THE AMERICAN
MINERALOGIST,
VOL. 44, SEPTEMBER_OCTOBER, 1959
STUDIES IN THE SYSTEM IRON OXIDETITANIUI{ OXIDE"
State
J. B. I{ecCuesNrvt axn Anxur-l N{uew,The Pennsylvania
LIni.aer
sity, LIni,aer
sity P arh, Pennsylaania
Assrnecr
Phase equilibria in the syiem iron oxide-TiOz in air have been investigated at temperatures extending from the liquidus to slightly below the solidus. Stability relations among
the phases spinel solid solution (FeO'FezOr2FeO'TiO),
hematite solid solution (Fe:Or
FeO.TiOz), pseudobrookite solid solution (Feror.TiOr-FeO.2TiOt, rutile solid solution,
liquid and gas are illustrated in at apparently binzry diagram obtained by projecting
ternary phase relations in the system FeO-Fe:OrTiOz into the the join FeO'FezOrTiO:.
INrnooucrroN
Titanium is relatively abundant in the earlh's crust; igneousrocks contain an average of approximately 0.64 wt.l6 of. this element. The most
important titanium minerals are ilmenite (FeO.TiOr), titaniferous magn e t i t e ( F e O . F e : O r - 2 F e O . T i O r ) ,r u t i l e ( T i O z ) a n d s p h e n e ( C a O ' T i O :
' SiOz).A great deai of interest has been shown in theselately, especially
in ilmenite and the titaniferous magnetites,becauseof their relation to
rock magnetism and their possibleuse in petrogenesisand geothermometry. These minerals will be important indicators of the environment at
time of their formation as the very complicatedphase relations in the
system FeO-Fe2O3-TiOz
becomebetter understood.Studiesof this system
will be most useful when completedfor an extensiverange of both temperature and Oz partial pressure.In the presentinvestigation a start on
the broad study has been made by examining equilibrium relations in a
region of relatively high temperatureand ()2partial pressure.
In previouspapersemanatingfrom this laboratory (I,{uan, 1955,1958;
IIuan and Osborn, 1956) the special difficulties involved in studying
high temperaturephaseequilibria in iron oxide containing systemshave
been pointed out. In all systems we have studied and describedpreviously, iron was the only elementpresentwhoseoxidation state was variable, while the others werenoble gastype elementswith constantoxidation
state such as Sia+,Al3+ and NIge+.In the case of the iron oxide-TiOz
system, additional complications may arise becauseTi also exists in
different statesof oxidation, viz. Tia+,Ti3+,Ti2F,and TiO.When attempting to study this system experimentally one should therefore keep the
possibility in mind that Ti as well as Fe may undergo changesin oxidation states unlessproper precautionsare taken.
* Contribution No. 58-39 from College of Mineral Industries, The Pennsylvania State
University, University Park, Pennsylvania.
t Present address: Bell Telephone Laboratories, Murray Hill, N. J.
SYSTEM IRON OXIDE-TITANIAM
OXIDE
927
,Fco
z9)
-rooooo
I
3l
;l
rl
- 200000
rooo
TEMP ,
1200
l40o
'C
Frc. 1. Diagram showing standard free energy of formation of oxides of Fe and Ti as a
function of temperature (solicl lines). The curves have been constructed from thermodynamic data tabuiated by Coughlin (1954). Dash curves indicate various ievels of O:
partial pressuresas labeled in the diagram (0.21, 10 u, 10-'0, 10 15,10-20atm.).
In orcler to evaluate the relative stabilities of the various oxidation
'Ii
ancl Fe it is inslructive to examinea diagram showing standstates of
ard free energiesof formation of the various oxidesof these elementsas
a function of temperature. Such a diagram is shown in Fig. 1, which has
been constructedon the basisof data tabulated by Coughlin (195a).The
J. B MACCHESNETAND A. ],TUA\
crystalline phasesrepresentedin the caseof iron are Fe2O3(hematite),
FeO.FezOa(magnetite), "FeO" (wiistite) and Fe (metallic iron), while
the crystalline phases encountered in the case of titanium are TiOz
(rutile), Ti3O5,Ti2O3,TiO and Ti (metaliic titanium). The AF" valuesfor
thesereactionsas labeledon the curvesin the diagram all refer to 1 mol
cf oxygen.Hence the AFo values are related to the partial pressureof Oz
of the gasphasein equilibrium with condensedphasesthrough the simple
relation AF': -RZ lnK: RT ln pctz.Curves representingconstant O:
pressuresof 0.21 (air) , 10-5,1010,10-15and 10-20atm. are plotted as dash
lines in Fig. 1. It is evident that the oxides of titanium are much more
stable than those of iron. In the caseof the pure substances,the lowest
oxide of iron, "FeO," can be reducedto metallic iron at 02 pressureswhich
are still high enough to prevent the structure of the highest oxide of
titanium, TiOz from decomposingto the structure of the next lower oxide,
TisO5,.
In a liquid phase the restrictionson the valencestatesof the ions
i m p o s e db y s t r u c t u r a rl e q u i r e m e n lasr e n o t c r i t i c a l a
, n d c o n l i n u o u sv a r i ations in the ratios of ions of different valencesare possible.Hence the
solid linesin the diagram of Fig. 1, pertaining to heterogeneous
equilibria,
would not be presentin the caseof a homogeneous
liquid phase.The com.
position of the liquid, expressedin terms of Fe3+/Fe2+and T'i4+/Ti2-*
ratios, would vary continuously as a function of temperature and 02
partial pressure.Becauseof the much greater stability of the higher
valencestatesof titanium than of iron, AFo for the reaction.
2 Fe3+1;,. f
Ti'+1n
:
2 Fe2+riq. *
Tia+rin
would be strongly negative.From the relationship LF" -- - RT ln K, it is
seenthat K would be numerically very large and hence Tia+/Ti2r very
much higher than Fesr/Fe2+.It is thereforeto be expectedthat in a liquid
made up from the oxides of Litanium and iron the titanium will remain
essentiallyas Tia+ unlessthe Fe3+/Fe2+ratio is very low. Hence, only if
extremely reducing conditionsare imposed on the system in an attempt
to obtain essentiallyall iron in the ferrous state is there seeminglyany
dangerof reducing titanium to any appreciableextent to lower oxidation
states than Tia+.
