Materials Transactions, Vol. 50, No. 6 (2009) pp.

1553 to 1557
#2009 The Japan Institute of Metals EXPRESS REGULAR ARTICLE

Experimental Measurement of Coefficient of Thermal Expansion for Graded Layers

in Ni-Al2 O3 FGM Joints for Accurate Residual Stress Analysis
Sae-hee Ryu1 , Jong-ha Park1 , Caroline Sunyong Lee1; * , Jae-chul Lee2 ,
Sung-hoon Ahn2;3 and Sung-tag Oh4
1
Division of Material and Chemical Engineering, Hanyang University, Gyeonggi-do, Korea
2
School of Mechanical and Aerospace Engineering, Seoul National University, Seoul, Korea
3
Institute of Advanced Machinery and Design, Seoul National University, Seoul, Korea
4
Department of Materials Science and Engineering, Seoul National University of Technology, Seoul, Korea

Functionally graded materials have composition gradients from one end to the other as the result of a gradual transition of the properties of
different materials. The residual stress caused by the difference of coefficient of thermal expansion can be minimized using functionally graded
material. Therefore, the gradient of the coefficient of thermal expansion should vary according to the compositional gradient. In this study, the
coefficient of thermal expansion of each compositional layer of Ni-Al2 O3 functionally graded material was measured using a dilatometer. These
measurements provided the material properties required to calculate the residual stress, using three-dimensional modeling for accurately
predicting crack positions, since it is difficult to measure residual stress experimentally. The measurement results showed the gradual increase of
the coefficient of thermal expansion from Al2 O3 -rich composition to Ni-rich composition. Finally, the results of calculating residual stresses
using the measured coefficient of thermal expansion showed that the crack positions were predicted more accurately than those using the
coefficient of thermal expansion calculated by the linear rule of mixtures. This was because the measured values include the effect of porosity of
the composite, whereas the linear rule of mixtures cannot account for the porosity of each layer. [doi:10.2320/matertrans.M2009028]

(Received January 26, 2009; Accepted March 31, 2009; Published May 20, 2009)
Keywords: functionally graded material, coefficient of thermal expansion, finite element method, thermal residual stress

1. Introduction Table 1 Compositions of samples for CTE measurement of nine-layer Ni-

Functionally graded material (FGM) provides a gradient in
thermal properties between two dissimilar materials whose
coefficients of thermal expansion (CTE) are significantly
different, and reduces the residual stresses due to sharp
differences in the thermal properties. The residual stress
is an important factor in the design of crack-free FGM joints,
but it is difficult to measure experimentally. Numerical
analysis techniques such as the finite element method (FEM)
are used to estimate the residual stress instead of measuring it
directly.
The thermal residual stresses are calculated using the CTE
of materials, assuming that the stresses are generated during
cooling. The accuracy of the CTE values directly influences
the calculation of residual stresses. In particular, since
thermal expansion is really due to the asymmetric curvature
of the potential energy trough, along with the increased
atomic vibrational amplitudes that accompany an increase in
temperature, it reflects the natural change of thermal proper-
ties with temperature.1) Generally, the CTEs of graded layers
using the values cited in published literature are used to
calculate the residual stresses of FGM using the rule of
mixtures.2) However, unexpected pores within graded layers
due to differential shrinkage of each material can result in
inaccurate calculations.3)
In this study, the CTE of each graded layer in Ni-Al2 O3
FGM was measured, and compared with the calculated CTE
using the linear rule of mixtures. The residual stresses in Ni-
Al2 O3 FGM were then calculated using these CTE values to
predict the position of cracks.
*Corresponding author, E-mail: sunyonglee@hanyang.ac.kr
Al2 O3 FGM.
Layer number Composition
1 100%Ni
2 85%Ni/15%Al2 O3
3 70%Ni/30%Al2 O3
4 60%Ni/40%Al2 O3
5 50%Ni/50%Al2 O3
6 30%Ni/70%Al2 O3
7 20%Ni/80%Al2 O3
8 10%Ni/90%Al2 O3
9 100%Al2 O3
2. Experimental Procedure
2.1 Material fabrication
The various compositions of FGM considered in this study
are given in Table 1. Ni and Al2 O3 powders were dispersed
ultrasonically in ethanol for each composition so that uneven
dispersion due to density differences could be reduced by
quick evaporation of the ethanol during the ultrasonic
dispersion.4) The mixed powders were grounded and sieved.
Nine specimens for CTE measurement were fabricated in
molds with 10 mm diameter. Nine layers of Ni-Al2 O3 FGM
were stacked in a mold with 24.5 mm diameter, and the green
bodies were cold pressed isostatically. Specimens were then
sintered in an argon atmosphere in a tube furnace as shown in
Fig. 1. The holding time at 150
C is needed to burn off
stearic acid that was used during stacking into a cylindrical
mold. The maximum sintered temperature was 1350
C with
a cooling rate of 2
C/min to minimize residual stress during
1554 S. Ryu et al.

