changes oxygen aluminides by vacuum induction, cold and
TRANSCRIPT
ISIJ International, Vol. 32 (1 992), No. 5, pp. 61 6-624
Changesin OxygenContents of Titanium AluminidesInduction, Cold Crucible Induction and Electron Beam
by VacuumMelting
Koichi SAKAMOTO.Katsuyuki YOSHIKAWA.Tatsuhiko KUSAMICHIand Toshio ONOYEMaterials Research Laboratories, KobeSteel, Ltd., Takatsukadai, Nishi-ku, Kobe, Hyogo-ken, 651 -22 Japan.
(Received on November25. 1991; accepted in final form on February 28. l992)
Fundamental studies on the process of melting titanium aluminides, TiAl, have been pursued from theviewpoint of contamination.
Three processes-vacuuminduction melting (VI M) with a calcia crucible, cold crucible induction melting(CCM)with a water-cooled copper crucible, and electron beammelting (EBM), -were investigated andcomparedto determine the behavior of impurity elements, especially oxygen.
Experiments using the conventional VIM method revealed that an increase in oxygen content from thecalcia crucible during melting wasunavoidable. With CCM,oxygen content did not change, while in EBM,it decreased abruptly with the evaporation of aluminum; the aluminumsegregation, however, was foundaccumulated in ingots. Thecleanest ingot of 0.02 mass"/. oxygenwasobtained using the combinedprocessesof EBand CCM(EB-CCM).Mechanical properties of the ingots were also examined, and elongation of
up to I .O'/• wasrecognized at ambient temperature.
KEYWORDS:titanium aluminide; vacuum induction melting; calcia crucible; cold crucible inductionmelting; electron beammeltingj impurity element.
1.Introduction
Titanium aluminides have recently received consider-able attention for their potential in high temperaturestructural applications-for instance, future aerospacematerials and automobile engine parts-because of their
high specific strength, Iow density and relatively goodenvironmental resistance.1~5) These aluminides, how-ever, have not yet been utilized as industrial materialsbecauseof their low ductility at ambient temperature andpoor workability.
Improvements in ductility have been sought largely
from the microstructural standpoint,6~ Io) and little hasbeen doneon the imputity element effect becauseof thedifficulties involved in the preparation of such a highpurity material. Further, knowledge on the control ofimpurity elements should lead to the advances in theindustrial application of titanium aluminide and is
therefore urgently needed.There are several melting processes for reactive and
refractory metals: vacuumarc remelting (VAR), plasmaarc melting (PAM), electron beammelting (EBM), and
vacuuminduction melting (VIM). A11 are commerciallyused nowadaysfor melting titanium and titanium al-
loys, although they have various advantages and dis-
advantages.
The cold crucible induction melting (CCM)processhas also recently received considerable attention for themelting of reactive and refractory metals. Called in-
duction skull melting, this process has a unique feature
in that a liquid metal is kept in a solid skull of the samemetal without contacting the crucible, and is stirred byan electromagnetic force.
In this paper, the processes of VIM, CCMandEBM-arefocussed on from the viewpoints of purity
andhomogeneity, which affect the mechanical propertiesof titanium aluminide. Report is madeon a comparisonof the behavior of their impurity elements, especially
oxygen.
2. Experimental Procedure
2.1. ApparatusFigures 1, 2 and 3 show schematic diagrams of the
VIM furnace with a calcia crucible, the CCMfurnaceand the EBMfurnace used in these experiments,respectively. TheVIMand CCMfurnaces are basically
similar, andconsist of several functional units: a vacuumchamber, an evacuating system, a gas introduction
system, a high frequency induction coil, a crucible andmolds.
Themain difference between the VIMand CCMis in
the nature of the crucible; the former uses acalcia crucible
and the latter a water-cooled copper crucible. Thecalcia
crucible used for the experiment is a commercial prod-uct (130 mmo.d. x 110mmi.d. x 200mmheight) and its
chemical composition is shownin Table l. It holds about
7kg of titanium aluminide charge and is poweredby a20-25kW, 3kHZ induction power source. The water-cooled copper crucible is multi-segmented and slits allow
@1992 ISIJ 61 6
ISIJ International, Vol. 32 (1992), No, 5
C(~)
~,\'l~]l .~ 8s8 .