With the above considerationsin mind we have chosento start our
study of iron oxide-titanium oxide equilibria by investigatingthe system
in air (16r:0.21 atm.). In choosingto work at a constant 02 pressureof
this level we are studying a section through the system Fe-FezOe-TiOz,
which is part of the ternary system Fe-Ti-O. n,Iethodsfor describing
phaseequilibria along an Or isobaricsectionthrough this systemare anal+ Tia+ and Tiz+ are chosen
as extreme valences of titanium in oxide phases. This choice
does not rule out the possible existence of intermediate valence states such as Ti3+; the
latter correspondsformally to a mixture of Tia+ and Ti2+ in the ratio 1:1.
SYSTEM IRON OXIDE-TITANIUM
OXIDE
ogous to those described for a iarge number of iron silicate systems'
(Nfuan, 1955). The most convenient presentanolably FeO-FezOs-SiOz
tion of phaserelationsis obtained by projecting the 0.21 atm. Ozisobaric
The diagram thus obtained has
section into the join FeO.FezOe-TiOz.
the appearanceof a binary system, but the diagram can be used and
interpreted correctly only if due considerationis given to its relation to
the ternary system.
Pnnvrous Wonx
Severalphaseequilibrium studiesin the pure iron oxide and pure titanium oxide systemshave been reported in the literature. The most complete discussionof phase relations in the system Fe-O is found in the
papersby Darken and Gurry in 1945and 1946.A diagram summarizing
some of their data which are particularly pertinent to our problem has
been presentedmore recently by llluan (1958). The Ti-O system has attracted considerableinterest in recent years. Andersson et al. (1957)
are making very comprehensivestudiesof the phasespresentin this system. The data obtained by that group will serve to refine the approximate diagram for Ti-O inferred by DeVries and Roy (1954) on the basis
of scattereddata available in the literature.
Someof the generalfeaturesof the ternary system FeO-FezOrTiOzare
fairly well established.Three solid solution seriesare known to exist in
the system: the cubic magnetite-ulvospincl(FeO Fezor-2FeOTiO2), the
hexagonal hematite-ilmenite (Fezos-FeOTiOz), and the orthorhombic
pseudobrookite(FezO:].TiOrFeO'2TiOt. In the first of these,magnetiteulvospinel,a continuoussolid solution exists at high temperatures,with
exsolution taking place below 600' C. (Kawai, Kume, Sasajima, 1954;
Vincent and Wright, 1957). The continuity of the hematite-ilmenite
solid solutionsseemsto be in some doubt. Ramdohr (1926, 1950, 1953)
reported that hematite and ilmenite form a continuous solid solution
seriesabove600oC., while Pouillard (1950)found that two solid solution
phasesexistedtogetherin equilibrium at 950oC. Thesewere describedas
03 with r
(Fe1-r2+Fe2rs+Tir-rn+)
Oawith y: 1/3 and (Fez-2,3+Fe,:+11"a*,
--lf 3, correspondingto a miscibility gap extendingfrom approximately
33 to 67 molf6. Nicholls (1955) in a recent review states that it is well
known from the work of Ramdohr (1926),as well as Posnjak and Barth
(1934), that completesolid solubilit.yexistsbetweenilmenite and hematite above 1050"C. The value of 1050"C. apparently is confirmedby the
work of Basta (1953)who showedthe miscibility gap to be absentat this
temperalure.
The most recent and most comprehensivework on the pseudobrookite
solid solution seriesis that of Akimoto and Nagata (1957).On the basisof
sealedtube runs they reported a complete solid solution seriesbetween
J, B. MACCHESNEY AND A. MUAN
F e z O r . T i O ra n d F e O ' 2 T i O za t 1 2 0 O "C . P o u i l l a r d( 1 9 5 0 ) ,h o w e v e r ,i n h i s
investigation of the system by a similar method could find no pseudobrookite among the phasesappearing at 950" C. Ramdohr (1950) reported that pseudobrookitegrows at the expenseof hematite-ilmenite
solid solution at 1100oC., and Basta (1953) found that pseudobrookite
appearsas a new phasein magnetite-ilmeniteintergrowths above 950' C.
On the basisof thesereports it seemsreasonableto concludethat a lower
temperaturelimit existsfor the stability of pseudobrookite.
Another interesting feature of phase relations in the FeO-FezOa-TiOz
system is the mutual solubility of ilmenite and magnetite solid solutions,
which is thought to give rise to the intergrowths of magnetite-ulvospinel
and ilmenite found in certain mineral deposits.The best evidencefor the
existenceof this solid solution is afiordedby Chevallierand Gerard (1950)
who succeededin producing a cubic solid solution crystal whosecomposition can be calculatedas conterining3Tmol/6 ilmenite. This is regarded
by Nicholls (1955)as a solid solution not betweenthe cubic and rhombohedral phasesbut betweenmagnetite and metastable'y-FeTiOa.
Reports on studies of phase relations close to the join FeaOr-TiOzin
the system FeO-FezOa-TiOzhave appeared sporadically in the literature
during the past 20 years. Junker (1936) made exploratory runs in air at
Iiquidus temperatures with mixtures containing from 0 to 60 Wt./6 iron
oxide. He concluded tentatively that a "eutectic" exists at about 50
wt./6 TiOl Ernst (1943) expanded the work, making runs in sealedtubes
as well as in air. He identified the primary phase on the iron oxide side of
the "eutectic" as pseudobrookite,which he believed to be stable at its
melting point (approximately 1500' C.) and which he thought possessed
considerablesolubility for TiOz and limited solubility for FeO. Ernst also
found evidencefor the existenceof a second"eutectic" betweenilmenitehematite solid solutionsand pseudobrookite.
