ISIJ International, Voi. 32 (1 992), No. 5, pp. 563-574Review

State of the Art of Superalloy Production

Application Using VIMIVARor VIM/ESRfor Aerospace and Other

Alok CHOUDHURYMetallurglcal Department, LEYBOLD

(Received on December9.

DURFERRITGmbH,Rueckinger StraBe 12, 6455 Erlensee, Germany.

1997.• accepted in final form on February 28. 1992)

Vacuuminduction melting is indispensable in the manufacture of Ni- and Co-basedsuperalloys and othersophisticated alloys becauseof their reactivity with atmospheric oxygen and nitrogen. The paper describesthe technology of melting and refining in a vacuuminduction furnace, the programmablefurnace controland metallurgical results. Thepaper also describes subsequent remelting processes like VARand ESRwhichmakeit possible to meet the very high quality requirements for aerospace applications.

KEYWORDS:VIM; VIDP; superalloys; vacuumarc remelting; electroslag remelting; ingot defects.

1. Introduction

Superalloys can be defined as iron-, nickel- orcobalt-based alloys having high strength and highresistance to oxidation properties at temperatures above650'C. The structure of superalloys is basically anaustenitic FCCy-phase with a number of secondaryphases like carbides, MC, M23C6.M6Cand gamma-prime (y'). In order to achieve a micro-structure allowinghigh strength and resistance to oxidation amongothefproperties at high working temperature the chemicalcomposition of superalloys is rather complex andincludes a variety of alloying elements of different

concentrations. Table I shows the main alloying ad-ditions and their structural effect in different types ofsuperalloys. I )

The superalloys also contain several other reactive

elements like Ti. A1, Nb, Ta, Zr, Band Hf in more or

Table l. commonranges of main alloying additions andtheir effects in superanoys.

ElementRangein wto/o

FeNi- andCo-base

Ni-baseEff ects

less high concentrations. As these elements are of pri-

mary importance for the properties of a material their

concentration must be kept within a narrow analytical

range in order to assure that the properties are re-producible. Accordingly, during melting of these alloys

every care must be taken to avoid any undesiredreactions, such as with air, increasing loss of these

elements, deterioration in oxide and nitride cleanliness.

The melting of superalloys should therefore be carried

out under vacuumor in an inertgas atmosphere.

2. Metallurgy of VacuumInduction Melting (VIM)

VIM is the most versatile me]ting process for theproduction of speciality alloys.2 ~ 6) This melting processallows the best control over the entire alloy chemistryincluding beneficial and harmful trace elements. Apartfrom this, the reproducibility of exact compositioncontrol from heat to heat is completely assured. Table 2shows, for example, the excellent chemistry control ofIN-71 8.7,8) Special features of vacuuminduction melting

can be summarizedas follows:

-fiexibility due to small batch sizes

- faster change of melt programmeCr

Mo, WAl

525

O-120-6

l9-3 O

C~I l0~.5

Oxidation and hot corrosion resis-

tance; carbides; solution hardeningCarbides; solution hardeningPrecipitation hardening; oxidationresistance

Precipitation hardening; carbides af-

fects amountof precipitate

Stabilizes austenite; forms harden-ing precipitates

Carbides; solution hardening; pre-cipitation hardening (FeNi- and Ni-base)

Carbides; solution hardening; oxi-

dation resistance

Table 2. Chemical composition control of IN-71 8based on

Ti

CoNi

Cb

Ta

0-60-20

0-5

O-12

~4

0-22

0~

o9

Element

lOO heats,

Specific Final

range rangein o/o in olo

Analytical Analytical acurracyfrequency in "lo*

in "/o* 1971 data 1980 data

C 0.02~.08 0.01~0.05 95 ~0.006 ~0.003Ti 0.80-1

. 15 0.90- I, IO 97 +0.05 +0.03

Cb 4. 75-5 .50 5.05--5 .40 99 ~O. i2 +0.08

Al 0.30~.70 0.5(}-•O 60 9i -~ OV•~ +o02

* Analytical frequency e,g. 950/0 mean*, t.hq.t the c.h_ emical com-position of 95"/o of the melts are within tllc final ['ange.

563 1992 ISIJ

ISIJ International, Vol. 32 (1992), No. 5

PRIMARYMELTiNG

e REFINE

e PURIFY

e CHEMCONTROL

SECONDARYMELTING

l ll I

I

e VIRGINCHARGEo SCRAP(REVERT)

e SCRAP(PROCESSED)[EF/AODHYDROMET]

Fig• 1.

FINALPROCESSING

ee

REMELTFOR:

MACROSTRUCTURALCONTROLMICROSTRUCTURALCONTROL

R~MELT/INVESTMENTCAST

ATOMIZE

REMELT/ATOMIZE

VAR

ESR

VADBR

CASTPRODUCTS

e EQUIAXED (+ HIP)

e DS-POLYCRYSTALLINEe DS-EUTECTICS

e DS-SINGLECRYSTALe DUO-STRUCTURE

(DIFFUSIONBOND)

NEAR-NETSHAPES

e pOWDER+ HIP

, pOWDER+ HIP+ FORGEe POWDER+ EXTRUDE-

ISOTHERMALFORGEe LASER-SPRAY

WROUGHTPRODUCTS

EXTRUDEFORGEROLL

e BILLET. BARSHEET-e [EXTRUDE.FORGE.ROLL

TOSHApE]

Processing routes for products cast from VIM-ingots or electrodes.

- isolation of the melt from air contamination: gasatmospherecan be selected at will, resulting in low loss

of reactive elements due to oxidationachievement of very closely reproducible composi-tional tolerances

- control of pressure above the melt: thus control ofpressure-dependent chemical reactions

- precise temperature control

- excellent homogenization due to inductive stirring

- removal of undesired trace elements with high vapourpressure

- removal of dissolved gases like hydrogenandnitrogen

- removal of oxide inclusions

- Iow level of environment pollution from dust output.Figure I shows the various uses and applications of

VIM.8) The tap weights range from a few kg to 15tdepending on whether the furnace is designed for pre-cision casting applications, for casting of forging ingots

or for the production of electrodes for further remelting.