~)
~)
C
(~
@C@CCR
Calcia crucible
VacuumchamberTundishMoldEvacuating systemCarbonSleeveInduction coil
C
Fig, l. Schematic diagram of the vacuuminduction meltingfurnace used in the experiment,
@@
@\:lLJ:
:a;
~)
R/
Molten Metal// Col•d Crucible
B*~Coil
Jo(j) Water-cooled CopperCrucible
Skull
RVacuumChamberRTundish &Heater
@Mold &Heater '=
@Evacuating System Water cooling
Frg. 2. Schematicdiagram ofthe cold crucible melting furnace
used in the experiment.
\,
View Port
Ingot
Fig.
HIL
EBgun
lL\IL\IL\ll\IL\tLILtht ol 6
/
ll tL
/1 il/1 tN,l, ,
! It c
o OaooC, O
oc, QooeQ:eO
OoC,.
Feeder
~>DP
Al
copper mold
Ti WatercooledCopperhearth
3. Schematic diagram of the electron
furnace used in the experiment.
Table
Material
CaO(Crucible)
1. Chemical(masso/o)
MgO
0,2~).3
composition
Si02
O, I~).2
beam
of the calcia
A1203
0.05~). l
melting
crucible.
Fe203
0.03~),06
CaO
Bal.
the induction fields to couple with the charge.11) Thiscrucible holds about 30 kg of tltanium aluminide and is
poweredby a 300kW, 3kHZinduction power source.TheEBMfurnace (Fig. 3) consists of two EBguns, a
vacuumchamber, an evacuating system and variouswater-cooled copper hearths and molds. The EBgunshave a two-stage differential exhauster which uses adiffusion pumpand can irradiate an electron beamstably
at a furnace pressure below I Pa. ThevacuumchamberIs evacuated with a rotary pump, a mechanical booster
pumpand a diffusion pumpand pressure in the chambercan be controlled with in the range of 10~4 to 100pa by
a methodwhich introduces various gases via a pressurecontroller.
2.2. Materials
Rawmaterials used in the VIMandCCMexperiments
were JIS No, I grade commercially pure titanium sheets
(20 > 20mmsquare x 2mmt)and commercially pure alu-
minum(lOmmip shot or 40mmblock). Table 2showsthe chemical composition of these raw materials.
As raw materials in the EBMexperiment shown in
Table 3two alloys having different aluminumcomposi-tion were prepared by cold crucible induction meltingusing JIS No. I grade commercially pure titanium sheets
(20 x 20mmsquare x 2mmt)and commercially pure alu-
minumingots (40 mmblock).
2.3. Procedure
In the standard VIMandCCMexperiments, after the
vacuumchamberwas evacuated to 8-l0~2 Pa with thethree pumps, the pressure in the vacuumchamberwascontrolled at 2.7 x 104Pa by introduction of argon gas.Melting experiments were carried out in a compositionof Ti-33-38masso/oA1 at a temperature of 1823l 893Kat 20-25kW in VIM and in a composition ofTi-33massoloAl at a temperature of 1853K at 300kWIn CCM.The change in chemical composition andimpurity elements of the molten metal was determinedby sampling with immersion type carbon sampler duringmelting. Chemical analysis wasmadeof the compositionand impurity elements on specified positions of the in-
got shownin Fig. 4 to learn the homogeneity of CCMingots cast into the carbon mold (180 mmi,d, x 250mmlong).
The EBMexperiment was carried out at an EBout-
put power of 36 and 50kWrespectively shownin Fig.
5. The hearth used in this experiment was a round-shaped, water-cooled copper hearth (150 cx 50hmm).The change in chemical composition and impurity ele-
mentsof the metal wasdetermined on specified positions
of the ingots shownin Fig. 6. At the sametime a crosssection was ground and etched to observe the so-lidified macrostructures and pool shapes.