ExpnnrlrnNrar Meruor
General Proced.ure
The quenching technique was used in this study. Preheatedmixtures
of iron oxide and titanium oxide wereheld in platinum containersin air at
constaht temperature for a period of time sufficient to ensure equilibrium
among gas, Iiquid and crystalline phases. These samples were then
quenchedrapidly to room temperatureand the phasesidentified by microscopic and r-ray techniques. Compositions of liquids at liquidus temperatureswere determined by chemicalanalysisof quenchedsamples.
Materials
Starting materials were reagent grade oxides ("Baker Analyzed").
After drying at 400o C for 18 hrs. the oxides were mixed in desired pro-
SYSTEM IRONOXIDE-TITANIAM OXIDE
931
portions, ground under alcohol and fired to melting in a gas-air combustion furnace. These samplesservedas starting materialsfor runs of long
d.uration for determination of subsolidus equilibria in the quench furnace, and occasionallyalso for quench runs in the liquidus temperature
region. In most cases,however, samplesfrom the gas-air furnace were
heated in air at temperaturesslightly above the expectedliquidus temperature prior to determination of liquidus temperatures. This pretreatment of the mixtures ensured a close approach to the equilibrium EezOzf
FeO ratios of the samplesand decreasedthe time necessaryfor equilibration runs in the quench furnace.
Platinum cruciblessaturated with iron were used to contain the mixtures during prefiring in the gas-air furnace, and small envelopesmade
from thin platinum foii served as containers for samples during the
quench runs. The use of platinum as material for containing iron oxide
bearing samplesdoesnot causeany complicationsin studies carried out
at relatively high levels of 02 partial pressures,such as that of ait (Po,
:0.21 atm.) used in the present investigation. For a more detailed discussion of this problem the reader is referred to a previous paper from
this laboratory(Muan, 1957).
Furnaces and, TemperatureConlrol
The quenching experiments were carried out in vertical tube furnaces
wound with platinum or platinum 20/s rhodium resistancewire. Temperatures were measured before and after each run with a platinum:
platinum-100/6 rhodium thermocouple calibrated at frequent intervals
against standards with melting points defined as follows: Ca\'IgSi2O6,
1391.5'; CaSiO3,1544' C. The temperaturesgiven on this basisare in accordanceboth with the GeophysicalLaboratory scaleand with the 1948
International Scale (Corruccini, 1949; Sosman, 1952). Furnace temperatureswere controlled by a secondthermocoupleinsertedcloseto the
heating element and connectedthrough compensatinglead wires to a
Celetray controller.
Samples
Eramination of Quenched.
Microscopic and r-ray methods were used for identification of phases
presentin the quenchedsamples.Examination of polishedsectionsin reflected light was found more informative than examination in transmitted
light under the petrographic microscope becauseof the opacity of most
phases.The r-ray technique was used for determination of compositions
of solid solution crystals, but could not be used as a tool for determination of liquidus temperatures becausethe liquids were exceedingly fluid
and gave rise to formation of dendritic crystals even during the fastest
quenching of the samples in mercury.
932
J. B. MACCHESNEY
AND A. MAAN
Divalent iron contents of samples quenchedfrom liquidus temperatures were determined by wet chemical analysis. The samples were
crushed in a Plattner mortar to pass a 100 mesh screen,after which
0.1-0.2 g. were heated to boiling in 20 ml. 1:3 HzSO+in a covered
platinum crucible. When steam evolved, 5 ml. Hl- was added and the
solution was allowed to boil for 5 10 minutes. The crucible was then
plunged into a 600 ml beaker containing 5 vo1./6 HzSO+saturated with
boric acid, and the solution was titrated with 0.05 n Kl.linOa.No analysis for total iron oxide content was carried out. Previous experience
from similar studies in our laboratories (l{uan and Gee, 1956; l\'{uan,
1957; Phillips and Muan, 1958) has indicated that the loss of iron by
alloyinq in the thin platinum foil used as containersfor the samplesduring the equilibrium runs is negligibleas long as the partial pressureof 02
of the gas phase is as high as that of air. Furthermore, the platinum
cruciblesusedfor pre-melting the starting mixtures in the more reducing
atmosphereof the gas-air combustion furnace were saturated with iron
under the conditionsprevailing in this furnacethrough repeatedprevious
use.
Rnsurrs
Results of quenching experiments are summarized in Table I and
illustrated graphically in Fig. 2. Two diagramsare used in this figure in
order to show the phase relations as clearly as possible.The upper diagram has the appearanceof a binary system with FeO.FerOr and TiOz
chosenas components.Stability rangesof the various phasesas a function of temperature can be read directly off this diagram. However, because the FeO/Fe2O3ratios in all phasespresent are not equal to 1:1
(molar basis) as in FeO.Fe2O3,true compositionsof condensedphases
cannot be representedin terms of the two chosen components. The
supplementary triangular diagram FeO-FegOe-TiOr
in the lower half of
the figure is presentedin order to show true compositionsof condensed
phases.Becauseof the restrictionsimposed by structural requirements,
the crystalline phase compositions will vary essentially along parts of
straight lines in this diagram, correspondingto binary solid solution
series.There are three such seriesin the system FeO-FezOa-TiOz,
indicated by light straight lines in Fig. 2. One is the magnetite-ulvospinel
(FeO.FezO3-2FeO.TiOz)series of solid solution crystals with spinel
structure, the secondis the seriesbetweenhematite (FezOa)and ilmenite
(FeO.TiOr) with hexagonala-R2O3structure, and the third is the series
with pseudobrookitestructure between the composition points Fe2O3
.TiOz and FeO 2TiO2.*
* In order to obtain experimently
compositions throughout these solid solution series,
one has to either vary the Oz partial pressure over wide limits or work 'lvith sealed contain-
Tesr,r I. Rnsur,ts ol QurNcnrnc
Temp. of
Quenching Run
PhasesPresent*
(c)
Temp
of
Run
for
Anal.
(c1
ExprnnnnNts
Initial Comp Comp. after Equil.