Melting in an induction furnace showssomespecific

differences as comparedwith electric arc furnace orwith ladle metallurgical practice. In spite of a relatively

intensive melt circulation the desired slag/metal exchangereactions are limited. Dueto the characteristic movementof the bath the slag is transported towards the crucible

wall, where it solidifies and becomesinactive.

The metallurgical processes such as desulphurization

and dephosphorization have, therefore, only a limited

effect. An additional disadvantage lies in the fact that

the rammedcrucible lining undergoes a higher erosion

due to slag attack. For this reason, the main metallurgi-

cal refining is primarily limited to the purely pressuredependent reactions, such as C-, O-, N- and H-removal,and the evaporation of trace elements with high vapourpressure as in the case of As, Cu, Pb, Bi, Te, SbandSn.

2.1. Slag/Metal-Reactions

The refractory lining of the crucible of a vacuum

Table 3. Typical refractories for VIM-crucibles for melting

of superalloys.

Refractory

type

Max. melt

temperature('C)

Density(g/cm3)

Thermal shockresistance

MgOAl203

MgO-spinellAl203-spinell

Zr02

l 800

l 900

1900

l 900

2300

2.8

3.7

3,8

3.7

5,4

GoodGoodPoor

Relatively goodPoor

Table 4. Reduction pressures of various crucible materials

with Fe-melts (0,lo/o C) at 1600 'C.

Crucible material Pco in Pascal

CaOZr02MgOAl203Si02

40. O

133.3

533.2533.2

81313.0

induction furnace has a numberof important aspectsrelevant to refractory/metal- and refractory/slag-interac-

tions. Someof the significant parameters influencing the

performance of the crucible are:

- reaction of metal or slag with the crucible lining

- erosion of the lining material

- chemical- and physical stability of the refractories.

Table 3 shows the VIM crucible refractories com-monly used today. Whenselecting the lining refractory

the behavior of the oxides under vacuumat elevated

temperature must also be considered. The main pre-requisite here is that these oxides remain stable whenin contact with the melt under the given operating

conditions, such as melt composition, pressure andtemperature. As the dissociation and reduction of the

C 1992 ISIJ 564

ISiJ International. Vol.

oxides result in the formation of CO, the reduction

pressure of these oxldes is an important indicator fortheir suitability. In Table 4 the reduction pressure ofvarlous crucible materials at 1600'C is listed. It is

apparent that CaOis the most stable crucible lining

material. But CaOis very sensitive to humidity andtherefore not suitable for crucible lining in industrial

furnaces. It is further evident from the table, that Zr02.MgOand Al203 are quite suitable for refractory liningin avacuuminduction furnace, with Zr02 the moststable

of the three.8) Apart from carbon, superalloys generallycontain relatively high concentrations of A1 and Ti. Thereduction of the lining material by Al and/or Ti mustthus also be considered. Arecent theoretical investigationof reaction kinetics regarded that an alumina cruciblewill be most suitable for melting superalloys.9)

Slag/metal reactions for desulphurization causepractical difficulties in vacuum induction furnaces.Desulphurization of the melt with a reactive basic slagin a vacuuminduction furnace is only carried out in

specific cases becauseof heavy slag attack of the cruciblelining. A crucible with rammedlining exhibits lowresistance to the refinery slag, which mayalso containsfluxes such as CaF2' Schlatter7) and Darmaralo) achiev-ed good desulphurization results in vacuuminductionfurnaces with steels and high temperature alloys by using

a lime bearing basic slag. Desulphurization primarily

occurs via the stable compoundsMgSand CaS.Desulphurizing agents, Iike Ca, Mgand Cecommoninthe secondary metallurgy of steel, have also been usedfor desulphurization of superalloys. With the additionof O.070/0 Ce in well-deoxidized melts of Inconel 901,sulphur-contents of less than 8ppmcan be achieved. Inthis case the sulphur is removed by the formation ofstable sulphur in CeS2/CaS.11,12)

2.2. Evaporation Reactions

For the production of critical aircraft engine com-ponents the concentration of certain trace elementssuch as Ag, Pb, Bi and Se must be kept very low in

superalloys,13) Figure 2shows the deleterious influence

32 (1992). No. 5

of these elements on stress rupture properties of certainsuperalloys.14'Is) Most of these harmful trace elements,fortunately, have relatively high vapor pressures and cantherefore removed by evaporatlon during melting in

vacuuminduction furnaces. Figure 3shows, howvarioustrace elements behave under vacuum. In NiCr-meltelements such as As, Sn and Sb cannot be removedvia

the gas phase at a pressure of 7 mbar, whereas the

elements Cu, Pb. Se and Bi are reduced under vacuumto concentrations well below 50ppmin a relatively shorttime. 16)

J(D

CL

~~cQ~5oc

1,o

o.8

0.6

0,4

0.2

(a)

200

Jc:_ 160o~.-

Q) 120*=~~:Dcc 80

40

(b)

Fig.

oBi

Te

Bi

Pb

Se

10 20 30Impurity Content, ppm

PbSe

Ag

O IO 20 30 40 50 60 70 80 90IMPURITrCONTENT,ppm

2. Effect ofvarious trace elements onthe 649'C/690 MPastress rupture of Inconel 718.

a) Iron b) 80'/, Ni and 20"/o Cr1600 'c As

Sn

0,1 0,1Sb

As:~s~o ~:~o

z zQ o_

h h0,01

H ~ o,olHz zLU uJ Se

o oz zo oo Bi oCu

o,Ool 0,001Te

Pb Bi

o 50 1oo 150 o 50 1oo 150

TIME (min) TIME (min)

Fig. 3. Evaporation of trace elements in a VIM-furnace at 7mbar.

565 =Q 1992 ISIJ

ISIJ International, Vol.

