Acta Materialia 50 (2002) 1869–1879

www.actamat-journals.com

Hot tearing of nickel-based superalloys during directional

solidification
J. Zhang ∗, R.F. Singer
Department of Materials Science, University of Erlangen–Nürnberg, Martensstrasse 5, D-91058 Erlangen, Germany

Received 10 April 2001; received in revised form 21 December 2001; accepted 12 January 2002

Abstract

The propensity to form cracks during directional solidification was studied in two Ni-based superalloys, CM247LC
and IN792 (with varying Ti and Hf contents). Quenching experiments were employed to freeze in the amount of
remaining liquid during different stages in solidification. It was found that alloys with a strong tendency to hot tearing —
and, therefore, bad castability — display a strong change in volume fraction of remaining liquid with temperature at
the final stages of solidification. A simple mathematical model shows that a strong change in the fraction of liquid
results in high strains and strain rates during solidification, and this leads to crack formation and bad castability. The
castability of IN792 can be improved significantly, and even be brought to CM247 levels, by control of Hf and Ti, as
these elements affect the change of liquid fraction during the final stages of solidification.  2002 Acta Materialia Inc.
Published by Elsevier Science Ltd. All rights reserved.

Keywords: Nickel alloys; Casting; Hot tearing

1. Introduction processing in certain alloys. The well known,

highly corrosion-resistant superalloy IN792 is
Directional solidification (DS) of Ni-based among this group [2,3]. It was confirmed by pre-
superalloys to columnar grain structures has been vious work that the poor castability of IN792 is
industrial practice for decades. Numerous ben- due to hot tearing, i.e. crack formation in the final
eficial effects were demonstrated due to the elimin- stages of solidification [3]. The most common sol-
ation of grain boundaries and the introduction of a ution for the problem of hot tearing is compo-
preferred grain orientation by DS casting [1]. sitional modification by addition of Hf [1,3,4].
Unfortunately, the insufficient resistance against However, a number of detrimental effects were
grain boundary cracking during solidification in found when Hf was added: (1) Hf is a reactive
some superalloys prevented the application of DS element that may cause reactions with the mold
and formation of brittle Hf-containing inclusions;
and (2) Hf lowers the incipient melting point and
∗
Corresponding author. Tel.: +49-9131-8527516; fax: +49-
9131-8527515. thus the solutionizing temperature, which limits the
E-mail address: jian.zhang@ww.uni-erlangen.de (J. possibility to homogenize the alloy and to dissolve
Zhang). coarse γ⬘-precipitates. Consequently, DS alloys

1359-6454/02/$22.00  2002 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 9 - 6 4 5 4 ( 0 2 ) 0 0 0 4 2 - 3
1870 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879

