7AD-AI43612 INFLUENCE OF CONTAINER INTERACTIONS ON THE VAPORIZATION /TH ERMOO NAMICS OF NICKE(U) MATENIALS RESEARCH LABS

ASCOT VALE (AUSTRALIA) F MANT MAY 83 MRL-884

UNCiASSIFIEO FOG1/ N L

00000000ND

I .0 ~28 ~

IIi.25 f .4 1.6

~ 4.L~ N 1E~I

~1

MRL-R-884 AR-003-295

AUST(Al 1A

DEPARTMENT OF DEFENCEDEFENCE SCIENCE AND TECHNOLOGY ORGANISATION

MATERIALS RESEARCH LABORATORIES

MELBOURNE, VICTORIA

REPORT

N MRL-R-884

C)

INFLUENCE OF CONTAINER INTERACTIONS ON THE

VAPORIZATION THERMODYNAMICS OF NICKEL

Peter L. Mart

CL-

Approved for Public Release

c-Cb '

84cMA, 19831

DEPARTMENT OF DEFENCE

MATERIALS RESEARCH LABORATORIES

REPORT

MRL-R-884

INFLUENCE OF CONTAINER INTERACTIONS ON THE

VAPORIZATION THERMODYNAMICS OF NICKEL

Peter L. Mart

ABSTRACT

This report describes studies to determine the nature of the

interactions between nickel and various refractory containers encountered inhiqh-temperature Knudsen cell vaporization systems. Riqh-temperature mass

spectrometric measurements, in conjunction with metalloqraphic studies, wereused to determine how these interactions influenced the vaporizationthermodynamics of nickel. such information is essential to permit thedetermination of valid hiqh-temperature thermodynamic activity data in multi-

component alloy systems.

The studies revealed that graphite is an unsuitable container formolten nickel. Due to a substantial solubility of qraphite in nickel, underthese circumstances the nickel activity and its enthalpv of vaporization aresignificantly reduced in comparison to pure nickel. Recrystallized aluminaand stabilized zirconia refractories did not react with molten nickel, butonly zirconia was stable towards the molybdenum and tungsten Knudsen celljackets.

Recommendations are made for suitable containers for studies of thethermodynamic properties of nickel-aluminium and nickel-aluminium-platinumalloys. These alloy systems are important because they constitute thealuminide protective coatings applied to hot end components of military gasturbines.

Approved for Public Release

POSTAL ADDRESS: Director, Materials Research LaboratoriesP.O. Box 50, Ascot Vale, Victoria 3032, Australia

SECURITY CLASSIFICATION OF THIS PAGE ONCLASSIF'IED

DOCUMENT CONTROL DATA SHEET

REPORT NO. AR NO. REPORT SECURITY CLASSIFICATION

MRL-R-884 AP-003-295 UNCLASSIFIED

TITLEINFLUENCE OF CONTAINER INTFRACTIONS ON THE

VAPORIZATION THERMODYNAMICS OF NICKEL

AUTHOR(S) CORPORATE AUTHORMaterials Research Laboratories

Peter L. MART P.O. P8x 50,Ascot Vale, Victoria 3032

REPORT DATE TASK NO. SPONSOR

MARCH 1983 DST 82/1 29 DSTO

CLASSIFICATION/LIMITATION REVIEW DATE CLASS IFICATION/RELEASE AUTHORITY

Superintendent, MRL

Metallurgy Division

SECONDARY DISTRIBUTION

Approved for Public Release

ANNOUNCEMENTAnnouncement of this report is unlimited

KEYWORDS

Nickel Thermodynamics EffusionVaporizing Enthalpy Refractories

Mass Spectroscopy MoltenKnudsen cell

COSATI GROUPS 0702

ABSTRACT

This report describes studies to determine the nature of theinteractions between nickel and various refractory containers encountered inhigh-temperature Knudsen cell vaporization systems. Hiqh-temprature massspectrometric measurements, in conjunction with metalloqraphic studies, wereused to determine how these interactions influenced the vaporizationthermodynamics of nickel. Such information is essential to permit thedetermination of valid high-temperature thermodynamic activity lata in mullti-component alloy systems.

The studies revealed that graphite is an unquitable container formolten nickel. Due to a substantial soluhility of graphite in ni-kel, underthese circumstances the nickel activity and its enthalpy of vapnrizatin aresignificantly reduced in comparison to pure nickel. PecrvstflliPd ilumini an!

stabilized zirconia refractories did not react with molten nijckol, h12 T 12\zirconia was stable towards the molybdenum and tunqsten KnuAson -P]l aket .

Recommendations are made for suitablp containprq frr tjh.-; ,tthermodynamic properties of nickel-aluminium and nickel-a mirim-r,.a r'

alloys. These alloy systems are important because thev rnltnt, th.,

aluminide protective coatings applied to hot end cnmp cn nt. r' mI I v I I

turbines.

SECLURITY CLASSIFICATION OF THIS 116AJ

INCLASSTFI Fr

CONTENTS

Pace No.

1. INTRODUCTION 1

2. EXPERIMENTAL CONSIDERATIONS 4

2.1 Apparatus 4

2.2 Quad-Cell Selection 4

2.2.1 Refractory Metal Quad-Cells 4

2.2.2 Graphite Quad-Cells 5

2.2.3 Ceramic Crucibles and Graphite Quad-Cells 5

2.2.4 Boron Nitride Crucibles and Graphite Quad-Cells 6

2.2.5 Ceramic Crucibles and Refractory Metal Quad-Cells 6

2. 2.6 Summary 6

2.3 Materials 6

2.4 Temperature Measurement 7

2.5 Mass Spectrometry 7

2.6 Thermodynamic Calculations 8

2.7 Gold Calibration 9

3. RESULTS 10

3.1 Container Interactions 10

3.1.1 Graphite Quad-Cells 10

3.1.2 Graphite Cells and Alumina Crucibles 10

3.1.3 Graphite Cells and Boron Nitride, Zirconia Crucibles 11

3.1.4 Refractory Metal Quad-Cells and Ceramic Crucibles 12

3.2 Gold Calibration Accesi - 12

3.3 Nickel Mass Spectrometry NITIS c" I 13

DTIC T'.T3.3.1 Appearance Potentials unai ]o:: 13

3.3.2 Nickel Vaporization Justil 14

By-

Av " s

Dist p

I ... . . 1 .i

[a

CON T EN T S

Page No.

4. DISCUSSION 16

4.1 Container Interactions and Vaporization Thermodynamics 16

4.1.1 Graphite Crucibles and Quad-Cells 16

4.1.2 Alumina and Zirconia Crucibles and Cells 16

4.2 Errors in Enthalpies of Vaporization 17

4.3 Activity Measurements in Multicomponent Nickel Alloys 18

5. CONCLUSIONS 20

6. ACKNOWLEDGEMENT 20

7. REFERENCES 21

I,,

INFLUENCE OF CONTAINER INTERACTIONS ON THE

VAPORIZATION THERMODYNAMICS OF NICKEL

1. INTRODUCTION

Basic thermodynamic properties of high-temperature alloys are offundamental importance to our understanding of their behaviour underoperational conditions. Because of the lack of reliable thermodynamic data,the development of structural superalloys and protective coatings used in thefabrication of components in gas turbine engines has proceeded somewhatempirically, and there is little doubt that the development has beeninefficient.

One of the Tasks of High Temperature Properties (HTP) Group, TaskNo. DST 82/129 entitled "High Temperature Properties of Metals", has as itsobjectives:

(1) to carry out basic research into the thermodynamic, kinetic andmechanistic aspects of high-temperature degradative phenomena suchas oxidation and hot corrosion in metals and alloys of current andpotential relevance to the Australian Defence Force; and

(2) to investigate and develop methods whereby the high-temperaturecorrosion resistance of such metals and alloys can be enhanced.

To facilitate the study of high-temperature thermodynamics, arotatable, four-compartment Knudsen cell (quad-cell) has been developed at MRLas an effusion source for the mass spectrometric study of high temperaturesystems. In a previous report [1) of the thermodynamic properties of thenickel-aluminium system, using the quad-cell mass spectrometry technique,difficulties were reported with the containment of liquid aluminium andaluminium-rich alloys. This prevented the determination of aluminiumactivities, but nickel activities were reported for the high nickelcompositions with nickel atom fractions, xNi, greater than 0.54.

' 4.

The present study was undertaken as a prel:minary to thedetermination of the activities of both nickel and3 aluminium over the complete

range of compositions.

It is well established [2,31 that vapour pressure measurements byKnudsen cell effusion are very susceptible to interactions between the sampleand the container. Where possible, the selection of a chemically inertcontainer is desirable, although a reactive container may be used if aprotective layer of reaction products is formed that provides kineticstability [4,51. A guide to the suitability of a particular container formetallic alloys is to examine its interaction with the separate components ofthe alloy. The absence of interaction with the pure elements indicates thatreliable thermodynamic properties are likely to be obtainable for the alloysin the same container. Conversely, interaction with either or both of the

component elements is likely to prevent the collection of reliablethermodynamic data. If the nature and the extent of the inter.ction can bewell characterized, then it may be possible to correct the thermodynamic dataderived for the alloy, but such a procedure is best avoided if a non-interactive system can be found.