2.3. GasReactions

The solubility of hydrogen and nitrogen in Ni-base-superalloy melts depends on chemical composition,temperature and pressure and is governed by Sievert's

law:

[o/oH]=KJ~!rr2""""'

"""""(1)

[oloN] =K~~:N2""-"'

"""-"(2)

Hydrogenis easily removedduring vacuuminductionmelting as this element strictly follows Sievert's law. Therate of hydrogen removal is considerably higher duringCO-boiling and stirring of the melt. Hydrogen-concentrations of less than Ippmare already attainedby the time the charge is completely molten.

Nitrogen solubility in Ni-base-melts principally obeysSievert's law, but the presence of various alloying ele-

ments, such as Cr, V, Ti, A1, Zr, etc., which appreciablyreduce the activity of nitrogen influences significantlyits solubility. Becauseof the formation of brittle nitrides

or carbon nitrides in superalloys containing strongnitride-forming elements like Ti, the nitrogen contentmust be kept to the lowest possible level. The removalof nitrogen by vacuummelting wastherefore the subjectof several investigations. 17~ 19) Simkovich19) investigatedthe influence of composition, pressure and temperatureon nitrogen-removal of Fe- and Ni-base-alloys. Theresults showed that under similar vacuum treatmentconditions alloys containing up to 15o/o Cr containedcomparable residual nitrogen content as the Cr freealloys. This is due to the fact that Cr is not a strongnitride-former and accordingly the rate of nitrogenremoval from a Cr containing melt under vacuumis

comparableto that from Cr free melts. Theconcentrationof nitrogen in alloys containing Cr along with strongnitride-forming elements A1, NbandTi wasmuchhigher.

Another important factor for the removal of nitrogenunder vacuumis the concentration of dissolved oxygenand sulphur in the melt. Since oxygen and sulphur aresurface active elements in Fe- and Ni-melts, they tend

IOO

~;~

~ 75

~~zo:~:~:

50

~

~ 25~

SUMMERYOFLrrERATURE

32 (1992), No. 5

o0,00 0,05 0,25O, IO 0,20O, 15

OXYGENCONTENTlN wr. %Fig. 4. Effect of oxygen on the reaction rate of nitrogen

removal from liquid iron under vacuum.

C 1992 ISIJ 566

to concentrate near the gas/metal-interface and this limitsthe transportation rate of nitrogen from the melt intothe gas phase. Figure 4 shows the effect of dissolved

oxygen on the rate of nitrogen-removal of Fe-melts.7)

For effective nitrogen-removal and to achieve very lownitrogen content in superalloys like Inconel 718, thefollowing practice is advantageous:

- selection of charging material with very low initial

nitrogen- and sulphur-content

- effective and prolonged boiling of the melt (CO-formation) also during the period of melt down in

order to reduce the dissolved oxygen content byC-O-reaction; subsequently, full deoxidation of meltwith Al along with Ar-purging at low pressure

2.4. Deoxidation via GasPhaseThe melt can be deoxidized in the vacuuminduction

furnace through the gas phaseaccording to the followingreaction:

[C] + [O] = {CO}gas"""""

"""""(3)

As the reaction product is a gas, the oxygen-contentat a given carbon-content is directly proportional toCO-partial-pressure at the top of the melt:

olo[C] x o/o[O]=Kx Pco """"""""'(4)

This deoxidation process, the so-called Vacuum-Carbon-Deoxidation (VCD)-Process, for Fe- and Ni-melts is ofdistinctive advantage for two reasons:

(1) the deoxidation product isagas andescapes frommelt without contaminating the melt with inclusions,

and(2) the reaction is highly pressure dependentand can

therefore be controlled by selecting proper pressureduring vacuuminduction melting.

The theoretical equilibrium value of [C] x [O]-product for Ni-melts is in the range of 10~7 to l0~8 atl.33Pa.20,21) For usual carbon content of 0.050/0 in

various superalloys, the theoretical oxygen content atequilibrium is in the range of 2ppm. In industrial meltsof Inconel 718 the oxygen content after VCD-treatmentat a pressure of I.33 Pa is in the range of 10 to 15 ppm.Thus, in actual industrial melts the C-O-reaction doesnot reach equilibriums owing to the inhibition ofCO-nucleation. Additionally, a continuous fiow of ox-ygen into the melt takes place due to the dissociationof the refractory lining of the crucible and leakages ofthe vacuumchamberat low pressure. Deoxidation of themelt by the C-O-reaction proceeds in two steps7,8).

- boiling phase, i.e, formation of CO-gaswithin the meltprimarily by heterogeneous nucleation with intensive

bath turbulence as a result of the formation of this gas

- desorption phase, in which no CO-bubblesform withinthe melt, with CO-formation taking place only at themelt/vacuum-interface.

The greatest reduction of carbon and oxygen takesplace during the first stage, the boiling phase. Duringthe desorption phase there is only a reduction ofcarbon, but not of oxygen, primarily owing to the

dissociation of the crucible lining. The melt musttherefore be deoxidized, finally by a strong deoxidizing

ISIJ lnternational, Vol.

agent like Al and the reaction product A1203must beremovedduring refining of the melt.

3. Improvements in Oxide Cleanliness

Oxide cleanliness plays a decislve role in superalloyswith their extremely high-strength properties at higherservice temperature. Figure 5 shows the deleterious

[h] 200

175

150

125

a)

Jo 100::

'lsoc 75

50

25

Fig.

- Udimet 500 Cast

- In-100 (PM)

UDIMET500 (Cast)(670'C1172 MPa)

IN 100 (PM)(732• C1689MPa)

50 250 ppm[O]150OxygenContent in ppm

5. Influence of oxygen on rupture life.

32 (1992), No. 5

infiuence of the oxygen content and thus of the oxidecleanliness on the rupture life of superalloys.15) An ap-preciable removal of the oxide-incluslons is achievedif there Is a llquid reactive slag in contact with themelt, which is capable of absorbing oxide Inclusions.