(Wt.%) (by analysis)
(Wt. %)
FerOr TiOz
FeO
FezOr TiOz
99
1555 Magn.(ss)
1515 Magn.(ss)
1477 Magn.(ss)fHem.(ss)
155.5 Magn.(ss)fl,iq
1515 Magn(ss)fHem.(ss)
1477 Magn.(ss)tHem.(ss)
97
l57l
1531
1520
1469
l44l
1420
Liq.
Magn.(ss)fl,iq. (tr.)
Magn.(ss)fHem.(ss)(tr.)
Magn.(ss){Hem.(ss)
Magn.(ss){Hem.(ss)
Hem.(ss)
1562
1554
1551
1547
1502
Liq.
Magn.(ss)([r.) +Liq.
Magn (ss)(tr.) +Liq.
Magn.(ss)(tr.) *Liq.
Magn.(ss)fHem.(ss)
1547
1542
1524
1516
1504
1468
1463
Liq.
Magn.(ss)*Liq.
Magn (ss)fl,iq
Magn.(ss)fHem.(ss)
Magn.(ss)fHem.(ss)
Magn.(ss)(tr.)+Hem.(ss)
Hem.(ss)
93
1529
1502
1470
1441
Magn.(ss)(tr.) +Liq.
Magn.(ss)(tr.) fHem (ss)
Hem.(ss)
Hem.(ss)
90
10
1529
1520
l5O2
l47O
1441
Liq.
Hem.(ss){Liq. (tr )
Hem.(ss)
Hem.(ss)
Hem.(ss)
1,530 88
12
1 ,565
25.5 69.5
95
5.0
23.6 64.+ 12.O
* Abbreviations used in this and succeeding tables have the foilorving meanings: Liq.
: liquid; Magn. (ss) : crystals of magnetite solid solution; Hem. (ss) : qly5tals of hematite
solid solution; P. Brook.(ss):crystals of pseudobrookite solid solution; Rut.(ss):6ry51615
of rutile solid solution: tr.:trace.
Trslr
Temp.
of
Temp. of
9"":tn
ing Run
I (continueil)
phasespresent*
(c)
(wt. 7d
Anal.
Fe:Or TiOz
(c ")
1534
1527
1520
1511
1508
1470
lMl
Liq.
Liq.
Hem.(ss)(tr.) +Liq.
Hem.(ss)fl-iq.
Hem.(ss)
Hem.(ss)
Hem.(ss)
1513
1511
1494
1486
1470
l44l
Initial Comp.
Rtn
ror
Comp. after Equil.
(v't.Vd fty anatvsis)
FeO
FezO: TiOr
85
15
Liq.
Hem.(ss)(tr.) +Liq.
Hem.(ss)fl,iq. (tr.)
Hem.(ss)fP.Brook.(ss)
Hem.(ss)*P. Brook.(ss)
Hem.(ss)fP. Brook.(ss)
1,518 80
20
1500
1496
1492
1470
Liq.
Hem.(ss)fl-iq.
Hem.(ss)*P. Brook.(ss)
Hem.(ss)fP. Brook.(ss)
75
25
1523
1520
1498
1492
Liq.
P. Brook.fliq.
P. Brook.fl,iq.
Hem.(ss)fP. Brook.(ss)
1,530
70
30
18.3 5I.7 30.0
1545
1531
1495
1492
1396
Liq.
P.Brook.(ss)*Liq.
P. Brook.(ss)fl,iq. (tr.)
P. Brook.(ss)*Hem.(ss)(tr.)
P. Brook.(ss)fHem.(ss)
1,545
65
35
20.4 M.6
1549
1534
1524
1504
1502
1486
l4l3
Liq.
P. Brook.(ss)fl,iq.
P. Brook.(ss)fl,iq.
P. Brook.(ss)-|I,iq.(tr.)
P. Brook.(ss)
P. Brook.(ss)
P. Brook.(ss)
62
38
15s5
1547
1542
1531
1492
Liq.
P.Brook.(ss)fl,iq.
P. Brook.(ss)*Liq. (tr.)
P. Brook.(ss)
P. Brook.(ss)
57.5 42.5
1551 Liq.
1547 P. Brook.(ss)(tr.) +Liq.
57.L 42.9
2 1. 5 5 8 . 5 2 O . 0
35.0
Tear,n I (conti.nueil)
Temp. of
Quenching Run
Phases Present*
(c')
Temp.
of
Run
for
Anal.
Initial Comp. Comp. after Equil
(Wt. %)
(Wt.%) (b)'analysis)
I.erOr TiOr
FeO
l-e)Or TiOz
(c)
1542 P. Brook.(ss)*Liq.
1537 P. Brook.(ss)fl,iq.
1528 P. Brook.(ss)
1492 P. Brook.(ss)
1400 P. Brook.(ss)
1553
1542
1492
1400
Liq.
P. Brook.(ss)
P. Brook.(ss)
P. Brook.(ss)
1556
1546
1542
1534
1492
1400
Liq.
P. Brook.(ss)(tr.) +Liq
P. Brook.(ss)fl,iq.
P. Brook.(ss)fLiq. (r.)
P. Brook.(ss)
P. Brook.(ss)
1556
1549
1531
1527
1400
Liq.
P. Brook.(ss)fl,iq.
P. Brook.(ss)fl,iq. (tr.)
P. Brook.(ss)
P. Brook.(ss)
1549
1543
1534
1516
1503
1492
1400
Liq.
Liq.
P.Brook.(ss)fl-iq.
P. Brook.(ss)fl-iq.
P.Brook.(ss)*Rut.(ss)
P. Brook.(ss)
iRut.(ss)
P. Brook.(ss)fRut.(ss)
1540
1518
1490
1400
P. Brook.(ss)(tr.)+Liq.
P. Brook.(ss)fl,iq.(tr.)
P. Brook.(ss)
{Rut.(ss)
P. Brook.(ss)
*Rut.(ss)
1.534
1525
1520
1517
1511
Liq.
Liq.
P. Brook.(ss)fl.iq.
P.Brook.(ss)fRut.(ss)fl,iq. (tr.)
P.Brook.(ss)fRut.(ss)
1552
1542
1534
1517
Liq.