containing Hf show little or no strength advantage behavior but quite different hot tearing resistance,
at all compared with the same alloy without Hf such as IN792 and CM247 [3].
modification in the equiaxed condition. It is of It is obvious that the most direct way to get
particular interest, therefore, to study the mech- insight into the hot tearing problem is by examin-
anism of hot tearing and to try to improve hot tear- ation of the microstructural evolution of superal-
ing resistance of superalloys without additions of loys in the freezing range, or rather the critical
Hf [3,5,6]. region of the freezing range. Unfortunately, very
The phenomenon of hot tearing of metals was limited information concerning the alloy IN792,
reviewed in detail by Sigworth [7]. Modeling work which is of large commercial interest at the
has been carried out that took into account surface moment, can be found at present. Hot cracking and
tension effects [8] or the combination of solid microstructural characterization of another superal-
deformation and interdendritic melt flow in the loy, IN718, was conducted at a rather detailed level
two-phase mushy zone [9–11]. However, little in previous work [20–23]. Wills and McCartney
attention has been paid so far to the multi-compo- [24] also studied the microstructural evolution and
nent systems, for example, superalloys. Neverthe- micro-segregation during solidification of several
less, it is obvious that a clear picture of the micro- other Ni-based superalloys. However, the different
structural evolution during solidification is composition of these alloys makes it difficult to
essential and the characterization of the mushy draw conclusions when considering hot tearing
zone is critical. Much work was therefore focused of IN792.
on the mechanical behavior of the alloys in the The present paper studied two types of Ni-based
mushy zone, and it was tried to evaluate the crack- superalloy, IN792 (with varying Ti content, and
ing susceptibility of the alloys in this particular with and without Hf addition) and CM247LC. The
condition [12–15]. The cracking behavior was microstructure at the end of the freezing range was
studied either by casting specially designed speci- investigated by quenching from various test tem-
mens where defined stresses are produced [13,14] peratures and subsequent optical metallographic
or by more standard mechanical testing methods examination [5,25]. Hot tearing susceptibility was
[3,15]. tested by DS casting of tube-like samples, a test
There is still certain confusion about the exact that has been developed and described previously
mechanism of hot tearing although this topic has [3]. CM247 and IN792Hf were chosen as reference
been studied extensively. It is generally believed materials because of their good castability. By
that a small freezing range that gives a small comparing the liquid fraction at different stages
mushy zone results in good castability [16]. Based during solidification of the alloys, guidance for
on this principle, researchers strive for smaller compositional modification of DS IN792 without
freezing ranges when trying to reduce hot tearing Hf addition was found and improvement in the
during DS casting of superalloys. There is also a castability of the alloy could be achieved. A gen-
general agreement that hot tearing occurs during eral principle for improvement of the castability of
the later stages of solidification when the freezing Ni-based superalloys is also suggested.
metal contains only a small fraction of remaining
liquid [10,11], although a different liquid fraction
for the vulnerable period is assumed by different 2. Experiments
authors [17–19]. For example, Pellini and co-work-
ers [18] concluded that the critical liquid fraction is 2.1. Materials
approximately 10%, while Clyne and Davies [19]
defined the critical region as the liquid fraction The compositions of the alloys used are listed
ranging from 10% to 1%. It is also quite obvious in Table 1. (All data are given in wt%.) Among
that the size of the freezing range cannot be the them, IN792Ti2, IN792Ti3, IN792Ti4.5 and
only material property governing hot tearing sus- IN792Hf were prepared in-house from the pure
ceptibility. Some alloys show similar freezing elements (99.99%) in a vacuum arc furnace. The
 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879 1871

Table 1
Compositions (wt%) of alloys used in the present experiments

Alloy Cr Co Mo W Al Ti Ta B Zr C Hf Ni

IN792 12.4 9.2 1.9 3.9 3.5 3.9 4.2 0.016 0.018 0.07 -- Bal.
IN792Ti2 12.4 9.2 1.9 3.9 3.5 2.0 4.2 0.016 0.018 0.07 -- Bal.
IN792Ti3 12.4 9.2 1.9 3.9 3.5 3.0 4.2 0.016 0.018 0.07 -- Bal.
IN792Ti4.5 12.4 9.2 1.9 3.9 3.5 4.5 4.2 0.016 0.018 0.07 -- Bal.
IN792Hf 11.9 9.0 1.9 3.8 3.5 4.0 3.9 0.012 0.09 0.12 2.0 Bal.
CM247 8.0 9.4 0.5 9.5 5.7 0.7 3.2 0.017 0.018 0.07 1.5 Bal.

other alloys were received from commercial ven- 2.3. Castability tests
dors. Differential scanning calorimetry (DSC;
Netzsch mode STA 409C) was employed to deter- Some selected alloys with Ti contents different
mine the critical temperature of phase formation from the standard IN792 were used to perform the
during solidification. A cooling rate of 5 °C/min castability tests. The alloys were cast in alumina
was employed in all measurements. The tempera- molds by a withdrawal method, described earlier
tures of phase formation were then used to select [3]. The alumina mold was designed to consist of
the quenching temperatures. an inner tube serving as a core and an outer tube
serving as an outer shell, thus giving a tube-like
superalloy casting with a wall thickness of 2.5 mm
2.2. Microstructural evolution and with a length of 苲140 mm. The stresses
developed in the present castability test are rather
The samples were quenched from different tem- severe compared with those in an actual casting,
peratures between the temperature of carbide for- because the alumina core does not yield at all. The
mation and the solidus. The experimental set-up of alloys were heated to 1520 °C and kept for 2 min
the quenching experiments is shown schematically before they were poured into the preheated mold
in Fig. 1. Alloys were cut into small cubic samples (mold temperature: 1500 °C). A withdrawal speed
about 3 g in weight and put into a ceramic crucible of 3 mm/min was used in all tests apart from two
for melting and quenching experiments. Three particular DS castings of IN792Ti2, where a with-
thermocouples were arranged in the furnace and drawal speed of 10 mm/min was employed. All of
one thermocouple was set inside the crucible near the DS processing parameters are set to meet
the test sample. Fig. 2 gives a schematic illustration industrial standards. The castability of the alloys
of the heating and cooling cycle. All samples were was evaluated by measuring a “crack ratio”
heated to 苲1430 °C and kept for 5 min to achieve obtained from evaluation of the length and width
complete melting. Then the furnace temperature of the cracks. The crack ratio r was then defined as:
setting was lowered to allow the temperature of the
冘
n

