EFFECTIVE ELECTROSLAGREMELTINGOF SUPERALLOYS

H. J. Klein and J. W. Pridgeon

Stellite Division, Cabot Corporation, Kokomo, Indiana

ABSTRACT

In order to take full advantage of electroslag remelting, it is necessary

to optimize the most important properties, such as structure, chemical
uniformity and inclusion control by careful choice of the remelt parameters,
since these properties essentially determine the acceptability of the final
part. The relationship between these properties and possible remelt vari-
ables such as melt mode, power level, slag chemistry and furnace atmosphere
for various types of superalloy are reviewed and discussed. From the pro-
cessing viewpoint, the differences between molten and solid slag starting
are discussed as well as the effects of ingot structure and ingot surface
on forgeability. The potential differences and similarities between the
electroslag remelting process and the vacuum arc remelting process are
considered and discussed.

In the past few years, electroslag re- showing electroslag versus a vacuum arc
melting has gained tremendous populari- furnace. In general, all furnaces are
ty as a valuable processing technique somewhat similar to this. However,
for superalloys, tool steels and other there are a number of variations. The
high performance matIsrials. The rea- mold can be short and movable resulting
sons for this are numerous and they in a net ingot withdrawal system or the
will be discussed in this paper. short mold can be stationary and the
ingot can be withdrawn. Figure 3 dem-
To set the stage, it is important that onstrates these concepts. The advan-
we review the history of electroslag tage of a withdrawal system is the
remelting. An adequate review is given simplicity of the mold and control of
by Hoyle(l) in the book "Electroslag the post-solidification ingot cooling
Refining." Table 1 depicts the histori- rate. Insulation or cooling can be ap-
cal development of the process. The plied to the portion of the ingot that
gestation period was long and torturous has been withdrawn. Another variation
going from principles written down in is the electrical power supply. Most
1892 to actual research in 1935 to full furnaces used for superalloy melting
scale production in a number of compa- utilize single phase alternating cur-
nies in the 19601s. rent. However, a number of direct cur-
rent vacuum arc furnaces have been con-
The growth of electroslag remelting verted to DC electroslag furnaces. DC
furnaces is shown in Figure 1, not furnaces do have advantages and dis-
including Russian facilities. The advantages as will be discussed later
number has grown from five in 1964 to in the paper.
approximately 65 in 1972. Of these 65
furnaces, approximately 10 are used Figure 4 shows a number of electrical
for superalloys. variations that can and are being used
to power electroslag remelting furnaces.
Equipment The twin electrode configuration in the
Armstrong setup is also known as the
Figure 2 is a general schematic diagram bi-filer technique. Superalloys known

B-l
 TABLE 1. HISTORY OF ELECTROSLAG
MELTING

1. Basic principle described in book by Nickoloi Slavyanov in 1892.

2. First research done on small scale by Armstrong in 1928 on use of resistance-

heated slag to aid melting.

3. First real definitive work done by Hopkins in 1935 and patent obtained in 1940.

4. Exploitation of ESR invention in late 1950's through Kellogg Company without

great success.

5. Hopkins joined Firth-Sterling in 1959 and in 1962 exclusivity was lifted and a
more open licensing policy was established.

6. Lectromelt took license in 1966 to build equipment.

7. Union Electric took license in 1966 to make steel.

8. In 1967, Stellite started melting solid solution $uperalloys in production and

to date has melted 25 million pounds of HASTELLOYalloy X, and 50 million pounds
total of superalloys and corrosion resistant alloys.

TABLE 2. SUPERALLOYSMELTED
BY ESR

Solid Solution Allovs

a. Nickel-base
HASTELLO?alloys B, C, c-276, G, S, and X
"Inconel" alloys 600, 601, 625, and 706
b. Cobalt-base
HAYNES*alloys No. 25, 188 and 31
UAYNESSTELLITE*alloy No. 6B

Precipitation Strengthened Alloys

a. Nickel-base
llInconel" alloys 718, 713, and IN-100
Rene' 41, 77, and 95
"Udimetlf 700
"Astroloy"
Rolls-Royce C-263
AF2-1DA

* "Hastelloy", "Haynes", and "Haynes Stellite" are registered trade marks

of Cabot Corporation.