Table 2.
Material
Ti
Al
Chemical composition of the raw material used in vacuuminduction melting and cold crucible melting.
C
0,00360.0039
Si
0.0060,002
A1
0.005Bal
Fe
0.012
O.OOl
Ca
0.00 10.00 l
o
0.0430.0019
N
0.003 l0.0012
(masso/o)
H
0.0020.00006
61 7 O 1992 ISIJ
ISIJ International, Vol. 32 (1992), No. 5
Table 3. Chemical cornposition of the raw material used in
electron beammelting. (masso/o)
Al o Fe Si
RMlRM2
41 .7
56.40.0320.043
0.0390.024
o.005
o.005
T-1
~
M-1~~
B.1~l
T-2~
M-2~
B.2~l
50mm4. Sampling locations for chemical analysis of an ingot
produced by cold crucible melting.
~~OO
,OU)cS
E~:
O=0
O
~~~00
U'u)':I
E::
OC:
OO
0.1 6 (a) mark[~l c
A+ Si
~0.1 2
0,08
A FeAl A oX NV H
0.04~-lC] ~~~o
+v ~~+v Vx VOo 10 20o 10
molten pool holding time
20
(min)
(b) mark..-.~-OO O o
o ca
o. 14
0.1
0.05
O
O\ol
O\oraw material o-_.e
o 10 20
Fig.
EB9un
Mctt' en mi~d;~~]!!1~tal/'_~;t:1L11\~
~$'_
~O,~~QUJ
~~OO
C"'O,5
EO
Fig. 7.
0,04
0,03
O IO 20
molten pool holding time (min)
Changein content of the impurities in molten titaniumaluminide in vacuuminduction melting with a calcia
crucible (Ti37masso/.). (a) carbon, silicon, iron,
oxygen, nitrogen and hydrogen, (b) oxygen andcalcium.
e
o 300kW 32.7massoloAI
(( O time (min)
Water-cooled Copperhearth
Fig. 5. Procedure of the electron beammelting experiment.
raw material o
0.02
oo
Area of molten pool Molten pool sample
~.Area of skull Icm
(Solidified metal ) Skull sample
Frg, 6, Sampling locations for chemical analysis of an ingot
produced by electron beammelting,
3. Results
3.1. Changein OxygenContent
The behavior of impurity elements in the molten
Fig.
O IO 20 30
mOlten pod holding time (min)8. Change in content of oxygen in molten titanium
aluminide in cold crucible melting (Ti-32.7mass~/.).
titanium aluminide in the VIMexperiment with a calcia
crucible is shown in Fig. 7. Impurity content did not
vary except for oxygenand calcium. Oxygencontent in-
creased to O.1-0.13masso/o with holding time, and cal=
cium content decreased to 0.02-0.05 masso/o after a mo-mentary increase at the early stage.
In CCM,oxygen content in the molten titaniumaluminide did not vary (Fig. 8), thus showing that this
process has the capability of preventing contamination
o 1992 ISIJ 61 8
ISIJ International, Vol,
from the crucible. In other words, the impurity contents,especially oxygencontent of titanium aluminide dependson the rawmaterial. CCMwasthus shownto be superiorto VIMfrom the standpoint of contamination from thecrucible.
In EBM, oxygen content decreased abruptly to0.01 masso/o at the early stage of melting irrespective ofthe initial composition as shownin Fig. 9.
(a)C]
0.03eo
36 kW50 kW
32 (1992). No. 5
~~~oo
u,u,,o
E
H.E-
O
0.02
0.01
+~:
Oo 2 4 6 8Molten pool holding time
e
3.2. Homogeneityof CCMIngot
As shownin Table 4, the ingot produced by the CCMprocess was very homogeneousbecause no segregationof composition was observed. This process obviouslygives good uniformity of composition.
