See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/226229107

The effect of graphite flake morphology on the thermal diffusivity of gray

cast irons used for automotive brake discs

Article  in  Journal of Materials Science · October 1999

DOI: 10.1023/A:1004643322951

CITATIONS READS

67 3,831

3 authors, including:

Ralph B. Dinwiddie Heda Wang

Oak Ridge National Laboratory Waikato Institute of Technology
120 PUBLICATIONS   2,413 CITATIONS    104 PUBLICATIONS   3,838 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Carbon-Carbon Aircraft Brake Materials View project

Additive Manufacturing - Metals View project

All content following this page was uploaded by Ralph B. Dinwiddie on 15 May 2015.

The user has requested enhancement of the downloaded file.

J O U R N A L O F M A T E R I A L S S C I E N C E 3 4 (1 9 9 9 ) 4775 – 4781

The effect of graphite flake morphology

on the thermal diffusivity of gray cast
irons used for automotive brake discs
R. L. HECHT
Vehicle Safety Research Department, Ford Research Laboratory, Dearborn, MI 48121-2053
E-mail: [email protected]

R. B. DINWIDDIE, H. WANG
High Temperature Materials Laboratory, Oak Ridge National Lab, Oak Ridge, TN 37831

Thermal diffusivity of automotive grade SAE G3000 (d) gray cast iron has been measured
as a function of graphite flake morphology, chemical composition and temperature. Cast
iron samples used for this investigation were cut from “step block” castings designed to
produce iron with different graphite flake morphologies resulting from different cooling
rates. Samples were also machined from prototype and commercial brake rotors, as well
as from a series of cast iron slugs with slightly varying compositions. Thermal diffusivity
was measured at room and elevated temperatures via the flash technique. Graphite flake
morphology of the various cast iron samples was quantified stereologically with image
analysis techniques. Several geometric features of the graphite flake morphology were
quantified. It was found that the thermal diffusivity of these gray cast irons increases with
carbon equivalent and has a strong linear correlation to graphite flake length. For gray iron
with the same chemical composition, a four fold increase in the graphite flake size results in
a 50% increase in thermal diffusivity. Amongst the commercial rotors, room temperature
thermal diffusivity varied from 0.156 to 0.200 cm2 /s. °C 1999 Kluwer Academic Publishers

1. Introduction The phase of the cast iron matrix influences diffusivity

Gray cast iron has been the primary material for au- a small degree, and ferrite has a higher diffusivity value
tomotive brake rotors (discs) because of its excellent than pearlite [3, 4].
thermal transport ability and its damping properties. A Most rotors on production passenger cars today are
brake rotor must be able to dissipate the frictional heat gray cast iron with Type A or B graphite flakes (AFS-
generated by application of the brake pads during a stop. ASTM A247 designation) embedded in a pearlite ma-
One way to improve the heat transfer ability of a rotor is trix. However, there are variations in chemical com-
to increase the material’s thermal diffusivity. Thermal position and processing between foundries which lead
diffusivity describes the rate of heat propagation dur- to microstructural differences between rotors produced
ing transient processes, and so is a fundamental material to the same specification. Solidification rate impacts
parameter for the design of brakes. Thermal diffusiv- graphite flake morphology, hence different casting
ity can be used to calculate thermal conductivity, the mold configurations (e.g. number of parts/mold) or
steady-state measure of how a material transmits heat. new component designs may result in slightly different
Gray cast iron rotors with improved thermal conductiv- graphit flake morphologies. Knowing how morphology
ity have shown increased resistance to heat cracks [1], impacts thermal transport allows assessment of how a
possibly improving the service life of the brake. If the variation in disc production will impact thermal man-
thermal transport ability of gray cast iron can be in- agement in the rotor.
creased without sacrificing other design criteria, brake Thermal diffusivity was measured at room and ele-
performance may be enhanced. vated temperatures for pearlitic gray cast irons to de-
Graphite inclusions have a significantly higher ther- termine whether subtle differences in morphology alter
mal diffusivity than the matrices of cast irons. Distri- heat transfer. In order to isolate the effects of graphite
bution of the graphite phase is the primary influence on flake shape from the influences of composition, test
thermal transport. Nodular iron, where graphite is in specimens were made from “step block” castings de-
spherical form, exhibits lower conductivity than com- signed to have 5 different cooling rates and hence dif-
pacted graphite, which in turn has a lower conductivity ferent graphite flake morphologies in each step. Steps
than gray iron with its flake graphite; as graphite flakes are labeled 1–5, with Step 1 being the thinnest step with
get longer, diffusivity and conductivity increase [2–5]. the fastest cooling. Variations in chemical composition