Active slags are normally not employed in a vacuuminduction furnace. This meansthat the oxides can onlyprecipitate at the crucible wall, which naturally, Iimits

the degree of removal of these inclusions.22)

As mentioned, the oxygen-pick up from the dis-

sociation and erosion of the crucible lining is animportant factor in the final cleanliness of the melt.

Consequently, it is of paramount importance to select

the right ceramic lining. Apart from that, Iong dwellingtime of the melt in the vacuum induction furnaceshould be avoided.23)

4. Process Technology, Furnace Conception and Auto-mation

The refractory crucible can be rammedor brick lined

with suitable ceramic material, preferably with Al2031MgO-spinell. After rammingand before sintering thecrucible should be dried at approx. 200'C with an electric

heater. The furnace is then charged with material,preferably low carbon unalloyed steel scrap, andheated-up by induction to approx. 800'C under air

atmosphere. After holding the charge at 800'C for about2hr, the furnace is closed and evacuated. The charge is

subsequently heated under vacuumto approx. 1700'Candheld for several hours for final sintering. Themeltingcycle for superalloys in a VIM-furnace consists of sev-eral steps of charging, melting down, refining, chem-

Charglng

1ooo

(T'

~)E

CL

o1)ECQ

~

1oo

ooo

1soo143

1oo

120

101ooo ;~

~':t_J

~90

s ~1 o= o-

cT~*~~

&~o)c;~

E:

o(D 60:~

.1 H500

30,oi

,oo

L~IDJneniSi

Temp.-measurementsarnplingArgon-stirring

1430

120

90

60

300

TSU:

o

!-oa,

J::

oo,:,

J,,Q~,

J

Fifst

Charge

r-lllIlll

o

Installed Power

r'~1r'Meftlng j ii

power ' " 'l II

r'~' 11I[

I Ii ii

I Ii ii

l lj ij

l li ii

l I' ii

lI. i

- . _.1

I,1

Ii

llll

Pressure II IIIl

[l

ll li

ll ii

ll ii

I'

1

Mekingdown

1'

'~ Temperature.1'1'("'~ ~ ~ ~ ~ ~ ~ ' . ~,.1"

'~;:Tlt'~"~

-. 1' '~'1"'~'~ I cb 9'.

o":: o'~:l J~o ~~oj ~~ ~~J~~3le -oR:

i ~q, ~io'1:~

ii ; i

~,, '

f 81 tl ~ I I I~l

l i!gl I-'- E i 'f~'~ i !l x' 'I' ' cql ilI ':" ' TS' :rsTI l!lJt~i L1

'l i jjl

II

I , r'~1li I ii' i i i~~~~~ !i i iE i . , I :1

2 3Ar-purging and/or stirring

Superheat and Refining

Alloying

4Time [h]

and ControlCasting

Fig. 6. Typical melting cycle for Inconel 718 in the VIM-furnace.

567 ~") 1992 ISIJ

l~lJ International, Vol. 32 (1992), No. 5

l (Il~ll __LL~~i~~= 7

~==1l~

~f Ir T~~l r~Tl I:1:~~r_,*.

_*-_~.::;:_

_~~~~~,*_~-_~:SJ J---JSt Js~1-

2-

istry correction and casting. A typical melting cycle for

Inconel 718 is shownschematically in the pressure-tem-perature-time-diagramme of Fig. 6. The basic chargingmaterial is comprised of alloying elements, except thereactive constituents. Further charging should be car-ried out under vacuumthrough a bulk charger locatedat the top of the furnace. During the melting down- andrefining-period relatively high outgassing of the chargingmaterial and a strong boiling due to C-O-reaction takesplace. It is necessary to adjust carefully the furnace

pressure in order to maintain a well-controlled boil,

which is very important for the removal of dissolved

gases, including oxygen. The boiling intensity decreasescontinuously as the oxygen content of the melt is re-duced. After completion of this refining period the meltis finally deoxidized and other reactive elements, Al,Ti, Zr, B, are added where necessary. After thoroughmixing and final chemistry check the proper pouringtemperature is adjusted and the melt is poured.

A typical vacuuminduction furnace is shown sche-matically in Fig. 7. As described earlier, the oxygencon-tent of superalloy melts at the end of the refining and de-oxidation period is usually in the range of 10 to 15 ppmbut it has been found that in the casting such as ofelectrodes for further remelting, the oxygen content is

higher. This is due to the fact that during casting via atundish ceramic contamination of the melt from the

eroded tundish material, caused by heavy turbulencetakes place. As the cleanliness of a VAR-ingot depends

on the cleanliness of the electrode, it is of vital impor-tance to use a clean electrode to get a cleaner ingot.

Anewdevelopment in this field of vacuuminductionmelting is the VacuumInduction Degassing and Pour-ing (VIDP)-furnace. The VIDP-furnace-design has anindependent melting and treatment unit, which allows

use of different casting technique modules (Fig. 8). Incomparison with a classical chambertype VIM-furnace(Fig. 7), the VIDP-furnace does not have any vacuumchamber. The furnace body itself is vacuum-tight.Accordingly, the melt chamber volume and thechamber-surface of a VIDP-furnace are muchsmallerthan those of a classical VIM-furnace. Similar to

Fig. 7.

Schematic diagrammeof a VIM-furnace.

Fig. 8. VIDP-concept with combined ingot/electrode-casting

and horizonal continuous casting.

VIM-furnaces, the casting of the melt can be carried outunder vacuumor under inert gas atmosphere. All hy-draulic systems for tilting and the flexible watercooled

power cables are outside the vacuumchamber, whichmakesthe equipment safer. During pouring the furnacebody with the coil and the crucible is tilted around avacuum-tight pouring tunnel. Themelt is poured into apreheated launder, which transfers the liquid metal intothe casting chamber. The casting chamber can bedesigned to incorporate ingot molds, Iadles, continuouscasting or atomizing equipment. Whereas the VIDP-furnace itself is comparable to the classical VIM-furnacein respect to melt treatment under vacuum, the special

feature of the equipment is the transport of the meltvia a relatively long launder. From the metallurgicalpoint of view, the launder is an important part ofthe VIDP-furnace for the cleaning of the melt fromnon-metallic oxide-inclusions. The launder is equippedwith two slag barriers. A ceramic filter just before theoutlet can also be inserted, if desired (Fig. 9). Thelaunderis conceived in such a way, that a quasi laminar fiowof the melt is assured. Accordingly, the removal ofoxide-inclusions by flotation during casting is optimized.

