The Role of Manganese and Copper in the Eutectoid

Transformation of Spheroidal Graphite Cast Iron

JACQUES LACAZE, ALINE BOUDOT, VALÉRIE GERVAL, DJAR OQUAB, and
HENRIQUE SANTOS

The decomposition of austenite to ferrite plus graphite or to pearlite in spheroidal graphite (SG) cast
iron is known to depend on a number of factors among which are the nodule count, the cooling rate,
and the alloying additions (Si, Mn, Cu, etc.). This study was undertaken in order to deepen the
understanding of the effect of alloying with Mn and/or Cu on the eutectoid reaction. For this purpose,
differential thermal analyses (DTAs) were carried out in which samples were subjected to a short
homogenization treatment designed to smooth out the microsegregations originating from the solid-
ification step. The effect of various additions of copper and manganese and of the cooling rate on
the temperature of the onset of the stable and metastable eutectoid reactions was investigated. A
description of the conditions for the growth of ferrite and of pearlite is given and shows that these
reactions can develop only when the temperature of the alloy is below the lower boundary of the
ferrite/austenite/graphite or ferrite/austenite/cementite related three-phase field. The experimental re-
sults can be explained if the appropriate reference temperature is used. The cooling rate affects the
temperature of the onset of the ferrite plus graphite growth in the same way as for the eutectic
reaction, with a measured undercooling that can be extrapolated to a zero value when the cooling
rate is zero. The growth undercooling of pearlite had values that were in agreement with similar data
obtained on silicon steels. The detrimental effect of Mn on the growth kinetics of ferrite during the
decomposition of austenite in the stable system is explained in terms of the driving force for diffusion
of carbon through the ferrite ring around the graphite nodules. Finally, it is found that copper can
have a pearlite promoter role only when combined with a low addition of manganese.

I. INTRODUCTION At high cooling rates, the eutectoid transformation tem-

perature range is depressed, generally resulting in the re-
THE microstructure of typical commercial spheroidal tardation of ferrite growth and a promotion of pearlite
graphite (SG) irons in as-cast state or after heat treatment
growth.[4,5] On the contrary, a slow cooling rate through the
consists of graphite nodules embedded in a ferrite shell and eutectoid transformation and/or a high nodule count pro-
of pearlite. This so-called bull’s-eye structure, according to motes the ferritic reaction. Indeed, Askeland and Gupta[6]
foundry jargon, is illustrated in Figure 1. The control of have shown that the relative amount of ferrite and pearlite
this microstructure is of practical importance because it de- is sensitive to changes in nodule count or cooling rate only
termines the mechanical properties of SG irons.[1] Among when the nodule count is low. When the nodule count is
other authors, Johnson and Kovacs[2] described how this
high, extremely rapid rates of cooling are required to sup-
microstructure evolves during cooling of a casting. The al-
press ferrite growth,[6] as discussed in terms of Fourier’s
loy first undergoes the stable eutectoid reaction in which solute ratio in a previous article.[7] In commercial castings,
austenite decomposes to give ferrite and graphite. Ferrite the cooling rate is determined, in practice, by the section
nucleates at the graphite/austenite interface and then grows size. Therefore, to control the as-cast structure of a given
symmetrically around the nodules; this reaction is hereafter component, the use of alloying elements promoting or in-
called the ferritic reaction. This growth is controlled by hibiting the ferritic or pearlitic reactions is an important
diffusion of carbon through the ferrite shell, which makes factor. Tin, Sb, Mn and Cu are among the most usual pearl-
it quite a slow process. Therefore, the temperature of the ite promoter elements. Johnson and Kovacs[2] and Kovacs[8]
metastable eutectoid could be reached before the complete compared the roles of Mn, Sn, and Sb and showed con-
transformation of austenite. Once initiated, the metastable vincingly that Sn and Sb segregate in a thin layer at the
reaction, also named the pearlitic reaction, proceeds quickly surface of graphite, which acts as a barrier to the transfer
because of the cooperative growth of ferrite and cementite. of carbon atoms to graphite nodules. Venugopalan[3] re-
This latter transformation is similar to the pearlitic reaction cently summarized the previous descriptions of the effects
of steels.[2,3] of alloying on eutectoid transformation in SG iron and re-
ported that the most frequently accepted role of Cu and Mn
is to depress the eutectoid temperature and to decrease the
ALINE BOUDOT and VALÉRIE GERVAL, Graduate Students, and
DJAR OQUAB, Research Engineer, JACQUES LACAZE, Research diffusion coefficient of carbon in ferrite. However, the work
Scientist, are with the Equipe Métallurgie Physique, Ecole Nationale of Ågren on diffusion in multicomponent alloys[9] predicted
Supérieure de Chimie de Toulouse, ENSCT, 31077 Toulouse cedex, that additions of substitutional elements at levels used for
France. HENRIQUE SANTOS, Professor, is with the Departmento de Mn or Cu in cast irons do not induce a drastic change in
Engenharia Metalurgia da Universidade do Porto, 4099 Porto codex,
Portugal.
the diffusion coefficient of carbon in ferrite.[7] In fact, Pan
Manuscript submitted September 19, 1996. et al.[4] agreed that the amplitude of the observed effects of

