Thermodynamic properties of solid nickel-platinum alloys

Thermodynamic properties of solid nickel-platinum alloys

THERMODYNAMIC PROPERTIES OF SOLID NICKEL-PLATINUM ALLOYS* and J. B. DARBY, JR.? R. A. WALKER71 The activities and free energies of formation of...

646KB Sizes 0 Downloads 109 Views

THERMODYNAMIC

PROPERTIES

OF SOLID

NICKEL-PLATINUM

ALLOYS*

and J. B. DARBY, JR.?

R. A. WALKER71

The activities and free energies of formation of solid Ni-Pt alloys at 1625°K were calculated from vapor-pressure measurements obtained with a torsion-effusion apparatus. Heats of formation at 298°K were determined by liquid-tin solution calorimetry. The entropies of formation at 1625°K were computed from the experimental free energy and heat-of-formation values, assuming Neumann-Kopp behavior between 298 and 1625°K. The heat-of-formation results appear to reflect the ordered structures that exist at low-temperatures in the Ni-Pt system. The entropies are believed to be largely configurational in origin. PROPRIETES

THERMODYNAMIQUES

DES

ALLIAGES

SOLIDES

NICKEL-PLATINE

Les activites et les 6nergies libres de formation des alliages solides Ni-Pt & 1625°K ont 6tB caloulees 21 partir des mesures de pression de vapeur obtenues avec un appareil de torsion-effusion. Lea chaleurs de formation & 298°K ont 6tB dbterminees par calorimetric B bain d’&ain. Les entropies de formation PL 1625°K ont 6tB calculbes 8. partir des valeurs exp&imentales de 1’Bnergie libre et de la chaleur de formation, en supposant un comportement de Neumann-Kopp entre 298 et 1625°K. Les rbsultats obtenus pour les chaleurs de formation semblent lies aux structures ordonnbes existant aux basses temperatures dans le systbme Ni-Pt. Les auteurs pensent que les entropies sont dans une large mesure liees aux configurations. THERMODYNAMISCHE

EIGENSCHAFTEN

FESTER

NICKEL-PLATIN-LEGIERUNGEN

Die Aktivitiiten und freien Energien der Bildung fester Ni-Pt-Legierungen bei 1625°K wurden aus den Ergebnissen van Dampfdruckmessungen in einer Torsione-Apparatur berechnet. BildungswBrmen bei Die Bildungsentropien bei 1625°K 298°K wurden in einem fliissig-Zinn-Lizjsungskalorimeter bestimmt. wurden aus experimentellen Werten der freien Energien und Bildungswarmen unter der Annahme eines Neumann-Kopp-Verhaltens zwischen 298 und 1625’K berechnet. DieErgebnisse fiir die Bildungswiirmen spiegeln die bei tiefen Temperaturen im System Ni-Pt existierenden geordneten Strukturen wider. Die Entropien riihren unserer Ansicht nach grijDtenteils von der Konfiguration her.

1. INTRODUCTION

An investigation of

the

Ni-Pd

system,

by

Bidwell

revealed an unusual behavior formation

same alloys from heat-of-solution

of the thermodynamic

properties

and

Speiser,(n

in the integral heats of

(AH) for this system.

The heat-of-forma-

tion values were small and varied from endothermic behavior

for nickel-rich

for palladium-rich

alloys to exothermic behavior The general similarity of

alloys.

elemental palladium and platinum, in terms of atomic size and electronic structure, led to the question Ni-Pt

of whether system

position

a

dependence.

properties

from

The

only

force

To compare

of Ni-Pd

and Ni-Pt

the integral thermodynamic system were required.

in the literature values

measurements

at

the thermodynamic alloys quantitatively,

properties

of the Ni-Pt

The present work was initiated to determine the free energies of formation for a series of solid Ni-Pt alloys by the torsion-effusion vapor-pressure nique, and to determine the heats of formation

techof the

* Received April 13, 1970. Work performed at the Argonne National Laboratory under the auspices of the U.S. Atomic Energy Commission. t -4110~ Properties Group, Materials Science Division, Argonne National Laboratory, Argonne, Illinois 60439. $Now at: Center for Metal and Mineral Technology, Nicol Hall, Queen’s University, Kingston, Ontario. ACTA

METALLURGICA,

VOL.