Liq.
Rut.(ss)(tr.) fl-iq.
Rut.(ss)fl,iq.
1. 5 5 5 5 6 . 0 4 4 . 0 2 t . 3 3 4 . 7 M . O
54.0 46.0
1 9. 7 3 3. 3 4 7. 0
1,560
50
r8.7 29.3 52.0
1,545
42
1,555
16.3 23.7 60.0
T,rnr.a I (eontinued)
Temp.
of
Temp of
phasespresent*
9"":.n
rngJ(un
(c)
Initial Comp
Run
lor
(Id/r.7;
Anal
FerOr TiOz
(c')
Comp. after Equil.
(wt.%) (by analysis)
1510 Rut.(ss)fP. Brook.(ss)
1552 Rut.(ss)(tr.)+Liq.
1552
1529
1516
1511
Rut.(ss)(tr.) +Liq.
Rut.(ss)fl-iq.
Rut.(ss)*P. Brook.(ss)
Rut.(ss)*P. Brook.(ss)
3 9 . 5 6 1 s.
37.0 63.0
1520 Rut.(ss)fl,iq.
1503 Rut.(ss)fP. Brook.(ss)
20.0 80.0
1520 Rut.(ss)fl,iq
1503 Rut.(ss)fP. Brook.(ss)
1 7. 5 8 2. 5
1520
1503
1470
l42O
15.085.0
Rut.(ss)*Liq.
Rut.(ss)fP. Brook.(ss)
Rut.(ss)fP. Brook.(ss)
Rut.(ss)fP. Brook.(ss)
1543 Rut.(ss).fl,iq.
1538 Rut.(ss)+Liq. (tr.)
1510 Rut.(ss)
1469 Rut.(ss)
1420 Rut.(ss)fP. Brook.(ss)(tr.)
1543
1510
1470
1420
Rut.(ss){Liq. (tr.)
Rut.(ss)
Rut.(ss)
Rut.(ss)
1543
1509
1470
1420
Rut.(ss)
Rut. (ss)
Rut.(ss)
Rut.(ss)
1543
1505
1473
1420
Rut.(ss)
Rut.(ss)
Rut.(ss)
Rut.(ss)
130 87.0
10
90
1543 Rut.(ss)
1420 Rut.(ss)
397
1543 Rut.
1500 Rut.
1420 Rut.
0
100
FeO
Fe:Or TiOz
SYSTEMIRON OXIDE-TITANIUM OXIDE
937
In contrast to these compositional variations of crystalline phases
along essentiallystraight lines, the compositionof the liquid at liquidus
temperaturesvaries along the irregularly shapeddash curve in the lower
diagram of Fig. 2. Light solid and dashedlines are conjugationlines connecting points representing approximate compositions of liquid and
crystalline phasescoexistingin equilibrium. In order to representcompositions in the upper "binary" diagram in terms of the chosencomponents FeO.FerO: and TiOz, a method must be adopted for projecting
true compositionsrepresentedby points in the ternary system on to this
join. In accordancewith previouspractice (X{uan and Gee, 1956; X'Iuan,
1958,Philiips and \Iuan, 1958)we have projectedall compositionsalong
straight lines originating at the O corner of the triangle representingthe
Fe-Ti-O system. These straight lines are the "oxygen reaction lines"
representingchangesin total composition of condensedphasesof mixtures as these react rvith oxygen of the atmosphere. For purposes of
illustration considerthe spinel phasesdesignaredrespectivelyby a" and
f" inthe upper diagram (Fig.2). Points o" andf" can be located on the
lower diagram by the intersections of the dash-dot curves with the
join. Proper projection of thesepoints along the approFeO.FezOa-TiOr
priate "oxygen reaction lines," which are very nearly parallel to the
FezO:]-FeO
side of the compositiontriangle, allows estimation of the true
compositionsof the correspondingspinel phases.Point a" is found to be
the compositionshown by the open circle at FeO'Fe2O3,while the point
correspondingLoI" Iies further along the spinel solid solution join. The
composition of the crystaliine phasesrepresentedat the points 8", c"
and d" of the upper diagram can be obtained by analogous methods.
The most notable features of the diagrams are as follows: Liquidus
temperaturesdecreaseas the first amounts of TiOz are added to iron
oxide, from 1594" C. for pure iron oxide (Darken and Gurry, 1946) to
1524" C. in a mixture containing 12 wt"./s TiOz, with magnetite (ss) as
the primary crystalline phasein this composition range. Hematite (-"s)
is the primary crystalline phase in the composition range 12-27 wt./6
TiOz, with liquidus temperaturesdecreasingfrom 1524 to 1493" C. A
liquid of the latter TiOz content is in equiiibrium with hematite (ss) as
well as pseudobrookite(ss) at what appears as a eutectic situation at
1493' C. This is in reality a univariant situation correspondingto the
intersectionbetween the 0.21 atm. Or isobar and a ternary univariant
line in the system FeO-FezOg-TiOz
along which two crystalline phases
coexistin equilibrium with a liquid and a gas phase.However, the existing degreeof freedom has been "used up" in our choosingthe Oz partial
pressureas that of air. As more TiOz is added to the mixtures, pseudobrookite is the primary crystalline phase, and liquidus temperatures
938
T. B. MACCHESNEY AND A. MUAN
I
I
I
I
I
I
I
p
U
t
l
F
E
U
0
F
(SS)+RUTILE (SS)
PSEUOOBROOKITE
lt
I
(
Frc. 2. Diagrams presented in order to show phase reiations in the system iron oxideTiOz in air. The upper half of the figure is a diagram with the appearance of a binary
SYSTEM IRON OXIDE-TITANIUM
OXIDE
939
increase.The liquidus curve goesthrough a maximum at approximately
1550' C. and 45 wt./e TiO2, and subsequentlydecreasesto another "eutectic," characterizedby the coexistenceat 1515' C. of the phasespseuclobrookite (ss), rutile (ss), a liquid containing 58 wt.l6 TiO2, and gas of
02 partial pressureequal to 0.21 atm. With rutile as the primary crystall i n e p h a s e ,I i q u i d u st e m p e r a t u r e si n c r e a s er a p i d l y a s t h e T i O z c o n t e n ti s
increased above 58 wt./6, reaching a maximum of approximately
1830' C. for the pure TiO2 end member (the latter based on data summ a r i z e db y D e V r i e sa n d R o y , 1 9 5 4 ) .