Accordingly, a study was made of the interaction of nickel andaluminium with a number of potentially suitable container materials. Thepresent paper primarily reports on the interaction with nickel; results of theinvestigations into the interaction with aluminium, and the detailed study ofthe nickel-aluminium alloys, will be reported in a future paper.

There are numerous literature reports on the vaporizationthermodynamics of nickel by a variety of techniques, including Knudsen masseffusion, Knudsen effusion mass spectrometry, Langmuir vaporization (both massloss and mass spectrometry), and gas transport. The earlier work has beenreviewed by Hultgren et al. [61 and the value selected _or the enthalpy ofvaporization of nickel was AH9 = 430.1 ± 2.1 kJ mol . This value

29(calculated by the Third Law method) was based mainly on gas transportmeasurements. Hultqren et al. disregarded the high values obtained from massspectrometry because of "the low absolute accuracy of this method" (6]. Itshould be noted, however, that these were Lamgmuir vaporization massspectrometry measurements rathen than Knudsen effusion mass spectrometry.Hultqren et al. also considered that the Knudsen mass effusion results ofNesmeyanov and Man [7,8] should be rejected due to the orifice area dependenceof the nickel vapour pressures. However Man and Nesmeyanov [81 calculatednickel vapour pressures inconsistent with their observed mass losses. Thecorrect pressures (calculated in the present work from their original results)are approximately two orders of magnitude lower than those they reported, butstill about twenty times higher than the accepted literature values [6]. Therecalculated pressures produce a mean enthalpy of vaporization (Third Lawmethod) which is significantly higher than than calculated by the authors andalso reported by Hultgren et al. [6] in their review.

Rutner and Haury [14) also performed a statistical analysis ofliterature data on the vapour pressure of nickel, they too rejecting the(incorrect) data of Nesmeyanov and Man [7,81. They calculated by the ThirdLaw method a best value for the enthalpy of vaporization,

2

0oo-"

= 424.5 ± 2.6 kJ mol Using the enthalpy funct ons of Hultqren et al.[6? this yields the value AH 2 9 8 = 426.5 ± 2.6 kJ mol

Since non-identical sets of experimental data were analyzed by theabove reviewers, the two best values are not strictly comparable. However,they are in agreement within their respective error limits, and forconvenience have been averaged to yield the value AH° = 428.3 ± 4 kJ mol298

In Table 1 the Knudsen effusion data from these reviews aresummarized, together with subsequent literature data. Third Law enthalpies ofvaporization of nickel have been calculated from the data of Alcock and Kubik

[10], and recalculated from the data of Man and Nesmeyanov [8]. With theexception of the results of the latter authors [7,8], the remaining literatureKnudsen effusion data agree reasonably well with the above averaged value forthe enthalpy of vaporization of nickel. In particular, the agreement obtainedby Knudsen effusion mass spectrometry [12] is satisfactory. This indicatesthat, with suitable containment, it should be possible to obtain reliablethermodynamic data for nickel using this technique.

TABLF 1. Knudsen Effusion Measurements of the Enthalpy of Vaporizationof Nickel

Sample/Cell AH0

298 -1Material Method Temperature (kJ mol Ref.

Ni/Mo Mass loss 1400-1527K 341.1 ± 0.8 7

Ni/Mo Mass loss 1320-1550K *393.1 ± 4.2 8

Ni/Alumina Mass loss 1463-1628K 430.0 9

Ni/Alumina Mass loss 1835-1882K *429.5 ± 0.3 10

Ni/Alumina Mass loss 1723-1873K ---- 11

and

Torsion

NiO and Ni/ Mass 1575-1709K **425.8 ± 2.0 12

Alumina spectrometric

Ni 63/Unknown Mass loss 1440-1600K 421.7 ± 0.3 13

* calculated in this work ** recalculated value [14]

3

2. EXPERIMENTAL CONS IDFRATfIlNS

2.1 Apparatus

In the present study a multiple Knudsen-cell mass spectrometrytechnique was used. In this a rotatable four-compartment Knulsen cell (quad-cell), developed at MRL, was the effusion source for a time-of-tliqht mass

spectrometer [15,16].

The schematic diagram of the quad-cell is shown in Fig. la. The

geometry of the high vacuum chamber, and of the pre-machined refractory metalquad-cells, limited the size of the crucible liners that could be used, which,

in turn, limited the ratio of the evaporating surface area to effusion holearea. The internal diameter of the cell compartment was 8 mm, the internaldiameter of cell liners was typically 5 mm, and effusion orific& Tiameter was

1 mm. This lead to ratios of the evaporating surface area (taken as

ueoometrical cross sectional area of the cell) to effusion orifice area of 64

for the cell alone, and 25 for the cell containing a crucible. While theformer figure is similar to that used by Ward [2], the latter ficgure is lowerthan is desirable (by a factor of four). However, it is qenerally aceptedthat the value for the coefficient of vaporization of nickel is close to unity

-], so that equilibrium conditions should have been maintained withi. thel>-ll even at a lower ratio.

As a result of the initial experiments, a second desiqin of iuad--]was manufactured, as shown in Fig. lb. In this desi(n the effusion orifice(1 mm diameter) was ultrasonically drilled in the ceramic liner (1.5 mm wallthickness), and the hole in the quad-cell was countersunk such that none ofthe effusing beam could contact the quad-cell. A ceramic lid was lapped and

polished to ensure a good seal to the top of the ceramic line., thusr r, ,-ntin,; escape of vapour within the quad-cell.

The choice of eIectron bombardment hfatinq of the Knudsen cc" tf)attain the higTh temperatures required (up to 200flK) necessitated that th-relIs ,were -7ade of, -r surrounded by, electrically conductive, materals. Thi

limited the choice of cell materials to the more commonly available r-friat:,)rvmetals (tunosten, molybdenum and tantalum), as we ll as high density

graphite. Refractory cell liners of recrystallized alumina and cal-ia-stabilized zirconia -ere available in the form of small crucibles, and w.roalso machined from boron nitride and graphite.

2.2 %_ad-Cpll Selection

2.2.1 Refractory Metal iad-Cel.s

Choice of refractory metal crucibles to contain liquid ni1k,, is

governed by the mutual solubility of the nickel and refractory metal.Molybdenum, tungsten and tantalum all show substantial solid solubility innickel [17,18], and are therefore unacceptable as crucibles and ouad--ellswithout liners.

4

The low values for the enthalpy of vaporization of nicko'l measuredby Nesmeyanov and Man [7,8] may be due to their use of molybdenum Knudsencells. Chatillon et al. [3] have shown that other 'parasitic' sources maycontribute to the molecular beam emitted from the Knudsen cell and sdmpled by

the mass spectrometer. These include climbing of the 'iquid sample up thecell walls, when the liquid wet-. the cell material, allowing free evaporation

to take place from the external !;orface surrounding the effusion hole. This

non-equilibrium evaporation can ca,:se the sampled flux to be much hiqher thanthat due to effusion alone, and may be partly responsible for the erroneously

high nickel vapour pressures calculated from the data of Nesmeyanov dad Man

[7,8]. Surface diffusion of the sample alonq the cell walls and through the

effusion hole can also contribute significantly to the evaporated material

[3].

Refractory metals may be suitable for use s -liad-cells if the

nickel is contained in an inert crucible liner, which will prevent the moltennickel from climbing the cell walls. The possibility still remains for

interaction between the nickel vapour and the exposed refractory m;tal surface

(for instance at the effusion hole). The solubility of nickel in molybdenum,

tungsten and tantalum is about one atomic percent at 1700K [17,181 and is

therefore likely to have little effect on the measured vapour pressure.

Notwithstanding the use of inert crucible liners, surface diffusion of nickel

may still contribute to the evaporation process and produce erroneously highnickel vapour pressures.

2.2.2 Graphite Quad-Cells

It is known that molten nickel can dissolve substantial amounts ofcarbon [171, which indicates that graphite crucibles are unlikely to be good

containers for nickel. The activity and solubi lity of carbon in Ni-C alloyshas been reported by numerous investigators over the temperature interval

1200-1500K, and these results are critically discussed by Bradley et al.

[19]. However, there are no reports on the activity of nicke in liquid

nickel saturated with carbon, and there is contention as to the stability ofthe Ni 3C phase which intersects this liquidus curve [17).

2.2.3 Ceramic Crucibles and Graphite )ad-Cell-

Either of the ceramics alumina or zirconia is suitable to contain

molten nickel since the free energy of formation of nick l oxide is

considerably less than that of the ceramic oxide [20]. However, the

interaction between the ceramic crucible and a graphite quad-cell is

considerable, as discussed by Jansson [211. At 1800K in vac'uum alumina willreact with carbon to form Al 4 C3 via the equation

2 Al 2 0 3 + 9 C --- Al4"1 -' r0c (,I) ()

with pCO = 427Pa. Oxycarbides may also be formed [22]. -;imilar thermodynamic

calculations using the data of Kubaschewski and Alcock [231 indicato that

--- - m I , i i . .. .i i n i - . .. . . . . . . . . . .. . .... . . . . .. TZ, .I Z3

zirconia will be, (-",n more lristahi - rrt f rmation, ird Inon-stoichiometrv of , i rcon al o -p rr-t f,,rm, oin of r- d(Iti r2_ w× i

evolution of CO.