18, DECEMBER

was attractive

platinum

calorimeter.

measurements The

for this investigation,

form a complete

Ni-Pt

in a

system

since nickel and

series of solid solutions

at

two high temperatures. (2) At lower temperatures, superlattices are formed near the compositions Ni,Pt and NiPt at 853 and 918”K, respectively.

The vapor

pressures

sufficiently

of

different

nickel

and

platinum

in magnitude

torsion-effusion

com-

thermodynamic

and the excess free-energy

solution

to

permit

are the

use of

the

technique.

of the

asymmetrical

published

electromotive

1273 and 1473°K. properties

similar

of this system

were the activities derived

the heats of formation

had

liquid-tin

1970

2. EXPERIMENTAL

Torsion-effusion The

PROCEDURES

method

torsion-effusion

apparatus

has

been

scribed(3-s) in detail and only a brief description presented.

A cylindrical

of high-purity eccentrically

alumina,

de-

will be

effusion cell was constructed and two orifices

in the cell allowed

positioned

the nickel vapor to

escape. The cell was suspended at the end of a fme tungsten wire, and the total torque produced was proportional to the total vapor pressure within the cell (assumed equal to the equilibrium vapor pressure of nickel).

The slightly irregular shape of the orifices

did not allow an accurate calculation of the FreemanSearcy correction factor 6, for an orifice of non-zero length. The constants were determined”) by comparing the measured vapor pressure of copper in the alumina cell with values determined in an independent experiment.@) 1261

ACTA

1262

Several

calibration

experiments

METALLURGICA,

were

that

induced

by

a small,

fairly reproducible additional Without

unidirectional

the furnace.

The induced

and corrections

rotation corrections,

was

torque

was

were made for the wire.

the largest error in vapor pres-

sure was only about 5 per cent for the platinum-rich alloys (and hence for small deflections), and only approximately 2 per cent for the nickel-rich alloys. The vapor

pressure

of nickel in equilibrium

the pure solid nickel, and in equilibrium of

solid

Ni-Pt

temperatures. computed ployed

alloys

at

several

of nickel and platinum,

the vapor-pressure

subsequently

with

with a series

determined

The activities

from

formation

was

to calculate

data,

were em-

the free energies of

of the alloys.

of an alloy was calculated

of the heat-of-solution

of the pure

components and of the alloy in liquid tin. twin-well, liquid-metal solution calorimeter earlier.cg)

The samples,

tained at a temperature

The was

which were main-

Ti (approximately

298”K),

were dropped into a bath consisting of 4 moles (65 ml) of liquid

tin maintained

proximately

698°K).

50 cal.

A sample

at a temperature

Tf (ap-

Each sample had a mass that

would yield a measured

3. EXPERIMENTAL

heat effect of approximately

was completely

of the surrounding

a larger

RESULTS

Torsiow-effusion The temperature of nickel

dependence

in equilibrium

of the vapor pressure

with the Ni-Pt

alloys

dissolved

block

was complete

was

determined from at least two independent

experiments

per alloy.

of the vapor

The temperature

pressure

of pure

dependence

solid nickel

separate experiments

was measured

in six

during the course of the present

investigation. The consistency of the data for pure nickel (Fig. 1) is indicative of the reliability of the apparatus

throughout

relationships

for

the investigation.

