 AlSb

1.63

// Sample No.

Addition wt%

C2

AT

(O

C)

0 . 5 0 AlSb 0 . 7 5 AlSb 1 . 0 0 AlSb 0 . 1 A1

2.30 1 94 1.96 2.32

0 . 5 A1

2.38

necessary to determine the temperatures of primary separation of AlSb from different melts. T h e results are contained in Table 111. TABLE I11 Three-component system: computed compositions and temperatures of primary crystallization (AlSb) Point No.

\Vt% Sb

\Vt% Bi

Temp.

(O

C)

(1

Point No.

W t % Sb

\lit% Bi

Temp. (' C)

Thermal Analysis on Heating Because of the uncertainty as to the true form of the solidus in the antimony-bismuth system two alloys were prepared, annealed for 120 hours a t 280°C, and then heated uniformly. Hysteresis effects are usually less marked on heating than on cooling. The results are given in Table IV. Thermal analys-is on heating Temperature of inflection on heating curve (" C)

Composition of alloy

Conjugate Liquids in Ternary System TABLE V Three-component system: composition of the two conjugate liquids in isothermal sections Equilibrium temperature ("

c)

Al-rich layer bYt% Sb

W t % Bi

Bi-rich layer W t % Sb

iVt% Bi

CAMPBELL AND WINKLER: Al-Sb-Bi SYSTEM

747

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/20/17 For personal use only.

From the totality of the preceding data, the solid model, of which Fig. 2 is a projection, can be constructed. This figure gives a complete picture of all equilibria.

FIG. 2. DISCUSSION OF RESULTS

The system as a whole is simple: it presents no features of unusual interest. The alloys of the system, however, have recently become of considerable commercial interest. The point of chief importance lies in the binary system antimony-bismuth. and is concerned with the true form of the solidus curve. Our X-ray work, details of which we do not give, since it was merely a repetition of previous work, was directed t o an examination of the question of the existence, or otherwise, of an unbroken series of mix-crystals between antimony and bismuth. We confirmed the results of previous workers: no lines indicative of the presence of a second solid phase ever occur on the X-ray powder photographs. Using annealed specimens, we found also that the lattice dimensions vary linearly with composition, especially if composition be expressed in atomic percent. On the other hand, the X-ray photographs of the unannealed specimens showed a characteristic structure. The appearance was not that of a mechanical mixture, in which the lines both of bismuth and of antimony would have appeared sharply defined, but rather an indefinite spread of the blackening between the lines. The back-reflection of the unannealed specimens was almost completely extinguished. With increasing annealing time (24 -+ 280 hours) the sharpness of the diffraction lines improved greatly and the lines of the back-reflection reappeared. I t is evident that an irregular lattice distortion

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/20/17 For personal use only.

748

CANADIAN JOURXAL O F

CHEMISTRY. VOL. 41,

1983

takes place, when bismuth is added to antimony and that this is arrested by too rapid cooling. On annealing, a superstructure is formed, corresponding t o the composition of the solid solution. The conclusion seems inescapable that bismuth and antimony form a continuous series of solid solutions and that therefore the solidus should be continuous. The problem remains of why the solidus determined by the ordinary method of thermal analysis on cooling should exhibit an apparently horizontal portion, while the solidus determined by the method of electrical resistance, on annealed specimens, should be normal in form. For the sake of convenience, we describe the abnormal form of the solidus as "Cook's solidus" and the other one as "Otani's solidus". We think we have an explanation for the discrepancy as follows: I t was observed throughout the present investigation that all alloys containing an appreciable amount of bismuth release or press out drops or globules of alloy on solidification, an evident consequence of the well-linown expansion of bismuth on solidification. Alloys containing less than 30 atomic% of bismuth showed no tendency to squeeze out alloy. Beginning with 30 atomicyo increasing quantities of metal appear on the surface of the ingot as the concentration of bismuth is increased (Fig. 3). The crl-stallization of alloys (of isomorphous metals) whose freezing points all lie between those of the pure metals is discussed by Desch (6, p. 26), who shours how the solidus may undergo an apparent lowering owing to slow diffusion within the solid solution but the existence of a horizontal solidus cannot be explained by slow diffusion only. MTe think, however, that the existence of extruded globules from the solidified alloy offers an explanation. Consider an alloy of composition X in Fig. 4. On cooling, the first crystals will appear a t a temperature T I ,with composition x'. If the behavior is ideal, on further lowering of temperature the composition of the liquid will change along the liquidus and that of the equilibrium solid along the solidus. At temperature T 2 ,the last infinitesimal amount of melt has the composition x". In the system containing antimony and bismuth, however, this last portion of the melt is squeezed out of the solidified ingot and, since it is no longer in equilibrium with the rest of the alloy, constitutes a new system. The final solidification temperature wiIl be some such temperature as T3. Following this reasoning, the curve Bi-Ts-Sb has been constructed for concentrations varying from 20 to 60% antimony, as shown in Fig. 4. The curve so obtained has a surprising similarity t o Cook's solidus. Moreover, bearing in mind that solid diffusion in antimony-bismuth alloys is slow, it can be seen that the experimentally determined solidus will be still lower. The conditions for the expulsion of the last melt are twofold. First, cooling must progress a t a rate too great for the establishment of equilibrium, and this is nearlj- always true in thermal analysis, where the rate of cooling has to be great enough for inflections t o show up. Secondly, the specimens must have a compact form, as with the ingots shown in Fig. 3. The specimens used by Otani were round bars of dimensions 5 x 9 0 mm. Here the expansion takes place mainly in one dimension without expulsion of liquid. This is why Otani finds higher temperatures for his solidus, even without annealing. The last melt is squeezed out a t the mechanically weakest point of the ingot and any temperature-measuring device introduced into the melt constitutes such a weak point, so that the last melt is squeezed out around the thermometer and the (erroneous) temperature T Qrecorded. As mentioned under Experimental Results, two experiments were conducted in an endeavor to obtain the true solidus temperature by thermal analysis, using rising temperature. Two ingots of different composition were annealed for 120 hours a t 280' C,

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/20/17 For personal use only.