alloy to reach the desired temperature, Ttest, for the r⫽ liwi / LC, (1)
quenching experiment. Cooling of the furnace was i⫽1

carefully controlled to ensure that the temperature where n is the number of the cracks, li and wi are
of the thermocouple inside the crucible fluctuated the length and maximum width of a crack, respect-
only within ±1 °C when the temperature reached ively, and L is the length of the casting tube, C the
the desired value. Samples were held at the preset outer circumference of the tube. This crack ratio is
temperature for 10 min before being quenched into believed to give a more quantitative description of
cold water. Argon protection was used throughout the castability of alloys than other rating
the experiments. methods [3,26].
1872 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879

microscopy (SEM; Philips XL30). The volume

fractions of liquid in samples were measured by
using an image analysis system with the software
OPTIMAS 6.2. Liquid composition at certain tem-
peratures was additionally examined using an
energy-dispersive X-ray analysis (EDX) accessory
of the scanning electron microscope. The surface
of hot tearing cracks from tube castings was stud-
ied by SEM.

3. Results

3.1. Microstructural evolution

The cooling curves from DSC analysis of IN792

and CM247 are compared in Fig. 3. Both curves
show the three main phase transformations during
solidification, namely (1) solidification of γ phase
(corresponding to the start of freezing or the
liquidus), (2) formation of carbides and (3) final
γ/γ⬘ reaction (corresponding to the end of freezing
or the solidus).1 More detailed results from DSC
measurements are listed in Table 2. A surprising
result is that the formation temperature of the γ
phase increased significantly when the Ti content
was lowered to 2 wt% (in IN792Ti2; the carbide
formation and γ/γ⬘ reaction peaks are too weak to
be detected in this alloy).
Some examples of microstructures for IN792
and CM247 are compared in Fig. 4. In IN792,
blocky carbides, apparently formed at high tem-
Fig. 1. Schematic illustration of the experimental set-up used perature, emerged when the sample was quenched
for quenching experiments to determine the fraction of liquid from 1302 °C, Fig. 4(a). It is also clearly shown
phase present in the alloy at different temperatures. The speci- in Fig. 4(a) that a liquid network was present in the
men is held in the furnace until equilibrium is reached. Sub-
sequently the ceramic tube holding the sample (the crucible) is
interdendritic areas at this temperature. The IN792
pushed downward rapidly to transfer the sample into the water sample quenched from 1252 °C [Fig. 4(b)] shows
bath. Ceramic fiber material serves to plug the hole at the bot- an interesting change compared with Fig. 4(a). Iso-
tom of the furnace during heating. It is pushed into the water lated liquid pools rather than continuous networks
during quenching. Two holes with a diameter of 4 mm are can be observed at this lower temperature, which
drilled near the bottom of the crucible to allow water to enter
the crucible and get into contact with the alloy during quench-
indicates that the change of liquid volume fraction
ing. in IN792 is quite significant. In alloy CM247, one
can observe that the liquid also exhibits a network

2.4. Sample examination

1
The final γ/γ⬘ reaction is also often referred to in the super-
Quenched samples were polished and examined alloy literature as the eutectic reaction. More detailed charac-
by optical microscopy and scanning electron terization of this reaction will be given elsewhere [27].
 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879 1873

Fig. 2. Schematic illustration showing the heating and cooling procedure employed during the quenching experiments. The material
is first heated to 1430 °C to melt it completely. It is then cooled to the temperature Ttest where the liquid fraction is to be determined.

Fig. 3. DSC results for superalloys IN792 and CM247 during cooling from 1430 °C (cooling rate 5 °C/min). Tγ, TC and Tγ/γ⬘ is the
temperature where γ phase, carbides and γ/γ⬘, respectively, begins to form in each alloy.