each

The linear

set of vapor-pressure

derived by the least-squares

method,

data,

are presented in of pure

within

good agreement

with the average value of 102.80 f

0.50 kcal/g-atom

reported

temperature

dependence

nickel obtained

by Hultgren

et aZ.(lo) The

of the vapor pressure of pure

in the present investigation,

shown in

Fig. 2, compares favorably with published data for the vapor pressure of solid(11-15) and liquid’16J7) nickel. The

activities

of nickel

of

culated

the vapor-pressure

from

sponding

1575,

in the Ni-Pt

temperatures

from 3 to 5 min and the exponential decay of the temperature of the calorimetric cell to the temperature imately

to provide

Table 1. The value for the heat of sublimation

The heat of formation

described

small pieces ratio.

nickel at 298°K is 101.85 f 0.20 kcal/g-atom, calculated by means of the third-law method, and is in

Calorimetry from measurements

1970

surface-to-volume

torque

of the cell and suspension

18,

0.004 in. foil into

performed

during the course of the present work, and the results indicated

VOL.

platinum

1625 and

activities

TEMPERATURE 14x)

I400

1380

1360

1675°K data.

alloys

at

were calThe

corre-

were determined

by

(“C)

1340

in approx-

150 min. II:

I

Alloy preparation Samples of the same alloy were employed the torsion-effusion

and the calorimetric

I

in both

experiments.

The alloys were prepared from pure nickel (99.999%) and pure platinum Mineral

(99.999%),

and Chemical

on a water-cooled

supplied by the United

Corporation,

copper

by arc melting

hearth in a helium-argon

atmosphere, followed by a homogenization anneal for 1 week at 1473°K. The alloys were rolled into sheets of approximately

0.1 in. in thickness,

a few hours at approximately

re-annealed

1473”K,

for

and given

a

final reduction to foil of about 0.004 in. in thickness. For the calorimetric measurements, samples with the appropriate mass were cut from the foil, rolled into a cylinder of about 1 in. in height and 0.125 in. in diameter, given a final heat treatment at 1473°K for 4 hr, and water quenched. effusion measurements

Samples for the torsion-

were obtained

by cutting the

FIQ.

1. The vapor pressure of nickel as a function of temperature.

WALKER

DARBY,

AND

JR.:

THERMODYNAMIC

PROPERTIES

TABLE 1. Vapor pressure of nickel in equilibrium Standard deviation Of log pNi (a x 104)

b

with Ni-Pt

Temperature

Ni-Pt

ALLOYS

range

Number of points

a

1.0 0.9

9.9180 9.8695

-21,255 -21,297

92 28

1572-1682 1574-1676

49 18

8:;

9.7148 9.8281

-21,375 -21,370

62 79

1600-1704 1574-1677

f ::

8:: o”.:

9.4524 9.3324 8.6622 9.3841

-21,221 -21,331 -21,823

78 92 101 97

1650-1769 1625-1703 1716-1836 1676-1795

;: 18

0:2 0.1

7.6326 6.8274

--21,034 19,708 -19,160

167 200

1755-1863 1809-1891

8: 36

= a + b/T.

integration of the Gibbs-Duhem equation with the aid of the u function,(ls) aNi = In YNi/Nrt2. The uncertainty in the nickel activities is estimated to be +4 per cent and for the platinum activities f5 per cent. The partial and integral free energies, heats of formation and entropies were calculated for 1625°K from the activities and the data are listed in Table 2. The integral free energies of formation are accurate to within f 115 Cal/g-atom. The integral heats of formation and the integral entropies have considerably larger errors than the free-energy values and are listed for completeness; estimated errors are f2.26

TEMPERATURE PC) 1500

(OK)

1300

I100

1000

TNESMEYANOV (1963)

\

Cal/g-atom-deg for the entropies and &3900 Cal/gatom for the heats of formation. Calorimetry