2.2.4 Boron Nitride Crucibles an i 2Iraph, tfe 1.al-yel_

Boron nitride is of marg inal sui >tbi litv as a container for liguidnickel, since it decomposes inconqruentlv to solid boron and oaseous nitroopn

at these temperatures. From JANAF thermochemical tahles [24] the pressure of

nitroqen from this dissociation is quit, biclh (ahout 10-2pa) at 2000K, and theboron formed reacts with the nickel to form Nip ahov,- 1700K [251.

Furthermore, thermodynamic calculations based on the- thermodvnamic data ofBockris et al. [261 indicate that boron nitride should be Quite unstable withrespect to carbon, and react to form boron carbide and n trouen oas at

temperatures as low as 1000K.

2.2.5 Ceramic Crucibles and Refractory Metal Quad-Cells

The thermodynamics of the reaction of alumina and zir - ia with the

refractory metals molybdenum, tungsten and tantalum has been r £t -d

[27,281. Under a vacuum of 10-2 Pa, alumina is ecuallv stahl molybdenum

and tunqsten with no interaction occurrina below about 2200K, Pe tantalum

is fairly unstable and forms qaseous tantalum oxides. Zircon- s slightlv

more stable than alumina, with no interaction occurrinq below c 2300K for

molybdenum and tunqsten, and about 2100K for tantalum. The Ca _abilizina

additive in zirconia may react with refractory metals and reduce the reaction

resistance of the zirconia ceramic, but the volatility of the CaO ensures that

no reaction occurs with molybdenum and tunqsten below 230K [271.

2.2.6 Summary

From the above considerations, it was decided that refractory metalauad-cells used alone would be unsuitable for containinc liquid nickel. The

use of an inert crucible liner was essential, and the combination of alumina

or stabilized zirconia crucibles with molvhdenum or tuntisten ouad-clls seemed

worthy of investiqation.

In the case of qraphite quad-cells, it was decided to check, hv

experiment, the validity of the thermodynamic considerations, and the

possibility of achievinq kinetic stability with various crucible linermaterials. Accordinqly, crucibles fabricated from gra-hite, alumina,

stabilized zirconia and boron nitride were used as liners in araphite nuad-

cells.

2.3 Materials

Nickel shot of nominal purity 99.9 wt % for Ni was used throughout,

and was spectroscopically and chemicallv analyzed at MRL. It contained 0.06

6

i mpuri ties . C-,, poor 5 :oi'34 ~ ~,:,Iiiration piirz,'-<

-.. 4 Tempratur- 4's >

with ani Irccn ' e 0; .,*~ ,> ,~/'''~,.g, ~Northrup modle If2 xip'rn in 'a(Tt ' '' - ' '

1 6) , and vertica.l cn'i r r ,il' ''

were mi n imjseji, by ai i i n ut

i ndependlent> con irv % heti in o 1)n 'unitformi ty in tne- ryi,'Ao. ly ,, + ''

tePmp e r at ure (Ir a in ti , "rw rquad-cells, for n~er r~0 t1"" 1.- ~variation is C: 'an t

Bu r.ey F± 1 for, :n ' o1,!", .,5)i- uIrthe symme rv ' th- icO7,1r iar-

qraphite is Sbjr ;"nI ' '1tI

for the cired '--r ,;~ i th vi5' '

ccmpartsie nt Km ~ i '' ' ;eue "j',o n~' "

extrapolati:1on nsrs-: F1 ;j

correctedI fo r th- t i no:lr- -41 *o n

whe rf A' r T'~ m' 1 '""'"

resne~ opr 'F' IFi~''r~ V 1 t- .. . . .

Ipyr(orit-or -an mi anNP_ t,-)i 1''" '; f' Imp i irn' m :;>

C2=( r_ 1 for i hi-I i'nly'/

2.5 Mass Spectrometrv,

Ion currents "I(:-~ moni toted for the- N!i ct :'6 78.3 n. iabhund anceP) d ur ingq n ickelI va por i -a tro n e xper imentsq a:s for I~ <1 - 1 1 FI c'.d u rino gold ( v apot i' ,,, on - e xpe r im(-en to-- e tth o -i 'ri 1 1'r o tf theapparatus. ELectror: -n-r-ileips - 30 eV anI 410 eV were, i'e Ti ther n oIrCie

of the mass spectromener in the first two oxperimnf-~ r-''i ev hereif ter35 eV was usdn il I xperimonts. A reqiilart - tri;) n 1') "A W'0 'used throuuihu.

Ion curr'-tt e'';i-' - -rhokr ,i' i ' C

current measured when, the -t fii-4 i'. heaxm wa i ' 0v'I'. f r'' t- iosource. This, r-nmpoenatod ft ,piirio~is r-ev,j )rat', i r i,, tl' t -,hnieldsi.

and other hot surfaces a I lac'ent to t!- quad-cel I. Howver, it does notdistinquish possible ,,irface-::reep of material from the cell orifice; ashutter between the cell and the ion source would he required to achievethis. Such a shutter would increase the distance between the quad-cell andion source, reduc-Lnj the amount of effusate enterinq the ion source, and hencfreducinq the sensitivity of detoction by the mass spectrometer.

2.6 Thermodynamic Calculations

For vaporization of nickel from different crucibles and quad-cells,

and for calibration experiments using qold, the enthalpy of vaporization wasmeasured since this provides a clear indication of possible chanqes inthermodynamic activity of the nickel due to interaction with the container, orchanges in the vapour pressure due to temperature nonuniformity.

The second-law enthalpy of vaporization is derived from the

expression

d in K

AH - R (3)T 1(!

T

where K is the equilibrium constant for the vaporization reaction and T ispthe absolute temperature.

d in P.1Hence AH - R (4)T 1(

T

where Pi is the equilibrium vapour pressure of species i inside the Knudsen

cell. The relationship between this and the mass spectrometric ion

current I isL

P. = +k I T (5)1 i 1

where k. is a calibration factor depending on the relative ionization cross

section, the gain of the electron multiplier in the mass spectrometer, and thei otope abundance, for the particular isotope of species i being measured. Italso i ncludeq a geometry-dependent sensitivity factor which is independent ofspecies. The latter factor incorporates the Clausinq factor, Kc, whichdepends on the physical dimensions of the effusion orifice and corrects for

the reduced ion intensity due to collisions between the effusing molecules andthe orifice walls. While the calibration factor can be calculated, enablingdetermination of absolute pressures Pi' this is unnecessary in the second-law

method.

d In (I+ T)I

Thus AH - R +. ± constant (6)T d(-)

T+ 'AH is calculated from the slope of a plot of ln(I T) versus y, and as such

T T

this iero t. rt ir. *-rr in

spec - es.

me t hod

where the han I-- ii ho tre,, 7n- ;yf-ntin rf sde.L

T

Valuec- of f-f for- nickel in' u i ar-~ tahiata !- the literaturevalues of AH ar:- ca.cu' ie-i fron each experimental poirt, is.;n

aionacouK~nthe (iata co-ve~r an insifffici,-nt, -npera tire ranie for

accurate use of th- ecn- wmethod. Whi le ca lcu'.at: 0 n hoth the- soc= '_

and third-law me.thodis is; desirahle, in practic- it was- not: exprmertallconvenient to de, termin, the hsolt vanour -uress-ires tor each' cell ani eaer.

experimental tpmpoeaturp, as reuiired in th- third-law method.

the s-amve riuaad-e 1 in) the oae xperimet, where :o t:,nsrItni zn fa to'

the etfusi-n orifi ce' h-,ive been Ihat ITe Lf o- innr rrn wilvary from run to ru n,si up~in on chances in the a Ii inMrel 'f the efs'mlecuilar beam with the ioni souirce o)f the mAss sno-trnmete r, ani -hances; in

the efficiency of the ionization and .iotect-io qvs teom -f 'he ma;ssspectrometer.

2.7 Gold Ca',ibration

Before c-arryinq out quantitative measurements of nickelvaporization, a transmission calihration expf-riment WaS n~erformelA to deotermine

the relative sensitivity factors for the four effusion holes of the o'iad-cell[15] . Gold was selected as the reference material, and contained in qraphitecrucibles in a qraphite quad-cell (Fig. 1a), in accordance with accepted

procedure for gold standard vapour pressure me-asurements [29], since gold doesnot react with graphite.

-- --- -

Theo ;nd-Iaw ,.nthi Ipy ni vip:ori.,,ition of gold wan also eterminedfor each compartment )f tho omuad-c] I. .iice the Iirqs t circumferential

temperature gradients in the present system were obtai nPd wi th graphite ,iuad-celIls, comparison of the _:alculated Ali values with Ie dC :epted literaturo

value woulA enable d check on the reliaiity of the system under "worst-case"

condi ti ons.

3. RES rIL'P:

3.1 Container Interactions

3.1.1 Graphite _iai-Ce lls

Linuid nickel contained in qraphite crucibles at about 1770K reacted

with the graphite, and when cooled the bond was sufficiently strong that the

solidified nickel could not be removed without breakina the crucible. AI ; ,the surface of the nickel! lost its shiny metallic appearance, and bpcame dull

grey in colour. Very little nickel vapour was found to have condensed on thes-ell walls.