1400 3hr

1200

1000
Temperature(°C)

800

2°C/min
600
4°C/min
400

200
30min
0
0 200 400 600 800 1000 1200

Time(min) Fig. 2 An axisymmetric finite element, showing stresses associated with

axisymmetric loading.
Fig. 1 FGM sample sintering conditions for CTE measurement.

cooling. The sintered FGM was cut with a diamond wheel,

and then the polished cross section was examined to detect
the presence of any cracks.

2.2 CTE measurement

The CTEs of the nine cylindrical specimens were
measured using a push-rod dilatometer. Each CTE was
determined from the change in the specimen length at room
temperature and 1000
C at a heating rate of 10
C/min. The
CTE was calculated from
1 L2  L1 1 L

¼  ¼ ð1Þ
L0 T2  T1 L0 T
where L0 is the initial specimen length, L1 and L2 represent Fig. 3 (a) Element condition and boundary condition and (b) PLANE82
the initial and final lengths as the temperature changes from geometry configuration.
T1 to T2 , and L and T are the change in length and
temperature, respectively. The CTE of each composition was 2   3
determined by averaging the measured CTEs over the various 1 0
6 1 1 78 9
temperature changes. The measurements were performed in 6   7>

T >
6 7>> >
>
6 1 0 7> >
an argon atmosphere to prevent any nickel oxidation. 61 1 7<

T =
6
6  
7
7> ð4Þ
6 1 0 7>

T >>
3. Calculation of thermal residual stress 61 7>> >
>
6 1 7: 0 ;
4 1  2 5
The residual stresses of the FGM were computed using the 0 0 0
finite element method (FEM); ANSYS. The residual stress of 2ð1  Þ
each layer was calculated individually. The thermal strain, This analysis assumes ideal joining and dispersion between
"T , of each isotropic layer was determined from the two materials. In addition, the elastic modulus, CTE, and
Poisson’s ratio of each layer are calculated based on the
f"T g ¼ f

Tg ð2Þ
linear rule of mixtures as follows:
where
is the CTE of the material and
T is the temperature Ec ¼ V 1 E 1 þ V 2 E 2 ð5Þ
change. The thermal stress, T , was calculated from the

c ¼ V1
1 þ V2
2 ð6Þ
thermal strain,
c ¼ V1 1 þ V2 2 ð7Þ
fT g ¼ ½Df"T g ð3Þ
where Ec ,
c , and c represent the elastic modulus, coefficient
where ½D is the stress-strain matrix derived from Poisson’s of thermal expansion, and Poisson’s ratio of the mixtures,
ratio () of the materials and the one-dimensional elastic respectively; V1 and V2 are the volume fractions of phases 1
modulus (E).5) and 2; and V1 þ V2 ¼ 1.7)
If the trivial relations r ¼ 0 and z ¼ 0 are omitted, the For finite element analysis, a two-dimensional eight-node
most general axisymmetric form of eq. (3) for the coordi- plane element (PLANE82) was used in the ANSYS. The two-
nates shown in Fig. 2 is as follows:6) dimensional axisymmetric model had 54,641 nodes and
8 9
> r > > 18,034 elements, and was meshed using 0.1 mm elements.
>
>
< > = ð1  ÞE Figure 3 shows (a) the elements and boundary conditions

¼ used for the finite element analysis, and (b) a schematic
>
>  >
z > ð1 þ Þð1  2Þ
: >
> ; diagram of PLANE82.8) Table 2 lists the material properties
zr of the FGM layers based on the properties obtained from the
 Experimental Measurement of Coefficient of Thermal Expansion for Graded Layers in Ni-Al2 O3 FGM Joints 1555

Table 2 Physical constants for calculating residual stresses.