The heats of solution at 298’K, obtained in the present work, for pure nickel or platinum in liquid tin are presented in Table 3 together with pertinent data reported in the literature. (1s-23) In order to compare the various sets of data, the heat contents for the pure elements over the temperature interval 29%914’K were evaluated from equations given by Hultgren et al.(lO) and were combined with the present results at 298°K to obtain the calculated heats of solution at 914°K. The smaller values for platinum reported by Oriani and Murphy(ls) are perhaps the result of incomplete dissolution of the solute ; the data reported by Geikent20)are in better agreement with the present work. The present results at 298°K for the heat of solution of nickel in tin are in good agreement with the values at 273°K reported by Leach and Bever.f21) The nickel results at 910°K reported by Day and Hultgren,(22) and at 913°K by Oriani and Murphy(23) are much smaller than the present data extrapolated to 914°K. The integral heats of formation of the Ni-Pt alloys at 298°K were determined from the heats of solution at 298°K (AH,,,,) by means of the equation AH = NNi AHZI, + N,t AHPt S&l

5.0

6.0

7.0

6.0

lOOOO/ TeK) Fra. 2. Comparative

data for nickel.

1263

alloys

NNI

Log p (mm Hg)

1700

OF

the

vapor

pressure

of

-

AH$T.

(1)

The heats of solution and the heats of formation at 298’K are listed in Table 4. The uncertainty in the heats of formation was determined as the maximum deviation of the individual measurements from the average value for each composition, and did not exceed f3 per cent and usually was less than f 1 per cent. The integral thermodynamic quantities for the Ni-Pt system, determined by using data obtained from torsion-effusion and calorimetry measurements,

ACTA

1264 TABLE 2. Thermodynamic

METALLURGICA,

quantities for Ni-Pt

NNi

aNi

apt

AQN~ A&t (Cal/g-atom)

0.9 0.8 0.7

0.843 0.691 0.528

0.012 0.037 0.083

-552 -1193 -2063

8::

0.359 0.233

0.170 0.289

:.: 0:e 0.1

0.131 0.076 0.046 0.016

0.463 0.622 0.731 0.880

alloys at 1625°K

AHN~ AHpt (Cal/g-atom)

18,

1970

calculated

ASNI ASPt (Cal/g-atom-deg)

from vapor-pressure AG (callg-at::)

data AS (Cal/g-atom-deg)

- 14,282 - 10,646 -8037

-260 -446 -584

+2 $2841 +2525

+0.1s 1-0.46 f0.91

$8.79 $8.30 +6.50

-1925 - 3084 - 3852

-235 +215 +357

1.04 2.03 2.59

--4704 3308

-5722 -4008

+137 -447

+2346 +9os

+2.12 +2.62

+4.0s +3.91

-4274 -4356

+438 +95s

2.90 3.27

-6563 -8322 -9943 - 13,353

-2486 -1533 -1012 -413

+1363 -2403 $7883 +6569

+ +5.96 2.56 + 10.97 + 12.26

14.16 +2.54 +0.93 +0.33

-4117 -3570 -2799 -1707

+1603 +2231 +1978 +763

3.52 3.57 2.94 1.52

TABLE 3. Heats

Solute

of solution

Experimental heat of solution at 298°K

+4274 +;I;; + 123

of pure nickel and platinum expressed in Cal/g-atom

in liquid-tin.

The data are

Heat content of solute from 298 to 914°K

Calculated heat of solution at 914°K

Heat of solution at indicated temperature

Ni

-7989

*

10

4557

- 12,b46

-7470 (273’K) Leach and Beverce” -9920 (910°K) Day and Hultgren’az’ -9464 (913°K) Oriani and Murphy’23’

Pt

-24,734

*

3b

4051

-28,785

-26.951 (900°K) Geiken’a@’ - 24,800 (914°K) Oriani and Murphy”O’

TABLE 4. Heat of solution and heat of formation values of Ni-Pt alloys at 298°K

NNi

VOL.