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/20/17 For personal use only.

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/20/17 For personal use only.

CAMPBELL A S D WINKLER: Al-Sb-Bi

200

61

20

40

60

80

SYSTEM

Sb

wt% A n t ~ m o n y

FIG. 4.

when it was observed that the (solidified) globules were incorporated again in the ingots. A 30-gauge iron-constantan thermocouple in a thin alundum tube was introduced into the center of the ingot and the heating curve recorded. The beginning of melting was found to lie a t temperatures which, while they do not lie exactly on Otani's curve, are evidently higher than the temperatures of Cook's solidus. REFERENCES 1. C. R. A. L ~ R I G H T . J. SOC.Chem. Ind. 11, 492 (1892). D. 4. ROCHE. lV1oniteur Scientifique, 7 (4), . 269 (1893). H. GAUTIER.Bull. Soc. dlEncour. Ind. Nat. 1 (5), 1315 (1896); Contribution 1'Etude des Alliages. Paris. 1901. pp. 93, 112. W. CAMPBELL and J. A. MATHETVS.J. Am. Chem. Soc. 24, 259 (1902). G. T A ~ I A N P Z. U . Anorg. Chem. 48, 54 (1906). G. TAM~IANN. Lehrbuch der heterogenen Gleichgewichte. Braunschweig. 1924. K. BORKEMANX.Metallurgie, 7, 573 (1910). M. \ O N SCHTVARZ. Metal1 -und Legieru~igskunde,78 (1929). F. S ~ U E R W A L 2. D . hletallkunde, 14, 458 (1922). \V. GUERTLERand A. BERGMANN. Z. Metallkunde, 25, 82 (1933). E. L o o ~ s RASSOW. Aluminium, 3, 20 (1931). E. H. DIX, F. KELLER,and L. A. WILLEY. Trans. AIPIE, Metals Div. 93, 396 (1931). J . VESZELKA.Mitt. berg.-huttenmann. Abt, ungar. Hochschule BergForstw. Sopron. 3, 193 (1931). METALSHANDBOOK. American Society for Metals. 1948. E. H. \VRIGHTand L. A. WILLEY. ALCOA Technical Papei No. 15. 1960. G. G. URAZOV.J. RUSS. Phys.-Chem. Soc. 51, 461 (1919). E. A. OWENand G. D. PRESTON.Proc. Phys. Soc. (London), 36, 341 (1924). R. F. BLUNT,H. P. R. FREDERIKSE, J. H. BECKER,and \\I. R. HOSLER.Phys. Rev. 96, 578 (1954). H, m7ELKER. Z. Xaturforsch. 7a, 744 (1952); 8a, 248 (1953). and F. H. SEVILLE. J. Chem. 2. C. R. A. WRIGHT.J. SOC.Chem. Ind. 11, 492 (1892). C. T . HEYCOCK Soc. 61, 893 (1892). W. CAMPBELL and J. A. WIATHEWS.J. Am. Chem. Soc. 24, 259 (1902). PECHEUX.Compt. Rend. 138 (2), 1501 (1904). A. G. C. GWYER.2. Anorg. Chem. 49, 311 (1906). M . HAYSEXand B. BLUMENTHAL.Metallwirtschaft. 10. 925 11931'1. L. \V. KEAIPFand K. R. VAN HORN. Trans. AIME, Inst. Metals Div. 133, 81 (1939). . a 1'Etude 3. K. H U T T ~ Eand R G. TAM~IAKN. Z. Anorg. Chem. 44, 138 (1905). H. G ~ U T I E RContribution des Alliages. Paris. 1901. p. 114. M. COOK. J. Inst. Metals, 28, 421 (1922). B. 0 ~ 4 ~ 1Scl. . Repts. Tohoku Imp. Univ. 13, 293 (1925). CHU-PHAYYAP. Trans. AIME, 93, 185 (1931). E. G. BOWEXand W. M. JONES. Phil. Mag. 13, 1029 (1932). \V. F. EHRETand M. B. ~ B R ~ M S O N . J. Am. Chem. Soc. 56, 385 (1934). E. SCHEIL. Z. Elektrochem. 49, 242 (1943). G. MASING, P. RAHLFS.and W. SCHAARUTACHTER. Z. Metallkunde. 40. 333 (1949). T . HEUMANV.2. Naturforsch. 5a, 216 (1950); Z. Metallkunde, 42, 182 (1951). 6. A. ZAPFFE. Z. Metallkunde, 44, 397 (1953). P. LiT. BRIDGRIAN.Proc. Am. Acad. Arts Sci. 84, 43. KASTEN.\Tiss. Verof4. C. R. A. WRIGHT. Proc. Roy. Soc. (London), 55, 137 (1894). GOTZ-WERNER fentl. Siemens-\Verken, Werkstoff Sonderheft, 53 (1940): G. Ritzau. Wiss. Veroffentl. SiemensWerken, Werkstoff Sonderheft, 44 (1940). 5 . A. 3.CAMPBELL and R. KARTZMARK.Can. T. Chem. 34. 1431 (1956). 6. C. H. DESCH. Metallography. 5th ed. 1942. 7. F. J. \VELCHER. The analytical uses of ethylenediamine tetracetic acid. p. 207. 8. \V. W. SCOTT. Standard methods of chemical analysis. 5th ed. p. 153. >

,