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 28A, OCTOBER 1997—2015

 composition of these alloys as measured by chemical anal-
ysis. The impurity level, expressed in weight percent, was
very low: 0.003Al, 0.01Ni, 0.035P, 0.005S, 0.01V, and less
than 0.005 for the other measured elements (As, B, Bi, Co,
Mo, Pb, Sb, Sn, Te, W, Zn, and Zr). The alloys were melted
in a 30-kg induction furnace, treated with a Fe-Si-Mg alloy
at 1500 7C in a first ladle, and then inoculated into a second
ladle with a Fe-Si alloy before being cast in a step mold.
Each step was cut into two halves, one for metallographic
observations and the other for machining the differential
thermal analysis (DTA) samples as close as possible to the
cut surface. Results related to samples from a 10 3 1023-
m-thick step will be used here. Full information on these
results has been reported previously.[10]
Fig. 1—Usual bull’s-eye microstructure of a spheroidal graphite cast iron, During the present study, new experiments were con-
with graphite nodules surrounded by a ferrite ring and pearlite in the ducted on a cast iron with 1 wt pct Cu added, which will
remaining volume. be called HCM cast iron, and also on a carbon steel whose
composition was the composition of the matrix of the cast
Table I. Alloy Compositions (in Weight Percent) and iron at the eutectoid temperature. The nominal composi-
Characteristic Transformation Temperatures (in Degrees tions of these alloys are also given in Table I.
Celsius) Calculated for the Austenite Matrix

Casting C Si Mn Cu Toa (7C) Ta (7C) Top (7C) Tp (7C) B. Differential Thermal Analysis
L1 3.58 2.57 0.03 0.005 823 801 794 785 The DTA experiments were carried out using a SE-
L5 3.66 2.47 0.43 0.011 813 775 794 783 TARAM DSC 2000 apparatus. The maximum possible
L7 3.62 2.46 0.41 0.22 811 771 791 777 cooling rate was 20 K/min in the temperature range for
L9 3.65 2.47 0.044 0.22 817 794 789 777 solid-state transformations. For each alloy, DTA traces
HCM 3.7 2.5 0.5 1.00 803 744 782 754 were obtained upon cooling during two series of experi-
Copper
steel 0.7 2.5 0.5 1.00 800 745 780 754
ments. In both series, the samples were first heated to 1100
7C and maintained at this temperature for 1 hour. It was
thought that, at this temperature, there would be a modifi-
Cu and Mn could not be explained in this way. Moreover, cation of the distribution of manganese, copper, and silicon;
recent studies[7,10] on SG cast irons containing a low level i.e., homogenization of the sample would occur.[10] The
of copper (0.22 wt pct) showed results in conflict with the holding time, t, at this temperature was calculated such that
pearlite promoter effect generally accepted for this element. the ratio Dgi t/L2 was on the order of 0.1,[12] where Dgi is the
Therefore, further study of the role of copper addition at diffusion coefficient of the i species and L the length over
higher level is of interest in understanding how it can act which the diffusion must proceed. The sample was then
as a pearlite promoter. cooled at a controlled rate down to 600 7C, then reheated
The experiments described in the present article were to 900 7C for 10 minutes before being cooled at another
achieved on a cast iron containing 1 wt pct Cu and on a rate down to 600 7C, then reheated to 900 7C for 10
steel having the composition of the matrix of this cast iron. minutes, and finally cooled at a different rate. The succes-
The results obtained on these alloys will be presented along sive cooling rates used were 10, 5, and 1 K/min in the first
with a recapitulation of the previous results.[10] As in the series and 20, 10, and 2 K/min in the second one. The time
previous work, the study of the decomposition of austenite during which the samples were maintained at 900 7C before
was followed by means of differential thermal analysis. The cooling was chosen in order to achieve full transformation
characteristic temperatures for the start of austenite decom- of ferrite and pearlite to austenite.
position are then discussed in light of a description of the In most cases, the DTA trace involved two peaks. The
growth conditions of ferrite and pearlite reported re- first peak (i.e., at higher temperature) corresponded to the
cently.[10] This approach is also used to analyze the role of growth of ferrite and graphite, and the second to the growth
manganese and copper in the decomposition of austenite. of pearlite, as previously inferred by Ekpoom and Heine.[11]
In previous studies, the temperature of the start of the fer-
ritic transformation was determined from the kinetics
II. EXPERIMENTAL TECHNIQUES curves calculated from the DTA traces and was defined as
A. Material and Casting Conditions the temperature at which a 1 pct volume fraction had trans-
formed. The temperature of the start of the pearlitic trans-
In previous studies,[7,10] four different alloys, each having formation was estimated on the DTA traces. The accuracy
approximately constant carbon and silicon contents such in the experimental determination of these characteristic
that they were close to eutectic composition, were used. temperatures was estimated to be 52 K for all of the cool-
The alloys differed in copper and manganese content, with ing rates, except at 1 K/min, where it was 54 K due to
one alloy having the basic composition chosen in the Fe- the smoothing of the DTA signal at lower cooling rates. In
C-Si system (alloy L1), one with added manganese (alloy the present investigation, both temperatures were estimated
L5), one with added copper (alloy L9), and one with both from the DTA traces, as illustrated in Figure 2. Because
copper and manganese added (alloy L7). Table I shows the the thermal arrest related to the ferritic reaction is small in

2016—VOLUME 28A, OCTOBER 1997 METALLURGICAL AND MATERIALS TRANSACTIONS A

Fig. 2—Example of a DTA trace (continuous line) and of its derivative
(dotted line) plotted vs temperature. The arrows show how the Fig. 4—Volume distribution of the nodules calculated from the data
temperatures for the start of the ferritic and pearlitic reactions were plotted in Fig. 3. The dark gray bars refer to the number and the light
estimated. gray ones to the standard deviation as estimated in each class.