The most notable feature of the subsolidusphase relations is the
marked influence of TiOr on the equilibrium decomposition temperature
of the sesquioxide(a-R:O3)structure of hematite to the spinel structure
of magnetite. This decomposition temperature in air goes up from
1390' C. for the pure iron oxide end member to a maximum of 1524' C.
as TiOr is added.
A summary of "invariant" situations in the system iron oxide-TiOzin
air is presentedin Table II, and *-ray diffraction data for pseudobrookite
and rutile are listed in Table III.
DrscussroN
One of the most interesting observationsmade in this investigationis
that of the very strong stabilizing influence of TiOz on the hexagonal
a-RzOgstructure of hematite relative to the spinelstructure of magnetite.
The addition of as little as lO wt./6 TiOz to iron oxide raisesthe temperature of the hematite to magnetite decompositionfrom 1390' C. to 1524"
C. in air. This observationis important for an understandingof stability
relations existing among iron oxide-Tior minerals and may also have
some important implications in technology.The changeback and lorth
from one of thesestructures to the other in responseto changesin temperature or 02 partial pressureor both has been recognizedin recent
system, showing the phases present as a function of temperature. Heavy lines are boundarl'
curves. Solid dots, open circles and crosses represent experimentally determined points on
liquidus, solidus and subsolidus boundary curves, respectively, based on data contained in
Table I. The lower, triangular diagram shows compositions of crystalline and liquid phases'
The heavy dash curve a b c d' e represents compositions of liquids along the liquidus
surface in air. Crystalline phases have compositions approximately represented by parts
of the joins FeO'Fe:Or2FeO.TiOr, FerOrFeO.TiO2 and Fe2O3'TiOrFeO'2TiOz. Solid
circles represent compositions of liquids as determined by chemical analysis. Light solid
and dash lines are approximate locations of conjugation lines connecting points representing
compositions of Iiquid and crystalline phases existing together in equilibrium in air. Light
dash-dot and dash-double dot lines are auxiliary construction lines as explained in the
text.
940
J. B
TAsr,n II.
MACCHESNEY
AND
A, MUAN
Suuuanv oF '.INVARTANT"* SrruarroNs rN nrn Sysrnu
Inox Oxron-TrOz rN Arn
Conrp. of Liquid (Wt. /6)
Phases Present
Temperature
FeO
Magn.fliq.
Magn.f Hem.
Magn.(ss)fHem(ss)*Liq.
Hem.(ss)fP. Brook.(ss){Liq.
P. Brook.(ss)fl,iq.
P. Brook.(ss)fRut.(ss)fl,iq.
Rut.fliq.
27
27
23
18
20
16
0
FezOs
73
73
65
55
35
26
0
TiO:
f C)
0
0
12
27
45
58
100
1,594
1,390
1,524
1,493
1,550
1,515
1,830
* The phase assemblages listed in reality represent
univariant situations'in the ternary
system Fe-Ti-O. Horvever, the existing degree of freedom has been "used up" in choosing
the O2 partial pressure of the gas phase, and hence the situations may be considered
"invariant.t'
years as an important causeof damage to refractoriesused in steelmaking furnaces(seefor instancea paper by Osborn, 1956).The stabilization
of the hematite structure relative to that of magnetite, causedby TiO2,
may be advantageousor disadvantageousas far as refractoriesbehavior
Ta,sr-a III.
Powonn X-nav DrllnecrroN
DATA lon PsnuooBRooKrrE eNo Rurtr,r Fe Ko RanratroN
Pseudobrookite
Rutile
(57.1Wt. /6kon Oxide,42.9Wt. Vo'tio,)
(TiOz)
d-Spacing
Intensity+
d-Spacing
Intensity*
4.92
3.49
2.75
2.46
2.41
2.23
2.20
1.97
w
3.25
2.49
w
r.87
r.76
1.67
1.64
I.JJ
S
S
w
W
w
w
2.19
1.69
1. 6 3
S
W
\'I
x{
M
M
w
w
w
w
* Abbrevialions used have the following meanings:
S: strong; M:
medium; W:
weak
SYSTEM IRON OXIDE-TITANIUM
OXIDE
H E M A T I T E( S S )
Frr Ot
WT.t
TlO2
Frc. 3. Sketch presented in order to show approximate stability relations among
sesquioxide and spinel phases in system iron oxide-TiOz at an Oz partial pressure of 10-2
atm.
is concerned.dependingon the conditions under which the refractory is
being used. Considerfor instancefirst a situation in which an iron oxidecontainingrefractory body exposedto air is subject to temperaturevariations in the interval 1300 1500' C. If pure iron oxide is present,a phase
changefrom hematite to magnetite or reversewill tend to make place
every time the temperaturecycle passesthrough 1390' C. If on the other
hand the iron oxide phase contains 10 wt./6 TiOz, the hexagonalhematite (ss) phaseis stable up to 1524oC. in air, and no phase changeswill
take place in the temperatureinterval 1300-1500' C. As Or partial pressure is reduced,the magnetite (ss) phaseswiil increaseits stability range
relative to that of hematite (ss). This correspondsto a dispiacementof
t h e t w o - p h a s ea r e a s p i n e l ( s s ) f h e m a t i t e ( s s ) i n F i g . 2 t o w a r d l o w e r
temperaturesas Oz partial pressureis decreased.Although quantitative
data are not yet available for the system iron oxide-titanium oxide at
lower levels of Oz partial pressures,data for iron oxide permits an estimate of the relations to be expected.The sketch in Fie. 3 illustrates the
942
J B. MACCIIESNEY
AND A. MUAN
situation that may be inferred to prevail where magnetite and hematite
coexistat an Oz partial pressureof approximately 10-2atm. Considering
the sametemperaturerange as mentionedabove, namely 1300 1500' C.,
pure iron oxide now remains as the single phase magnetite throughout
this range, whereas iron oxide containing I0 wt./e TiO2 begins to transform from a-Rzoa to spinel at 1380oC. Hence at this level of Oz partial
pressure(10-' atm.) the presenceof TiOz may result in damaging phase
transformationsthat did not occur in the presenceof pure iron oxide as a
constituent of the refractory body.