B-2
to have been melted by ESR are shown TABLE 3. ESR PARAMETERS
in Table 2.
VA gsJ Parameter
Another variation in electroslag fur- Power
naces is the starting technique. X AC
Either liquid slag starting is used or X X DC
some version of a dry slag starting Polyphase
X
technique such as arcing against a
X Armstrong
mixture of slag and chips or slag,
chips and exo-thermics, or resistance Electrodes
heating of a bar or coil that is sur- X Variable shape
X
rounded by slag. Licluid slag starting Change-over
X
has advantages because it gives a good Multiple
X
butt end surface free of gas and is
more reproducible than dry starting. Start
X X Arc
ESR Parameters X Resistance
X Molten slag
Many investigators argue that ESR has
an advantage over VAR because there Special Techniques
are more parameters or degrees of free- X Continuous withdrawal
dom in the electroslag process. Con- X Electromagnetic
X
versely, one can argue that there are stirring
so many parameters that it is impossible X Tubular ingots
to get reproducible control of the pro- X Composites
cess. Table j presents an example of X Multiple ingots
the degrees of freedom of ESR versus
VAR. In melting of superalloys, -the X Slag refining
main parameters to control are:
X Vacuum purification
1. Slag chemistry

2. Power input - melt rate With alloys :ontaining reactive ele-

ments such as Zr and Ti, some produc-
3. Electrode diameter/mold ers add Zr02 and Ti02 to the slag to
diameter get the benefit of mass action, i.e.,
Ti + 2O+TiO2 where the equilibrium
In melting superalloys, slag chemis- constant
tries are usually based on the calcium
fluoride, calcium oxide, alumina system. &m2.
Calcium fluoride by itself is satis- aTi aO
factory except it does not possess
high enough electrical resistivity to In other words, the addition of TiO2
obtain good I2R heating. The addition to the slag retards the oxidation of
of CaO is mainly for economical rea- the Ti by driving the reaction to the
sons, i.e., CaO is cheap and also for left. This is basically a naive ap-
desulfurization. Up to 30% can be proach and mechanistically is wrong in
added to calcium fluoride without many eases. In summary, for effective
raising the mel-ting point. The elec- electroslag remelting of superalloys to

trical resistance of the CaF2-CaO take place, the slag should have in
system is essentially the same as CaF, general the following characteristics:
thus, if higher resistance is needed,
the addition of a third constituent 1. It should be essentially non-
such as Al203 is made which effectively reactive with the alloy.
raises the electricaIL resistance with- 2. It should form a smooth layer upon
out raising the melting point with solidification on the mold wall to
additions up to 20%. provide a good ingot surface.
R-3
3. It should react with non-metallic g = aceeler&33n due to
inclusions in the melt to remove gravity
them from the metal.
4. It should possess enough electrical r = radius of the inclusion
resistance to obtain efficient
power utilization during melting. Pm = density of the molten
5. It should be inexpensive. metal