10(i21)/10 and 10i/10. It can be seen that this distribution has

two different peaks that merge in class 11, i.e., for an equiv-
alent diameter equal to 10 mm. In accordance with previous
authors,[14,15] we considered that the peak at the smallest
sizes should be disregarded, because most of the corre-
sponding measurements were in fact related to inclusions
or pores. Thus, the actual distribution of nodules in the
material is represented by the second peak at larger diam-
eters. The volume distribution was then calculated from the
area distribution using the Saltykov’s method.[13] The vol-
ume distribution thus calculated with the data reported in
Figure 3 has been plotted in Figure 4, considering only the
classes with index higher than 11.
The experimental standard deviation measured in each
class is also reported in Figure 3 along with the distribution.
Because of the correlation between the number of nodules
in the different classes of the distribution, the standard de-
viation of the volume distribution has to be estimated by
Fig. 3—Example of area distribution of nodule number. The numbers of again applying Saltykov’s transformation to the variations
sections are plotted vs the equivalent diameter of the sections selected in the area distribution. This leads to the calculated standard
according to the Saltykov’s classes. The dark gray bars refer to the number variations in the volume distribution that have been plotted
and the light gray ones to the standard deviation as measured in each along with the volume distribution in Figure 4. From the
class.
measurements performed in this study, the relative error on
the total nodule count was estimated to range from 15 to
the present case, the expected accuracy in the determination 20 pct for a 95 pct confidence interval.
of this temperature is not as good as before, maybe 55 K. An electron microscope equipped with an energy disper-
sive spectrometer was used to quantitatively characterize
C. Metallography the distribution of iron, silicon, copper, and manganese in
the as-cast and homogenized states of the HCM iron and
Measurements of the number, size distribution, and frac- the steel. The procedure selected for this study was similar
tional area of graphite nodules were made with an image to the one used previously to study the distribution of al-
analyzer. This was carried out on 25 fields at a magnifi- loying elements in alloys L5 and L9.[10] The analysis was
cation of 430 times in the previous study and 500 times in performed in the spot mode at 20 kV with a beam current
the present one. These magnifications were chosen so that of 2 nA. Reference standards of pure elements were used
there were a sufficient number of nodules in the field and for the quantitative analysis. In most cases, measurements
at a large enough magnification so that other particles that were taken using a grid of 30 points by 10, which covered
were not nodules were easily detected and eliminated. Fig- a total area of about 8 mm2. Such a large area is expected
ure 3 shows an example of the size distribution of the nod- to give all the features that a random plane may cut through
ules section, plotted according to the Saltykov’s classes.[13] the solidification microstructure. This procedure thus allows
These classes are such that class i contains sections with a precise description of the distribution of alloying elements
equivalent diameter, expressed in micrometers, of between in the material to be made.[16,17] The weight percentages of

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 28A, OCTOBER 1997—2017

Table II. Microstructural Features of the HCM Cast Iron

f g (Pct) NA (mm22) NV (mm23)

As-cast 10.2 145 4860
DTA sample 11.8 174 6050

each element were calculated by a program that accounted

for ZAF corrections. The conditions used for the analyses
were such that the statistical errors for a measurement were
0.11 wt pct for silicon, 0.15 wt pct for manganese, 0.62 wt
pct for iron, and 0.22 wt pct for copper.

(a)
III. RESULTS
A. Metallographic Examination
The results of the measurements of the fraction of graph-
ite and of the number of graphite nodules per unit area in
the as-cast material and DTA samples of the HCM cast iron
are reported in Table II. Values of the number of graphite
nodules per unit volume, as estimated by Saltykov’s
method, are also given in the table. It can be seen that the
homogenization treatment and maybe also the thermal cy-
cles led to a small increase in these microstructural features.
A slight increase in the size of the graphite nodules between
the nonhomogenized and homogenized samples was also
observed. These results are in agreement with previous ar- (b)
ticles.[7,10]
The extent of the ferrite and pearlite phases in the ma-
terial was examined after etching the samples with Nital.
Figures 5(a) and (b) show micrographs of HCM cast iron
samples cooled, respectively, at 2 and 10 K/min. It is seen
that the matrix is essentially pearlitic, although ferrite halos
developed around a limited number of graphite nodules.
More significantly, it is noted that there are numerous fil-
aments of ferrite imbedded in pearlite. This ferrite formed
without any particular relation to the graphite nodules and
is certainly intergranular ferrite. This mixture of intergran-
ular ferrite and pearlite has exactly the same appearance as
the microstructure of the steel sample illustrated in Figure
5(c), and differs significantly from the usual bull’s-eye mi-
crostructure illustrated in Figure 1. These observations led (c)
us to the conclusion that the eutectoid transformation of the Fig. 5—Micrographs of DTA samples etched with Nital. (a) and (b) are
high copper cast iron proceeds as in the case of the steel from samples of the high copper cast iron and correspond to cooling at 2
for most of the matrix. This is in line with the results of and 10 K/min, respectively. (c) is from a DTA sample of the steel, which
Lalich and Loper,[5] who reported that the transformation was cooled at 2 K/min.
proceeds into the matrix in samples with a low level of
ferrite. These filaments correspond to the proeutectoid fer- steel. The dense cloud of points in each figure corresponds
rite described by Kovacs,[8] where the word ‘‘proeutectoid’’ to measurements made on the matrix. In Figure 6, the
should be understood here with respect to the metastable points with low copper, silicon, and manganese content cor-
phase diagram. respond to the measurement made on a graphite nodule or
Microprobe measurements were carried out to discover at the interface between a nodule and the matrix. As silicon
if the homogenization treatment at 1100 7C for 1 hour had and copper segregate negatively during solidification of cast
been effective. The microprobe results have been reported iron, the highest silicon and copper values correspond to
by plotting the correlation between the silicon and copper regions solidified near the start of solidification, and the
contents vs the manganese content for all of the counting lowest values correspond to regions solidified near the end
points. Because of the small interlamellar spacings of pearl- of solidification. The opposite is true for manganese, as this
ite, measurements made on this aggregate were expected to element segregates positively so it is highest in content at
give values that were an average of the ferrite and cementite the end of solidification. In the case of the steel, it is seen
compositions. The graphs in Figure 6 present these corre- on the graph relating to the as-cast state that Mn and Cu
lations for the HCM cast iron in the as-cast and homoge- segregate positively during solidification, while silicon seg-
nized states. Figure 7 presents the graphs related to the regates negatively.