A cross-sectional view of such a crucible is shown in Fig. ?a, where

a reaction zone is visible in the underlying graphite. Figs. 2h and 2c of the

same specimen under parallel illumination show more clearly the erosion of thecrucible by the melt, the graphite precipitated from the melt, and graphite on

the surface of the melt. Chemical analysis of the nickel residues was not

considered feasible due to the difficulty of removing and analysing a very

small amount of material from the crucible. Electron probe microanalysiscould be carried out on a sectioned and polished sample, but the existing MN[facility is unable to guantitatively determine carbon composition. Comua'i.-:se

with pure nickel standards might enable an estimate of carbon compositio:. h%

difference, but was not attempted.

Attempts were made to measure the lattice parameter of the f.c.c.

nickel residue from melts in graphite crucibles, so that the carbon

concentration could be determined from the relationship between latticeparameter and carbon content [301. However, Debye-Scherrer X-ray powderdiffraction patterns using Cu-Ka radiation produced very diffuse high an.']eo

f.c.c. reflections which could not be indexed.

3.1.2 Graphite Cells and Alumina Crucibles

Several experiments were conducted with a graphite quad-cell in

which the walls of some cells were covered with a coating of alumina powder,

deposited from a slurry in methanol and then dried at about 420K. The

underside of the graphite lid was similarly coated so that no graphite wasdirectly exposed within the cell. Nickel was placed directly in the cell or

in alumina crucibles. A cell with exposed graphite walls and nickel in a

10

qraphi t- cruci olp ' "An rfeene At 15'73K .I fthe F 11, s ,(-n tl aro n i alIum ina i is pla yd i , '1r -':.'nder,t mass pt- or urr pe-,K "t M - 8correspond; in; c, m ,. t Jin'; t- rn the- or i f roe No) suhp~t. was obse-rvedfrom the aliumina-tre"" V7 . >:.mi larly, peaks cc~rrtespondinq to,CC) + Alot irii A! + worc, idn 1only for rho-r c(ortri, Iii no alumina.

22Furthermore, tht, e ' uticen w-<. ,ifti-cient ly 3re.,r that al I hoc'?iround peakswere i ncreaiped f )r the clls C'0:1t,11i 1 _IIIIg urn n, nTi c1,at inq that Knudsenconditions n.o lonrqter prevailod in! :iat no re li ableO thermodyn' amic data could4be obtained tr )n such cells. "'ell residues showed. siuns of reaction, with athin light blue coo nqn on, the nickel res:- Atues and the al imi na crucibles,irrespective of whether the, (-eIl walls we;ro al,.mina crated o)r left exposed.

In addition, there was a thic-k orey reaction :ave-r on the i-7). shot placeddirectly in the alumina-coated -7ellig 3). A fi n- m-ta 1 1c lacework hadqrown beneath tht- al m na-coated lids ofA two ccl :-; 4) and 5ZEM

examination rec> a odriicsrtue(Figj. 5I. FIner pdispersive X-rayanalysis (Figl. '-) showe-d the preseonce of nickel an( ,-:u-, and area, scans

and spot ana lys-s indlicated( -hat the aI imi ni ur wasi fairy ix' :tr-ly

distribuit>i throuohout the laco-work, andA was not the resul It of U sc'rete

particles o)f alic': '7 adh-ring- the ni ckpl. -I >vpro1 1, there was evidenice ofsubstant ial M- -ti 11 s" - w',11 t) 7, t'lln -,1'nc' I cimpartrenit a' t eMrpratlures

well below th.' e1 I'' ~ '-'l

In .contri-t, tnorp wa-; no s;ii of coa ctransport in the all-graphite cellI , a I th-iiuh t i "- *e -h ,i ''ci - -i~. PeferenceP to the Ni-Cphase diaciram ['7j !ho-w,; rho't t li ):ke 'Oiti mist becls to that of

the eutectic' -mom 0s1tinn~ -it ir"-i,'-i.( an] "tu1v, buit Ili- residlue was niotchemica I I ana L~

3.1.3 Graphite eVIp- Kj it-',Ct'ui"rcne

Boron nitride woo; fouind t,- o b nosuitable, As a -nntai ocr for nickelat temperatures above 1 900K, hc-th becauise the, bron ni t-iri adlheredl to the

gjraphi t-, and~ h.-(can -- 'ne Iro wa-3 qn i o 1 ant mon '0 ri -n decr-nops itioni

product detected( byv the mass spoctr-,mctor, whico l I -. r-r it) -' i

intensities and higih backg;round levels. A cross-soectionr -f the boron, ni tridiecrucible which was, broken away from t.L graphite c-l1 base i!; shown in Fi q. 6.

Some reaction with) the nic~kel melt is evident in thin rsso-'in

Zi r-oniai c-ricilln were also ulnsatisfactory as, confa ne-rs for nickel

in graphi te qtiai-cel Is eveni though there was no visiblo re act ion with the

nickel (ti(c. 7). Peac?-ion betweeni the zirconia and graphiteocurd

however, beas uhbstanti ~il ('.0 was evolve)] at temperatures above 1800OK, the

crucibles turned girey i nd itatl no 5nubstnoirchinotri( r( 2c X 7t (in-i the cuile

of ten develope-d cracks whic-h 01lorwe-i lePakagle of the molten nickel . Th ereaction between the ziri-onia andl graphitef: could be minimic-ed by inseorting a

molybdenum fo-il between them, allowing somt, mass spentrometric mlsrements tobe made.

3.1 .4 Re f racto ry Metal , ad-ce I Is and Cerar ni ( YLi hl.'s

Both alumina and zirconia wor, found to be s;; t,&lf- -airer, f .rmolton nickel in molybdenum quad-cells, with no visibl - roaction with thenickel (Fig. 8) or with the molybdunum interior of the '- . 1. Howe/.er, th,.cell design shown in Fig. la was unsati,7factory since the equilibrium vapoilrpressure of nickel produces condensation of nickel throughout thf cll, whichsubsequently reacts with the molybdenum, is transportei by surfatce creep intoand through the effusion orifice, which leads to the eventual blockage of the

orifice.

The cell design shown in Fig. lb was successful in overcoming theabove problems. However, when held for extended periods at temperatures above1800K, vaporization of the alumina occurred from the external surfacesurrounding the orifice, where the alumina was not covered by the molybdenumquad-cell. This resulted in a gradual shortening of the length of theeffusion hole, and a consequent change in the transmission properties of thecell. There was no substantial improvement by replacing the molybdenum quad-cell with one made from tungsten.

When comparative tests were conducted using alumina and calcia-stabilized zirconia crucibles in adjacent compartments of a tungsten quad-cell, the zirconia crucibles were significantly more resistant to vaporizationin the temperature range where alumina vaporized appreciably.

3.2 Gold Calibration

As described in Section 2.7, a transmission calibration experiment

was performed using gold to determine the relative sensitivity factors for thefour effusion holes of a graphite quad cell. The sensitivity factors were

found to all fall within the range 1.00 ± 0.05, and this held for most of teh

mtuad-cells used in this study.

+ 1A second-law plot of ln(I T) versus - is shown in Fig. 9 for one of

u T 0

the four effusion holes, and the calculated AH and AH enthalpies ofvaporization for all four effusion holes are s~'own in able 2. The enthaipy

functions of Hultqren et al. [6] were used. The close agreement between thefour compartments, and with the recommended literature value [291, AH 0

--1 298

367.02 _+ 0.88 k mol7 , indicates that reliable second-law thermo2ynam9c data

can be obtained from the system, provided that apprcpriate temperaturecorrections are applied.

A further gold calibration was carried out during the course of thenickel vaporization experiments, to verify that the enthalpies of vaporization

for nickel were free from substantial temperature errors. Attempts to carryout simultaneous measurements on nickel and gold in adjacent compartments of a

graphite quad-cell were unsuccessful due to the vapour pressure of gold beinglabout five times that of nickel in the temperature range 1800-2000K. Thiscaused rapid depletion of the gold and condensation problems within the

furnace. Therefore the nickel vaporization and calibration experiments werecarried out sequentially, with minimal disturbance to the system when the gold

12

was added to the quad-cell. The second-law plot is shown in Fiq. 10 and theenthalpy of vaporization is given in Table 2, in good agreement with theliterature value.

3.3 Nickel Mass Spectrometry

3.3.1 Appearance Potentials

Electron energies usually are chosen to be only a few volts abovethe appearance potential (A.P.) of the ion in question (A.P. for Ni + is 7.633ev (31]), to minimise possible fragmentation effects. In the present work,the ionization efficiency curve of 58 Ni+ for nickel in a graphite cell at1973K was quite complex at electron energies between 9 eV and 20 eV (Fig.11). A similar curve was obtained for nickel in an alumina cell at 1973K,indicating that the complex behaviour is probably not dictated by theinteraction of nickel with its container. At ambient temperature, the 40 A+ionization efficiency curve had the usual shape (Fig. 12) and yielded anappearance potential of 15.2 ± 0.5 eV, in good agreement with the literatureionization potential of 15.755 eV [31]. When the electron energy scale iscorrected by 0.5 eV, the linearly extrapolated appearance potential for Ni +

obtained from Fig. 11 is 7.3 ± 0.5 eV, in reasonable agreement with theliterature value.