3.3. Production of Low-oxygenHomogeneousIngot ofTitanium Aluminide
3.3.1. EB-CCMProcessThe results previously noted suggest that low-oxygen
homogeneoustitanium aluminide can be obtained byCCM-remelting of materials produced by EB. The fiowof this process is shownin Fig. lO. TheEB-CCMprocesshas two melting steps. Firstly, in EBM,after titaniumand aluminum melt down, the melts are held for
deoxidation and a small amountof aluminum is addedby the feeder to adjust to the desired composition. In the
next step, the ingot withdrawn in EBMis remelted by
(Ti, AI )
10 12
(min )
~oo
(o,nc:I
E
F:
C
O
EB(b)
0,04
o,02
o
• 36 kW
O IO 20
Molten pool holding time (min )Relation betweenoxygen content and holding time in
electron beammelting of titanium aluminide with (a)
O.043mass"/., (b) 0.032masso/o of initial oxygencontent.
CCM
,EB
~Z? ~oooe_o
~gun AIoot
ll\eooIt\ //I~
/ / i~ Feeder
// IltLL Illl11\ /' I o c ooo
alQr
tl~llI~ ,
llView port i g ::~>DP
HIL 1'~~
///L[~ ~\copper mold
Ti WatercooledIn90t copper hearth
,Molten Metal// Cold Crucible
"I-,~~,.
Fig. 9.
Coil
..._~- Skull
Watercooling
,Low-Oxygen&Homogeneity
TiAl Castings
Fig. 10. Schematic diagram of the EB-CCMprocess.
Table 4. Chemical composition analyzed from several positions of the CCMingot. (massolo )Al Si Fe Cu C o N H
T- lT-2M-lM-2B- lB-2
34.7
34.4
34.8
34.5
34.8
34.8
0.0060.0070.0050.0050.0050.005
O, 193O, I050.0750.0640.0670,070
o,0020.0020.002
o.002
o,002
0.002
0,0040.003
0.0040.0040,003
0.004
0.040O. I070.043
O.042
0.0430.047
0,00280.00340.00290.003 10.0038O.0030
0.00050.0022O.O0080.00100.00080.0007
61 9 C 1992 ISIJ
Table 5.
IS]J
Changes in titanium aluminide
EB-CCMprocess. (masso/o)
International, Voi,
content by
32 (1992), No. 5
Table 6. Chemical composition of the evaluated ingots fortensile test. (masso/o)
Al O C H N Al O Fe Si CRawmaterial
After
EBAfter
CCM
Ti
Al
TopBottom
MeltCast
Bal.
35.3
37.3
33.6
33.8
0.0850.019
0,0120.010
0.02 I0.024
0.0039
0,0090.0 1O0.0090.007
0.0020.00006
0.00050.0003
0.00440.0012
0.00420.0044
0.00530.0054
TAlTA2TA3
33.6
34.2
33.2
0.021
0.0170.019
O. 15O,lO
O,24
0.007 0.0090.009 0.0060.007 0.010
~Ez~:o)=a)
,,,
~~
700
600
500
400
300
o
Ti - 33.6 massaloAl
ll'll20mm
Fig. I l. Macrostructure of the central cross section of an ingotarld sampling position for tensiie test.
CCMto obtain a homogeneousingot.
Chemical compositions of the raw materials, the ingotafter EBMand that after CCMare shown in Table 5.
Homogeneoustitanium aluminide ingots having a lowoxygen content of 0.021 masso/o were produced by the
EB-CCMprocess. As the Table 5shows, the aluminumand oxygen contents after EBare different from thoseafter CCM.This is because pure titanium is added tothe titanium aluminide melt during CCMto adjust the
chemical composition.
3.3.2. Mechanical Properties of EB-CCMTitaniumAluminide
The mechanical properties of EB-CCMtitaniumaluminide ingots cast into a carbon mold (80mmi.d.
x 130mmlong) were tested. Test pieces were cut froman ingot vertical to the columnar-grained structure asshownin Fig. 11, and machinedinto smoothbar tensile
specimens with 4mmgauge diameter and 20mmgaugelength. The chemical composition of the evaluated test
piece is shownin Table 6. Nometallographically visible
O
Fig. 12.