2018—VOLUME 28A, OCTOBER 1997 METALLURGICAL AND MATERIALS TRANSACTIONS A

Fig. 6—Correlation between silicon (closed circles) and copper (empty Fig. 7—Correlation between silicon (closed circles) and copper (empty
circles) contents vs the manganese content in the case of the HCM cast circles) contents vs the manganese content in the case of copper steel: (a)
iron: (a) in as-cast state and (b) after DTA experiments including in as-cast state and (b) after DTA experiments including homogenization
homogenization for 1 h at 1100 7C. for 1 h at 1100 7C.

On the samples after the homogenization treatment, a

grouping of the points is observed. In the case of the steel,
homogenization was really efficient so that the clouds of
points are just slightly larger than the scattering domain
expected from the standard deviation of the measurements.
In the case of the cast iron, it is clear that all chemical
heterogeneities have not been eliminated. It is seen that the
grouping of the points that is observed is mainly because
the segregations associated with the areas solidifying last
have been smoothed out. There also seems to be a small
decrease in the maximum silicon and copper values and an
increase in the minimum manganese values, which could
indicate that the distribution of alloying elements had also
been slightly changed near the nodules. These observations
are in agreement with those made previously.[10]

B. DTA Results
1. DTA results on samples with low level additions
(series L) Fig. 8—Example of DTA traces obtained with a sample of alloy L9 (Fe-C-
The main characteristics of the DTA traces are all similar Si base alloy with 0.2 wt pct Cu added) cooled at 1, 5, and 10 K/min.
The initial temperature was 900 7C.
with respect to the cooling rate. A typical set of traces ob-
tained on a sample of alloy L9 cooled at various cooling
rates, 1, 5, and 10 K/min, is shown in Figure 8. As ex- is only one peak on the trace when the sample is cooled at
pected, the curves were shifted toward lower temperatures 1 K/min, while two peaks are present at 5 and 10 K/min.
and the thermal effect was enlarged when the cooling rate Moreover, the relative size of the second peak grows higher
was increased. More importantly, it is observed that there in correlation to the increase in the cooling rate. From the

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 28A, OCTOBER 1997—2019

 lower temperature than for the base alloy. At about 740 7C,
there was a clear change in slope, indicating the beginning
of pearlite growth, which then proceeds very rapidly until
complete transformation. On the contrary, adding 0.2 wt
pct Cu did not strongly affect the shape of the DTA trace
when compared to the base alloy, although this shifted the
ferritic and pearlitic reactions to slightly lower tempera-
tures, in agreement with the literature. Similar effects of the
additives to those observed on Figure 9 were noted for the
other cooling rates investigated.
Figure 10 shows the DTA traces for the base alloy L1
and alloys L5 and L7 with, respectively, manganese added
and both copper and manganese added. It is observed that,
when copper was added together with manganese, it can-
celed the effect of manganese, virtually returning the ferrite
transformation to that of the base alloy L1, apart from the
temperature range where the transformation proceeds. Add-
ing copper at low levels definitely countered the effect of
manganese in regard to the onset and the kinetics of the
Fig. 9—Comparison of the DTA traces obtained at 5 K/min on samples
from alloys L1 (base), L5 (0.4 wt pct Mn added), and L9 (0.2 wt pct Cu austenite decomposition. This discrepancy with previous
added). The initial temperature was 900 7C. works that described copper as a pearlite promoter element
might be related to the fact that the amount of copper added
to the base alloy in the present study was much lower than
in most previous studies. In fact, Pan et al.[4] and Lalich
and Loper[5] also studied such low amounts in the frame-
work of wider investigations. Unfortunately, kinetics data
relating to the corresponding case studied by the former
authors are lacking. On the contrary, the full results of Lal-
ich and Loper concerning SG cast iron with 0.3 wt pct
manganese and various amounts of copper are given in their
Figures 19 and 20. While their data for 0.5 and 1 wt pct
copper show a strong decrease of the ferrite growth rate,
the curve for a 0.3 wt pct copper addition shows the op-
posite effect with which the preceding results are in good
agreement. It was thus proposed that the pearlite promoter
effect of copper should be related only to large additions
(typically more than say 0.5 wt pct) of this element.[7]
2. Effect of high level addition of copper
The curves relating to the base alloy L1 and to the alloys
L7 and HCM with both Mn and Cu added are compared
Fig. 10—Comparison of the DTA traces obtained at 5 K/min on samples in Figure 11, again for a cooling rate of 5 K/min. Although
from alloys L1 (Fe-C-Si base alloy), L5 (0.4 wt pct Mn added), and L7 the manganese content was not strictly the same in these
(0.4 wt pct Mn and 0.2 wt pct Cu added). The initial temperature was 900 last two alloys, the effect of copper when added at a level
7C.
of 1 wt pct is clearly seen. It is worth noticing that the
DTA trace obtained with the HCM cast iron closely resem-
knowledge of the phase transformations in cast iron, it is bles the trace obtained in the case of alloy L5 with man-
self-evident to associate the first peak with the ferrite ganese alone, but is further shifted to lower temperatures.
growth and the second with the pearlite reaction.[11] It was The ferritic reaction started at about 735 7C and proceeded
observed that the sensitivity of the onset of the ferritic re- slowly, as can be seen by comparing this DTA trace with
action to the cooling rate was similar whatever the alloy those of the two other alloys. On the contrary, the pearlitic
composition,[10] and the data obtained were in good agree- reaction, which starts at about 710 7C, seems to be as fast
ment with the experimental results of Ekpoom and as in the case of alloy L5. Similar results were obtained for
Heine.[11] the other cooling rates investigated, with the ferritic reac-
In Figure 9 is plotted the DTA trace, which were ob- tion being increasingly less evident as the cooling rate was
tained at a cooling rate of 5 K/min for the base alloy L1 increased. This is in line with the metallographic observa-
and for alloys L5 and L9, with Mn and Cu added, respec- tions, which showed a corresponding decrease in the ferrite
tively. The slight intermediate bump at about 745 7C on the volume fraction.
trace related to the base alloy is attributed to the magnetic Finally, DTA experiments performed on the eutectoid
transformation of ferrite at the Curie temperature. It is first steel showed features that were quite similar to those ob-
noted in this figure that the Mn addition strongly affects tained on the HCM cast iron, except that the initiation of
the shape of the DTA trace. As expected from the literature, the austenite decomposition was less marked. This is illus-
the start of the ferritic transformation was slower and at a trated in Figure 12, where the DTA traces for the cast iron