The power input, amount of slag and pine = density of the inclusion
the electrode diameter/mold diameter
(fill ratio) control the pool shape = viscosity
1
and thus affect segregation. Increasing
the -fill ratio, slag volume and voltage For a typical melt rate of 1,000 lbsfhr
independently or in concert lowers the for a 20" diameter ingot one would ex-
pool depth and leads to less segrega- pect to lose inclusions having a radii
tion. greater than 8 microns by flotation.
One must remember that Stoke's law as-
Refining Reactions sumes a spherical in~clusbon, and also
no stirring and no surface tension
In applying the Electroslag remelting effects.
technique successfully to superalloys
an understanding of the potential re- Inclusion removal at the molten metal -
actions and refining mechanisms, is molten slag interface c50uld also take
necessary, These may be divided for place as a droplet contacts the molten
convenience into flotation or mechani- metal. In the AC mode a vibratory
cal removal and chemical and electro- mechanism called the "electro capillary
chemical reactions. There has been a effect" has been proposed (2) to account
significant .amount of work done in the for the elimination of inclusions. This
past in an attempt to determine not was sug:ges$ed to be the result ,of a
only the mechanisms of inclusion re- change in the surface en'ergy as the cur-
moval but at which oi" the reaction zone rent direction reverses causing the
or zones the electrode tip, tlhe metal interface to vibrate and thus removing
droplet, or the molten-metal-molten the inclusions.
slag interface, inclusion removal takes
place. Removal of inclusions by flota- The importance of inclusion removal by
tion could take place either in the the flotati'on mechanism has been point-
droplet during transfer to, the mpltee ed out by Me.dovar ('4) and a number of
metal pool or after it becomes part of other authors (5). They stress the im-
the metal pool., It, has been concluded portance of 'maintaining a flat, shallow
in the past that. there is insufficient pool in order to decrease the probabtlity
time for removal of inclusions from the of Gnclusions beZag trapped by the grow-
droplet as it descends to the molten ing dendrites.
metal pool. One can approximate, using
Stoke's law-, as &mm in %igure 5 the The seeo,nd sit& for. inclusion removal
ssze. of an inclusion which mi;ght be is the electrode tip. During droplet
expeActe8 to float out from the molten formation the greatest surface to vol.-
metal pool during electroslag remelting. ume ratio is seen. This increases the
The appropriate Stoke's law equation is,: probability of an inclusipn coming into.
co.ntact with the molten slag and thus
being absorbed into it.. Vachugov (6)
apd reCently .Loyd. (7): has showi? that
about 5.0%of the inclusion reduction
wh-e Tee
"cakes place- at the elec-trade tip.. In
d..
1 .-_ the Be.rmina"l velocity this ease they were eansidering removal
of silicate inQu&lons in, m41d steel.
There is, however, some questions as to
the mechanism of inclusion removal at in the electrode reductions of from
the electrode tip; that is, whether it 20-60s are still seen. Table 4 shows
is by dissolution of the inclusions the initial and final sulfur contents
into the molten metal and subsequent for a number of superalloys which have
reaction with the slag or by absorption been electroslag or vacuum are remelted.
due to mechanical contact as mentioned It is noted that the sulfur levels in
previously. The former seems to be the ESR ingots are in general below the
quite probable for a low melting point levels reported for the primary melt
inclusion such as sulfides. It has and the VAR ingots. Due to the low
been proposed however by Kay, et al initial sulfur level of superalloys it
(8) that in fact due to the high cur- is questionable whether the proposed
rent densi-ty at the electrode tip suf- reaction for the elimination of sulfur
ficient temperatures are attained such from the slag, thus maintaining a low
that the dissolution of alumina in- sulfur activity in the slag, is of im-
clusions occurs as shown in Figure 6. portance in this case. Even if total
The inclusions in the electrode are removal of sulfur from the ingot was
much larger than in the region which accomplished there would be an insigni-
has been mol-ten indicating that dis- ficant increase in the activity of sul-
solution takes place at the electrode fur in the slag.
tip. The current density in Kay's
ease was 223.9 amp/cm2 while the normal TABLE 4. SULFUR CONTENTSFOR
current density seen in the large scale SUPERALLOYSELECTRO-
pr;duction electrodesis about 10 amp/ SLAG REMELTEDAND
em . 'Thus, the same reasoning of a VACUUMARC REMELTED
high current density resulting in ex-
cessively high temperatures at the Alloy Melt Mode s.
electrode tip may not apply to produc-
tion ESR units. HAYNESSTELLITE Electrode 0.010
alloy No. 6B ESR ingot 0.005
The dissolution of oxide inclusions at VAR ingot 0.010
the molten metal tip implies that the
degree of inclusion removal of oxides HASTELLOY Electrode 0.002
as well as sulfides is the dependent alloy B ESR ingot 0.002
upon chemical reactions through bulk VAR ingot 0.002
diffusion. Thus the removal and re-
distribution of inclusions is control- HAYNESalloy Electrode 0.008
led by the following slag metal reac- No. 31 ESR ingot 0.007
tions. VAR. ingot 0.009
(MxOy) inc 2 x lMl + y [Oli Modified Electrode 0.005
llAstroloyfl ESR ingot 0.003
x [Ml + y LOI 2 (MsOy) slag VAR ingot 0.005
and
(MxSy) inc z x [Ml + S
During electroslag remelting there
iSI + (O=) slag 2 (S=) *[ 01 exists the possibility of losing easily
oxidizable elements or the introduction
As can be seen the removal of oxygen of inclusions into. the ingot by chemical
by dissolution may be inhibited by the reactions. In considering this it
introduction of oxygen as sulfur is should be remembered however that it is
removed. desirable to lose elements such as sili-
con, titanium or aluminum during ESR if
For steels it has been reported that they were in the electrode initially as
sulfur losses from 40 to 85% can be inclusions. However, the elimination of
realized with the use of basic slags. losses by oxidation of aluminum and
For superalloys where great pains are titanium or other reactive elements from
taken to insure a low sulfur content liquid solution during ESR is of prime
B-5
importance. The oxidation and sub- to enable accurate calculation of the
sequent loss of reactive elements is appropriate activities.
determined by the oxygen potential in
the slag. Two of the most important In the foregoing discussions we have
variables affecting the slag's oxygen neglected potential electrochemical
potential are: slag composition and reactions, i.e., assumed that the re-
the atmosphere above the slag. The melting was accomplished using the AC
presence of impurities in the slag mode and that no rectification occurred.
which are easily reduced by the re- Some electroslag remelting however is
active elements in the superalloys carried out using converted vacuum arc
will most certainly cause a loss of remelting units with DC power source and
reactive elements. In addition, it thus one must consider in this case the
is possible for elements which have potential electrochemical reaction.
more than one valence state to act as Hoyle, et al (l), Medovar, et al (4),
an oxygen pump by transferring the oxy- and Cameron (9) have reported that a
gen from the atmosphere to the molten significant amount of rectification
metal interface as shown in the follow- takes place during AC ESR; therefore,
ing equation for iron oxide: even in this mode it is necessary to be
aware of potential electrochemical re-
2(FeO) + O2 2 (Fe203) actions. Medovar, et al (4) has re-
ported differences in degrees of refine-
Slag-Atmosphere Interface ment depending upon the melt mode use.
They report that the greatest decrease
(Fe20j) -I- Fe 2 (3 FeO) in oxygen content is seen with reverse
Molten MetalSlag Interface polarity, i.e. electrode positive and
that the removal of sulfur is enhanced
Similar reactions can be proposed for by the AC mode. Holzgruber (10) sug-
the oxides of manganese or titanium in gests that electrochemical effects due
the electroslag remelting of superalloys . to current density considerations may
It is thus necessary to eliminate all become insignificant in production ESR
impurity oxides from ESR slags which units. He in fact, as shown in Figure 7,
are to be used for melting superalloys. reports a reversal in the oxygen content
Another source of oxygen is surface of ESR ingots for decreasing melt mode
oxidation of electrodes just above the as the size of the ingot increases. He
slag bath. The use of an argon cover ascribes this to the difference in cur-
over the slag will eliminate electrode rent density and thus an insufficient
oxidation and reduce the po2 above the EMF to induce electrochemical reactions.
slag bath thus eliminating or mini- He concluded the reversal was due to
mizing the oxidation. It has been the electrochemical reaction only taking
shown by Medovar (4) that the loss of place at the electrode tip for large
reactive elements is essentially eli- diameter ingots with relatively small
minated by employing an inert atmos- fill ratios. Electrochemical reactions
phere. such as the elimination of oqgen, how-
ever, depend upon the dissolution of
In addition to this, the proper choice the inclusions as discussed previously.
of the remelting flux can minimize It must be realized however that harmful
elemental loss since the loss is con- reactions such as the introduction of
trolled by the following reactions oxygen into the ingot during electroslag
remelting do not depend upon dissolution
x [Ml + y LO] 2 (M,Oy) of inclusions and most in fact are en-
hanced if inclusions are formed.
Thus if the activity of the oxide in
the slag is increased by the addition The presence of electrochemical reactions
of M 0 the thermodynamic driving force and their effect in electroslag remelting
is lzsgened. At this time the theory has been demonstrated by Klein (11) for
of slags is not sufficiently developed HASTELLOYalloy X. The material used

