Enthalpies of formation of binary phases in the systems FeTi, FeZr, CoTi, CoZr, NiTi, and NiZr, by direct reaction calorimetry

CALPHAD Printed

Vo1.7, in

the

No.1,

pp.

l-12,

1983

036~-59~6f83~0~000~-~~~03.00~~ Cc) 1983 Fergamon Press Ltd.

USA.

ENTHALPIES

OF FORMATION OF BINARY PHASES IN THE SYSTEMS FeTi, FeZr, CoTi, CoZr, NiTi, and NiZr, BY DIRECT REACTION CALORIMETRY

J. C. Gachon Laboratoire Universitd F 54506 This

ABSTRACT

paper

was

and J. Hertz

de Thermodynamique Metallurgique de Nancy 1, Faculte des Sciences Boite Postate 239 VANDCEUVRE LES NANCY CEDEX

presented

at

the

CALPHAD

X (Vienna

1981).

The enthalpies of formation of the binary compounds in the systems AB where A is taken in the group VIII and B in the group IV A are measured by direct reaction calorimetry at high temperatures. Checks are made to ensure that the stoichiometry and the structure of each phase is obtained,

Introduction The binary compounds of transition metals are of great interest both for the theoretical study of Unfortunately their thermodynamic knowledge is the solid state and for many industrial purposes. far from good and only a few experimental data are tabulated. In order to provide an experimental data base covering the transition metal alloys usable for diagram computation and for solid state theoretical study we decided to measure the enthaipy of formation of the binary compounds AxBl-x with A taken in the group VIII and B in the group IV A. Such a selection is directed towards a theoretical study but in fact the method we have used is suitable for many other binary (or ternary) compounds. we shall summarize the results In this paper, after giving a survey of the experimental method, concerning the systems FeTi, FeZr, CoTi, CoZr, NiTi, and NiZr. The whole set of data is examined in view of characterising the accuracy of the experiments.

E%perimental Method The

calorimeter.

The calorimeter (figure 1) has been devised for measuring heat effects at temperatures ranging from 800 to 1800 K. it is heated in a vertical furnace (SETARAM 2400) specially built for the calorimetric applications. The support is made from a massive piece of alumina sawed off a tube (outside diameter 20 mm, cell is located in the cylindrical part (height inside 15 mm) as shown in figure 1. The calorimetric 9 cm) which warrants the symmetry of the thermal exchanges. The above part is a half cylinder with two alumina two bore tubes cemented on its edges to drive the output wires up to the top of The dissymmetry of this part does not alter the measurements as shown by the support. comparative experiments made with symmetrical devices. from 17 (PtRh6%, PtRhSO%) thermocouples connected in series and The calorimetric cell is made supported by alumina two-bore tubes at two different levels around the working crucible. This disposition coupled with the length of the junctions (about 1 cm) ensures that if the part of the heat transfer between the sample and the caforimeter passing through the thermocouples iS All the junctions of one type are maximal, ii) the effect of the filling of the crucible is minimized. around the working crucible while the junctions of the other type (all at the same level) surround the inert crucible. The integral of the calorimetric signal is proportional to the heat transfer A calibration provides the converting factor; we will between the sample and the calorimeter. examine this latter point in the second section.

Received

27 July

1982

1

J.C.

Gachon and J.

Hertz

3

ENTHALPIES OF FORMATION OF BINARY PHASES

The samples are introduced in the vacuum put under argon, are driven down in the cylinder of the support. The crucible temperature is given working crucible. The calorimetric recorded. Preparation

of the

by the Signal

lock of the calorimeter tight entrance crucible by an alumina tube taking of one e.m.f. is electronically

junction of amplified,

and, Place

the cell in integrated

after in

the

being half

touch with the and graphically

samples.