fn contrast to this stabilization by TiO2 of the hematite phase,which
contains essentiallyall iron as Fe8+,TiOz added to liquid iron oxide
stabilizesFe2+relative to Fe3+.This can be seenby an inspectionof the
Iower half of Fig. 2. The heavy dash curve representscompositionsof
liquids along the liquidus surface,hence temperature is a variable in addition to composition.Becausethe decompositionof oxide of Fe3+to the
oxide of Fe2+is a strongly endothermic reaction, Fe3+is stabilized relative to Fe2+as temperature as decreased,provided 02 partial pressureand
total compositionin terms of ratio TiOz to total iron oxide remain constant. The stabilizing effect of TiO2 on Fe2+becomesapparent by following the curve d b c d ein Fig. 2 from point o on the iron oxide join toward
e as TiOz content increases.The first part of this curve (ab c) is pointing
approximately toward the TiO, corner of the diagram; such a straight
line correspondsto constanf f'sz+/I'sa+ratios of the liquids. This situation
arises because the compositional efiect of TiOz shifting the equilibrium
towards Fe2+,is balancedby an equal effect in the oppositedirection of
the decreasein liquidus temperature as TiOz is added to iron oxide (see
upper part of Fig.2). As the TiOz content reachesapproximately 27
wt./6,liquidus temperatures start rising rapidly (see upper half of Fig.
2). This causesa discontinuity in the slopeof the isobaric curve at c, the
curve c-d bending sharply to the right. The rapid increasei1 psz+/ps3+
ratio of liquids with increasing TiOr content at point c is the combined result of increasing liquidus temperature and the compositional efiect of
TiOz on the equilibrium. As the TiOr content approaches that of the
maximum on the liquidus curve (see upper half of Fig. 2) , the effect of
temperature dies out, and as the liquidus temperatures decreasewith
further increasein TiOz content, the Ozisobar swingsback towards point
d. Here the TiOz content is that of the pseudobrookite-rutile"eutectic,"
and further increase in TiOz content causes a very sharp increase in
liquidus temperatures.Hence a discontinuity in the slopeof the Or isobar
occursat point d, the curve bendingsharply to the right as TiOr content
increasesbeyond this point.
The binary joins between the composition points of the stoichiometric
end members have been taken to represent compositions of solid solution
SVSTEM IRONOXIDE-TITANIUM OXIDE
943
crystals in the three series magnetite-ulvospinel (FeO'FezOr2FeO
. TiOr), hematite-ilmenite(FezOr-FeO'TiOz) and pseudobrookite(FezOr
'TiOrFeO'2TiOt in the lower half of Fig.2. This is an approximation
made for the purpose of plotting approximately the conjugation lines
giving relations between compositions of crystalline and liquid phasesexisting together in equilibrium. In the case of the magnetite-ulvospinel
solid solution, data are available to indicate that this spinel phase contains less Fe2+than correspondingto the stoichiometric ratio (Darken
and Gurry, 1946;Chevallierand Gerard, 1950).Compositionsof only two
crystallinephases,one from the hematite solid solution and one from the
pseudobrookitesolid solution field, were determinedby chemicalanalysis
in the present investigation. The observedcompositionsfall reasonably
closeto the expectedjoins. In the caseof rutile which has dissolvedsome
iron oxide in its structure, the mechanism of substitution is not well
known. \\'ork on the system FeO-TiOz (to be published) has indicated
that both Fe2+and Fe3+ions can be accommodatedby the rutile lattice'
Therefore,conjugation lines in the region whererutile (ss) is the primary
join.
crystalline phase have been drawn to the FeO'FezO3-TiO2
Suuuanv
Phaseequilibria in the system iron oxide-TiOzin air (?or:0.21 atm.)
have been studied by the quenchingtechnique.
The relationsare representedby a projection into the join FeO'FezOaTiOz of compositions actually located in the ternary system FeO-FezOaTiOz. Points of intersection between the 0.21 atm' Oz isobar along the
liquidus surface and liquidus univariant lines in the ternary system look
like eutectic and peritectic points in the apparently binary diagram which
results. The main features of the diagram as well as temperatures and
phase assemblagesat the "invariant" situations are summarized as follows: Hematite is the stable phase of pure iron oxide up to 1390' C. At
this temperature hematite and magnetite coexist in equilibrium in air,
while magnetite is the stable phase from this temperature up to the liquidus temperature of 1594oC. N'Iagnetite (ss) is the primary crystalline
phase in the composition range 0-12 wt./6 TiO:, with liquidus temperatures decreasingto a minimum of 1524' C. At the latter temperature a
"peritectic" situation exists with the following phasespresent together
in equilibrium: Spinel (ss), hematite (ss), Iiquid, gas (1e,:0.2t atm.).
Hematite (ss) is the primary crystalline phase in the composition range
12-27 wt./6 TiOr. At 1493' C. a "eutectic" situation prevails, characterizedby the coexistencein equilibrium of hematite (ss), pseudobrookite
(ss), liquid and gas (?or:0.2I atm.). Pseudobrookite(ss) is the primary
crystalline phase in mixtures between 27 and 58 wt.o/6TiOz, with the
liquidus temperature reaching a maximum of 1550' C. at 45 wt.To TiOz.
914
J. B. MACCHE,]NEYAND A. MUAN
Pseudobrookite(ss), rutile (ss), a liquid containing 58 wt.o/sTiOz and
gas exist together in equilibrium at 1515" c. in another "eutectic" situation. Rutile (ss) is the stable phasein mixtures containing more than 58
wt./6 TiO2, with liquidus temperaturesincreasingfrom 1515. C. to the
melting point of pure TiO2, approximately 1830. C.