B-6
for the investigation had been previously the molten metal pool resulting in an
electroslag remelted in the AC mode and increase in the number of inclusions
was essentially inclusion free. Upon in the ingot as exhibited in Figure 8.
remelting in the AC mode no change in In superalloys the presence of large
reactive elements or inclusion content amounts of deoxidents such as aluminum
was seen but for the DCSP mode a signi- essentially eliminates the large over
ficant loss of aluminum and silicon potential reported by Mitchell (12) for
and an increase in oxygen content of the iron by keeping the oxygen activity at
ingot was observed as shown in Table 5. a very low level in the molten metal
The loss of aluminum was a result of bath. This implies that the difference
oxygen being deposited at the anode, in melt rate and pool depth as reported
the molten slag - molten pool interface, for the DC SP modes may be a result of
and then reacting with the aluminum other reactions. One possible source
and/or silicon in the molten metal by is the oxidation of the species electro-
the following reaction: chemically plated out at the cathode by
the oxygen in the air. This reaction
(O=) 2 [O]+ 2e anode taking place at the electrode tip
could result in a very efficient use of
2 [AS + 3 lOI f (A1203) the energy from this exothermic re-
action and thus increase the melt rate.
The inclusions thus formed are either In addition the oxidation of aluminum
dissolved into the slag thus resulting in the molten metal bath also is highly
in a loss of aluminum or are trapped in exothermic and would contribute to the