The cause of the complex ionization efficiency curves is at presentunknown, and warrants further investigation. One possibility, that energeticions were formed in the vicinity of the cell by electron bombardment of theeffusing beam, was eliminated by appropriate experimentation. The nickel ioncurrent monitored with the cell at 1973K did not change appreciably when theelectron beam he~fting was switched off momentarily, indicatina that there waslittle contribution of energetic ions to the measured ion current.

However, for the purpose of the present study, it was possible toavoid the complex region by using electron energies of -35 eV, where the ioncurrent is fairly insensitive to slight changes in electron energy. Themeasured 58Ni+ signal may therefore contain some contribution fromfragmentation of nickel vapour species other than Ni(g). A survey of theliterature on gaseous positive nickel ions [31] indicates that NiO(g) andNi(CO) 4 (g) are the only two likely sources of fragment ions, and could arisefrom the relatively high carbon and oxygen impurity levels of the nickel(Section 2.2).

The appearance potential of Ni + formed by fragmentation ofNi(CO) 4 (g) is 16.0 ± 0.3 eV which is attributed [32] to the probable process

Ni (CO)4 (g) -+ Ni+ + 4 CO (q) (10)

The appearance potential of Ni + formed by fragmentation of NiO (q) can beestimated from the relationship

13

where IP (Ni + ) is the iofli-7.2 ,ic7 rootential 1-,f nre 7.61-1 n' -< N;

is the dissociation -enoroy f i .1 1 )2 + r-, ~ I~' a1'ia tfd y t titreorand Haury [ 14]1 from t1e Aia a .4' *,;r !1-v, Burns and ln~iri;, 1 2. The-calrulated amnparanc- ;)oten!tia I, V t N, -rr-qn-nS wr t- thg. .unseP

o f a n i nc rea sed N i + ior)n -- 1rrIe-n t ) or ved ( i n the pTrese-,n t- r riia t i ,n Pef fi c i P r-,/

curve (Fiq. 11). The Ni+ from Ni r'))t~"na~n.itIhr an arpearanrepotential which corresponds close 1 v wi th the if-xt 7or~st -1 Tv-r-,,isPd Ni+ inn

current. Further studies are nPcesszary (, ,i3 errnev- ' r vapr~n

species are actually presont and in what concentration -r- sioIn was observed

in the present work of Nio + or Ni(CO). + ions . winter-, andi Kispr '32] have-reported appearance potentials, and -ibundance is i fuinction o)f elecrtron -nprcov

for all the sinqly-chiarqTed Ni (CO) + o*n-s, with !, va-rir, r,) -niirn to, four.

From thermodynam-1c consisleration-, it is exoected thrth p-s -:l Tpssure r.fNiO (q) and Ni(CO) 4 ((1) above essential ly purr, Niun) Fshod hr verv -mall at2000K, and so should their correspord'nq fraqmentation co(-ntrib utions h

observed Ni +ion current.

3.3.2 Nickel Vaporization

Second-law plots )f Ioll T) versusi ar- shown Fn4,,-;. 1 J-!0 1C,

Experiments 1-6, the details- of which are orveon in -0-1- 3. The eoriplots were analyzed by the method of least !soiarres .rA th- co cii ateil

enthalpies of vaporization are also prresented in Tal) e l. The errrs 1 jv-n

for individual values of AH are 95* co-nfidence- l~i-, anl the- tH va 1,11es Ar;,

ascribed to( the mid-poi nt J~ the state I temperatuire rroce. The-iva'o

arac'cul- -td hV eaua t ion

Ir Fxnorimentn, 1-3, nickel win ce-n i nr)- -, '

us qc;riptii tof riuad--- I . -h.- -amro (-,-I1 wan. jisei in 'nr 'c tb'tt the relative trarsni ~ni on fac-tors wore tn t - 'i tr-!!1 Tn pel-

(-,a. 15) comrartmert R was zir(o--sli Iihrate-d hv h--i ino i k ntei-t

ruci hip for abouit three bouir, at 205 (1K with the lid -nr 'rornn

the name- time, co-mpartment- - ro~ntai ni no an ermrtv -rnit-hi- was tc-~~l

leIaned by leavino i tn lid off. in the stibsr-iient ...ns. sutc - ry run,fresh nickel waq added to) compartrent (7, while P wa-; left alo)ne after onsirno7tnat sufficipnt nickel remained in the cuil. Tlhe 1-a1lit'aed nhi In" o

vaporization of nickel from the nre-(eouilihraite. co.(mpartmen-T-t- was S iarAicanT1lxhiqh,-r than that fromn the compartynfot nntainin(T fresh Trajphile and nickel.71irthermore, th- ion current was aproxi mate v '.) tire at-eater rin!n theore

eo-Uilihrated compartment R. A seond qonld calihrat-i-t (isino co)mrartment '

was performed to vprify the accuracy oif the calcuilated nhaieas

described in Section 3.2, and these results oire shown in Vijo. 10 an-i Tabe '

In Experiment 4 (Firi. 16), nickel was -can' in an iliiminacrucible inside a molybdenum quadl-cell, with the effiision orifice defined, Lx,%

the alumina crucible as in Fin. lb. The- calculated orthalipy is- sionificantI'.

hiaher than that for nickel in cqraphi fe, both new an,! np-oriiihr irtod. flnl vion currents measured for l iquid nickel have bee-n inloedin the eritha Ipv

14

calculation, althouqh the sensitivity was sufficient t,) measure sItronq ioncurrents below the nickel melting point, 1726K (Fiq. 16). An increased slopein the second-law plot should be observed for the ion (:urrents from solidnickel compared to liquid nickel, reflecting the enthalpy of melting ofnickel, 17.47 kJ mol- [6]. Insufficient data points were recorded toverify AH , and the dashed line shown is an extrapolation of the liguid nickeldata. m

The results of Experiment 4 are confirmed in Experiment 5 (Fiq. 17)in which nickel was contained in graphite and alumina crucibles respectivelyinside adjacent compartments of a molybdenum quad-cell. The alumina cruciblearrangement was identical to that used in Experiment 4, while the compartmentcontaining the graphite crucible had the effusion orifice defined by themolybdenum as in Fig. la. This introduced a possib .1r :,omplication in that thenickel vapour from the graphite crucible was containe, within a molybdenum

quad-cell, rather than a wholly graphite cell. The chayr' in enthalpyobserved for this compartment during the experiment will be discussed inSection 4. Since the relative transmission factors were not measured for thetwo effusion holes, it is not possible to accurately compare the magnitudes ofthe ion currents from each compartment. However, the observation that the ioncurrent from the cell with the alumina crucible is three times greater thanthat from the compartment containing a graphite crucible is believed to besignificant.

Although zirconia crucibles were unsatisfactory as long-termcontainers for nickel in graphite quad-cells (Section 3.1.3), usefulinformation was obtained from short-term mass spectrometry anneals. Fig. 18shows second-law plots (Experiment 6) for nickel vaporized from zirconia andgraphite crucibles contained in a single graphite quad-cell. The quad-cell,of Fig. la design, had effusion orifices with identical transmissionfactors. The ion currents from the zirconia crucibles were approximatelythree times greater than those observed from the graphite crucible. Theexperimental arrangement within the cell was similar to that shown in Fig. la,except that a slot was cut in the extended crucible, opposite the effusionhole. In cell D the zirconia crucible was fitted with a zirconia iid, whilethat in cell B was not. The ion currents observed were marginally higher incell D, reflecting the smaller area of graphite exposed to the nickelvapour. The calculated enthalpies of vaporization are shown in Table 3, andare significantly greater for the zirconia crucibles compared to the graphitecrucible. In subsequent anneals of this system, the ion curronts from cell Cgradually increased and became closer to those of cells B and D, whichremained essentially constant at any particular temperature. This confirmsthe gradual equilibration of nickel and graphite that was observed inExperiment 3. Insufficient data was collected in these subsequent anneals toverify whether the enthalpy of vaporization from cell C increased in accordwith the ion current.

15

4. -[ISCUSS ION

4.1 Container Interactions and Vaporization Thermodynamics

4.1.1 Graphite Crucibles and Quad-Cells

It is clear from the metallographic evidence that significantinteraction occurs between molten nickel and graphite crucibles. This has a

pronounced effect on the nickel vapour pressure and enthalpy of

vaporization. The mean AH 0 calculated from Experiments 1-3 for nickel infresh graphite crucibles an quad-cells is 366 ± 6 kJ mol - I , significantly

lower than the literature average value of 428.3 ± 4 kJ mol - . Pre-

equilibration of the graphite and nickel at a much higher temperature in an

attempt to form a kinetically stable reaction layer was partially successful,

in that the measured AH0 falls midway between the fresh graphitc value and

the literature value. 298

Although chemical analysis of the nickel residues was not carried

out, estimates of the carbon concentration in liquid nickel may he made from

the relationship reported by Shunk r17]

-8 57log Nc = 89) - 0.462 (12)

T(K)

which is valid for the temperature range 1623-2023K, where Nc is the atom

fraction of carbon. The calculated carbon composition for the averaqe

temperature of each mass spectrometry run is shown in Table 4. There is

insufficient range in the calculated compositions to establish any

relationship between carbon composition and nickel enthalpy of vaporization.