Recent results have shownthat all oxides with diametcrs

C 1992 ISIJ 568

ISIJ International, Vol, 32 (1992), No. 5

VI DP-Furn ace

Filter ~A TapphgAAlterholder Filter S,8g Barrier Stag Barrler

Ill

~~

llb IIa I

lNozzle ~B

Ceramlc launder

cross-sectlonA-B

Frg. 9. Schem2ltic of a VIDP-launder.

Ceramlc launder

I Tapplng Zonell Caimlng ZoneIll Castlng Zone

of more than 15~mhave been removedduring castingT'ia a launder.24)

The use of a programmable logical control systemin combination with process computers for the auto-mation of the vacuum induction melting improvesmelt reproduclbility. With such a system, increasinglystringent metallurgical requlrements of the future canalso be met. Closer analysis tolerances can be achievedby the acquisition, storage and appropriate processing ofall the data necessary for the metallurgical process.

The basic concept involves combining the individual

modules, such as the opened and closed loop controls

of the furnace, computational modules and thealloying- and charge-calculation operation, in order to

automate the vacuum induction process to varyingdegrees up to 100o/o, depending on requirements. Suchsystems have been in use for some years now as asuccessful means for monitoring and controlling the

vacuumsystem or for recording melt parameters, for

detection purposes and for the diagnosis of faults.

Using charge- and alloying calculations, it is quite aneasy matter nowadaysto achieve the required chemicalanalysis at minimal cost. In the ideal configuration, the

process computer is directly linked to the computer ofthe analysis system so that immediately after the analysis

has been made, the additions can be calculated andpossibly even automatically weighed and charged to the

melt. If the alloy calculation system also features a chargecalculation facility, the charge materials and quantities

(scrap) can also be optimized at the start of the process.The complete calculation system facilitates accuratecomputatlon of the alloying elements starting from theinitial quantlties and ending with the adjustment values

569

for fine correction at the end of the fining period.

5. Remelting Process

Growing demands for superalloy cleanliness andstructural homogeneity in particular, cannot be met by

vacuuminduction melting and casting alone. This hasled to the practice of remelting of already melted andrefined materlai (known as electrode) in a watercooled

copper crucible. Principally three proccsses, vacuumarcremelting, electroslag remelting and electron beamremelting, combinedwith a controlled solidification in awatercooled crystallizer have been developed and in

commercial use. The newest development in this respectis the application of a plasma torch for remelting

purposes. Figure 10 showsthe various process combina-tions for the production of superalloys.

5.1. VacuumArc Remelting (VAR)The vacuumarc remelting process was the first com-

mercial remelting process for superalloys. It wasused in

the late 1950s to manufacture materials for the aircraft

industry. Theprimary feature of vacuumarc melting andremelting is the continuous melting of a consumableelectrode (manufactured in a vacuuminduction furnace)

by meansof a dc arc under vacuum.Themolten materialsolidifies in a watercooled copper mold.

The basic design of a VAR-furnace has remainedlargely unchangedover the years; however, significant

advances have been madein the field of control andregulation of the process with the object of achieving

a fully automatic melting procedure. This, in turn, has

had a decisive posltive influence on the metallurgical

properties of the products. The manufacture of homo-

1992 ISIJ

ISIJ International, Vol. 32 (1 992). No. 5

PlasmaCold

Crucible Melting

and

Atomization

l VIM

L~~L~_*P/M I~~J LY~~~L

Ceaniiness + Structure Structure

,~, [i]l

LVAR

EBCHR PCHA

C!eanliness + Structure

Plasma

Cold Crucible

Melting

and

Casting

Pressingor

HIP

Hot Forming

working

Structurai Part

Fig. lO. Various process combinations for superalloys.

geneous ingots with minimal segregation requires care-ful machining of remelting parameters. Of these, themelting current density and melt rate have the greaterinfiuence on the melting bath geometry and conditionsof solidification.

Theprimary benefits of melting aconsumableelectrode

under vacuumare:

- Removal of dissolved gases such as hydrogen andnitrogenMinimizing the content of undesirable trace elementswith high vapor pressure

- Improvementof oxide cleanliness

Achievementof directional solidification of the ingotfrom bottom to top in order to avoid macrosegregationand to minimize microsegregation.Oxide inclusion removal is optimized because of the

relatively short reaction paths during melting of the hotelectrode end and becauseof a good drop dispersion in

the plasma arc. Oxide removal is achieved by chemicaland physical processes. Less stable oxides or nitrides arethermally dissociated or are reduced by carbon presentin the alloy andare removedinto the gas phase. However,in superalloys and in high-alloyed steels the non-metallicinclusions, e.g, alumina and titanium carbonitride, arevery stable. The removal of these inclusions duringremelting takes place only by flotation. The remaininginclusions in the solidified ingot are small and evenlydistributed in the cross section. The solidification

structure of an ingot of a given composition is a functionof the local solidification time and the temperaturegradient at the liquid/solid interface. To achieve adirected dentritic primary structure, a relatively hightemperature gradient at the solidification front must bemaintained during the entire remelting period. Accord-ingly, the growth direction of the cellular dentrites

corresponds to the direction of the temperature gradient

or the direction of the heat flow at the momentofsolidification at the solidification front. Thedirection ofthe heat flow is always perpendicular to the solidification

front or, in the case of curved interface, perpendicularto the tangent. The growth direction of the dentrites is

a function of the metal pool profile during solidification.

C 1992 ISIJ 570

Fig. Il. Top longitudinal section of cast -50"/. HCl/50*/.ferric~hloride grain etch.