(c*lgtom)

AHzws (Cal/g-atom)

0.9

-8988 -9023 -9021 -9014 - 9003 -9028

-676 -641 -643 -650 -661 -636

0.8

- 10,007 -9988

-1331 -1350

0.7

-11,127 -11,122 -11,174 -11,158

-1885 -1890 - 1838 - 1854

0.6

- 12,519 - 12,567 - 12,566 -12,522

-2168 -2130 -2121 -2165

0.5

-14,144 -14,154 -14,145 -14,148

-2218 -2208 -2217 -2214

0.4

-lb,872 - lb,877 - 15,883

-2164 -2159 -2153

0.2

-20,129 -20,138 -20,165 -20,211 -20,157

-1256 -1247 -1220 -1174 - 1228

0.1

-22,285 -22,313 -22,297

-775 -747 -763

are summarized in Table 5. The entropy values at 1625°K were calculated by means of the GibbsHelmholtz relationship from the tabulated free energies at 1625°K and from the heat of formation values at 298°K by assuming AC, = 0 between the two temperatures. The entropy data listed in Table 5 are preferred to those presented in Table 2, since large uncertainties are involved in deriving entropy values from the temperature coefficient of the free energies. 4. DISCUSSION

The activities at 1625°K for the Ni-Pt system (Fig. 3) are in good qualitative a&reement with the activities determined by Schwerdtfeger and Muan,(24) over the temperature range 1273-1473”K, and with the activities obtained by Alcock and Kubik’l’) from 1843-1905°K. The activities of nickel and platinum exhibit negative deviations from ideality and are consistent with the formation of superlattices in the Ni-Pt system at low temperatures. The integral free energies, heats of formation (calculated from calorimetric measurements), and entropies of formation, obtained in this investigation, are presented in Fig. 4. The integral heats of formation for the Ni-Pt system are exothermic over the entire range of composition, in contrast to the endothermic values obtained for the nickel-rich alloys in the Ni-Pd

WALKER

AND DARBY,

THERMODYNAMIC

JR.:

PROPERTIES

OF

Ni-Pt

1265

ALLOYS

TABLE 5. Summary of integral thermodynamic quantities for the Ni-Pt alloys

NNI

(oa$z’atom)

(oa$!tom)

ASS (Cal/g-atom-deg)

A@* (Cal/g-atom)

ASxd$ (Cal/g-atom-deg)

:.: 0:7

--3084 1925 -3852

-1340 -651 -1867

0.78 1.07 1.22

- -869 1463 -1877

0.6 0.5 :.:

-4274 -4356 -3570 -4117

-2146 -2214 -2159

1.31 1.32 1.20

-2102 -2121 -1936

+0.13 +O.OS +0.01 -0.03 -0.06 -0.14

0:2 0.1

-2799 -1707

- 1225 -762

0.97 0.58

-1186 -655

-0.02 -0.07

* From present work at 1625°K. t From present work at 298°K. $ From AB at 1625”K, and AH at 1625°K (assuming AC, = 0 between 298 and 1625°K).

system.(l) Similar differences between the Fe-Pt and Fe-Pd systems were attributed(24*25) to the greater

transformation

in the nickel-rich

tendency

rections

be small,

for clustering

the formation

in the palladium

of ordered

structures

alloys or to

in the platinum

alloys.

contribution

from

the integral

at 1625°K were

free energy

1625°K and the heat of formation

values

data at 298°K.

would

formation only

The integral entropies of formation calculated

the

that

arises

from

since

the

at the Curie temperature of pure nickel is The calculated en20 cal/g-atom.(26)

tropies of formation

(Fig. 4) are very close to the ideal

entropy

at 1625”K,

In

excess entropies of formation The

of mixing interpretation

of

and,

entropies

configurational

determined

suggests that the assumption may not be valid. However, the errors in the heats of formation calmeasurements

to the heats of formation

The magnetic

magnetic

so that

+ AS vib+ AS,,,f.

contribution

is and (2)

to the entropy

AS,,,

are so

great that the change in sign may not be real. corrections

contributions,

AS = AS,,,

the

of formation

the algebraic sign of the heats of formation

in this investigation

therefore,

are small.

generally made in terms of vibrational,

culated from the vapor-pressure

cor-

of trans-

140 f

Kopp behavior was assumed, although the disparity in employed

The

the heat

at

the absence of adequate specific heat data, Neumann-

by the two techniques

magnetic

alloys.