The initially high value of AH observed in Experiment 5 is subject to lari-Terrors since it is only based on three data points. Furthermore, ther-

possibility of initial equilibration of toe nickel vanour with the mrclyhir o:quad-cell used in this experiment. However the-3 ibsequent AH is not~Tsignificantly different from that obtained from graphite cu-cel , anIt Ibelieved that the interaction with graphite dominates the nickel ertha ,

vaporization.

The evidence from FDperiment 3 for the establishment of a parti illistable pre-equilibrated layer indicates that care muut he taken in as, 1io, ;

the calculated carbon composition to the averaqe temperature of each ruin, -.the assumption of the establishment of rapid eauilibrium between the iraphil .

crucible and the melt. No obvious relationship exists between nickel enthil:

of vaporization and carbon composition calculated from the maximum temperatur,,

reached in each run.

4.1.2 Alumina and Zirconia Crucibles and Cells

Alumina crucibles were shown to be incompatible with qraphite quoi-

cells, through reduction oF the alumina and evolution of Co. The exact natur-

16

of the reaction wa not characterized, but probably fol lows reaction (1Rinehart and Behr,-n [33] hav- shown that Al 4 C3 vaporizes incongruently in thetemperature range 1300-1600K to give Al(ci) and graphite a, reactionproducts. The equilibriun vapour pressure of Al(q) at 1800K extrapolated from

their data is 26 Pa.

The obsorvation of r-act'i-n products in the cell residues whennickel was vaporized from A1 2 0 3 /graphite systems further hignligh-s theincompatiblility. There was significant metallic transport throughout thecell, even though the nickel shot had not meltei (anneal temperature was

1573K). Such transport couLd occur if the Cr pres:zire was sufficiently highto enable volatilization of nickel as Ni(CO)4 (q), although this species was

not observed in the mass spectrometric analysis. The obse rvatioo of aluminiumin the dendritic nickel lacework provides evidence for the presence of Al(g)

in the vapour, from reduction of the alumina by the 'iranhi te,as discussed

above.

The metalloqraphic evidence indicates that there is no interaction

between molten nickel and alumina or zirconia crucibles. with appropriate

quad-cell design, it has been shown (Table 3) that the calculate enthalpy ofvaporization from alumina cells is in excellent agreement with the literatur.,value. However, the vaporization of alumina from the vicinity of the effusion

orifice, when held for extended periods at high temperature, does provide alinitation to accurate vapour pressure and hence activity measurements, due tothe changing transmission factor for the orifice. The available evidenceindicates that calcia stabilized zirconia crucibles will not show this

limitation. A detailed study of the vaporization of nickel from zirconiacrucibles, utilisinq both the second-and third-law methods, has not yet beenpossibLe. It awaits the availability of crucibles of the required oeometry

and density.

4.2 rrors in Fnthalpips of Vaporization

The errors given in Tables 2 and 3 for the individiai vauIesof AH are 95 confidence limits, calculated from tho slope of th,ln( I. ) versus 1!T least-squares plot. This gives a measure ,of the

1uncertainties caused by random errors. The errors attributed t,- the meanvalues are the standard deviations. Johnston and Burley [i6] have discussed

the role of temperature measurement as the maior source of systematic errors,and the effect of a uniform temperature uncertainty across the range of

temperature measurement. It can be shown that the percentage error in thecalculated seconA-law enthalpy is approximately twice the percentagle error inthe true temperature, and about equal to the differential error in temperaturedivided by the temperature interval (x100) [34]. The diff,-rontial temr-rature

error is the difference between the absolute temperature errors at either endof the temperature range under study, and is a more important source of errorthan a uniform absolute error across the temperature range. Thus a uniform10K error in T over the temperature interval 1900 to 210()K produces an errorin AHT of about 1%, whereas a differential error of 10K over that 200Kinterval causes an error of approximately 5%. Although this- is )f the same

magnitude as the random errors (at the 95% confidence level) In Tableq 2 and3, the latter errors can be reduced by measuring a greater number of data

1 7

tD)ints ini i ndividual dis t-~rrinations, of- Aii Indeedl the : tinda r i !,,%:T

th( e~t cC0 f the comhinei measurements from four c-umpartmiits -8.1

I ow er . The above considerations therefore hiqhlight the need to ri ni:,-

diffe~rential temperature errors in accurate second-law measuremetnts. '4

exce!llent agreement between the experimentally-measured and lite rature- v.

fo)r th,, enthalpy of vaporization of gold indicates that there were, no -1,iifferential temperature errors present in this work.

4.3 Activity Measurements in Multicomponent Nickel Alloys

As was discussed in the introduction, the present study was

undertaken as a preliminary to the determination of both nickel and -1:!-In: §

a-tivities across the complete compositional range of the nickel-aliwro.111

Vs te m. In the previous study [11 of the hiqh nickel compositions

(- Ni 0.54) it was reported that materials problems prevented'- the,

dletermination of aluminium activities. In the liqht of the prese:: r-'sil

it is necessary to consider whether the reported nickel activities iniyDn,,--iderably in error.

The thermodlynamic activity aNi of nickel in a binary all~oy is 0

Ly

P Ni

P Ni

Whiere P a 'nd P N* are the vapour pressures of nickel in equilibriur wifth

11loy an~ puire nickel,, respectively. When pure nickel is in one crmi:-.f he uadcel an th aloyin another, and when the effusio e

-rn;'artme-nts have identical transmission factors, thpn the nicke. i

si mcd': hv the- ratio of the ion curreznts-

Ni

Ni r)TNi

I. r" I III I tNi ire- the, mas-spertrometric ion currenits 1(r .

p;t.' nickeL, res:pec tively [151. The ion cu7,-rrent for nic-keJ vipo~ri'e- '

'ir,--na -7ru'iJhe (Experim-nt fi) was approximately threec times- -hif

.r nic7k-l vipor ized from a qratphitp crucible, both :iile

i: lentico-l -ompartments of a graphite quad-cell. The activitv t '

i s msi ne in the- graphite crucible, therefore, was appr'XI ma"

r thatf of r' nickel contained in the adjacent zirconia nr

Tnieit is likely that the ion current for nickel vapo-rized f-rorl

zirronia crucible- was somewhat reduced due to the interaction of tlv'

with the expused inte-rnal graphite walls of the quad-cell. Therefr.-

nnkel- activi ty of nickel saturated wi th graphite would be some-what- 1'

ait this temperature (-1890K).

The same method was used to determine nickel activities- in ni

11,1minium alloys in graphiteo quad-cells- [11. The activities; so

wor, in reality psejudo-activities, in that both the pur- nick-] and the nr:ke]

in tho' all,-v iindi;rwent reaction with the graphite. 11 , actual value of th,-pseudo-activity may correspond to that of the alloy in a non-reactive system,provided that only one component of the alloy (either nickel or aluminium)roact with the crucible. However, where both components of the alloy reactwith the crucible, as is indicated in the case of nickel-aluminium alloys ingraphite [], the pseudo-activity for nickel then depends on the complox

interactions between nickel, aluminium and carbon, across the compositionrange of the alloy.

The thermodynamic properties of the ternary nickel-aluminium-carbonsystem are likely to be quite different from that of the binary nickel-

aluminium system. The phase boundaries have been determined at 1000 0C [35].

No ternary Ni 3AIC x phase was found, but y'-Ni3 Al exhibits a solubility for

carbon of up to 7-8 at.%. ITne phase -NiAl was also found to dissolve carbonto about 3 at.*, but no solubility was detected in the phases Ni 2 Al 3 andNiA1 3. A study [36] of the effect of carbon on ordering of y'-Ni3 Al inrapidly solidified Ni3AI-C alloys verifies that only an ordered N3AICx

(x = 0-0.34) sinqle phase is formed.

The nickel activities reported in Il] refer to the ternary Ni-Al-Csystem, and, until they can be reproduced using a non-reactive crucible/quadcell system, their relevance to the binary Ni-Al system is somewhat

uncertain. In such a non-reactive system, aluminium activities should then be

simultaneously calculable. The nickel and aluminium activities should beself-consistent using the Gibbs-Duhem method of calculating the activity ofone component from that of the other. Such consistency would be evidence of

the absence of spurious container interactions.

The results from the present work indicate that stabilized zirconiacrucibles contained in molybdenum or tungsten quad-cells, with the zirconjadefininq the effusion orifice, are likely to be suitable containers for

nickel. Preliminary studies also indicate that such a system should be stable

to liquid aluminium, and hence suitable for the thermodynamic study of thenickel-aluminium system over the complete range of compositions.

The above quad-cell/crucible combination should also be suitable touse in the recently commenced study of the nickel-aluminium-platinum system.

This ternary system is of great interest since it constitutes the second-generation aluminide protective coating that has been applied withconsiderable success to the first-stage turbine blades in an Australianmilitary gas turbine engine [37]. The platinum addition is extremelyefficacious in reducing the degradation of the coating by high temperaturecorrosion [37]. Because little is known of the thermodynamic properties ofthis important system, the quad-cell mass spectrometry technique is being uel

to determine the individual component activities and hence the thermodynamic

parameters of the Ni-Al-Pt system.