0022–2461 °
C 1999 Kluwer Academic Publishers 4775
were considered by cutting material specimens from metallography, avoiding the use of wet grinding to
production and prototype ventilated front rotors of sim- prevent graphite flake pullout. Metallographic mounts
ilar dimensions, labeled here US #1, US #2, US #3, were observed by optical microscopy in an unetched
German, Japanese, Prototype US and Prototype Eu- condition. Image analysis was performed with a Nikon
rope. In order to isolate the influences of composition, optical microscope and Image Pro Plus software to
8 cast iron alloys with slight compositional variations, quantify the structure, size and shape of the graphite
were cast in a large cylindrical mold. Test specimens flakes. Chemical composition of each cast iron alloy
were cut from these castings designated as Alloys 1–8. was determined on a JOEL optical emission spectrom-
eter (OES), and LECO combustometric analysis was
used to determine the weight percents of C and S.
2. Experimental
Room temperature thermal diffusivity was measured by
the flash technique [6] using a xenon flash system at the 3. Results and discussion
High Temperature Materials Laboratory at Oak Ridge The chemical composition in wt % of the alloys ap-
National Laboratory. To perform this measurement, one pears in Table I. Fig. 1 shows that for the rotor and cast
face of a sample of known thickness is subjected to an iron alloy samples there is a positive, almost linear de-
intense, brief heat pulse, and the time for the pulse to pendence of thermal diffusivity with increasing carbon
propagate to the opposite face of the sample is recorded. equivalent (CE), where CE = % C + 1/3(% Si+ % P).
The diffusivity, fi, is calculated from the time it takes This increase is expected since increasing the % C or
for the rear surface to achieve 1/2 of its maximum tem- CE is the most direct way to improve the graphitization
perature rise, in gray iron. Amongst the brake rotor samples, there is a
25% increase in diffusivity with a 0.4% increase in CE.
fi = C D 2 /t0.5 (1) The general linear trend seen in part of Fig. 1 extends
beyond the carbon % range of the rotors studied here
where C is a dimensionless parameter, D is the speci- (3.43–3.79% C) to 2.5–4.0% C [1]. Other research has
men thickness, and t0.5 is the half rise time of the rear demonstrated the effect of increasing thermal diffusiv-
surface temperature. Analysis software determines the ity with increasing % CE from 3.6 to 4.8% CE [11].
value of C using Koski’s parameter estimation tech- In order to isolate the effect on thermal diffusivity of
nique [7] applied to Clark and Taylor analysis [8]. Clark the graphite shape from the composition (or CE), “step
and Taylor’s analysis includes the effect of radiation block” castings were produced to gray iron G3000 spec-
heat losses. ification. A step block casting is shaped like a staircase;
Specimens used for the diffusivity measurements step heights were 3.175, 6.35, 12.7, 25.4, and 50.8 mm,
are disk-shaped, nominally 12.7 mm in diameter and with each step being 124 mm wide and 56 mm deep.
3.18 mm thick. Before testing, each specimen was mea- Because of the difference in step thickness, each step
sured and weighed to determine its density and thick- experiences a different solidification and cooling rate,
ness, and coated with carbon to maximize the energy producing different graphite flake morphologies in ma-
absorption. Five diffusivity measurements of each spec- terial with the same composition. Although all of the ro-
imen were made at room temperature; between 3 and 8 tor and alloy samples have a fully pearlitic microstruc-
specimens were tested for each alloy and each test con- ture with only trace amounts of ferrite, the step block
dition, assuring a statistically reliable and representa- castings contain small amounts of ferrite in the pri-
tive value for each sample and for each alloy. marily pearlitic matrix. The thinnest step, Step 1, con-
The elevated temperature thermal diffusivity was me- tains approximately 4.9 vol % ferrite, Step 2 contains
asured on a system equipped with a moveable Nd:glass 3.8 vol %, and the amount of ferrite drops to a minimum
laser, a cryogenically cooled InSb IR detector, and four in Step 5 of 0.4 vol % ferrite.
furnaces [9]. A low temperature aluminum block fur-
nace was used to achieve the test temperatures of 200◦ –
500 ◦ C. Oxidation of the cast iron samples was mini-
mized by flowing ultra high purity argon through the
furnace. Each specimen was measured three times at
each elevated temperature. The Koski parameter esti-
mation technique applied to Cowen’s analysis [10] was
used to determine the value of C in Equation 1 and to
generate the diffusivity results for the elevated temper-
ature results.
Automotive disc brakes operate at high temperatures
and are subject to the influences of environment. An-
other goal of this work was to determine how the pres-
ence of oxide scale effects thermal diffusivity. Rotor
samples were oxidized in a furnace at 500 ◦ C for 1, 10
and 270 h. Thermal diffusivity was subsequently mea-
sured at room temperature.
To quantify the graphite flake morphology, diffusiv- Figure 1 Room temperature thermal diffusivity increases as a function
ity specimens were mounted and prepared by standard of carbon equivalent (CE). Error bars are ± one standard deviation.