Table 2
Critical temperatures (°C) measured from DSC analysis in different alloys at 5 °C/min during cooling

Alloy Tγ TC Tγ/γ⬘ ⌬T ⌬TCTR

IN792 1339 1315 1245 94 70

IN792Ti2 1386 -- -- -- --
IN792Ti3 1323 1305 1251 72 54
IN792Ti4.5 1338 1302 1250 88 52
IN792Hf 1340 1314 1245 95 69
CM247 1384 1342 1268 116 74

Tγ, TC and Tγ/γ⬘ is the temperature where γ, carbide and γ/γ⬘ begins to form, respectively. Tγ corresponds to the liquidus temperature
of the alloy, Tγ/γ⬘ to the solidus temperature. ⌬T, equal to Tγ minus Tγ/γ⬘ or Tliqidus minus Tsolidus, is the total freezing range. ⌬TCTR,
equal to TC minus Tγ/γ⬘ or TC minus Tsolidus, is the critical temperature range. The meaning of ⌬TCTR as well as its definition in the
present paper are discussed in Section 4.
1874 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879

Fig. 4. Optical micrographs comparing the alloys IN792 [(a), (b)] and CM247 [(c), (d)] that were quenched from various tempera-
tures. The remaining liquid at test temperature is clearly visible due to its darker color. The micrographs (a) and (c) were taken at
ca. 15 °C below the respective carbide formation temperature of the two alloys; micrographs (b) and (d) at 7 °C above the respective
γ/γ⬘ formation temperature. It is apparent that the fraction of remaining liquid changes much more strongly in IN792 than in CM247
in the corresponding temperature region. IN792 is an alloy with a castability inferior to that of CM247.

morphology at a quenching temperature 15 °C which have bad castability, changed much more
below the temperature of carbide formation, Fig. than that of the other three alloys studied, CM247,
4(c). However, the liquid film at most grain bound- IN792Hf and IN792Ti2, which have good cast-
aries is thinner compared with that in the micro- ability. [The alloy IN792Ti2 displays a slope of
structure of IN792 at the same stage. When the zero in Fig. 5(b). As mentioned above, we were
temperature drops to 1275 °C, the liquid in CM247 unable to measure TC and Tγ/γ⬘ in this alloy, i.e.
still remains a network, although the network is no we assume that the alloy was almost completely
longer well connected due to the occasional contact solidified in the temperature range investigated in
of dendrite arms [Fig. 4(d)]. Fig. 5(b). The data are therefore not relevant for
The liquid volume fractions at different tempera- the conclusions below and were not included in
tures of the alloys are compared quantitatively in Fig. 10 in Section 4.
Fig. 5(a) and (b). The approximate slope for each It is not clear from our results whether the absol-
curve is also indicated in the diagram. In the inter- ute level of volume fraction of the remaining liquid
val from carbide formation to γ/γ⬘ reaction (ca. 60 before final γ/γ⬘ formation does also have an effect
°C for all alloys investigated), the liquid volume on castability. IN792 and IN792Ti4.5, which have
fraction of the alloys IN792 and IN792Ti4.5, bad castability, exhibit a similar fraction of remain-
 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879 1875

Fig. 6. Concentration of the alloying elements in the quenched

liquid at different temperatures in IN792. The concentration is
normalized with respect to the nominal alloy composition. Ti,
Ta and Cr show severe segregation in the sense that their con-
centration C is much higher than their nominal concetration C0.

ever, to findings by other authors in complex

superalloys, where the data indicate a value of
C/C0 close to 1 or lower [28,29].