In this Report s iqni Fiait i rtoraictions have been shownoweri ncke andi varinus refractoryi rontainers commonly usedl in nu

t--euerairl- Kniudsen cell vaporization systems. in particular, qraph'.oonto be an kinsut table container, considerably reducinq the nick,-

and enthalpy of vaporization. Recrystuillized alumina and stabi lized 7i-

refractorips wpre stable to molt-en nickel, enablinq the determinati:iThermodynamic diata. Recommendations have been made for Asuitablo

Ir i-e 'a-(-ell Sysqtem for the study of th~e thermodynamic Propectlni -ke1-aluminium and nickel-aluminium-platinum alloys.

6. iCKNOWLFDY;EMFNT

The author is grateful to Mr F.A. Griffo foi carryinq out:';HieJsctio-n optical mic-roq)ra nhs of the crucibles, and for yen 1f-

*±e iiza~or. ff1iinc-y curve- for nickel vaporized from an c:n

7. R EF FP N .(

Tunfa' can,~ aId pi I mer, L.1) (9). "ua-ce '. mass spectrome try:

ict hi, dvnr'nur. propnrt i os of I iqida t lid n 1 urn-ni a'k I aI oy sTtiu 7ori . ffig/i Pri;, 1-2, 26 1-266.

2. ard, T.W. iind Mulford, R .N . R. (,1967) "Stu:ly of some of the param-t.,rsaf foct inq Knudson effu-sion . I. Fxp~erimeital tests of the validityf '(in cos 1 no law as a functiorn of cell and sample qeOine(:trica, -id,

3. (ba t i In, AI., libher t, m. , a nd Pa ttore t, A. (1979q). !'h e r 71o dyn ar-,irInA purysi cu-chemical bnhav4i our or thfe interactions between YKnudse-

efuion-c-ils an,! th- ;ystems under investigation: analys is bvii ott'~n~Itie m s spoctrurnetrv" . 'B-,.tpf!ia: FPobl icat;On,5< : r s's'- 'f tho ! 0 th' rnt1li s '7.arh.c;pcj.s: ccir rn

r~~ ~ )n a ono i temp('riture . r andI gass NBS

4. Clark, N.J. and~ Mairt, 11.1,. (1981 1 . "'Fungisten bronze-contiinerinte-ractions- it Ili 1i temipratures". Mat. Res. Bull., 16, 127-'3>.

5. Mart, P.L. anl I larK', N.JT. 1982). "Vaporization of tungsten brunzes:cimic I i t l n nr,)n ze s . H',:li Tt rn . Sc i ., 1 5, 1- 1 6.

6. Hultqrean, R., F. -- 11 , P.O, awkins, fl.T., Gleisor, M., Kelley, K.Y.arWaqma n, r) 9) 7 3) ."ne'udva I urs of tethe rrolynamic

.'-o ~Metals Par,, Ohio, ASM, 350-357.

7. Nejrn-!inrv, An. 1 196 3) . "Yajor prossz.re of thoe 'l'ieents", I Trans ihy I- T . , lfoseairch Lirni ted, London, 300-397.

a,.N nij Man !i .r. (11960).Rs a eaL rc(""-, 4.

3~Kovtun, -,.P., Kruqlykh, A.A. and Pavlov, V.S. (1962). "Vapour pressir-

andi the- evaporation coeffic-ient- of nickel". t7krain. pjF "z. 714), 436-438.

10 Alcock , C. P. and Kuhik, A. ( 1968)., "Thermodynamic bephaviour of liqiiiron-cobalt and nickelI-platinulm alloys". T ra ns. Tns t . M i n. We

77, C220-C224.

11. Lindscheid, H. and La~nge, K.W. (1970). "Vapour pressure measu-;,rements ot

binary alloys containinki two vclatile componts." 7 . N!"t<>i,

61, 193-200.

12. Grimley. R.T., Burns, P.P. and lnqhram, M.G. (1961). "Therriodynami cs ofthe vaporization of nickel oxide". J7. Chemn. Phys., 35, (2),551 -154.

13. Vrestal, J. and Kurera, J7. (1971). "Determination of nickel vaporpressures by a n effusion method using nicke 1-63". .7 -2 . Fnurn.,17, (5), 158-160.

21

'TI i --

-"H Ih r. Ir nr t ir- %,a p, r: z ot 1

In HI N Ar~~v' 7nn 7'la 71

P 'r I U ~ ' T 10W , lr *', qey ar r,'

9'~~~~~ I; IaA -r (- Iie' Nr) A '8 7) ?'", 1 m r '

Y'~ 1-~ 23)! 2 7 a rIiv a rotl Y

fMe -ha i 7a I nq;i neers, 2 22- 23T;.

N-~W Y rk , a ndsipees<I p* P. 1 ' 65, Firn; spni me n t

:-n-r i Fl-fit2 1, Newd York.

* ~ ~ p ~ '' 3,Totnaker, J.M. and 1Hon-, F.H. (1 982r.-C a 11coy s f r om 9) tof 1 2 1 5 C"-

T.Ad~o-Wesley, !,onon 21 4-215.

r an; * A . 9?71 "ThePr-,O(chePIP t rv of Ic m et a3

r ,,- Io 'aco11 m" .- . Zan O . -7 , -

-. Ton ~ r. a' ln , M . S. 1 C

o:,xide and carbon. TheP .%. 20 3- 4 3In;

* . .-. * ini Alcock. C.R. ( 19793).

olhr;, 5th ed.,Prao '''('x f r,"

I, a - P a r Proh + , P . (1971. AA troTherc.* -*'-- 37.

* Yirn:intn' C . (197 1) 1Iron a nd nicrkelI ,tah Li ty ir i , .cr-I~an-t gi th ref rac tory compound s ", T,) rnq5)' k'fe . 1 1-' 2

]'* 1' M. (3, Wh ite , J. L1. and Mackenz i, 'T. n . 1 959)Ph-s coflefc 3measurfements at hijqloirano, i

7', r:don, 25-26.

27. YuL n, P. F. a nd Ma rkh':I iya , T. P. ( 1967) "pe-ac t ion thf, rm)iyI1:-zr) ~2' Mjqn, and~ can wi th ref rac--tory metalIs" A?-n'.:'of Pefractor4-:s i;, 440-446.

nl-iiw' itL :.,: it i t1 wi th i i 'imi n: r' ,xi i iic In. c si I i-rn gic il"

Pa',i].',t)(I T.' r) M-ii ,.r-) ) AvIv; a l nt rprt.

3. Rub) h Pc. a nd CohIen i, M (1 96 7) "Metastahlp pxtenn:ions of carhorisolIubi I i t v in nickpi an! -obaltI". -rnaMetail'ircica -1, 73-74.

3. Fr ank I in, J1. T_ , Dillard, R 8,(ost-nstock, H.M., Herron, J.T. , Draxi, K.anid t'i.-l, F. H. (193) . "Ionization potepntials;, appearance-

Tpotenti als, and h-ats of fo-rmation of qase-ous poiieions".

NSR~-N8~26.

3-_. Xi nter , ':.F. and K2 ser, R.W. (134)(). "A mass speI tromp trio

_nvpstiqation of nirkel tetra-carbonyl,, aoli iron or~tacarborjvI.rc.(Thon.,3, (5), 699-702.

3. R:nohairn, C.H. and; Behrens, ;,.r,. (1980). "Vaporization thermod~vnamicsof a iminium carbide ". 1. C -ern-. 7ho-rn'Lur. 12, 205-215.

4. ,),,-kprmanr, P.J. and Thorn, R.J. (1961). "Vaporization of ()xides". in

'rs:ncerai-c- sc 'ence, Vol I ( Burke, J. E. pd., Pe roamrr.',o ,42.

5 . O;c huster, J.C. and Nowotny, I) . ( 1982). "The ternary syste-m. nickF-a 11 n. r. n-carbon" ois: Cher._ 11'3, 163-170.

H6 -an r). k. and Chico, W.K. (1983). "Carbon effect on orderina rf , -%1.in rapidly solidifi-I Ni 3 Al-C- alloys". Scrirfa '7,~'v

jnrs~nG.P . and Pic-harris, F.G. ( 1981) "Hiob-tenvorature corrosior

i n ii ita ry c as t iirb i no, an iitial1 metalinooraphi c comnariso.-he durabilitv of two protective coatinqs (PWA 73 and RT22) for

fircg-staue- turbine- hIades inr TF30-P- 3 enqines, W1) ."7, -1aterials Reseairch iahnratorices, Melbourne, Victoria.

2:prf

TA 13L E 2

Enthalphy of Vaporization of Gold from Graphite Quad -Cel1s

T Ranqe 4 H * AHT 298Compartment (K)

(kJ mol - 1) (kJ mol -

)

Calibration No. 1

A 1688 -1933 344.8 ± 24.4 370

343.4 ± 23.9 368

C 345.5 ± 24.5 370

D 346.2 ± 28.6 371

mean 345.0 ± 1.2 370

Calibration No. 2

P 1667 - 1890 345.4 ± 23.9 370

* 95% confidence limits on individual values, ± standard deviation on m-an.

r v .. . . . . . . .. .....