4776
T A B L E I Chemical compositions (wt %)

Rotors
Cast iron alloys
Proto. Proto. Step
Japanese German US #1 US #2 US #3 Europe US blocks Alloy 1 Alloy 2 Alloy 3 Alloy 4 Alloy 5 Alloy 6 Alloy 7 Alloy 8

Carbon 3.69 3.79 3.51 3.45 3.43 3.78 3.79 3.64 3.39 3.42 3.40 3.61 3.60 3.63 3.50 3.70
Manganese 0.56 0.74 0.88 0.65 0.71 0.59 0.51 0.68 0.65 0.71 0.84 0.83 0.67 0.70 0.89 0.89
Phosphorus 0.03 0.05 0.01 0.02 0.03 0.03 0.04 0.05 0.09 0.11 0.10 0.09 0.08 0.10 0.11 0.11
Sulfur 0.10 0.09 0.10 0.08 0.13 0.10 0.16 0.01 0.12 0.12 0.12 0.12 0.13 0.12 0.12 0.15
Silicon 2.14 2.01 1.87 2.30 2.43 2.16 1.90 2.20 2.12 2.26 2.33 2.23 2.30 2.28 2.26 2.25
Copper 0.05 0.19 0.95 0.10 0.28 0.08 1.20 0.10 0.27 0.29 0.29 0.31 0.33 0.29 0.18 0.20
Nickel 0.03 0.10 0.01 0.08 0.12 0.08 0.07 0.01 0.13 0.13 0.13 0.13 0.13 0.12 0.12 0.12
Chromium 0.05 0.12 0.24 0.30 0.31 0.25 0.14 0.31 0.21 0.40 0.39 0.22 0.37 0.20 0.19 0.43
Iron Base Base Base Base Base Base Base Base Base Base Base Base Base Base Base Base
Carbon 4.41 4.48 4.14 4.22 4.25 4.51 4.44 4.39 4.12 4.20 4.20 4.38 4.39 4.42 4.29 4.48
Equivalent

T A B L E I I Microstructural characteristics of gray cast iron diffusivity specimens

ASTM
Average flake Maximum flake Surface/Volume Aspect Fields of view
Sample length (mm) length (mm) (1/mm) ratio analyzed Flake type Size class

Step 1 (3.18 mm) 14 195 0.88 2.6 27 D, A 3–4

Step 2 (6.35 mm) 42 383 0.45 4.4 20 A 2
Step 3 (12.7 mm) 58 556 0.40 6.2 20 A 2
Step 4 (25.4 mm) 89 685 0.37 5.4 19 A 1
Step 5 (50.8 mm) 94 881 0.19 6.5 10 A 1
German 47 547 0.39 3.2 10 A, B 2
Japanese 71 710 0.34 4.8 10 A 1
Proto. Europe 43 575 0.41 3.1 20 A, B 2
Proto. US 52 655 0.44 4.4 20 A 1–2
US #1 51 403 0.32 3.8 10 A 2
US #2 46 469 0.38 3.1 10 A 2
US #3 49 564 0.39 3.5 20 A 2