0.1 0.30.2
Ocontent (massolo )Effect of oxygen content on tensile strength oftitanium aluminide ingots (Ti-33.6masso/.).
1.o
~::oo
=0
c:l
a, 0.5=OUJ
o
Ti -33.6 massoloAl
Strain rate : 2x 10-4 sec~1
C 1992 ISIJ 620
o
Fig. 13.
0.1 0.2 0.3
Ocontent (massolo )Effect of oxygen content on elongation of titaniumaluminide ingots (Ti33,6 massolo).
porosity was detected in the entire sample. The tensile
test was conducted at 2.0 x l0~4sec~1 strain rate at
room temperature. Measurementof elongation wascarried out using a strain gauge. For comparison, twokinds of specimens containing oxygen of 0.054 and0.29 masso/o were prepared, 5-10 specimens of eachingot were tested.
The tensile test results in Fig. 12 show the effect of
oxygen content on the tensile strength of titaniumaluminide castings, and those in Fig. 13 the effect of
oxygencontent on elongation of the alloy. Theerror barshowsminimumand maximumvalues, and solid circle
the meanvalue. Below O.05masso/o oxygen, elongationis seen to dependonoxygencontent, while tensile strength
does not have such dependence.Theseresults agree with
ISIJ International, Vol.
the data reported by Degawaet al.12)
4. Discussion
4.1. OxygenBehavior in VIM4.1,1. ThermodynamicStudy of Calcium-oxygen in
Molten Titanium AluminideTherelationship betweencalcium and oxygencontents
is shownin Fig. 14, where the temperature was kept at
l 823-1 843 and 1873-1 893 K. The two were apparentlyin equilibrium becauseof the linear relationship betweenthem plotted on log-log graph paper. The solubility
products were obtained as [o/oCa] ' [oloO] =2.70 x l0~3
and 5.14x 10~3 at 1823-1 843 and 1873-1 893K, re-
spectively from the methodof least squares.Thermodynamical considerations are madeas fol-
lows: Assuming that equilibrium phase is calcia, the
apparent reaction and its equilibrium constant K arerepresented by Eqs. (1) and (2), respectively.
CaO(s)= Ca(in TiAl) +0(in TiA1) ..........(1)
K=ac* ' ao/ac*o-" -" ' "" -"
"(2)
If K' is taken as Eq. (3), Eq. (2) is transferred to Eq. (4)
with activity coefficient fi assuming the dilute solution
K' = [o/oCa] ' [oloO] ........................(3)
K=j~* [o/o Ca] ' fo[ o/o O]/ac*o
=fc. ' fo ' K'/ac*o" " " ' --"-
"(4)
Equation (5) is given by taking logarithms of both sides
of Eq. (4)
log(K' ac.o) =10gK + Iogfc.+ Iogfo -""""(5)
Onthe other hand, activity coefficient f is represented
by Eqs. (6) and (7)
logfc* = (eg~[oloCa] +eg*[oloO]) ..............(6)
logfo = (eg*[oloCa] +eg[oloO]) ................(7)
c*Assummgthat e~~ and e8 in a composition of
O~~OO
1.o
0.3
marko
~:~8~;~t•
-
~,
Temp.1823- 1843K1873- 1893K
(~-
'/•Ca)('1•O) = 5.14 x I 0-3
32 (1992), No. 5
0.1
0.070.05
0.03
0.01
(%Ca)('1•O) = 2.70 > 10-3
Oo
OO1 0.03 0.1 0.3 1.O
olo CaFig. 14. Relation between calcium and oxygen contents in
molten titanium aluminide in vacuum induction
melting with a calcia crucible (Ti-37masso/o).
621
Ti-3338massoloAl and using the relation, Eq. (8) is ob-tained as follows.
log. fc. + Iog fo =eco'[oloCa] +eg.[oloO].. . . .. . .