2020—VOLUME 28A, OCTOBER 1997 METALLURGICAL AND MATERIALS TRANSACTIONS A

Fig. 11—Comparison of the DTA traces obtained at 5 K/min on samples
from alloys L1 (base), L7 (0.4 wt pct Mn and 0.2 wt pct Cu added), and
HCM cast iron (0.5 wt pct Mn and 1 wt pct Cu). The initial temperature
was 900 7C.

Fig. 13—Schematic isopleth Fe-C section of the Fe-C-Si diagram in (a)

the stable system and (b) the metastable one. The temperature Ta in
diagram (a) is the temperature at which the ferritic reaction can start when
the product phase inherits the silicon content of the parent austenite. The
temperature Tp in diagram (b) is the temperature at which the pearlitic
reaction can start when the two-phase mixture of ferrite and cementite
inherits the silicon content of the parent austenite.

and the steel obtained at 1 K/min are compared. This dif-

ference is due to the fact that ferrite formed both around
Fig. 12—Comparison of the DTA traces obtained when cooling samples graphite nodules and at grain boundaries in the cast iron,
of the HCM cast iron and of the 1 wt pct Cu steel at 1 K/min. The initial while it is only intergranular in the case of the steel. The
temperature was 900 7C. characteristic temperatures estimated from the DTA traces
for the HCM cast iron and the steel have been reported in
Table III. Similar results related to the alloys of the series
Table III. Effect of the Cooling Rate on the Temperature L have been reported previously.[10]
(in Degrees Celsius) for the Start of the Ferritic and
Pearlitic Reactions for the HCM Cast Iron and the Copper
Steel*
IV. DISCUSSION
Cooling HCM Cast Iron Copper Steel
A. The Ferritic Reaction
Rate Ferritic Pearlitic Ferritic Pearlitic
(K/min) Reaction Reaction Reaction Reaction One particular feature of the growth of ferrite during
1 744 719 730
cooling of cast irons is the fact that no redistribution of
2 748-746 716-716 740 726 substitutional elements at the ferrite/austenite interface has
5 738 711 726 been reported in the literature. In other words, ferrite in-
10 725-723 702-705 733 714 herits the alloying content in substitutional elements of the
2 738 713 parent austenite. This is the so-called paraferrite mentioned
10 722-722-723 698-698-698 724-726 709-707 by Venugopalan.[3] An isopleth Fe-C section of the stable
20 687 714 696 phase diagram, as schematically shown in the upper part
*The starting temperature was 900 7C for the first four lines, and of Figure 13, is thus relevant for describing the austenite
1100 7C for the others. phase before its decomposition and the ferrite phase dur-