.i, :i _py of Vapori zationi of Nickel tr-m Vari ous _ -b1__and ',iad -Cell1

Compartment T ranqe AH *AH29gxpt, T 29 a

(crucible) (K) (kG mol- l ) )kJ m - I

I. Graphite quad-cel, graphat- crucihles

A 1790 - 2015 333.0 ± 18.7 367

B 335.0 f 15.5 3f9C 337.1 ± 24.4 371D 339.3 ± 24.1 373

mean 336.1 ± 2.7 3.7,

2. Graphite quad-cell, graphite crucibles

A 1827-1972 325.5 ± 7.6 359B 325.3 ± 15.6 3%

C 329.6 ± 15.6 363

D 339.4 ± 14.7 373mean 330.0 6.6 364

3. Graphite ga-cell graphite crucibles

B 1753 - 2020 351.5 ± 9.6 385

(pre-eiuilibrated)

327.6 ± 20.4 36inew)

4. Molvhdenum auad-celI, alumina crucible

1734 - 1895 404. 2 ± 17.4 43(1

. Molybdenium quad-cell, alumina and oIraphi-t crtc hblos

Q 1736 - 1918 3)3.1 ± 3.3 425

(al r i. na)

C 1721 - 1825 (4G0 59)

(graphite) (pts. 1-3)

1728 - 191i9 332.3 ± 13.0 ml

(pts. 4-10)

6. Graphite quad-cell, zirconia and graphite crucihles

C 1794 - 1995 331.8 ± 20.5 36W5(graphite)

B 356.1 2 1.1 31)0

(zirconia)

D 355.' ± 38.5 389(zirconia + lid)

95% confidence limits on individual values, + standard dPviati un on- i n.

T A B I, E 4

Calculated Carbon Composition of Nickel and Enthalpy of

Vaporization of Nickel

T average Nc AH TExp t.T

(K) average (kJ mol - )

Graphite quad-cells, graphite crucibles (new)

1. 1903 0.117 mean 336.1 ± 2.7

2. 1900 0.117 mean 330.0 ± 6.6

3. 1887 0.116 327.6 ± 20.4

6. 1895 0.116 331.9 ± 20.5

Molybdenum quad-cell, graphite crucible

5. 1773 0.108 (400 ± 59)

(pts. 1-3)

1823 0.111 332.3 ± 13.0

(pts. 4-10)

* 95% confidence limits on individual values, ± standard deviation on means.

ud-cc] I ith arucible containing sample (after Johinstcvn

,U.Eid-:elil wih r:er~imiI I hiille Eli

1' I

F1) 1 X

FIG. 3. Nickel in graphite qu.id-cell annealed at 1573K(magnification 4 X).

Top L. - Al,0 3 coating and crucible, metallic lacework,

blue deposit.

Top P. - A1 2 03 coating, metallic deposit, reaction layer

on Ni.

Bot L. - Graphite walls and crucible, Ni melted.

Bot R. - Graphite walls, Al 2 03 crucible, blue depositthroughout cell.

r'IG. 4. Metal lic ]cework r l net t, .1 1el, F1q. 3)(maqn~fic1ti)n ).

41, 'A

4'8 A I

NNi (L,)

61

0

4e

FIG A.Sann leto Ka)rp ndEegyDsers

X-a nlss(aeae) fmtli aeoki

Fig. 4.d

Ni (LA

pA.

.

T, °C

1600 1500 1400

17 2

516

I"• 3

15C

6

4

14 1

13 I I I 1 I I5.2 5.4 5.6 5.8 60

104

T (K)

.;old cilibration No. I. Second-liw plot of In (I T vArso;I for gjold wvaporized from (riphite quai-cell.

T

T, 0C

1600 1500 140022 ]

21

4

I- 7

20 2

35

88 6

18 , I A I A I I I

5.2 5.4 5.6 5.8 6,0

104

T (K)

FIG. 10. G old calibration No. 2. Second-law plot of In (I TI vor'u.I for gold vaporized from graphite quad-celJ. ,>0u

T

6 TI

5-

4

Z 2

0 I'5 10 15 20 25 30 35 40

Electron Energy, eV

FTC;. 11. ionization efficiency curve for nickel vij (.i i,,

qraphite quad-cell at 1973K.

I " -- IIIImJ.I"1 I I

10

S 6

- 4

2

010 15 20 25 30 35 40 45

Electron Energy, eV

I. Trinzation efficien( -,ur','r i.r Irn it wnii2nt

tomperature (9f

T. °C

1700 1600

21 1 1 I

20

B+C4

ZA

2

18 3

5.0 52 54 56

T(K)

Seccnrd-law plot'; fol n]ri k l v i ,P ied I 1 q,1i , kcrucibles in i qrapli te Iutd-&c 1] I 'i. i Il I ,energy A0 eV.

-' . . .. . . . .. . ' .. .. ... , , , , - - - " ". ... . . . . . . II I II I I I I I I II I I . . . . . .

T, 0C

1700 1600211

4 ABC

2 D

202

193

6

18

50 5.2 5.4 5.6

104T (K)

iI.14. Second-law plots for nickel vaporized from qcaipl.itccrucibles in a graphite quad-cellI. I 'n i zi nelectron energy 40 eV.

T, O0

1700 1600 1500

21C

(new graphite)

220

6

8

19 (pre-equild

4

3

17 1 55.0 5.2 5.4 56

T(K)

1;.15. Second-law plots for nic:kel vaporizedl f i

crucibles (B pre-equiflibyited, C new) in A; ij !itt

quad-cell. Ionizing electron energy >A'

T

1600 1500 1400

2: T 1

20 12g

2/

7/z/

141

177

180

0

10

52 5.4 5.6 58 60C

T (K)

FI';. 16%. Second-law plo)t for nickel vaporized from an a11umIM,crucible in a molybdenum quad-cell. Ionizing elejctT('II

enemqy 35 eV.

T 0C

1600 1500 140021 1

50

20 -

C1graphite//

19 6 3/j

C8 Balumina

18 -2

7

101

17 -Ni M.Pt.

1615.2 5.4 5.6 5.8 6.0

T (K)

fI.17. Second-law plots for nickel vaporized from graphite(C) and Alumina (B) crucibles in a molybdenum quad-cell. Ionizing electron energy 35 eV.

T. OC

1700 160022 1 1

4

C

(graphite)21

3 (ZrO2 )

B

20 0

(ZrO2 + lid)2

50

19

18

17 -

5.0 5.2 5.4 5.6

104

T(K)

FIG. 18. Second-law plats for nickel vaporized from graph-ite(C) and zirconia (B and D) crucibles in i qrdiphitequad-cell. D with zirconia lid, B without.Ionizing electron energy 35 eV.

A.A

(MRL-R-884)

DISTRIBUTION LIST

MATERIALS RESEARCH LABORATORIES

Director

Superintendent, Metallurqy Division

Dr G.R. Johnston

Library (2 copies)Peter L. Mart

DEPARTMENT OF DEFENCE

Chief Defence Scientist (for CDS/DCDS/CPAS/CERPAS) (1 copy)Army Scientific Adviser

Air Force Scientific Adviser

Navy Scientific AdviserOfficer-in-Charge, Document Exchange Centre (17 copies)

Technical Reports Centre, Defence Central Library

Director of Quality Assurance Support (DQAS)

Deputy Director, Scientific and Technical IntelligenceJoint Intelligence Organisation

Librarian, Bridges Library

Librarian, Engineering Development EstablishmentDefence Science Adviser, Australia Hiqh (Summary Shee's only)

Commission, London

Counsellor Defence Science, Washington, D.C. (Summary Sheets only)Librarian (Through Officer-in-Charge), Materials Testing Laboratorips,

Alexandria, NSWSenior Librarian, Aeronautical Research Laboratories

Senior Librarian, Defence Research Centre, Salisbury, SA

DEPARTMENT OF DEFENCE SUPPORT

Deputy Secretary, DDS

Head of Staff, British Defence Research & Supply Staff (Aust.)

OTHER FEDERAL AND STATE DEPARTMENTS AND INSTRUMENTALITIES

NASA Canberra Office, Woden, ACTThe Chief Librarian, Central Library, CSIROLibrary, Australian Atomic Energy Commission Research Establishment

MISCELLANEOUS - AUSTRALIA

Librarian, State Library of NSW, Sydney NSWUniversity of Tasmania, Morris Miller Lib., Hobart, Tas.

(MRL-R-884)

DISTRIBUTION LIST

(continued)

MISCELLANEOUS

Library - Exchange Desk, National Bureau of Standards, USA

UK/USA/CAN/NZ ABCA Armies Standardisation Representative (4 copies)

Director, Defence Research Centre, Kuala Lumpur, Malaysia

Exchange Section, British Library, Lending Division, UK

Periodicals Recording Section, Science Reference Library,

British Library, UK

Library, Chemical Abstracts Service

INSPEC: Acquisition Section, Institute of Electrical Engineers, UK

Engineering Societies Library, USA

Aeromedical Library, Brooks Air Force Base, Texas, USA

Documents Librarian, The Centre for Research Libraries, Chicago, Ill.

Defense Attache, Australian Embassy, Bangkok, Thailand