.(8)
Equation (10) is obtained by the relation between eg.
and eg• shownas Eq. (9) whenMc*, Moare the atomicweights of calcium and oxygen respectively.
o = (Mc./Mo)eco" + 1/230 {(Mo Mc.)/Mo} (9)ec*
eg. =2.505eg*-6.54 x 10~3.
..........(lO)
Equation (1 l) is obtained by Eqs. (lO), (8) and (5).
10gK' = Iog(K' ac.o) +eg"([o/oCa] +2.505[o/oO]).(1 l)
Consequently, Eqs. (12) and (13) are determined fromthe relation between logK' and [o/oCa] +2.505[oloO] at
1823-1 843 K.
eg*=1.2.....
..........(12)
log(K ac*o)= ~2.8........
..........(13)
Equatron (14) rs obtamedwhenac.o = I .
K=0.0016.....
..........(14)
Equation (15) is determined by Eq. (14).
AG'=98.3kJ/mol (1 833K) ..............(15)
4. I .2. Mechanismof OxygenContaminationThe mechanismof oxygen contamination is consid-
ered as follows:
dissolution : CaO(s)H'Ca (in TiAl) +O(in TiA1).(16)
...(17)evaporation : Ca(in TiAl)H,Ca (g)T••••••••••••••••-
(18)dissolution : 11202(g)~.O(in TiAl)..................
That is, the calcia on the inner surface of the crucible
and the molten titanium aluminide are considered to
react partially, and calcia dissolves into the moltentitanium aluminide. (Eq. (16)) Calcium in moltentitanium aluminide, in contrast, vaporizes becauseof its
high vapor pressure, andoxygen is retained and increases
with holding time (Eq. (17)). Dissolution reaction Eq.(18) shows the contamination which dissolves oxygeninto the melt from atmosphere.
Considerations of the mass transfer rate of calcium
and oxygen in molten titanium aluminide follow:
d[O]/dt= (dissolution rate by Eq. (16))
+ (dissolution rate by Eq, (18)) ........(19)
d[Ca]/dt = (dissolution rate by Eq. (16))
+ (evaporation rate by Eq. (17)).....(20)
According to a study of evaporation of manganesefrommolten iron by Wanibeet al.,13) the evaporation rate is
represented by first order equations concerning the
contents of the evaporating elements. The total masstransfer rate is therefore considered to be high at low
pressure such as under this experimental condition.
Calcium compoundsare deposited in the inner surface
of the chamberas shownin Fig. 15. Thesecompounds
C 1992 ISIJ
ISIJ International, Vol, 32 (1992). No. 5
5.00K
>d-85
=cD~,::
2.5
1o,oo
caC03
50.002e
1o0.00
Ca(OH)2caC03CaOCaF2
Fig. 15.
X-ray diffraction pattern of the deposits oninner surface of a vacuum chamber in
vacuum induction melting with a calcia
crucible.
~:::oo
u)U'CE
E
F.E
44 (a)
4240~:~eA•\~38 \36 e\e34 \ 50 kW~e32
30 o
A 36 kW
z~~~
2 4 6Molten pool holding time
8 10
(min )
Temperature2500 2000
(K)1800
(X 10-6)
(b
~E~e
o
o~:*-,~
*o
:~co>UJ
1oo
50
10
60~~~.'
c,,c,) 50':I
E 40
F:
.S 30
25
(b)
e36 kWe-~_
e\\ e-
o 1O 20
Molten pool holding time (min )Relation betweenaluminumcontent andholding timein electron beammelting with (a) 41.7mass"/., (b)
56,4masso/* of initial aluminium content.
5
1
11'•~1:V"'
/'. *
lc)L,
(;)1~:/
P:~
Kusamichi et al.
A.Mitchefl et al.
--- R.G.Wardet al.
-•- This study
Fig, 16.
diff-~:usl o\ncontrol~ --l*~~*\~~\ c1.'