The pool depth increases with melting rate. Thus thegradient of the dentrites, with respect to the ingot axis,

increases with melting rate. In extreme cases, the growthof the directed dentrites can cometo a stop. The ingot

core then solidifies nondirectionally in equiaxed grains,

which leads to segregation and microshrinkage. Even

ISIJ International, Vol.

in the case of directional dentritic solidification, themicrosegregation increases wlth dentrite armspacing. Asolidification structure with dentrites almost parallel tothe Ingot axis (as shownin Fig. Il) yields optimal results.

However,this is not always possible. Agoodingot surfacerequires a minimumenergy input and, accordingly, aminimummelting rate. Optimal melt rates and energyinputs dependon ingot diameter. This means, that the

necessary melting rate for large-dlameter ingots cannotbe maintained for crystallization parallel to the ingot axis.

Theingot diameter for several superalloys like Inconel718 or Waspaloy is therefore limited and is less than500mm.Figure 12 showsmelting rates for various steels

and alloys as a function of ingot diameter. These areempirical values that were obtained from experience in

operation. Thesemelting rates gave low microsegregationwhile achieving acceptable surface quality.

In spite of directional dentritic solidification, suchdefects as tree ring patterns, freckles and white spots canoccur in a remelted ingot. This can lead to rejection ofthe ingot, particularly in the case of superalloys. Treering patterns can be identified in a macroetchedtransverse section as light-etching rings. They usually

represent a negative crystal segregation. Tree ring pat-terns seem to have little effect on material properties.

Theyare the result of a wide fluctuation in the remeltingrate. In modernremelting plants, however, the remeltingrate is maintained at the desired value by precise controlduring operation, so that it exhibits no significant

12

I~ Io'~E

,~o)8_(D(Ts6

~42

Fe-base

Superalloys (Ni-, Co-base)Inc. 718Waspaloy

1OOO400 600 800Ingot diameter [mm]

Fig. 12. Typical melt rates for different grades.

32 (1992), No. 5

fiuctuation.

Freckles and white spots have a muchgreater effect

on material propertles than tree ring patterns, especiallyin superalloys. Both defects represent an important causeof the premature failure of turbine discs in aircraft

engines. Freckles are dark-etching circular or nearlycircular spots that are generally rich in carbldes orcarbide-forming elements. The formation of freckles is

usually a result of a high pool depth andmost frequentlyin a rotating liquid pool. The liquid pool can be set in

rotation by stray magnetic fields during remelting.Freckles can be avoided by maintaining a low pool depthand by avolding the dlsturbance of magnetic fields

through the use of a coaxi_al current supply.White spots are typical defects in VAR-ingots. They

are recognlsable as light-etched spots on a macro-etchedsurface. They are low in alloying elements, e.g. titani-

um and niobium in Inconel 718. There are several

mechanismsthat could account for the formation ofwhite spots25).

- Unmelted dentrite clusters dropping from the castelectrode

- Particles which falls from the crown which formsaround the edge of the mold

- Particles disintegrating from the solidified edge (shelDof the ingot top.A11 three of the above mentioned mechanisms,

Individually or combined, can be considered as sourcesof white spots. This indicates that white spots cannotcompletely be avoided during vacuumarc remelting. Tominimize the frequency of occurence of these defects, thefollowing conditions should be observed during remelt-ing:

- Use the maximumacceptable metal rate permittedby the ingot macrostructure

- Usea short arc gap to minimize crown formation andto maximlze arc stability

- Use a homogeneouselectrode free of cavities andcracks.

As mentioned earlier, the melt rate is an importantfactor in the quality of the ingot macrostructure. AmodernVAR-plant is therefore equipped with a load cell

system to measure the weight of the electrode of aparticular interval of time. Theactual values of the meltrate are compared by computer with the desired set

StlrringColl CurrFrequ8ncy

Voltage,PulsesResistSwing

Melt Rate

MeltlngCurrent/Pawer

Hot Tapping ValueStart Value

Meit PhaseValue

Slope 1

Slope 2 Slope 1Slope 3

Slope 3Slope 1 Slope 2 Hold

Slope 2 \Alternative Sel ection

Slope 3 Slope 1\

Slope 1f Slope 3

!Start

Slope 2 Haid

TimeBaseStart-Up

Weight BaseMelt Phase

571

Hot Tapping Value

TimeBaseHot Topping

Fig. 13.

Control pararneters and setpoint functions.

(('-') 1992 ISIJ

IS]J International, Vol.

values. Any difference between the measuredmelt rate

and the desired value is eliminated by the properaccommodationof the power input. Figure 13 showsthe

melt rate and the melting current at start-up, duringsteady-state melting and during hottopping. Start-upand hottopping are usually controlled based on time.

The melting phase is controlled based on weight. Thehottopping begins whena preselected residual weight is

reached. A computer controls the melting parameters,which are stored in the form of recipes in the computer.

5.2. Electroslag Remelting (ESR)Thetheory behind this process wasknownin the 1930s,

but a general breackthrough for this process took morethan 30 years. Intensive studies carried out in the SovietUnion. Germany,United Kingdom, Austria and Japanafter World War 11 madethe use of the electroslag

remelting possible on a production scale.26~28) In

contrast to the vacuumarc remelting (VAR), the re-melting in the ESR-processdoes not occur by striking

an arc under vacuum.The ingot is builtup in electroslag

remelting in awatercooled moldby melting aconsumableelectrode immersedin a superheated slag usually undernormal atmosphere. The heat required is generated byan electrical current (usually ac) flowing through theliquid slag, which provides the e]ectrical resistance. Asthe slag.temperature rises abovethe liquidus temperatureof the metal, the tip of the electrode melts.

Themolten metal droplets fall through the liquid slag

and are collected in the watercooled mold. During theformation of the liquid film, the metal is refined andcleaned of contaminants, such as oxide-and sulphide-particles. The high degree of superheat of the slag and,partially, of the metal favors the slag/metal-reaction.Melting in the form of metal droplets greatly increasesthe metal/slag-interface surface area. The intensivereactions between metal and slag result in a significant

reduction in sulphur and non-metallic inclusions. Thereduction of non-metallic inclusions is better than withthe VAR-process. The remaining inclusions are verysmall and are evenly distributed in the remelted ingot.