TABLE 5. RESULTS OF CHEMICALANALYSIS FOR SELECTED

ELEMENTSAl, Si AND 0 COMPARED TO ELECTRODE
VALUES OF 0.15, 0.45 AND 0.002 RESPECTIVELY

Electrode Diameter
b-3/4 Inches 6 -Inches
Melt Slag Weight
Mode Position 10 14 10 A- 18
Al 0.14% 0.15 0.08 0.15 0.10 0.13
Butt End Si 0.41 0.45 0.39 0.43 0.42 0.45
0.002 0.002 0.002 <o-o02 0.002 0.003
AC
Al 0.10 0.08 0.08 0.10 0.11 0.11
Hot Top Si 0.44 0.42 0.21 0.42 0.43 0.42
0 0.002 <0.002 0.002 0.003 0.002 0.003
Al <O.Ol <O.Ol 0.03 0.07 0.11 0.11
Butt End Si 0.23 0.025 0.21 0.39 0.27 0.39
0 0.020 0.019 0.007 0.004 0.004 0.002
DSCP
Al <O.Ol <O.Ol <O.Ol 0.05 0.06 0.08
HJt Top Si 0.20 0.017 0.23 0.36 0.35 0.38
0 0.014 0.013 0.009 0.003 0.005 0.002

B-7
heat input of the system and a deeper inclusions are more effectively removed
pool. in ESR than in VAR and moreover the re-
maining ones are significantly finer
Thus it is extremely important in elec- and more randomly disposed. For the ESR
troslag remelting to consider the pos- and VAR ingots 86.1% and 49.3%, respec-
sible electrochemical reactions and their tively, had inclusion ratings of D l/2 T
resultant effects on the superalloy in- or less. This result was significant at
got. The choice of the improper remelt the a. = 0.01 confidence level.
practice will result in an ingot of
significantly inferior quality. Secondly, the pool shape of the ESR in-
got is significantly shallower than that
Hot Workabilitv of a correspondingly VAR ingot resulting
in a vertically oriented macrostructure
One of the major reasons for the suc- which is more suitable for side forging.
cess of electroslag remelting is the The result of pool depth measurements
improvement in the hot workability of (14) for ESR and VAR experimental ingots
the ESR ingot over that of VAR or direct with similar melt rates are given in
forged ingots. This difference has been Table 6 for HASTELLOYalloy B, F and
shown to be significant as illustrated HAYNES alloy No. 31. As can be seen
in Figure 9 by Kelley (13) for llUdimet't the ESR ingot has a much shallower pool
alloy 700 in an investigation sponsored and thus a more vertically oriented
by the Air Force. A doubling of the macrostructure. The effect of slag
hot working range over that of the direct weight, melt mode and fill ratio on pool
forged or VAR ingot was seen. The test depths in electroslag remelting was
which has been used to measure hot shown by Klein (11) as seen in Table 7.
workability with great success at As can be seen increasing the slag
Stellite is the Gleeble test. The weight results in decreasing the pool
Gleeble unit can be programmed such that depth. An increase in fill ratio while
the specimen being tested experiences not d,ecreasing the total pool depth re-
the same thermal cycling as in forging. sults in a much flatter pool with a
At the appropriate time a high strain more vertically oriented maerostructure.
rate tensile test is performed with the The shallowest pool is obtained using
reduction insarea of the sample being the DCRP melt mode while the DCSP re-
taken as a measure of its hot work- sults in the deepest with the AC mode
ability, A similar hot workability in- being as expected inbetween.
crease was seen by Pridgeon, et al (14)
for a number of other alloys including The other two important faetors effect-
HASTELLOY alloy B and F and the Gleeble ing the hot workability of superalloys
curves are shown in Figures 10 and 11. are the sulfur content and the chemical
homogeneity, i.e., segregation. It is
The increased hot workability as a re- well known that the presence of sulfur
sult of electroslag remelting is due to can cause incipient melting and thus
a number of f&tors. In ESR the slag poor workability. The use of ESR re-
above the molten metal pool may absorb sults, in general, in lower sulfur con-
ineLusions as previously discussed and tents which should result in better
thus one does not find the porous in- workability with corresponding increases
clu.sion rich surface layer around the in the Forge Shop recovery. It has also
circumferance of the ingot as normally been shown that segregation is signifi-
seen in VAR ingots. As a result, in cantly reduced in electroslag remelting.
general, VAR ingots must be ground to By employing good remelting practice
remove this layer before forging thus the massive segregation commonly known
generating a significant amount of low as "freckles" sometimes showed in vacuum
grade grindings. arc remelted ingots are eliminated. This
is especially important when remelting
Keiley, et al, (15) as seen in Figure 12, alloys which are quite prone to "free-
has shown for HASTELLGYalloy X that kle" formation such as "Inconel" alloy
718.
B-8
 TABLE 6. POOL DEPTHAND COLUMNARGRAIN GROWTH
ANGLEMEASUREMENTSFOR ESR AND VAR INGOTS