No

were made for

0.6

0

0.2

0.4

0.6

0.8

I.0

NNI

FIQ. 3. The activities of nickel and platinum at 1625’K. Open circles are experimental nickel activities, closed circles are the calculated platinum activities, closed triangles are experimental nickel activities reported by Schwerdtfeger and Muante*) (1273-1473”K), crosses are experimental nickel activities of Alcock and Kubik”‘) (1843-1905°K).

-6m1 0.0





a2



1

0.4



1

0.6





0.8



1

1.0

NNi

FIQ. 4. Integral free energies, heats of formation and entropies of formation of Ni-Pt alloys at 1625°K. Broken curve represents ideal entropy of formation at 1625OK.

is

ACTA

1266

METALLURGICA,

equal to $ ACgmas dT/T. In the absence of specificheat data for the Ni-Pt system, the magnetic contribution must be estimated and is assumed to be small for the Ni-Pt ment.

Weiss

and

magnetic entropy the form AS,,,

alloys,

from the following

Tauer(s’)

have

of formation

shown

argu-

that

the

of a binary alloy is of

N, ln (Pi + 1) -

NB ln (rug +

111 (3)

where p, and ,uB are the moments of the pure elements, ,L@‘Y and &uor are the moments of the elements in the alloy, and N, and NB are the mole fractions of the component elements. The rate of change of magnetic platinum dilution atom, solute

with

moment

nickel,@)

of a constant

per atom, upon alloying

is consistent

magnetic

with

moment

simple

sumption,

the contribution of

equation

the

of AS,,

Ni-Pt

to the entropy of

alloys,

estimated

(3), is small and has a maximum

approximately

0.05 Cal/g-atom-deg

40 at.% Ni.

The

changes

entropy

upon alloying

are determined

from Neumann-Kopp behavior. specific-heat data, the following

from

value of

in an alloy

taining

in

con-

vibrational

by deviations

In the absence of assumptions can be

made : (a) assume no deviations from Neumann-Kopp behavior,

thus

formation

become

AS,,

= 0;

temperature,

as suggested

obtained

the two

by

investigation, Ni-Pt

system

if the heats

of

with an increase

in

by comparing

the values

employed

in this

would be positive.

to form at low

(b)

techniques

then AS,,

The tendency

or

less negative

ordered

structures

temperatures

exists

in the to some

extent at higher temperatures, as indicated by the negative deviations in the activities from Raoult’s Law. AS~~~~ will be less than ideal by an amount depending upon chemical ordering amall quantity equal

to zero,

the extent of in the system.

high-temperature Since AS,, is a

and if AS,,, is assumed approximately the entropy

of formation

is almost

exclusively

determined by ASooni [equation (2)]. Then ASconi would be very close to ideal and, hence, the tendency to order chemically is small. Conversely, if ASvib is positive, ASzzefBB would be negative to

yield the integral entropy close to ASideal. It is not possible to distinguish between the two alternatives at this time in view of the inaccuracy associated with the high-temperature heats of formation obtained from the vapor pressure measurements. In conclusion, the heats of formation

are

for the Ni-Pt

1970

exothermic

composition,

over

the

entire

range

of

which is in contrast to the results for the

Ni-Pd system. The contrast in behavior also is observed between the Fe-Pt and Fe-Pa systems and probably

reflects the ordering tendency

systems containing

platinum.

tion of the Ni-Pt

system,

which

are close to ideal,

mainly

from contlgura-

The source of the difference

the thermodynamic cannot

that exists in

The entropies of forma-

to be derived

tional contributions.