All the preparations were made in a glove box under pure argon much as possible. Table 1 gives the characteristics of the metals Characteristics METAL

SUPPLIER

Ti Zr Fe co Ni

Alpha Ventron Alpha Ventron Koch Light Alpha Ventron Alpha Ventron

TABLE 1 of the pure FINENESS 44 177 6-8 44 1

(*I in order we used.

to

avoid

oxidation

as

metals. (11)

PURITY

(mass

%)

99 99 99 99.9 99

The metal powders were introduced into the glove box just after being received from the suppliers and were stored in it . The powders (Fe, Co for example) which were supplied, under air, were put under vacuum for at least 48 hours before being introduced into the glove box in order to remove a quantity of powder was part of the adsorbed gases. In addition, before using it, the necessary pestled alone in a mortar in order to remove another part of the physisorbed oxygen. Then the right amount of the second component was added and the two powders were mixed together for ten that, the mixture was compacted with a hand powered press under a pressure of a minutes . After few hundred megapascals. The compact was broken into pieces of about 0.1 gram, weighed and then packed in individual boxes under pure argon. Then the boxes were themselves packed under pure argon to permit their transport up to the calorimeter. Measurements Each experiment included i) the calibration, ii) the measurements. The calibration is made with alumina cold samples dropped, at room temperature, into the working crucible kept at the temperature of the experiment To. Three samples were dropped before the measurements and three afterwards. For each alumina sample the signal was integrated. Using the heat content tabulated in (1) the sensitivity of the calorimeter was calculated.As the variation of the level in the crucible was not important, the sensitivity did not systematically vary during the experiment and thus the mean value of the six results was taken as the calibration coefficient of the calorimeter. Each sample of compacted mixed powders was removed from its box in air and immediatly introduced into the vacuum tight entrance lock of the calorimeter at room temperature, then put under argon and dropped into the crucible kept at the temperature T chosen just below the melting point of the compound under study. The rise of temperature pn the sample induced a reaction of alloying and so the calorimetric signal is the sum of the heat contents of the pure metals and of the enthalpy of formation of the compound. Figure 2 shows the two parts of the signal corresponding to a) the cooling of the crucible by the cold sample; b) the exothermic heat flow due to the formation of the compound. The reactions were achieved in general in less than two to five minutes. However, for a few phases with low melting temperatures, we got very slow reactions (figure 3) which remained incomplete. It was the case for CoO.33TiO.67 and CoO.7STiO2S. Taking into account the magnitudes of the two heat quantities involved in the reactions, the results of the measurements were often about zero. This leads us to notice two things: i) a zero result is very favorable as the accuracy of the calorimeter calibration has no effect on the accuracy of the result; ii) the determination of the enthalpy of formation of the compound depends heavily on the values of pure metals heat contents. We used the data tabulated in (1) and (2) but, as a proof, we measured heat contents of all the pure metals in our calorimeter, using the same procedure and products as for the alloy measurements. We will discuss in detail the results we got in section 3. Checking

of the

products.

After each experiment the products were checked by X-ray diffraction and by electron microprobe anlysis. The X-ray diffraction was used to compare the structure of the product with that given in (*)

Oxygen

pressure

about

lo-15

atm.

3.C.

4

nlv

gachon and

J.

Hertz

’ 5 I

0

1

15 t (mn) m I

10 I

b

coQ&7 0,138lg

Q5.

a I FIG.

3: Calorimetric occurs.

the literature and to search for oxide the product and allowed us to verify fourth section.

signal

when an incomplete

reaction

traces. The microprobe analysis gave us the stoichiometry of its homogeneity. We will discuss the results we got in the Experimental results TABLE

Heat Metal

Ti Zr Fe co Ni

Room Temperature (K) 295 297.5 295.5 296.5 296.5

content

2

of the pure

Crucible Temperature (K) 1492 1700 1500 1499 1499

metals Measured Heat content (J.mol-l) (Stand. dev.) 39800 (1400) 45029 ( 1000) 42400 (2000) 43624 (521) 37400 (1400)

Tabulated Heat content (J.mol-1) 39823 46246 46191 43632 39222

The results show that the tabulated and measured values are coherent exept in the case of Fe, but for this metal (and also for Ni at a lower level) the samples were badly compacted and so the loss of weight of the samples was not negligible (5% for Fe). The two main points are: i) noticeable oxidation is avoided, and ii) the coherence between the tabulated data and our measurements is satisfactory.

5

ENTHALPIES OF FORMATION OF BINARY PHASES

Enthalpies

of formation

Table 3 and figures and computed data.

of the compouds.