ESR VAR
Columnar Grain(2) Pool Columnar Grain
Growth Angle Depth Growth Angle
Alloy Inches Degrees Inches Degrees

HASTELLOYalloy B 2.7 23.0 3.3 43.0

HASTELLOYalloy F 2.6 20.0 3.4 45.0
E-IAYNESalloy No. 31 2 .l 15.5 2.8(?) -
HAYNESSTELLITE 2.2 13.0 2.8(3) -
alloy No. 6B

(1) 30/32-inch was added to each ESR measurement to allow for the liquid alloy
present below the bath surface next to the solidified slag layer.

(2) Growth angle was: measured as the angle between the direction of growth
and the longitudinal axis of the ingot.

(3) Depth was estimated from shrinkage defect in the top of the ingot. The
columnar grain size was too small to obtain a good measurement of growth
angle.

TABLE 7. POOL DEPTHMEASUREMENTS FOR 8-INCH

DIAMETER "H.ASTELLOY"ALLOY X INGOTS

Electrode Diameter
b-3/4 Inches 6-Inches
Slag Weight
MU Position 10 lb. 14 18 10 - 14 18
AC Center 2.38" 2.13 2.00 2.50 2.30 2.15

Difference 1.01 0.96 0.90 1.10 1.03 0.93

DCSP Center 3.09 2.28 2.88 4.78 4.43 4.06

Difference 1.81 1.09 0.44 2.91 2.44 1.81
DCRP Center 1.97 1.78 1.65

Difference 0.97 0.92 0.88

B-9
In can be concluded from the above dis- less than the standard deviations of
cussion that as a result of a combina- the corresponding properties reported
tion of a number of favorable factors for the VAR material and all but that
the hot workability of ESR ingots is for the yield strength are significant
superior to that of either the direct at the a, = 0.01 confidence level.
forged or VAR ingots.
Cook, et al (16) have reported chemical
Properties analysis and mechanical property data
for a number of superalloys. It was
In order for the ESR material to be the object of their investigation to
marketable its microstructure and prop- prove the equivalence of the ESR pro-
erties must be superior or at least duced material to that of VAR; thus
comparable to that produced from a VAR a limited number of tests were run and
ingot, In selecting a processing tech- for the alloys under study: HAYNES
nique for a particular type of material alloys No. 600, 625, 750 and ~605
one considers not only the average prop- (HAYNES alloy NO. 25) their results
erty of the material produced from that show that the ESR superalloy is at
ingot but also the reproducibility of least equivalent to the VAR vacuum-arc
that particular property of interest, remelted material.
i.e. standard deviations. The reproduci-
bility of this particular property is a A comparison of the mechanical properties
function of the inclusion level and the of "Astroloy" from material produced by
homogeneity of the ingot and its pro- ESR and VAR is shown in Table 9. The
cessing history. The superior chemical properties of this alloy like in Cook's
uniformity and reproducibility of the investigation are essentially equivalent.
ingot as a result of the electroslag The same is true for the stress rupture
remelting compared to the VAR process properties shown in Table 10.
was statistically shown by Kelley, et
al (15). For a large number of pro-
duction heats of HASTELLOY alloy X they
determined the variation between the It has been shown that to successfully
primary and remelt chemistries as shown apply the electroslag remelting techni-
in Table 8. For seven of the ten ele- que to superalloys an understanding of
ments considered the material produced potential reactions and a knowledge of
from the ESR ingot exhibited a smaller the processing parameters is necessary.
or equal deviation from ladle analysis Among the most important considerations
than material produced from a corres- are:
ponding VAR ingot. This greater chemi-
cal uniformity of the ESR ingot had a 1. Effect of slag chemistry on retention
significant effect on the resultant ofoxidizable elements.
mechanical properties as shown in Table
9. As is seen the average mechanical 2. Role of electrochemical processes
properties of the HASTELLOYalloy ESR in inclusion control.
material were greater than the corres-
ponding VAR properties. Statistically 3. Control of atmospheres above slag
the average ultimate tensile strength to minimize oxygen potential.
and reduction in area results are signi-
ficantly different at greater than the If the proper choice of electroslag
a, = 0.01 confidence level. While the remelting conditions is made the re-
other two properties are also reported sultant ingot has been shown to have
higher, they are not statistically superior workability and product
significant at the a = 0.01 level. superior or at least equivalent mechanl-
Probably most important, however, is cal properties to material produced by
the standard deviation since most de- vacuum arc remelting.
sign criteria is based on the -3a limit.
The standard deviations in the mechanical
properties of the ESR material are again