Another feature of the ESR-process, as in vacuumarcremelting, is the directional solidification of the ingotfrom bottom to top.

The special feature of the ESR-process is that acontinuous transport of liquid metal through the slag

takes place. Durlng this transport, the slag and the metalcompositions chang according to the kinetic andthermodynamic conditions. To perform its intendedfunction, the slag must have somewell-defined proper-ties, for example:

Its melting point mustbe lower than that of the metal.It must be electrically efficient.

-Its composition should be such that the desiredreactions like removal of sulphur and oxides areensured.

-It must have suitable viscosity at remelting tempera-ture.

Slags for the electroslag remelting are usually

composedof calcium fluoride (CaF2), Iime (CaO) andalumina (A1203). Magnesia(MgO), Titania (Ti02) and

32 (1992), No. 5

silica (Si02) are also added, depending on the alloy tobe remelted.

Oneof the primary advantages of the ESR-processis the good desulphurization of the metal. The final

desulphurization is determined by two reactions. Thefirst is the metal/slag-reaction, in which sulphur is trans-ferred from the metal to the slag:

[S] + (CaO)= (CaS)+ [O]........

..........(5)

The second reaction is the slaglgas phase-reaction. Inthis case, the sulphur absorbed by the slag is removedby the oxygen of the gas phase in the form of gaseoussulphur oxide:

(CaS)+3/2 02gas= (CaO)+S02g.~" "" "" "• -

(6)

It is evident that a saturation of the slag with sulphurdoes not take place, therefore, the desulphurization ca-pacity of the slag remains intact throughout the entire

remelting process. With a highly basic slag (CaO/Si02greater than 3), more than 800/0 of the sulphur can beremoved.

As mentioned previously, the ESR-process is usuallycarried out under a normal air atmosphere. Oxidationof the metal is unavoidable. Oxygencan be transferredInto the metal in several ways:

- Oxidation of the electrode surface above the slag bath

- Oxidation of the slag surface of elements with var-iable valences such as iron and manganeseRust attached to the electrode surface

- Transfer of oxygendue to desulphurization accordingto Eq. (5).

This results in losses of easily oxidizable elements suchas titanium, aluminium and silicon. During remeltingof Fe-based alloys this oxidation is usually compensat-ed by a continuous deoxidation of the slag, preferablywith aluminium. In case of superalloys such as Inconel718 or Waspalloy with relatively high titanium andaluminium content the deoxidation of CaF2-CaO-Al203-slag does not help muchto assure the desired

narrow analytical range of these elements in the remeltedingot. The introduction of oxygen into the superalloymelt is primarily caused by the ESR-slag constituentsFeO, Si02 and Al203.

Due to the very low FeO- and Si02-content of theslag the reaction between titanium and these oxides is

of minor Importance, so that the actual oxidation oftitanium takes place according to the equation:

3[Ti] +2(Al203) =3(Ti02) +4[Al] ............(7)

By applying the law of massaction, the equilibriumcondition can be expressed in a simplified mannerby theequation:

Table 5. Al- and Ti-contents after electroslag remelting ofInconel 718.

Electrode ESR-ingotTop Bottom

Ti in ol.

Al in olol .05

0.57l .04

0.521.04

0.55

O1992 ISIJ 572

ISIJ International, Vol. 32 (1 992), No. 5

[Ti] 3_ K

(Ti02)3..........(8)

[Al]4~ (Al203)2 ""'

In order to avoid aTi-loss in the melt, the Ti02-contentshould be adjusted for given Al203-content of the slagto achieve equilibrium for a given [Ti]/[Al]-ratio of thesuperalloy (balanced slag system). Table 5 shows the

content of Ti and Al in an ESR-ingot remelted with abalanced slag but without any deoxidation of the slagduring remelting. All superalloys with [Ti]/[Al]-ratio of

3or less can be remelted with a balanced slag withoutany problems. For higher [Ti]/[Al]-ratio in the alloy,

such as A286, a deoxidation, even of the balanced slag,

wil] be necessary during remelting under air atmo-sphere.29)

The solidification of an ESR-ingot is governed by the

same thermodynamical laws as in the case of VAR-process. The melting rate here is also the essential fac-

tor influencing the pool depth and, accordingly, theingot structure. In order to achieve a primary structurefree from any macrosegregation, a directed solidification

must be assured. The melt rate depending on ingotdiameter is of similar order of magnitude as in theVAR-process. Here also the ingot diameter of somesegregation-sensitive super-alloys is limited to max.500mm.As in a VAR-ingot, structural defects, e.g, treerings, patterns and freckles, can occur in an ESR-ingot.The reason for occurrence of these defects and meansfor their avoidance are the sameas described in theVAR-process. It is, of course, important to note thatwhite spots normally do not occur in an ESR-ingot, asthere are no crowns above the solidifying ingot and noshelf formation. The relics of unmelted dentrltes fromthe consumableelectrode can, of course, fall down, butthey have to pass through the superheated slag and aremelted. Significant advanceshave been madein the last

few years in plant design, especially in the area of processcontrol and coaxial power supply. Figure 14 showsschematically a modernESR-furnace. Figure 15 showstwo modernESR-furnaceseachwith two melting stationsfor remelting of superalloys. The ESR-processhas been

'extensively automated in a manner similar to thataccomplished in the case of VAR-furnaces. Theremeltingprocess takes place under fully automatic control. Table6showsa comparison between the features offered bythe ESR-processand those by the VAR-process.22)

6. Summary

The increasing demandfor very narrow analytical

range of the chemical composition, very low concentra-tion of harmful trace elements and g~ses along with

Fig.

12345614

~. ~*~~f**~~;~~i~*',~*~"' s**^** ~^~s

23

45

11

12

RANIDrive System 7 Pivoting DriveElectrode RAM 8 ElectrodeXY•Adjustment 9 Mold AssemblyLoad Cell System 10 Coaxial BusTubeSliding Contact 11 BasePlate

4BusTubes 12 Multi Contacts

Schematicof an ESR-furnacewith astationary mold.