4 to 9 give all the results

we got and parallel

them

with

available

experimental

AHNmol-'-3000

-50 ooo*.

-60000

Fe

50 (at Ye r; 1 FIG.

4: System

FeTi,(

1

*our work; X(6)~7);+(10); 0 fi5);.~(163.

Discussion Experimental

results.

As the experimental results on FeTi, CoTi, NiTi, and CoZr systems have already been discussed elsewhere (3, 4) we shall just recall here the main points. In the system FeTi, Feg.67Tig.33 was found pure while Feg.SgTig.66 showed traces of Feg.67Tig.33 . Our values contradict Kubaschewski’s earlier work (5) and thus the computation of Kaufman (6), but are in agreement with the measurements of Dinsdafe et al (7). For the system CoTi only two phases led us to usable results: ~og.5gTig.5g and Cog,67Tig.33. Cog 67Tig.33 was found hexagonal as proposed by (8) and not cubic as quoted in (9). For the system N’Ti the equiatomic phase showed traces of its two neighbors but in very few quantities while the two other phases were found pure. The experimental results from (10) agree with ours except for the phase Nig.76Tig.25. As the checks we made do not show any trace of oxide or wrong phase we think our value is reliable. In the system FeZr only one phase was studied because of the uncertainty on the phase diagram. The result we got is not far from

J.CI

Gachon

and

J.

Hertz

-2oooc AH/Jml.-3OOOl

-4ooot

I I 50(oto. % Ti)

-60001

!

0

FIG.

5: System

CoTi.

( l our work; x(6);

1 o(l5);

&(16))

that of the earlier work of Schneider et al (111, although the techniques used in the two determinations were very different. The structure of the phase Feg.6TZrg.33 that we found is not the one quoted in (12), the X-ray pattern showed some lines similar to those of the Cl5 structure but with other lines that can neither appear in a Cl5 structure nor in a C36 like the one proposed in (13). These lines cannot be attributed to any oxide but are probably due to traces of neighboring phases and, as the microprobe analysis confirmed the homogeneity and stoichiometry of the sample, we take our enthatpic result into account, In the system CoZr four phases were studied and only one was obtained pure (Cog 66Zrg2g). The three others contained traces of unwanted neighboring phases but in so few quanti’ties that these wrong phases could hardly change the results of our In the system NiZr four phases have been studied. No experimental data were measurements. available to check our results; in addition the structures of the phases are not well estabished, so the X-ray checks of the products were difficult. The conclusion that could be retained is that the phases we wanted were obtained pure or with a few traces of their neighbors as already seen in other systems; so we think our results can be taken into account.

ENTHALPIES OF FORMATION OF BINARY PHASES

7

_20000 AH/J molf -3oooc

-4oooc

-5oooc

-6000C

II

Td

5O(ot.% FIG.

6: System

NiTi.

(

l our

work;X(6);+(10);

Tl o(15);

A (16))

Accuracy

of the experimental

results.

Our measurements can be affected by some different factors: i) a bad calibration, ii) non-accurate heat content, iii) loss of weight of the compacted samples, iv) oxidation. As we saw in table 2 the coherence between the tabulated heat contents and the measured ones is satisfactory and also The calibration method gives results with a spread of shows the absence of significant oxidation. of the about 3 to 5%, but as quoted above the heat quantities we measure are only a fraction enthalpy of formation, so the uncertainty of the calibration coefficient applies only to this fraction and finally the influence on the result is no more than 1 to 2% in the worst case. For the same reason the loss of weight of the samples has nearly no effect on the measurements, as we will show below: Let us call mg the mass of the sample when weighed, ml the mass of the sample when introduced

J.C.

Gachon and J.

Hertz

-30000

t

-40000

-5000 I -60000' FC

I. 5O(rJt.% Zf) FIG.

7: System

FeZr.

( l our work~~li);o(l5))

in the calorimeter, H2 the heat content of one mole of [2] temperatures and’AH the enthalpy of formation of M. The heat quantity measured in the calorimeter is: content (mOH~)(l/M). The enthalpy of formation of the

AH' = (m,H:

+ mlAH

-

mOH~).(l/mO)

AH' is the one we found. The difference (mO-m,)(l/mO)(AH+H:).