B-10
 TABLE 8. DEVIATION OF BILLET FROM
LADLE ANALYSIS

VAR
Analysis
Element Error 3 (%) Od (%I 8 (%I ad (%)

Cr 0.27 0.05 0.20 0.09 0.24

Fe 0.27 Luw 0.18 0.03 0.13

MO 0.15 -0.01 0.06 0.02 0.07

CO 0.07 0.02 0.06 0.02(5) 0.07
Si 0.04 -0.011* 0.02(5) 0.007 0.03
c 0.01 0.000 0.01 -0.002 0.01

Mn 0.04 0.010 0.04 -0.002 0.02

W 0.04 -0.020x 0.03(5) -0.016* 0.03

S 0.003 0.0000 o.ooo(6) 0.0002 O.OOl( 5)
P 0.004 0.0002 0.002 0.0008 0.003
n 8: 57

d = Billet Minus Ladle Analysis

* 8 f O(a = 0.01 for yes decision)

B-11
 TABLE 9. HOT ROLLED BAR ROOMTEMPERATURE
TENSILE TEST RESULTS (CODEDDATA)

U.T.S. 0.2% Y.S. Elong.

{ksi) (ksi) 0

ESR x + 0.42 - 0.34 + 2.52 + 0.77

(n = 107)~ 2.2 2.0 2.9 2.4

VAR x
(n = 101)a
o .69
- 3.4
- 0.88
2.3
+ 1.96
4.7
- 0.33
3.9
S.M. %'
(n = 104)~
-I- 0.27
3.3
+ 1.22
2.9
- 4,.48
3.4
- 0.44
4.5

Tests ST A sig, a $ia. & Sig. a

X (ESR) vs. 2 (VAR) Yes 0.01 No 0.05 No 0.005 Yes 0.01

CT (ESR) vs. cs (BAR) Yes 0.01 NO 0.05 Yes 0.01 Yes 0.01
x (ESR) vs. z (S.M.) No 0.05 Yes 0.01 Yes 0.05 Yes 0.01
a (233~) vs. cr (S.M.) Yes 0.01 Yes 0.01 Yes 0.01 Yes 0.01
x (VAR) vs. x (S.M.) Yes 0.02 Yes 0.01 Yes 0.01 No 0.05
0 (vm) vs. 0 (S.M.) No 0.05 Yes 0.01 Yes 0.01 Yes 0.05

* Data transformed to normalize distribution

B-12
 TABLE 10. C0APARISONOF TENSILE PROPERTIES
OF ESR AND VAR. "ASTROLOY"

Temp. U.T.S. 0.2% Y.S. Elong.

("F) i?xi) bsi) 0
ESR RT 206,500 149,600 21.8 21.2
VAF? RT 207,300 150,100 22.1 22.5

ESR 1400 162,500 127,500 19.4 32.9

VAR 1400 158,000 124,700 17.4 28.0

TABLE 11. STRESS RUPTUREPROPERTIESOF ESR

AND VAR "ASTROLOY"

Temp. Stress Life Elong.

0. (psi) (hours) C%\

ESR 1400 85,000 40.4 23.2

VAR 1400 85,000 41.6 21.0

B-13
 REFERENCES

1. W. E. Duckworth and G. Hoyle, 16. R. L. Cook, G. W. Reese, Jr.,

Electroslag Refinirq, Chapman and and P. M. Gadsby, 3rd Int. Symp.
Hall Ltd., London, 1969. on ESR, 142, (1971).

2. m, Aug. 1969.