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 28A, OCTOBER 1997—2021

 cooling with respect to the corresponding Ta temperature.
The characteristic temperatures T7a and Ta for the alloys
investigated have been calculated with the THERMOCALC
software[19] using the data compiled by Uhrenius;[20] they
are reported in Table I. It was assumed that the graphite
phase was pure carbon, such that the content in species i
of the parent austenite if chemically homogeneous is given
by
r ggg 1 rg z (12gg) o
wgi 5 z wi
rg z (12gg)
where r g and rg are the density of the graphite and austen-
ite, respectively; 7wi is the nominal i content of the cast
iron; and gg is the volume fraction of graphite. With r g 5
2300 kgzm23, rg 5 7000 kgzm23, and gg 5 0.13, one has
wgi ' 1.049 7wi.
Figure 14 shows the effect of the cooling rate on the
undercooling of the start of the ferritic reaction. The ex-
Fig. 14—Plot of the undercooling for the start of the ferritic reaction, perimental points have been plotted with various symbols
estimated with respect to Ta, vs the cooling rate. Circles, squares, according to the alloy composition and to the initial tem-
diamonds, up triangles, and down triangles represent, respectively, alloys perature in the austenite range, which was either 900 7C
L1, L5, L7, and L9 and HCM cast iron. Open symbols represent data (empty symbols) or 1100 7C (full symbols). As expected,
obtained during cooling from 900 7C and closed symbols that from 1100
7C. the undercooling for the onset of the growth of ferrite in-
creases with the cooling rate. A similar effect was observed
for the eutectic reaction, in which case, the kinetics are
ing its growth. With decreasing temperature, while re- controlled by carbon diffusion from the liquid to the graph-
maining above the upper limit of the three-phase ite nodules through the austenite shell.[21] The data related
ferrite/austenite/graphite field (denoted T7a on the figure), to the base alloy (L1) and to the alloys with either Mn or
the composition of the austenite around the graphite nod- Cu added (L5 and L9) are very close to each other and can
ules follows the austenite/graphite equilibrium line. At tem- be extrapolated to a zero undercooling at zero cooling rate.
perature T7a , the ferrite phase becomes stable according to This gives confidence to the description of the conditions
the phase diagram, but its growth is subjected to the con- for the ferritic reaction upon cooling, which was detailed
dition of nonpartitioning of the substitutional species. This previously. The undercoolings for alloys containing both
growth is controlled by redistribution of the carbon rejected Mn and Cu are about 5 to 10 K below the former data, but
at the ferrite/austenite interface, either to the parent austen- it should be emphasized that the effect of the cooling rate
ite in the case of intergranular ferrite or to the graphite on the absolute change in the undercooling is identical to
phase when ferrite forms halos around the nodules. the effect observed on the other alloys. Thus, it appears that
Intergranular proeutectoid paraferrite is generally not ob- this discrepancy is just a matter of a slight underestimate
served in cast iron, except in some special cases, e.g., when of the reference temperature Ta for these alloys. This is
a high level of copper is added, as in the HCM alloy pres- certainly due to thermodynamic interactions between Cu
ently studied, or when Sb or Sn is added.[8] Its growth in- and Mn, which are not accounted for in the data compiled
volves the very slow diffusion of carbon in the parent by Uhrenius.[20] Finally, it is observed in Figure 14 that the
austenite, so that the final amount of proeutectoid ferrite undercooling depends on the initial temperature, as seen
will always be quite low. Thus, one should focus on the when comparing the points represented by empty and full
most usual case, where ferrite develops as rings around the symbols for the 10 K/min cooling rate. This sensitivity is
graphite nodules. In this case, it is admitted that the decom- usually related to the limited kinetics of graphite growth
position of austenite is controlled by the diffusion of carbon during cooling of the alloy above the T7a temperature.
through the ferrite shell, from the austenite to the graphite Figure 15 shows isopleth Fe-C sections limited to the
phase, which acts as a carbon sink.[2,3,18] For this diffusion iron-rich part of the diagram and calculated, respectively,
to occur, the carbon content (this is equivalent to consid- for a ternary Fe-C-Si alloy with 2.5 wt pct Si and a qua-
ering the chemical potential in this case) at the fer- ternary Fe-C-Si-Mn alloy with 2.5 wt pct Si and 1 wt pct
rite/austenite interface must be higher than at the Mn. The interrupted lines on the figure are the calculated
ferrite/graphite interface. The upper part of Figure 13 shows extrapolation of the ferrite/austenite boundary of the ferrite
that this condition is fulfilled only at temperatures lower field. The addition of Mn stabilizes austenite with respect
than the lowest temperature of the three-phase austen- to ferrite, i.e., shifts the ferrite/austenite boundary to a lower
ite/ferrite/graphite field, denoted Ta in the figure. carbon content. On the contrary, the ferrite/graphite bound-
From the preceding description of the decomposition of ary is not affected by the Mn addition. When combined,
austenite upon cooling of cast iron, it is concluded that the these effects lead to a strong decrease of the driving force
temperature that should be used as the reference tempera- for the growth of ferrite rings in alloys with Mn additions.
ture for the start of the ferritic reaction is Ta and not T7a . It This is clearly the reason for the well-known effect that Mn
therefore appears useful to express the experimental tem- has on the kinetics of the ferritic reaction, and this does not
peratures for the onset of the ferritic reaction as under- involve a change of the carbon diffusion coefficient as gen-

2022—VOLUME 28A, OCTOBER 1997 METALLURGICAL AND MATERIALS TRANSACTIONS A

 Fig. 16—Plot of the undercooling for the start of the pearlitic reaction,
estimated with respect to Tp, vs the cooling rate. Circles, squares,
diamonds, up triangles, and down triangles represent, respectively, alloys
Fig. 15—Fe-C isopleth sections of the iron-rich part of the stable diagram, L1, L5, L7, and L9 and HCM cast iron. Open symbols represent data
calculated with, first, 2.5 wt pct Si and, second, with 2.5 wt pct Si and 1 obtained during cooling from 900 7C and closed symbols that from 1100
wt pct Mn. 7C. Plus signs and crosses refer to measurements made on the DTA traces
with the copper steel during cooling from the same two temperatures.