3.2. Castability tests

The results of the castability tests are summar-

ized in Fig. 7. Castability data of alloys IN792Hf
and CM247 from previous work are also included
in Fig. 7 [3]. The effect of Ti content in IN792 is
Fig. 5. Fraction of remaining liquid in various alloys as a quite dramatic. Castings without any crack or with
function of temperature. The temperature interval investigated only tiny cracks near the chill end were obtained
for each alloy ranges from its carbide formation temperature to
the γ/γ⬘ formation temperature. Alloys with bad castability are
when the Ti content was dropped to 2 wt% com-
shown in (a), and alloys with good castability are shown in pared with the original 3.9 wt% in IN792. How-
(b). The slope of the curves for alloys with good castability is ever, cracks reappeared as Ti content was increased
clearly smaller. to 3 wt%. Little difference was observed between
the 3 wt% Ti alloy and alloy IN792 with 3.9 wt%.
ing liquid before final γ/γ⬘ formation as CM247 and A further increase in Ti content to 4.5 wt% resulted
IN792Hf, which have good castability. This parti- in even more severe cracks than in IN792. As
cular fraction of remaining liquid ranges from 2 to shown above, Ti is strongly partitioned to the
5% in all four alloys. liquid and affects the solidification behavior very
The composition of the remaining liquid in much.
IN792 at different temperatures according to EDX The typical fracture surface from a castability
examination is shown in Fig. 6. Ti, Ta and Cr are test is shown in Fig. 8. It can be observed clearly
highly segregated and partitioned to the liquid that the crack formed along the interdendritic areas.
when compared with their nominal concentration. No cleavage planes are typical for crack formation
Al, on the other hand, is partitioned to the solid. in the solid that appears [3]. We therefore conclude
The latter fact may be surprising in the light of that there was a more or less continuous film of
what is to be expected based on the binary and remaining liquid present on the dendrite surface
ternary phase diagrams. It does correspond, how- when the crack formed.
1876 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879

Fig. 7. Results of the castability tests for various alloys (see Table 1). Higher crack ratios correspond to more severely cracked
samples (see text), i.e. inferior castability. The withdrawal speed during directional solidification in the castability test is also indicated.
If the Ti content is increased from 2 wt% (IN792Ti2) to 3.9 wt% (IN792) or 4.5 wt% (IN792Ti4.5), castability is lowered.

situation will occur when the secondary dendrite

arms begin to contact each other and bonds begin
to form between them.
The upper end of ⌬TCTR, Th, can be determined
by Gleeble testing [3]. A cylindrical test specimen
with threaded ends is loaded with a special spring
mechanism that allows the application of very
small forces (10 to 30 N). The specimen is gradu-
ally heated to a temperature above the solidus. The
small force eventually causes the specimen to frac-
ture and the temperature at fracture is recorded.
Heck et al. [3] determined Th as 1319 °C for equi-
axed and 1322 °C for directionally solidified
CM247, and 1273 °C for equiaxed and 1291 °C
Fig. 8. Typical fracture surface in a casting of directionally
solidified IN792 superalloy. Note the rounded appearance corre-
for directionally solidified IN792, respectively.
sponding to dendrite arm surfaces and the complete lack of One can see from Fig. 3 and Table 2 that these
cleavage planes. This is interpreted as cracking at a temperature temperatures are close to, but somewhat below, the
where the dendrite arms were still covered by a more or less formation temperatures of carbides in both alloys.
continuous film of liquid rather than cracking in a completely Since no Gleeble testing was carried out in the
solid state.
present investigation, we will take TC, the forma-
tion temperature of carbide, as an indication for Th
4. Discussion below. It is interesting to note that Th corresponds
to a liquid fraction of ~10–12% in IN792 as shown
4.1. Critical temperature range (⌬TCTR) for hot in Fig. 5(a). [The assumption that the upper end
tearing of the critical temperature range Th is equal to the
temperature of carbide formation TC may appear
The critical temperature range (⌬TCTR) for hot somewhat arbitrary. We like to point out that any
tearing to occur corresponds to the lower end of definition where Th corresponds to a certain per-
the freezing range, where the solid fraction is high centage of ⌬T would have given the same result.
enough to prevent feeding and to allow the devel- TC in our case corresponds to approximately 70%
opment of stresses. In microstructural terms, this of ⌬T: (TC⫺Tsolidus) / ⌬T⬵0.7.]
 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879 1877