Elastic modulus Measured CTE

Poisson’s ratio Measured CTE
Thickness & Linear rule of & Linear rule of
Layer Contents & Linear rule of values
[mm] mixtures mixtures
mixtures [106 /
C]
[MPa] [106 /
C]
1 100%Ni 0.72 207000 0.31 19.700 19.700
2 85%Ni/15%Al2 O3 1.69 220950 0.2965 17.825 17.350
3 70%Ni/30%Al2 O3 1.69 234900 0.283 15.950 15.000
4 60%Ni/40%Al2 O3 2.02 244200 0.274 14.700 13.900
5 50%Ni/50%Al2 O3 2.25 253500 0.265 13.450 12.200
6 30%Ni/70%Al2 O3 1.21 272100 0.247 10.950 9.500
7 20%Ni/80%Al2 O3 1.21 281400 0.238 9.700 8.733
8 10%Ni/90%Al2 O3 1.24 290700 0.229 8.450 7.966
9 100%Al2 O3 1.64 300000 0.22 7.200 7.200

Literature values (www.matweb.com)

Fig. 5 Microstructure of 60% Ni/40% Al2 O3 . Pores are indicated by white

Fig. 4 Comparison of CTE calculated by the linear rule of mixtures using
circles, where the white area is Ni and the black area is Al2 O3 .
measured values of pure materials and measured directly by the
dilatometer.