\ '>4
'/ e\\Qc
'l\'~4
\\'\(~
*\
,\,\
\e'/\,\\ Q^
~"I~:r/\\,~c~
'i4\\,~
were formed from calcium and H20, C02contained in
atmosphere. The contribution of dissolution reaction(Eq. (18)) is negligible, because the dissolution amountof oxygen calculated in consideration of oxygen partial
pressure and the volume in the chamber is less than20-30ppm. Therefore, the masstransfer rate of oxygenis considered to dependon the dissolution reaction (Eq.(16)), and the masstransfer rate of calcium to dependon the evaporation reaction (Eq, (17)) morethan on thedissolution reaction (Eq. (16)). Oxygen content andcalcium content were determined from the equilibriumof calcium and oxygen (Eq. (1)) described previously.Therefore, it is necessary to retard evaporation reactionby high pressure melting and low temperature melting to
~'~C' 1992 ISIJ 622
3.5 4.0 4.5 5.0 5.5 6.0
1/T (K-1 )X 104Fig. 17. Relation betweenevaporation rate ofaluminum from
molten titanium aluminide with 41.7 masso/, of initial
aluminium content and temperature of the moltenpool, and comparison with results by other in-
vestigators,
prevent contamination by the calcia crucible.
4.2. ChangesAluminumand OxygenContents in EBM4.2, I .
Changesof AluminumContentTheevaporation of aluminumis a problem in melting
of titanium aluminide by EBM,because this melting is
carried out under a high vacuumcondition caused byheat efficiency which dependson the energy loss due tothe collision betweenelectron and remaining gases. Thechanges of aluminum content in the molten titaniumaluminide is shown in Fig. 16. Aluminumcontent de-
creascd by evaporation with holding time and evapo-ration rate increased as the EBpower, that is, the melt
temperature wasraised. Thealuminumevaporation rateis thus represented by the first order Eq. (21).
ISIJ International, Vol. 32 (1 992), No. 5
~~OO
(n(b,u
E
H.E
O
~~~00
(b(oca
E
F:
,S
O
Fig. 18.
0.03
0.02
0.01
o
(a)
0,05
0,04
0.03
0.02
0,01
e 36 kWo 50 kW
-f'p.Q) . . .... :
30 32 34 36
[Al Jin TiAl
(b)
ei:
e
38 40 42 44
(mass'/• )
: [],:
O 30 40 50 60
[AI Jin TiAl (massolo)Relation between aluminum and oxygen contentswith (a) 41.7masso/oAl and 0.032massoloO, (b)
56.4masso/oAl and 0.043 masso/oO of the initial
cornpositions in electron beammelting.
d[Al]/dt= -k(A/V)[Al] .................(21)
10g([Al] t2/[Al] tl) = - k(A/V)O.4343(t2 - tl) ....(22)
Here, [Al] : alumlnum content in titaniumaluminummelt (masso/o)
k: constant of evaporation rate (m/s)
A: surface area of melt (m2)V: volume of melt (m3).
The constant of evaporation rate can be calculatedfrom Eq. (22) and the melt volume estimated from the
cross section of the ingot shown in Fig. 6. Figure 17shows the relationship between the evaporation rate ofaluminumfrom molten titanium aluminide and the melttemperature. Theresult of this study correlates quite wellwith that on the evaporation rate of aluminum fromTi-6A1~,V melt by Kusamichi et al. 14) and Mitchell andTakagi
.
15)
4.2.2. Mechanismof DeoxidationOxygencontent decreased abruptly at the early stage
of aluminum evaporation as shown in Fig. 18. Themechanismof deoxidation is considered to be as follows:
Al (in TiAl)H,AI (g)T••••••••••••••••••••..............