The lower end of ⌬TCTR, Tl, is more difficult to

define. As the inspection of the fracture surfaces
showed that the cracks were formed close to, but
somewhat above the solidus (Fig. 8), we will
assume Tl⬵Tγ/γ⬘ in the following discussion.
It is usually assumed that bad castability is asso-
ciated with a large ⌬TCTR. This is because a large
⌬TCTR implies a large volume of material and a
long time interval where cracks may occur:
⌬TCTR ⌬TCTR
⌬l ⫽ , ⌬t ⫽ (2)
G Ṫ
where ⌬l and ⌬t are the length and the time where
hot tears may form during directional solidifi-
cation, respectively, and G is the temperature
gradient, Ṫ the solidification rate. As Table 2 Fig. 9. Schematic illustration of a unit volume of material dur-
shows, however, the bad castability of the alloys ing solidification used for derivation of Eqs. (3)–(6). The liquid
IN792 and IN792Ti4.5 cannot be explained by a with a fraction fl(T) is assumed to distribute homogeneously
between the solid alloy. z is the direction of solidification.
large ⌬TCTR. This conclusion does not change if
we take ⌬T, the total freezing range, as a measure
for the critical temperature range rather than where as is the thermal expansion coefficient of the
⌬TCTR, as defined above. solid alloy, which is assumed here to be constant as
T changes. When feeding is impossible, the strain
4.2. Hot tearing mechanism
el resulting from the solidification and thermal con-
The present results indicate that another effect traction of liquid can be written as (here again, el
exists that can influence the hot tearing tendency refers only to the linear strain along the x direction
of an alloy during DS processing. As Fig. 5 shows, in Fig. 9):

冕 冕
the alloys with bad castability display a stronger
change of liquid volume fraction with temperature, el ⫽ b1/3 dfl(T) ⫹ al fl(T)dT, (4)
df/dT, within ⌬TCTR.
The parameter df/dT is very important because where b is the volume shrinkage coefficient of the
it determines the strain and the strain rate that liquid during solidification, and al is the thermal
result during solidification. expansion coefficient of liquid that again is
Hot tearing occurs in a material whenever the assumed to be constant. Provided the thermal
strain caused by shrinkage during solidification expansion coefficients as and al are equal, the total
cannot be accommodated by elastic and plastic strain e can be written as:
deformation of the alloy. In the ⌬TCTR discussed
above, thermal contraction and solidification
shrinkage take place simultaneously as temperature
drops. Take Fig. 9 as an example, where the liquid
冕
e ⫽ es ⫹ el ⫽ a dT ⫹ b1/3 dfl(T). 冕 (5)

film with a volume fraction of fl(T) at temperature The total strain within ⌬TCTR can then be obtained
T is assumed to distribute homogeneously between by integration of Eq. (5):
the solidified sections. Then, the strain in the solid e ⫽ a(Th⫺Tl) ⫹ b1/3[fl(Th)⫺fl(Tl)], (6)
alloy es due to thermal contraction along the x
direction can be expressed as: where Th and Tl are the boundary temperatures
of ⌬TCTR.
冕
es ⫽ as [1⫺fl(T)]dT, (3) Assuming Th and Tl to be equal to the tempera-
tures TC and Tγ/γ⬘, respectively, the total strain
1878 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879