As the

product

Zr

the sample between room [l I and crucible one mole of the alloy having a molar mass (Hf + AH)(m,/M) and the calculated heat mole of alloy is then calculated:

using the

mass m0. The enthalpy

between

the two values is then:

of the

two

first

brackets

AH is the real value whiie

is generally

less than

0.1% and

the last bracket less than one third of AlI this difference is less than 0.3%. The most important error is then the one due to the spread of the results and it can be estimated to be about *6%, as we found testing the reproductibility of the measurements, The lack of earlier experimental results for many systems does not allow us to systematically check our results. The data already tabulated are all obtained by the same kind of technique originally used by Kubaschewski (S), except for the measured and the enthalpy of system FeZr for which the activities of the components were formation of the intermediate phase calculzted (11).

9

ENTHALPIES OF FORMATION OF BINARY PHASES

FIG. Comvarison

with

the computed

8: System CoZr. (

l

our work;v(l4);

0(15);&16))

data.

Three sets of data from different origins are quoted in Table 3. The first, here called CALPHAD results, is taken from computer calculation of phases diagrams (6, 14) and the values depend heavily on the former experimental or estimated data used as starting bases. This is the case for FeTi and CoZr where the experimental data.were undoubtedly far from the truth. The second set under study is that of Miedema et al (15) which use a semi empirical description involving the electronic density at the boundary of atomic cells in the pure metals, and electronegativity parameters. To compare their values with our own we have calculated for each compound the difference between the measured and computed enthalpies of formation and taken the mean value. We found: ~~~~~~~~~~~~~~ 2 +9700 J mol-1 and the standard deviation from this mean value is nearly the same: 8300 J.mol-1. These two results show that Mledema’s values are too exothermic in general but that the difference between experimental and computed data cannot be easily estimated. The third set is from Bennett and Watson (16) who used a theoretical model based on a rectangular Id’ band model originally proposed by Friedel (17). As above we have calculated the differences between each couple of results and we found: (AHexp-AHcomp

) = -3700

J mol-1

J-C.

10

Cachon

and

J.

Hertz

-20000 AH/Jmol-' -30000

-50

ooc

-60000‘. NI

FIG.

9: NiZr

system.

(eour

work;~(l5);A~l6)}

and the standard deviation from this value is 6100 J.mol -1. If the comparison of the results based on only five compounds has any meaning we can deduce from these figures that the Bennett and Watson model, when used for equiatomic alloys, gives accurate results but with a spread whose tendency is impossible to foresee.

This attempt to measure the enthalpy of formation of the binary compounds of transition metals shows the efficiency of our method of direct reaction at high temperature between powders of the to ternary compounds if needed. The main components. This procedure can also be applied and ii) the .absence of reliable structural results difficulties are i) the prevention of oxidation, about some of the compounds. From another polnt of view the comparison between the experimental If Mledema’s model gives only and theoretical data shows that the models become very accurate. semi-quantitative results the trends it indicates are in agreement with the experiments. The Bennett and Watson model is more accurate than Miedema’s and becomes really quantitative but is limited only to equiatomic alloys. We think that the comparison between the experimental and theoretical for theoreticians but also for approaches of the solid state study is now not only profitable experimentallsts.

: Throughout

NOTE

fee bee

discus.

see

cub. discus.

D8a

tetrag. Cl6 bee 82 observ. A2 fee Cl5

see

Temperature of reaction

-36800 -51500 39500 -32400

-29800

1596 1230 1405 1670 1479

-41000

-33000 -42200

-29700

-42900

-34000

-29300

34100

-44300

-27600

(1000) (2000) (500) (3000)

( 1500)

(1600)

(2000) (1000)

(1700)

(1000)

(2000)

(500)

(600)

(500)

(1000)

-30000 at 773

K ($)

-24700*3500 at 1023 K (11)

-268OOi2000 at 298 K (10) -33900~2000 at 298 K (10) -347OOi2000 at 298 K (10)

-20292§

“20292~2000 at 298 K (5) -278OOk.4000 et 1513 K (7) -25400~2000 at 1413 K (7)

-18520$

-29330$

-19020$ -26750$

-287949 -28125* -39957§ -37133f -370288 -37331*

-421545 -39246* -364105 -38167’

-25694s -23025*

CALPHAD results

Earlier experimental works

OF THE COMPOUNDS.