Commercial image analysis software was used to Length” is the maximum value of L observed for a
measure the graphite flake morphology parameters such particular iron over the all the fields of view. It should
as flake length, area, major shape axis, minor shape axis be noted that only flakes fully visible in the field of
and shape perimeter on unetched mounts. These param- view were measured and tabulated as is necessary for
eters are defined in the schematic pictured in Fig. 2. accurate assessment of average length values. A three
Table II lists average microstructural characteristics of dimensional property important for heat transfer is sur-
the graphite found in each step and in the brake ro- face area/volume; in order to estimate this property for
tors. The “Average Flake Length” is simply the average the graphite from two dimensional images, the average
value L from all the flakes measured for a particular graphite shape perimeter, P was divided by the area
iron over all the fields of view. The “Maximum Flake enclosed, A, and this value is included in Table II as
“Surface/Volume.” The “Aspect Ratio” was defined as
major axis/minor axis or X /Y .
Examples of unetched optical images used for mea-
suring graphite parameters are shown in Fig. 3. Actual
diffusivity samples were mounted, polished and used
for the microstructural measurements, hence the num-
ber of fields of view analyzed was limited by the size
of the samples. The number of fields of view analyzed
for each sample type is listed in Table II; this corre-
sponds to measuring between 5000 and 12000 flakes.
Alloys 1–8 were not used in the microstructural study
because they were all cast in the same large cylindrical
mold which resulted in very large flake sizes, e.g. max-
imum flake lengths of 1000–1400 „m. With such large
Figure 2 This schematic shows how the parameters used to characterize
graphite flake morphology are defined. A commercial image analysis
flakes sizes statistically valid assessment of the graphite
software program was used to measure the dimensions defined here: P, morphology was not possible on the small diffusivity
A, L, X , and Y , for thousands of flakes in each diffusivity sample. samples.

4777
Figure 3 Optical images of gray iron from Step 1, Step 3, Step 5 and US #2. Step 1 contains Type D and Type A graphite; other samples have only
Type A graphite flakes.

The average flake length-diffusivity relationship is length, as seen in Fig. 4. The step block specimens,
shown in Fig. 4a, and maximum observed flake length which all come from the same casting, illustrate the
vs. diffusivity is plotted in Fig. 4b. In Fig. 4, the dif- graphite length effect clearly because compositional in-
fusivity value of a particular sample is plotted against fluences are reduced. The rotors vary in composition but
the flake parameter measured on that same sample. For still show a strong trend of increasing diffusivity with
the step blocks it was found that both the average flake maximum flake length. For gray irons with a similar
length and the maximum flake length exhibited a vir- flake size, CE is a better predictor of diffusivity. (Flake
tually linear relationship with diffusivity; longer flakes size does not necessarily increase with increasing CE.)
result in greater diffusivity. For the step blocks, a lin- Consistent with the diffusivity-length relationship, it
ear correlation of diffusivity and average graphite flake was also noted that as the aspect ratio of the graphite
length has an R 2 value of 0.90 (Fig. 4a) while the lin- flakes increases, thermal diffusivity increases. Fig. 5
ear fit with maximum flake length was not as good, shows this remarkably linear relationship for the step
with R 2 = 0.71 (Fig. 4b). Considering Fig. 4 and the block samples (R 2 = 0.99), and again the rotor diffu-
data in Table II, the presence of Type D (interdendritic) sivity values appear to be influenced by more than just
graphite, as well as the shorter Type A flake size in graphite shape, and do not neatly follow the linear in-
Step 1, drastically reduced thermal diffusivity in those crease.
samples. Although the matrix also contributes to diffu- Figs 4 and 5 confirm that the length and shape of
sivity, and ferrite is a better heat conductor than lamel- the graphite effects heat transport. Graphite in gray
lar pearlite, the small amount of free ferrite present in iron is an interconnected three-dimensional network;
Steps 1 and 2 is not enough to compensate for the lower what is known as a “graphite flake” in a two dimen-
diffusivity resulting from the interdendritic graphite sional metallographic image is a curved basal section
and shorter Type A flakes. through the crystal or an edge-on view of the basal
Diffusivity specimens from each rotor were analyzed planes [12]. The thermal transport ability of graphite
to determine if the graphite flake length also accounts in the basal plane significantly exceeds its value along
for the variation found in the rotors’ diffusivity. For the the C-axis. Thermal conductivity along the C-axis is
rotor samples, maximum flake length showed a stronger approximately 84 W/mK but along the basal plane is
agreement with the step blocks than the average flake 293–419 W/mK [12]. A gray iron alloy with longer