within ⌬TCTR for the tested alloys can be calculated solidification due to a stronger change in liquid
based on the experimental results and Eq. (6). The fraction in ⌬TCTR.
thermal expansion coefficient a⬵2 × 10⫺5 K⫺1
and the volume shrinkage coefficient b⬵5.1% are 4.3. Compositional modification of IN792
taken from the literature [30,17]. The measured
crack ratio of different DS superalloys as a func- Based on the results discussed above it seems
tion of the calculated e is shown in Fig. 10. that, in order to improve the castability of Ni-based
It is clearly seen from Fig. 10 that the alloys superalloys by compositional modification, the
IN792, IN792Ti3 and IN792Ti4.5, with a higher content of elements that affect df/dT in ⌬TCTR must
df/dT in ⌬TCTR, are subjected to larger strains than be controlled. Hf and Ti seem to be such elements.
the alloys CM247 and IN792Hf. This leads to a Both elements strongly segregate to the melt. It
high crack ratio, i.e. bad castability. The reason could be that segregation of an element to the melt
why large strains can lead to cracking is rather is a first indication to what their effect on solidifi-
obvious. The strains in Fig. 10 are much larger cation behavior might be. Future work, therefore,
than what the material can accommodate by elastic will focus on the effects of Ta and Cr, which have
deformation. Consequently, strains will result in been found in this paper to partition to the melt,
flow of the liquid or plastic deformation of the too.
solid, which may induce the formation of hot tears.
It should be noted that larger strains for a spe-
cific alloy also imply larger strain rates in the Acknowledgements
present case. This is because ⌬t according to Eq.
(2) is about the same for the alloys under investi- The authors would like to thank the Deutsche
gation. High strain rates may lead to bad castability Forschungsgemeinschaft for financial support.
for two reasons: (1) there is less time for feeding
and (2) there is less time for stress relaxation to
occur. References
In summary, the present study shows that alloys
with bad castability suffer from both a higher strain [1] Sims CT, Stoloff NS, Hagel WC. In: Superalloys II. New
and a higher strain rate during the later stages of York: John Wiley & Sons; 1987. p. 97.
[2] Rösler J, Konter K, Tönnes C. In: Kissinger RD, Deye DJ,
Anton DL, Cetel AD, Nathal MV, Pollock TM, editors.
Superalloys 1996. Warrendale (PA): The Minerals, Metals
and Materials Society; 1996. p. 515.
[3] Heck K, Blackford JR, Singer RF. Mater. Sci. Technol.
1999;15:213.
[4] Doherty JE, Kear BH, Giamei AF. J. Met. 1971;23:59.
[5] Zhu Y, Radavich JF, Zheng Z, Ning X, Lou L, Xie X et
al. In: Pollock TM, Kissinger RD, Bowman RR, Green
KA, McLean M, Olson S, editors. Superalloys 2000. War-
rendale (PA): The Minerals, Metals and Materials Society;
2000. p. 329.
[6] Lin TL, Huang S, Sun C. In: Reichman S, Duhl DN,
Maurer G, Antolovich S, Lund C, editors. Superalloys
1988. Warrendale (PA): The Minerals, Metals and
Materials Society; 1988. p. 345.
[7] Sigworth GK. AFS Trans. 1996;155:1053.
[8] Holt M, Olson DL, Cross CE. Scripta Metall. Mater.
1992;26:1119.
Fig. 10. Measured crack ratio r as a function of calculated [9] Katgerman L. J. Met. 1982;2:46.
total strain e in different DS alloys. Crack ratio r and total [10] Rappaz M, Drezet JM, Gremaud M. Metall. Mater.
strains e are obtained from the equations indicated on the dia- Trans. 1999;30A:449.
gram. [11] Farup I, Mo A. Metall. Mater. Trans. 2000;31A:1461.
 J. Zhang, R.F. Singer / Acta Materialia 50 (2002) 1869–1879 1879

[12] Lewandowski MS, Overfelt RA. Acta Mater. [23] Thompson RG, Mayo DE, Radhakrishnan B. Metall.
1999;47:4695. Trans. 1991;22A:557.
[13] Warrington D, McCartney DG. Cast Met. 1989;2:135. [24] Wills VA, McCartney DG. Mater. Sci. Eng.
[14] Instone S, StJohn D, Grandfield J. Int. J. Cast Met. Res. 1991;145A:223.
2000;12:441. [25] Sponseller DL. In: Kissinger RD, Deye DJ, Anton DL,
[15] Lin CS, Sekhar JA. J. Mater. Sci. 1994;29:5005. Cetel AD, Nathal MV, Pollock TM, editors. Superalloys
[16] Flemings MC. In: Solidification Processing. New York: 1996. Warrendale (PA): The Minerals, Metals and
McGraw-Hill; 1974. p. 254. Materials Society; 1996. p. 259.
[17] Campbell J. In: Castings. London: Butterworth-Heinem- [26] Ross EW, O’Hara KS. In: Antolovich SD, Stusrud RW,
ann; 1991. p. 219. MacKay RA, Anton DL, Khan T, Kissinger RD, editors.
[18] Bishop HF, Ackerland CG, Pellini WS. AFS Trans. Superalloys 1992. Warrendale (PA): The Minerals, Metals
1957;65:247. and Materials Society; 1992. p. 257.
[19] Clyne TW, Davies GJ. Br. Foundrymen 1981;74:65. [27] Zhang J, Singer RF. Z. Mettalkd. 2002; in press.
[20] Yaman YM, Kushan MC. J. Mater. Sci. Lett. [28] Sung PK, Poirier DR. Metall. Mater. Trans.
1998;17:1231. 1999;30A:2173.
[21] Radhakrishnan B, Thompson RG. Metall. Trans. [29] Taha MA, Kurz W. Z. Metallkd. 1981;72:546.
1993;24A:1409. [30] Hermann W. Ph.D. dissertation, University of Erlangen–
[22] Vincent R. Acta Metall. 1985;33:1205. Nürnberg, Erlangen, Germany, 1995.