The most significantresult of the studiesmade at subsolidustemperatures is the marked stabilization of the hematite structure reiative to
that of magnetiteas Tior is added to iron oxicle,the equilibrium temperature in air increasingfrom 1390" c. for the pure iron oxide end member
to a maximum of 1524oC. In the liquid, TiOz stabilizesFe2+relative to
Fe3+.
Acxuowr.ooGMEN't's
This work was carried out as part of a researchproject on steel plant
refractoriessystems,sponsoredby The American rron and steel rnstitute. The authors are indebted to E. F. osborn for constructivecriticism
of the manuscript.
RplnnnNcns
Axruoro, S., aNo Nec,lre, T. (1952), The TiFezOrTi2FeO5 solid solution series: Nature,
179,37-38.
ANoonssoN, S , Cor.r.6N,B., Kuvr.rmsrronNe, U., ANI Xl.scNor.r.r,A. (1952), phase analvsis studies on the titanium-oxygen system: Acta Ckem Scond.,ll,164l-1652.
Basre, E. Z. (1953), PhD. Thesis, Bristol.
Cunvarrrnn, R., aNo Grnenn, J. (1950), SynthEse de titanomagn6tites: Bul,l. Soc. Chim.
Franee, 17. 576-581.
connuccrNr, R. J. (1949), Difierences between international temperature scales of 194g
and 1927: J. ResearchNatl. Bur. Stand.ard.s.43.133-136.
CouclrN, J. P (1954), Contributions to the data on theoretical metallurgy. XII. Heats and
free energiesof formation of inorganic oxides: [/. S. Bur. Mines,lilaskington, D. C.
Bull 542.
DanxrN, L. S., a.No Gunnv, R W. (1945), The system iron-oxygen: I, The wiistite field
and related equiiibria: J. Am. Chem. Soc.,67,1398-1412.
Damnx, L. S. (1946), The system iron-oxygen. II, Equitibrium and thermodynamics of
liquid oxide and other phases:-I. Am,. Ckem..Soc., 68, 798-816.
DrVnres, R. C., ewn Rov, R. (1954), A phase diagram for the system Ti-TiO, constructecl
from data in the literature: BulI. Am. Ceram. 9oc.,33,370-372.
Enwsr, Th. (1943), Uber Schmelzgleichgewichte im System FezOrFeO-TiOz and Bemerkungen iiber die Minerale Pseudobrookite und Arizonite: Z. Angezu. Min
, 4,
394-409.
JuNrcn, E. (1936), Zur Kenntnis des Verhaltens des Titandioxycles beim Erhitzen und
seinesVerhaltens zu Fe2O3,Na2O und MgO: Z. Anorg. u. AtIg. Ctzem.,228rg7-lll.
Kewar, N., Kum, S., aNo Sasepul, S. (1954), Magnetism of rocks and solid phase transformaLion in ferromagnetic minerals: Prc. f apan. Acad ,3O,588-593.
MueN, A. (1955), Phase equilibria in the system FeO-FegOrSiO2:J. Metals, 7, September
7955;Trans. Am. Inst. Mi.ni,ng Met. Engrs.,2O3,965-976.
SYSTEM IRON OXIDE-TITANIUM
OXIDE
945
\{ueN, A., eNr oseonN, E. F. (1956), Phase equilibria at liquidus temperatures in the
system MgO-FeO-FezOrSiOi: J. Am' Ceram' 50c.,39, t2I-40.
MueN, A., eNo Grn, C. L. (1956), Phase equilibrium studies in the system iron oxideAlzO: in air and at 1 atm. Oz pressure: J. Am. Ceram. S0c.,39, 207-214'
MuaN, A (1957), Phase equilibria at liquidus temperatures in the system iron oxideAlrO3-SiO,in air atmosphere: J. Am' Cerom- Soc.,40, l2l-133.
\{r,ex, A. (1958), Phase equilibria at high temperatures in oxide systems involving
changesin oxidation stales'.Am. J. Sc ,256, 17l-2O7.
Nrcuolls, G. P. (1955), The mineralogy of rock magnetism: Adntanc' Physics,4, 113190.
osnonN, E. F. (1956), Importance of the Spinel Phase in the Steel Industry, pp. 135-150
oI "Regional' Technicol Meetings," Americon Iron oni| Steel Instihtte, New Yorh'
Puu-r.les, 8., aNo Mr.'aN, A. (1958), Phase equilibria in the system cao-Iron oxide in air
and at 1 atm. Or pressure: J. Am. Ceram. Soc.,4l, M5-454.
PosNlan, E., eNo Banrr, T. F. W. (1934), Notes on some structures of the ilmenite type:
Zeit. Kri.st., 88, 27l-280.
Pourllenn, E. (1950),Sur le comportementde I'alumine et I'oxyde de titane vis-a-vis
desoxydesd.efer:Ann. d,eChim ,5, 16+-214.
Reuoonn, P. (1926), Beobachtungen an Magnetit, Ilmenit, Eisenglanz und Ueberlegungen iiber das System FeO, FezOs, TiO:: Neues Jahrb' M i n . , A b t . A , B e i I . - B d ' . '
s4t 320-379.
R,unnorrn, P. (1950), Die Erzmineralien und ihre Verwachsungen: Akademie verlag.
Berlin.
Reuooun, P. (1953), Ulvospinel and its significancein titaniferous iron ores'.Econ. Geol.,
48. 677-687.
SosueN, R. B. (1952),Temperature scalesanclsilicate research:Am. J ' Sc',Bowenvolume'
Part 2,517-528.
VrNcnNr, E. A., lxn Wucnr, J. B (1957), Heating experiments on some natural titaniferous magnetit es: M'in. M ag., 31, 62+-688.
vrNcnNt, E. A , aNl Pu[.Lres, R. (1954), Iron titanium minerals in Iayered gabbros of the
Skaergaard intrusion, East Greenland: Geochim Cosmochim.Acta,6, l-26'