4778
 (a)

(b)

Figure 4 (a) The average flake length measured on step block and rotor samples exhibits an increasing relationship with the room temperature thermal
diffusivity. Line is a fit through the step block sample results only. (b) For both the step blocks and the rotor samples, room temperature thermal
diffusivity increased as the maximum observed graphite flake length increased. Linear correlation shown is for step block samples only.

(or higher aspect ratio) flakes therefore has more of flake length or that other factors limited the diffusivity
the basal planes available for heat flow, and hence that of the gray iron from these large castings. It should be
cast iron will exhibit a higher diffusivity (and thermal noted that these castings contained significant poros-
conductivity). ity which would reduce diffusivity. The large volume
Cast iron slug samples (Alloys 1–8) all had aver- of the mold (a cylinder with approximately 160 mm
age flake lengths of 80–100 „m, and maximum flake diameter and 300 mm height) means that significant
lengths of 1000–1400 „m due to the drastically slower segregation could occur during solidification. Chem-
cooling rate in this mold configuration compared to ical analysis was not performed on the same samples
the cooling rate in the commercial rotors. However, that were used to measure diffusivity, hence the average
these gray irons did not follow the trends observed in compositions listed in Table I might vary for individ-
Figs 4 and 5. Diffusivity values were comparable to ual samples. Another possible cause for the discrep-
the commercial rotors (Fig. 1.) even through the flakes ancy relates to oxidation; oxidation can occur along the
were longer. This suggests that either there is a limit to graphite-pearlite interfaces and longer flakes mean that
how much the diffusivity can be improved by increased oxidation could reach greater depth below the surface.

4779
Figure 5 For the step block samples, room temperature thermal diffusivity increased linearly with graphite flake aspect ratio; this relationship did not
exist for the rotor samples. Error bars represent ± one standard deviation as in previous figures.

Figure 6 Thermal diffusivity of gray cast iron decreases with temperature. The difference in thermal diffusivity between gray iron variations diminishes
at elevated temperature.

At this time, it is not clear why the graphite structure with results from Ref. 13, whose authors measured dif-
analysis for these gray irons is not consistent with the fusivity at 300 ◦ C.
other data. Since brake rotors operate in a corrosive environ-
Elevated temperature diffusivity was measured by ment and at high temperatures, the effects of oxide scale
a laser flash system at temperatures between 200◦ to on the diffusivity were investigated. Room temperature
500 ◦ C for Steps 1–4 and several of the cast iron rotors. thermal diffusivity dropped following 270 h of expo-
Diffusivity decreases with temperature as shown in sure to 500 ◦ C, as shown in Fig. 7. Insignificant changes
Fig. 6, where the average diffusivity value for each tem- in the pearlitic matrix would be expected following so
perature is plotted for several gray irons. Gray iron’s short an exposure at 500 ◦ C, and the graphite flake mor-
ability to transport heat decreases as its temperature phology would not be affected at all. This diffusivity
increases. Fig. 6 shows that the difference in diffusiv- change can be attributed to the growth of oxide scale on
ity for different cast irons is less significant at elevated the sample surfaces. The largest influence was observed
temperatures. The results agree reasonably well with in the German rotor specimen which experienced a 14%
elevated temperature diffusivity values reported for drop in diffusivity. Although there was no measurable
“coarse flake” gray iron from 200◦ –700 ◦ C [3]. Data weight gain to quantify the amount of scale developed it
at 200 ◦ C reported in Ref. 3 is included on Fig. 6, along was visible on the sample surfaces. The German rotor