erally accepted.[3] Calculations made for a section with 1

wt pct Cu in place of Mn showed a similar trend, although must thus be cooled down to the lower limit of the three-
the decrease of the Ta temperature and of the driving force phase field, where a mixture of ferrite and cementite may
was much less pronounced. The combined effect of copper have the same composition as the parent austenite. It is
and manganese will be discussed in Section 3. assumed for simplicity that, during cooling in the three-
phase field, the composition of ferrite nuclei follows the
B. The Pearlitic Reaction extrapolation of the austenite/ferrite line. The intersection
of this line with the lower boundary of the three-phase field
From the results by Al-Salman et al.[22] on silicon steels, gives the point at which pearlite can actually start to grow
it is known that, above about 700 7C, pearlite grows with under the preceding assumptions. The corresponding tem-
full partitioning of silicon between ferrite and cementite. perature is noted Tp on the figure. The values of the char-
Tewari and Sharma[23] were able to reproduce the results of acteristic temperatures T7p and Tp were evaluated again
Al-Salman et al.[22] with the model developed by Hillert.[24] using THERMOCALC and are given in Table 1.
They assumed that cementite does not dissolve any silicon The effect of cooling rate on the undercooling for the
and that pearlite inherits the composition of the parent aus- onset of the pearlitic reaction, expressed with reference to
tenite. This latter point seems to be verified also in the case the Tp temperature, is illustrated in Figure 16. The results
of cast iron, as it has not been reported that the pearlite obtained with samples from the high copper steel have also
reaction leads to any build up of chemical heterogeneities been plotted. The data show a similar rate of increase in
at the relevant scale, i.e., the scale of the solidification mi- undercooling with the cooling rate for all alloy composi-
crostructure. tions. On the basis of a growth rate of pearlite of the order
The lower part of Figure 13 presents a schematic isopleth of 0.1 to 5 microns per second depending on the cooling
Fe-C section of the metastable phase diagram, which may rate, the 20 to 60 K undercoolings observed are in agree-
be used for the description of the conditions for the pearlitic ment with the data obtained by Al-Salman et al.[22] on a Fe-
reaction. In the following description, it is assumed that 0.84C-1.99Si alloy (the Tp temperature of which is 770 7C,
ferrite nucleated when the temperature of the material as calculated with THERMOCALC). It is of some impor-
reached the stable three-phase field, although it may not tance to note that the undercooling for the start of the pear-
have grown as emphasized previously. Cementite also be- litic reaction is much higher than the undercooling for the
comes thermodynamically stable with respect to the parent onset of the ferritic reaction in the investigated range of
austenite at the temperature T7p in the figure. As was the cooling rates. This is certainly due to the fact that nuclea-
case of ferrite, proeutectoid cementite is generally not ob- tion of pearlite is more difficult than nucleation of ferrite
served in SG cast irons at temperatures in the three-phase in cast irons.
field, because this would involve partitioning and long- Finally, Figure 17 shows the isopleth Fe-C section of the
range diffusion of substitutional species. Thus, ferrite and stable (continuous lines) and metastable (interrupted lines)
cementite must grow cooperatively in pearlite as only short- diagrams for 2.5 wt pct Si, 0.5 wt pct Mn, and 1 wt pct
range diffusion is necessary. However, pearlite could not Cu, i.e., the nominal alloying content of the HCM cast iron
inherit the composition of the parent austenite when grow- and of the steel under study. The Ta and Tp temperatures
ing at temperatures in the three-phase field.[10] The alloy are close to each other, but with Tp about 10 K higher than