3. 33 Magazine, Jan. 197'2.

4. B. I. Medovar, Yu. V. Latash, B. I. "Hastelloy", tlHaynes", and "Haynes
Muksimovich and L. M. Stupak, Stellite" are registered trade marks
Electroslag Remelting, U. S. Depart- of Cabot Corporation.
ment of Commerce, No. 22, 217
(1963).
5. W. Holzgruber, P. E. Schneider, K.
Peterson, Trans. of Int. Vat. Met.
Conf. (1968).

6. G. A. Vachugov, V. V. Khlynov and

G. A. Khasim, Stal, 6, 483 (1967).

7. G. W. Loyd, T. A. Owen and L. A.

Baker, J. Australian Inst. Met.,
& No. 1, 17, (1971).
8. D. A. R. Kay and R. J. Pomfret,
J. Iron and Steel Inst., m, 962
(1971).
9. J. Cameron, M. Etienne and A.
Mitchell, Met. Trans., S, No. 7,
1839 (1970).
10. W. Holzgruber, P. Machner, and
E. Ploeckinger, Proc. A Vat. Sot.
(1969).
11. H. J. Klein, Electric Furnace COnf.
Proceedings, 21, (1970).

12. A. Mitchell and G. Beynon, Trans.

Met. Trans., 2, 3333, (1971).

13. E. W. Kelley, Tech. Rep. AFML-TR-

69-114, (1969).
14. J. W. Pridgeon, M. L. Pochq R. T.
Gross and V. Sharma, Trans. Vat.
Met. Conf. 525, (1968).
15. T. N. Kelley, D. A. Junker, R. C.
H. Chen, 3rd Int. Symp. on ESR,
125, (1971).
B-14
B-l 5
 I
I AC or DC

DC O

CONSUMABLE ELECTRODE
.-
LAIR ATMOSPHERE

I
-COOLING
VACUUM
ATMOSPHER
or

WATER
IN

-MOLTEN FLUX
ARC GAP

-SOLIDIFIED FLUX SK-IN

-MOLTEN METAL POOL-

-COPPER CRUCIBLE-
“LIVE”

-SOLIDIFIED METAL

WATER JACKET-

COPPER BASE PLATE-

Electroslag Remelting Vacuum Arc Remelting

Figure 2. Comparison of ESR and VAR furnaces.

 Consumable
Electrode ‘1

Atmosphere

Cooling
-Water

Molten
Solidified Flux
Flux Skin 4

Short-Jacketed
Molten
Copper Crucible
Metal Pool -4

Cooper
Base Plate ,

Withdrawing
Mechanism \

Figure 3. Electroslag Remelting using withdrawable ingot mechanism.

B-17
 A.C. D.C.
Single Single A.C.
Phase Phase Armstrong

LJ
- c

A.C. Three Phase

Figure 4. Power supplies for electroslag refining.

B-18
 -1
1x10

_ Rate of Rise of Pool for

4-inch Dia. Ingot

_ Rate of Rise of Pool for

20-inch Dia. Ingot

VI
3
-
”
-c

TERMINAL VELOCITY (CM/SEC)

Figure 5. Stoke’s Law Calculation for Al203 inclusions in superalloys

Figure 6. Dissolution and reprecipitation of

inclusions at the electrode tip.

B-20
+oSul-ysr~ UI juawo3 ua6Xxg
B-21
 -
co
B-22
1oc

9a

ESR

---B,L--L,---
\
\
20 / /’ \
/’ \
&--.A- \
10 \
\
\
0I I I I I I I P I
1750 1800 1850 1900
1950 2000 2050 2100 2150
TEST TEMPERATURE, OF

Figure 9. Gleeble Hot Ductility of U-700 Alloy.

 I I I I I I I .
I

C
O-Longitudinal
ESR ---
0 -Transverse
9c
Strain Rate - lin/in/sec O-longitudinal
VAR-
0 8a m -Transverse
0 C
m
G 70
5
E 6a
.-+0
i 50
w Q
r;: of 40
z
Q) Y\
p”
L’ 30
I\
I” \
20
I \
10 II \
0
1600 1700 1800 1900 2000 2100 2200 - 2300

Test Temperature, OF

Figure 10. Gleeble Hot Ductility of HASTELLOY@ alloy B

 0
DaJv jo UO!+>npaa +Ua3lad
B-25
 .-
L
t;”
IW cj
l-
--I cu 4
a -
2
IV
sl
ii
B-26