quantities

be discerned

until

for Ni-Pt

appropriate

in

and Ni-Pd specific-heat

data are available. ACKNOWLEDGMENTS

The authors wish to thank Mr. R. E. von Massow for helpful discussions and Mr. A. P. Paulikas and Mr. S. D. Smith for their assistance with the measurements.

per nickel

i.e. essentially no magnetic moment on the platinum atoms. With this simplified as-

formation

system

18,

are considered

= R[ln (,uyl”r + 1) + In (,u$‘~Y + 1) -

VOL.

REFERENCES (1965). 1. L. R. BIDWELL and R. SPEISER, Acta Met. l&61 2. M. HANSEN and K. ANDBRKO, Constitution of Binary Alloys, 2nd edition. McGraw-Hill (1958). Report ANL-6657, Argonne 3. K. M. MYLES, USAEC National

Laboratory

(1963).

4. K. M. MYLES. Acta Met. 13. 109 119651. Am. Inst. Min. 5. A. T. ALDREG and K. M.-%YLEs: T&s. Engrs 250, 736 (1964). 6. R. D. FREEMAN and A. W. SEARCH, J. them. Phys. 22, 762 (1954). 7. K. M. MOLES, Trans. Am. Inst. Min. Engrs 242, 1523 (1968).

8. K. M. MYLES and J. B. DARBY, JR., Acta Met. 16, 485 (1968).

9. J. B. DARBY, JR., R. KLEB and 0. J. KLEPPA, Rev. scient. Instrum. 87, 164 (1966). 10. R. HULTOREN, R. L. ORR, P. D. ANDERSON and K. K. KELLEY, Selected Values of Thermodynamic Proper&z of Metals and Alloys. John Wiley (1963). 11. A. N. NESMEYANOV, Vapour Pressure of the Elements. Academic

Press (1963).

12. H. A. JONES, I. LAN~MVIR and G. M. J. MACKAY, Phya. Rev. 30,201 (1927). 13. G. P. KOVTVN, A. A. KRU~LICH and V. S. PAVLOV, Ukr. fiz. Zh. 7(4), 436 (1962) 14. H. L. JOHNSTON and A. L. MARSHALL, J. Am. them. Sot. 02, 1382 (1940). G. BRYCE, J. them. Sot. 2, 1517 (1936). ::: J. P. MORRIS, G. R. ZELLARS, S. L. PAYNE and R. L. KIPP, U.S. Bureau of Mines, Report of Investigation No. 5364 (1957). 17. C. B. ALCOCE and A. KUBIK, Trans. Inst. Min. Metallurgy 77[C](745), C220-C224 (1968). 18. R. W. CURRY and L. S. DARKEN, Physical Chemistry of Metals. McGraw-Hill (1953). 19. R. A. ORIANI and W. K. MURPHY, Acta Met. 10, 879

(1962). University 20. G. W. GEIKEN, USAEC Report UCRL-17615, of California (1967). 21. J. S. LL. LEACH and M. B. BEVER, Trans. Am. Inst. Min. Engrs 215, 728 (1959). 22. G. F. DAY and R. HULTOREN, J. phys. Chem. 66, 1532 (1962). R. A. ORIANI and W. K. MURPHY, Acta Met. 8,23 (1960). ii: K. SCHWERDTFE~ER and A. MUAN, Acta Met. 13, 509 (1965).

25. C. B. ALCOCK end A. KUBIK, Acta Met. 1’7, 437 (1969). 26. 0. KUBASCHEWSKI and E. LL. EVANS, Metallurgical Thermochemistry. Pergamon Press (1958). 27. R. J. WEISS and K. J. TAUER, Phys. Rev. 102,149O (1956). 28. J. CRANQLE and D. PARSONS, Proc. R. S’oc. A255, 609 (1960).