(23)
2A1(in TiAl) +O(in TiAl)->Al20 (g)T• • ••••
•(24)
According to the change of aluminum and oxygencontent, the main reaction is aluminumevaporation (Eq.(23)), that is, evaporation as aluminum monooxidepartially occurs. The fact that the evaporation ofaluminumcontributes to the deoxidation was reportedin titanium by Yahata et al. 16)
5. Cor]clusion
Fundamental investigations of melting processes intitanium aluminides have been pursued from the view-point of contamination for the purpose of developing
a manufacturing process. The obtained results includethe following:
(1) Oxygencontent increases to O.1-0,13 masso/o andcalcium content decreases to 0.02-0.05masso/o withincrease in holding time in a liquid state in a calciacrucible. The solubility products were obtained as[oloCa] ' [oloO] =2.70 x l0~3, 5,14 x 10~3at 1823-1 843,
l 873-1 893K, respectively.(2) In EB melting, the oxygen content decreases
abruptly to approximately 0.01 masso/o in the early stageof aluminumevaporation.
(3) In CCM,the oxygen content does not vary, andit is possible with this process to produce a super clean,
homogeneousingot of TiAl.(4) Ahomogeneoustitanium aluminide ingot having
low oxygen content of O.017-0.02lmasso/o can beproduced by the EB-CCMprocess.
(5) Below O.05masso/o oxygen, elongation is de-pendent on oxygen content, while tensile strength is notdependent.
REFERENCESl) T. Tsujimoto: J. Jpn. Inst. Light Met., 36 (1986), 162.2) R. L. Fleisher, D. M. Dimiduk and H. A. Lipsitt: Ann. Rev.
Mater. Sci., 19 (1989), 231.3) Young-WonKimandF. H.Froes:HighTemperatureAluminides
& Intermetallics, ed, by S, H. Whang.C. T. Liu, D. P. PopeandJ. O. Stiegler, The Minerals, Metals, and Materials Soc.,
Warrendale, PA, (1990), 465.4) H. A. Lipsitt: High-Temperature Ordered Intermetallic Alloys,
MRSSymposiumProceedings, Vol. 39, ed. by C. C. Koch, C.T. Liu andN. S. Stoloff, Material ResearchSoc.. PA, (1985), 351.
5) M. Yamaguchi,S. R. Nishitani and Y. Shirai: High TemperatureAluminides & Intermetallics, ed, by S. H. Whang,C. T. Liu, D.P. Popeand J. O. Stiegler, The Minerals, Metals, and MaterialsSoc., Warrendale, PA, (1990), 63.
6) H. A. Lipsitt, D. Shechtmanand R. E. Schafrik: Metal/. Trans.,
6A (1975), 1991,
7) T. Fujiwara, A.Nakamura,M.Hosomi,S. R. Nishitani, Y. Shirai
and M. Yamaguchi: Philos. Mag. A, 61 (1990), 591.8) Y. YamabeandH.Kikuchi: Bull. Jpn. Insl. Met., 30(1991), 37.9) C. McCullough, J. J. Valencia, C. G. Levi and R. Mehrabian:
Acta Metall., 37 (1989), 591.
lO) N. Fujitsuna, H. Ohyamaand Y. Ashida: ISIJ Int., 31 (1991),
l 147.
623 C 1992 ISIJ
[SIJ International. Vol. 32 (1 992). No. 5
l l)
l2)
13)
l4)
P. G. Breig andS. W.Scott: Mat. Man.Proc., 4(1989), No. I,73.
T. Degawa,K. Kamataand Y. Nagashima:Proc. Int. Symp,onIntermetallic Compounds(JIMIS-6), ed. by O. Izumi, The Jpn.Inst. Met., Sendai, (1991), 1003.
Y. Wanibe, T. Shimada, K. Ito and H. Sakao: Tetsu-to-Hagan~,
69 (1983), 1280.
T. Kusamichi, H. Kanayama.H. Matsuzaki and T. Onoye:
15)
l6)
Electron BeamMelting and Refining State of the Art, Part 2,
Bakish Materials Corp., NJ, (1989), 137.
A. Mitchell and K. Takagi: Electron BeamMelting and Refining
State of the Art, Bakish Materials Corp., NJ, (1984), 89.
T. Yahata, K. Miyoshi and M, Maeda: CAMP-ISIJ, 3 (1990),
l646.
C 1992 ISIJ 624