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 28A, OCTOBER 1997—2023

 V. CONCLUSIONS
Using DTA, it has been observed that the cooling rate,
in the range of 1 to 20 K/min, affects the onset of the ferrite
growth in SG cast iron. The temperature for the start of this
transformation in a base Fe-C-Si alloy is decreased by al-
loying with a low level of Mn or Cu. However, when both
of these elements are added together, the effects counteract
one another. This confirms previous observations,[7] which
suggested that, when copper is added at a low level (0.2 wt
pct), the decomposition of austenite to ferrite plus graphite
is favored, although this occurs at a lower temperature than
in the base alloy. At a higher level (1 wt pct), the well-
known pearlite promoter effect of copper when combined
with manganese was experimentally confirmed.
A description of the conditions for the growth of ferrite
and of pearlite has been given, which shows that these
reactions can proceed upon cooling only when the tem-
perature of the alloy is below the lower limit of the re-
lated ferrite/austenite/graphite or ferrite/austenite/cementite
three-phase field. The experimental results might then be
conveniently understood when the appropriate reference
Fig. 17—Fe-C isopleth section of the stable (continuous lines) and
metastable (interrupted lines) diagrams calculated with 2.5 wt pct Si, 0.5 temperature is used to plot the effect of the cooling rate on
wt pct Mn, and 1 wt pct Cu. the undercooling of the reaction. The effect of the cooling
rate on the temperature of the onset of the ferrite growth is
thus found to be similar to that observed for the eutectic
Ta. It is seen from Figures 14 and 16 that, at a cooling rate reaction, with a measured undercooling that can be extrap-
of, say, 10 K/min, the undercooling for the onset of the olated to a zero value when the cooling rate is zero.
growth of ferrite is about 20 to 30 K, while it is 40 to 50 Concerning the growth kinetics of ferrite, it has been
K for the pearlitic reaction. Thus, taking into account the shown that the effect of Mn addition is due to an efficient
10 K difference between the two reference temperatures, it decrease of the driving force (expressed in terms of a dif-
appears that the decomposition of austenite may start at ference of carbon content in ferrite at the ferrite/austenite
about the same temperature (or time upon cooling) in both and ferrite/graphite interfaces) for the diffusion of carbon
the stable and metastable systems. This is indeed what was through the ferrite ring. Copper has a similar effect, al-
observed by Pan et al.[4] in the case of alloys with 0.5 or 1 though it is much less marked. Manganese decreases
wt pct Cu. In agreement with the present study and previous strongly the temperature gap between the stable and met-
works,[4,5] lower cooling rates lead to a slightly larger astable three-phase fields, while copper increases it. Alloy-
amount of ferrite, while higher cooling rates give a nearly ing with 1 wt pct copper and 0.5 wt pct manganese brings
fully pearlitic matrix. the reference temperatures for the stable and metastable eu-
In relation with Figure 15, we emphasized the fact that tectoid reactions close to each other and appears to be a
Cu affects the Fe-C isopleth section in the austen- good compromise for obtaining fully pearlitic SG cast irons
ite/ferrite/graphite system to a much lesser extent than does without the presence of coarse cementite that high levels
Mn. The opposite is true in the austenite/ferrite/cementite of manganese promote.
system, where Cu significantly lowers the Tp temperature,
while Mn leaves it nearly unchanged. At 1 pct Cu and with-
out Mn, the metastable three-phase field is thus shifted by ACKNOWLEDGMENTS
about 20 K to lower temperatures with respect to the three- This study was financially supported by Péchiney Elec-
phase field in the stable diagram. This temperature gap ex- trométallurgie. The authors thank Claude Bourgraff
plains the observed fact that copper alone could not give (LSG2M, Ecole des Mines de Nancy, Nancy, France) for
fully pearlitic structures in usual cooling conditions.[4] Such the kind help with the DTA experiments.
a structure could indeed be obtained by means of a low
level addition of Mn to cast irons previously alloyed with
1 wt pct Cu, as illustrated previously. Manganese is an REFERENCES
efficient pearlite promoter because it depresses Ta without
affecting Tp and decreases the driving force for the growth 1. K.E. Spray: Metals Handbook, ASM, Metals Park, OH, 1981, vol. 4,
pp. 545-51.
of ferrite as rings around the graphite nodules. However, 2. W.C. Johnson and B.V. Kovacs: Metall. Trans. A, 1978, vol. 9A, pp.
high additions of Mn are generally avoided because they 219-29.
favor the presence of coarse eutectic cementite precipitates, 3. D. Venugopalan: in Fundamental and Applications of Ternary
which are detrimental for mechanical properties. Alloying Diffusion, G.R., Purdy ed., Pergamon Press, New York, NY, 1990,
with both Cu and Mn thus appears to be the best compro- pp. 173-83.
4. E.N. Pan, M.S. Lou, and C.R. Loper: AFS Trans., 1979, vol. 87, pp.
mise for obtaining a pearlitic structure without resorting to 819-40.
species such as Sn or Sb, which are difficult to control in 5. M.J. Lalich and C.R. Loper: AFS Trans., 1973, vol. 81, pp. 217-28.
practice. 6. D.R. Askeland and S.S. Gupta: AFS Trans., 1975, vol. 83, pp. 313-20.

2024—VOLUME 28A, OCTOBER 1997 METALLURGICAL AND MATERIALS TRANSACTIONS A

 7. J. Lacaze, S. Ford, C. Wilson, and E. Dubu: Scand. J. Metall., 1993, 17. J. Lacaze and G. Lesoult: in Nature and Properties of Semi-Solid
vol. 22, pp. 300-309. Materials, and J.A. Sekhar and J.A. Dantzig, eds., TMS, Warrendale,
8. B.V. Kovacs: AFS Trans., 1981, vol. 89, pp. 79-96. PA, 1991, pp. 105-42.
9. J. Ågren: Acta Metall., 1982, vol. 30, pp. 841-51. 18. B.F. Brown and M.E. Hawkes: AFS Trans., 1951, vol. 59, pp. 181-
10. J. Lacaze, C. Wilson, and C. Bak: Scand. J. Metall., 1994, vol. 23, 200.
pp. 151-63. 19. B. Sundman, B. Jansson, and J.-O. Andersson: CALPHAD, 1985, vol.
11. U. Ekpoom and R.W. Heine: AFS Trans., 1978, vol. 86, pp. 281-86. 9, pp. 153-90.
12. J. Ågren: Scand. J. Metall., 1982, vol. 11, pp. 3-8. 20. B. Uhrenius: Hardenability Concepts with Applications to Steel, TMS-
13. E.E. Underwood: Quantitative Stereology, Addison-Wesley, Reading, AIME, Warrendale, PA, 1977, pp. 28-81.
MA, 1970. 21. J. Lacaze, M. Castro, N. Aichoun, and G. Lesoult: Mem. Et. Scient.
14. J.M. Theret and J.C. Lecomte: in Quantitative Analysis of Rev. Met., 1989, pp. 85-97.
Microstructures in Materials, Science, Biology and Medicine, J.L. 22. S.A. Al-Salman, G.W. Lorimer, and N. Ridley: Acta Metall., 1979,
Chermant, ed., Dr. Riederer Verlag GmbH, Stuttgart, 1978, pp. 153- vol. 27, pp. 1391-1400.
60. 23. S.K. Tewari and R.C. Sharma: Metall. Trans. A, 1985, vol. 16A, pp.
15. T. Skaland and O. Grong: AFS Trans., 1991, vol. 99, pp. 153-57. 597-603.
16. E.A. Feest, G. McHugh, D.O. Morton, L.S. Welch, and I.A. Cook: 24. M. Hillert: in Mechanism of Phase Transformation in Crystalline
Solidification Technology in the Foundry and Casthouse, The Metals Solids, Monograph No. 33, The Institute of Metals, London, 1968,
Society, London, 1983, pp. 232-39. pp. 231-47.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 28A, OCTOBER 1997—2025