4780
 Figure 7 Room temperature thermal diffusivity of gray irons dropped with increasing exposure to 500 ◦ C. Error bars represent ± one standard
deviation.

samples have greater flake sizes than the other sam- Office of Transportation Technologies, as part of the
ples and therefore may be more vulnerable to deeper High Temperature Materials Laboratory User Program,
oxide intrusion. Results in Fig. 7 suggest that if ox- Oak Ridge National Laboratory, managed by Lockheed
idation occurs while in service gray iron brake rotors Martin Energy Research Corp. for the US Depart-
may experience a significant drop in their heat transport ment of Energy under contract number DE-AC05-
ability. 96OR22464.

References
4. Conclusions 1. Y . J I M B O , T .MIBE, K. AKIYAMA, H. MATSUI, M.
It was shown that thermal diffusivity is influenced by Y O S H I D A and A . O Z A W A , SAE International SAE 900002
changes in chemical composition of gray cast iron, and (1990) 22.
that a roughly linearly increasing relationship exists be- 2. K . B . P A L M E R , J. BCIRA 8(540) (1960) 266.
3. J . O M E R O D , R . E . T A Y L O R and R . J . E D W A R D S , Met.
tween room temperature diffusivity and carbon equiv-
Tech. April (1978) 109.
alent (CE) within the carbon range of gray cast iron. 4. D . F I T Z G E O R G E and J . A . P O P E , Transactions of the North
Figs 4 and 5 show that graphite flake morphology and East Coast Institution of Engineers and Shipbuilders 75 (1959) 284.
geometry affect diffusivity in G3000 standard automo- 5. R . L . H E C H T , R . B . D I N W I D D I E , W . P O R T E R and H .
tive gray iron. For a constant CE, as in the step block W A N G , SAE Technical Paper Series 962126 (1996).
6. W . J . P A R K E R , R . J . J E N K I N S , C . P . B U T L E R and G . L .
castings, average flake length, maximum flake length
A B B O T T , J. Appl. Phys. 32 (1961) 1679.
and aspect ratio for Type A graphite flakes show a strong 7. J . A . K O S K I , in Proceedings of the 8th Symposium of Ther-
linear correlation with thermal diffusivity; the longer mophysical Properties, Gaithersburg, MD, June 1981 (ASME, New
the flakes the greater the diffusivity. For the compo- York, 1981) Vol. 2, p. 94.
sition range within the rotors tested, maximum flake 8. L . M . C L A R K and R . E . T A Y L O R , J. Appl. Phys. 46 (1975)
714.
length proved to be a reasonable indicator of diffusivity.
9. H . W A N G , R . B . D I N W I D D I E and P . S . G A A L , in “Ther-
Diffusivity decreases in gray cast irons as the tempera- mal Conductivity 23,” edited by K. E. Wilkes et al. (Technomic
ture increases up to 500 ◦ C. At elevated temperatures, Publishing Co., Lancaster, PA, 1996) p. 119.
the difference in diffusivity observed between alloys de- 10. R . D . C O W A N , J. Appl. Phys. 34 (1963) 926.
creased, and the materials had similar values at 500 ◦ C. 11. S . Y A M A D A and T . K U R I K U M A , Paper F98T683, Pro-
ceedings from 1998 FISITA World Automotive Congress, Paris,
These data suggest that higher diffusivities and thermal
September–October 1998.
benefits may be achieved through control of the casting 12. H . T . A N G U S , “Cast Iron: Physical and Engineering Properties,”
process to produce gray cast iron rotors with long Type (Butterworths, London, 1976) p. 126.
A graphite flakes for a particular alloy composition. 13. P . F E R D A N I , R . A . E M M E T T and J . D . H O L M E , Paper
25, Leading Through Innovation T&N Symposium, Wurzburg, In-
dianapolis, 1995.
Acknowledgement
Research partially sponsored by the Assistant Sec- Received 19 February 1998
retary for Energy Efficiency and Renewable Energy, and accepted 23 March 1999

4781

View publication stats