3

-31000 (1300)

Enthalpy of format&on (Standlev.)

1708

1290 1512

1760

1513

1475

1202

1432

1490

1514

1440

TABLE OF FORMATION

this table, units are J.mol-1, and K. 5 Kaufman’s states of reference for pure components (from reference 6). * Experimental states of reference, corrections with data from reference 18. $ Reference 14.

Nio33Zr0.67 Nio_5oZro,5o NiO.78Zr0.22 NiO.83ZrO.i 7

NiZr

Ni Zr

bee

Zr

Coo.67Zr0.33

Coo.80Zr0.20

fee

co

Coo.33Zr0.67 CoosoZro.50

cozr

Fe bee Zr bee

Feo.67Zro.33

I32

complex

C36

hexag.DO24

bee

Ti bee

NiozoTio.50

Nio.75Tio.25

fee

Ni fee

Laves

bee 82

Nio.33Tio.67

Ti bee

Coo.67Ti0.33

fee

co

Coo.5oTio.50

FeZr

NiTi

CoTi

Laves

Ti bee

Feo.67Tio.33

Cl4

bee (62)

Fe fee

Feo.5oTio.50

FeTi

Crystal structures Components Compounds

Phases

Systems

ENTHALPY

-39000

-53000 -73000

-37000

-57000

-44000 -60000

-34000

-43000

-58000

-45000

-42000

-47000

-25000

-28000

Miedema and Niessen (15)

-48300

-47900

-30800

-30700

-26800

Bennett and Watson (16)

g G:

5

z

%

z z

i

0

E a z VI

m z

J.C.

12

Gachon and J.

Hertz

A&now ledgments We are grateful

to Mr.

W. Courtney

for

his help

to

review

the

english

manuscript.

BIBLIOGRAPHY

1. 2.

1. BARIN, Substances. 1. BARIN,

0. KNACKE, Supplement. 0.

0. KUBASCHEWSKI, Thermochemical Properties of Springer Verlag, Berlin, Germany, 1977. Thermochemical Properties of Inorganic substances.

KNAcKE,

Springer

Verlag,

Berlin, 3. 2. 6. :: 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

Germany, 1973. J. c. GACHON, M. NOTIN, J. HERTZ, Thermochem. Acta, 1981, J. c. GACHON, M. DIRAND, J. HERTZ, J.L.C.M., 1981, 85, 1. 0. KIJBASCHEWSKI, w. A. DENCH, Acta Met., 1955, 3, 329.

Inorganic

L. A. P. A. 0. A. R. H. T. A. R.

48, 155.

KAUFMAN, H. NESOR, Calphad, 1978, 2, 55, and 2, 81. T. DINSDALE, F. H. PUTLAND, T. G. CHART, private communication, 1981. DUWEZ, J. L. TAYLOR, Metal Trans., 1950, 188, 1173. E. DWIGHT, Trans. Am. Sot. Met., 1961, 53, 479. KUBASCHEWSKI, Trans. Faraday Sot., 1958, 54, 814. SCHNEIDER, H. KLOTZ, J. STENDEL, G. STRAUSS, Pure Applied Chem., 1961, P. ELLIOTT, W. ROSTOKER, Trans. Am. Sot. Met., 1958, 50, 617. J. WALLBAUM, Z. Krlst., 1941, 103, 391. G. CHART, F. H. PUTLAND, Calphad, 1979, 3, 9. R. MIEDEMA, A, K. NIESSEN, private communication, 1981. E. WATSON, L- I+. BENNETT, Calphad, 1981, 5, 25. J- FRIEDEL, in The Wsics of Metals, edited by J. M. Ziman. Cambridge University Cambridge, England, 1969. L. KAUFMAN, H. BERNSTEIN, Computer Calculation of Phase Diagrams. Academic New York, USA, 1970.

2, 13.

Press, Press,