Fig. 15.

Modern ESR-furnace.Alloys, Hereford/UK)

(Courtesy of Wiggin

573 o 1992 ISIJ

ISIJ International, Vol. 32 (1 992). No. 5

Table 6. Specral features of the Automatic Melt Control (AMC)system

Features VAR ESR

Completeautomatic operation of the furnace by preset melt profiles

Automatic start-up and hottopping by preset current-time profile

Automatic start-up and hottopping by preset power-time profile

Melt rate control maintaining a preset melt rate profile

Alternatively selectable current control maintaining a preset profile

Alternatively selectable powercontrol maintaining a present profile

Preset profiles for arc gapcontrol (voltage, pulse rate); arc gap is controlled by slave processor board fully

integrated in AMCPrdset profiles for immersion depth control (resistance, swing)

Computer-calculated basic speed of feed based on melt geometry and melt rate

Preset profiles for stirring coil (current, reversal time)

Preset profile for operating pressure (optional)

Monitoring of selectable limits for melt rate and melting powerFree selection of four melt parameter signals onto four analog outputsColor-plotter software package for four free selectable melt parameter signals

Function keys to influence programmeexecution such as hold, run, step-forward, break-off, start

hotto ppingPrinted melt record containing process data as well as special events; events are printed with elapsed

actual timeVideo terminal designed for simple dialog for data entry; graphic display available

Established melt profiles are stored on hard disk and on floppy disk for backup purposesIntegrated diagnostic maintainance programmeHost computer interface with ownCPU

xX

XX

X

Xxx

XXX

X

XXxX

>

Xx

X

X

xXxX

X

>

XXx

an exceptional cleanliness in sophisticated superalloys

makes the vacuum induction melting of these alloys

absolutely necessary. Vacuuminduction melting is aversatile process and allows independent control of

temperature and pressure. A stirring of the melt, either

inductive or by Ar-bubbling influences the masstrans-

port to a great extent, which accelerates the desired met-allurgical reaction favourably. A relative new devel-

opment of the classical VIM-furnace is the VIDP-furnace with a long launder. Pouring the melt via laun-

der by creating a quasi-laminar flow greatly improvesthe degree of oxide removal.

In spite of the large metallurgical potential of VIM-furnaces the solidification of the ingots or electrodes

cannot be controlled, which results in unacceptable in-

homogeneity of the primary structure. Hencea second

step of remelting with controlled directed solidification

in awatercooled coppermold is inevitable. Thesecondaryremelting of VIM-cast electrodes is carried out either in

a vacuumarc furnace or an ESR-furnace. With propercontroll of the process, especially of the melt rate,

macrosegregation-free ingots can usually be producedby both means.

l)

2)

3)

4)

REFERENCESM, J. Donachie, Jr: Superalloys Source Book. American Soc.

Met., Metals Park, Ohio, (1984), 3.

C. Le Chevallier et al,: Rev. Met., (1980), 791.

H. Burghardt and G, Neuhof: Steelmaking, VEB-Verlag ftir

Grundstoffindustrie, Leipzig, (1982), 391.

W. Esmarch: Wissenschaftliche Ve,'bffentlichung Siemens, lO(1931), 173

5)

6)

7)

8)

9)

lO)

l l)

l2)

13)

l4)

l5)

l6)

l7)

l8)

l9)

20)

21)

22)

23)

24)

25)

26)

27)

28)

29)

W. Zschintzsch: Die neue Giesserei, 39 (1952), 381,

F. Neumannand F. Hegerwaldt: Stahl Eisen, 94 (1974), 53.

T. Schlatter: J. Met.. 24 (i972), 17.

J. W. Pridgeon et a/.: Metall. Treat., (1981), 261.

K. Yeumand D. R. Poirier: Proc. 9th Int. Conf. VacuumMetallurgy, SanDiego, Ca., (1988), 68.

F. N. Darmara: J. Met., 19 (1967), 42.

R. E. Bailey et a!.: Proc. 3rd Int. Symp. of Superalloys, SevenSprings, Pa., (1976), 109.

M. F. Rothman:Trans. VacuumMetall. Conf., (1977), 49.

R. T. Holt andW.Wallace: Int. Met. Rev. Review203, TheMet.Soc., (1976), 1,

W. B. Kent: J. VacuumSci. Tech., Il (1974), 1038.

P. Hupfer: Fachberichte Huttenpraxis Metallverarbeitung, (1986),

773.

P. P. Turillon: Trans. 6th ICVM, (1963), 88.

G. K. Sigworth Trans. Mel. Soc. CanadianInst. Mel., (1977), 104.

W. A. Fischer: Arch. Eisenhtittenlves., 33 (1960), l.

A. Simkovich: J. Met., 253 (1966), 504.

K. W.Krone: Ph. D. Thesis of the Technical University of Aachen,Germany,(1969).

M. Gard et a/.: Proc. Con. Int. Sur Les Applications DesTechnique Strasbourg, (1967), 177.

A. Choudhury: VacuumMetallurgy, ASM-Int.. Metals Park,Ohio, (1990), 37,

W.H. Sutton and E. Manner: Proc. 6th ICVM,SanDiego, Ca.,

(1979), 340.

A. Choudhuryet al.: Paper presented at ATS-SteelmakingDay,Paris, (1990).

A Mitchell: Proc. Vac. Met. Con. on Special Metals Melting,Pittsburgh, (1986), 55.

W, Holzgruber and E. Pl6ckinger: Stah! Eisen, (1968), 638.

A. Choudhuryet a/.: Stah/ Eisen, (1968), 1193.

A. Choudhuryet al.: Arch. Eisenhtiltenwes., (1971), 299.

A. Choudhuryet al.: Proc, Vac. Met. Con. on Special MetalsMelting, Pittsburgh, (1986), 141.

C 1992 ISIJ 574