literature, the measured CTE values, and the values calcu-

lated from eqs. (5)–(7) that were used in the finite element
analysis.
Generally, researchers use the maximum tensile stress and
maximum principle stress theories to estimate the failure of
brittle materials such as ceramics, and the maximum
deformation energy theory (von-Mises stress theory) to
estimate the failure of ductile materials such as metals.7–10)
Therefore the paper used the maximum tensile stress for
Al2 O3 and the von-Mises stress for Ni to estimate their
failures.
Fig. 6 Residual stresses of Ni-Al2 O3 FGM obtained using the linear rule of
4. Results and Discussion mixtures. The peaks of stress occur at 100% Al2 O3 (Layer 9), 60% Ni/
40% Al2 O3 (layer 4), and 80% Ni/15% Al2 O3 (Layer 2).
Figure 4 compares the measured CTE with values from the
linear rule of mixtures. The measured results showed a
gradual increase in the CTE as the Ni content increased, as small pores marked by white circles, and these pores can
expected. However, as the samples moved from an Al2 O3 - cause large deviations from the calculated CTE values.
rich composition to a Ni-rich composition, the measured Figure 6 shows the calculated tensile and von-Mises
values deviated more from the measured values combined stresses obtained using the linear rule of mixtures. The
with the linear rule of mixtures. These differences between failure of FGM sample can be predicted from the peaks of
the measured and calculated CTE values could be caused by stress in this plot. The figure indicates that the peak stress
unexpected porosity in the sintered body. points are at 100% Al2 O3 , 60% Ni/40% Al2 O3 , and 85% Ni/
Figure 5 shows the microstructure of the fabricated 60% 15% Al2 O3 , and suggests that cracks will form at those
Ni/40% Al2 O3 layer. The microstructure includes much positions in the FGM.
1556
Fig. 7 Residual stresses of Ni-Al2 O3 FGM obtained using measured CTE
values. The peaks of stress occur at the 100% Al2 O3 (Layer 9), 50% Ni/
50% Al2 O3 (Layer 5), and 80% Ni/15% Al2 O3 (Layer 2).
Figure 7 shows the calculated tensile and von-Mises
stresses obtained using the measured CTE values. According
to the figure, the stress peaks are located at 100% Al2 O3 , 50%
Ni/50% Al2 O3 , and 85% Ni/15% Al2 O3 . These three stress
peaks correspond to the expected failure positions.
Figures 8 and 9 show micrographs of the FGM fabricated
using the compositions given in Table 1. Figure 8 shows the
surface of the sample, where cracks located between 100%
Ni and 85% Ni/15% Al2 O3 , and cracks at 50% Ni/50%
Fig. 8 Exterior of fabricated Ni-Al2 O3 FGM. The micrographs show the magnified cracks at 100% Ni and 85% Ni/15% Al2 O3 .
Fig. 9 Cross section of fabricated Ni-Al2 O3 FGM. The micrograph shows the magnified crack at 100% Al2 O3 .
S. Ryu et al.
Al2 O3 were observed. The cross-section of this sample,
shown in Fig. 9, has an additional crack in the 100% Al2 O3
layer.
The crack positions of the fabricated sample matched
the points of peak stress shown in Fig. 7 using the measured
CTE values. The maximum stress points calculated using
the linear rule of mixtures did not accurately predict the
crack in Layer 5 (50% Ni/50% Al2 O3 ). Therefore, the
calculation of residual stress using the linear rule of mixtures
was unable to predict crack positions accurately in the
porous composite, unlike the calculation using measured
CTE values. The calculations using the measured CTE
values were more accurate because they included the porous
effects of the composite. Thus, the porosity of the composite
must be reflected in the property of FGM fabricated by
pressureless sintering to improve the accuracy of stress
calculations.
5. Conclusions
The CTE for graded layers in a Ni-Al2 O3 FGM were
measured to calculate residual stresses. These values were
compared to those obtained using the linear rule of mixtures.
The results showed that the estimated residual stresses
obtained using the measured CTE values were more accurate
for predicting the crack positions in a fabricated FGM sample
than those obtained using the linear rule of mixtures. This
 Experimental Measurement of Coefficient of Thermal Expansion for Graded Layers in Ni-Al2 O3 FGM Joints
was because the measured CTE values already include the
porosity of the composite, whereas the linear rule of mixtures
cannot account for the porosity of each layer. Therefore, the
measured coefficient of thermal expansion must be used in
order to predict the crack positions of porous FGMs
accurately, otherwise a new mixture rule is required to
reflect the porosity of each layer.
Acknowledgment
This work was supported by the Korean government
(MOEHRD, Basic Research Promotion Fund #KRF-2007-
311-D00516), the Korea Science and Engineering Founda-
tion (#R0I-2008-000-11015-0) and Engineering Research
Center, ERC (Micro Thermal System Research Center) of
Seoul National University. The authors would like to thank
Professor Jae-sung Lee at Hanyang University for assistance
with the cold isostatic press. The authors would also like to
thank Professor Yong-ho Choa and Han-Bok Song at
Hanyang University for support with the tube furnace.
1557
REFERENCES
1) W. D. Callister, Jr.: Materials Science and Engineering: An Introduc-
tion, (John Wiley and Sons, Inc., 2003) pp. 659–661.
2) S. H. Ryu, J. H. Park, C. S. Lee, J. S. Lee, J. C. Lee, S. H. Ahn, D. K.
Kim, J. H. Chae and D. H. Riu: Korean J. Mater. Res. 18 (2008) 547–
551.
3) L. M. Pines and A. H. Bruck: Acta Mater. 54 (2006) 1457–1465.
4) J. H. Park, S. H. Ryu, C. S. Y. Lee, Y. H. Choa, J. C. Lee and S. H. Ahn:
Proc. Multi-functional materials and structures, Hong Kong, China,
July 27–31, (2008).
5) D. L. Logan: A Firsr Course in the Finite Element Method: Third
Edition, (BROOKS/COLE; 2002) pp. 532–536.
6) R. D. Cook, D. S. Malkus, M. E. Plesha and R. J. Witt: Concepts and
Applications of Finite Element Analysis: Fourth Edition, (John Wiley
and Sons, Inc., 2002) pp. 508–510.
7) J. C. Lee, J. H. Park, S. H. Ryu, H. J. Hong, D. H. Riu, S. H. Ahn and
C. S. Y. Lee: Mater. Trans. 49 (2008) 829–834.
8) ANSYS. Release 11.0 Documentation for ANSYS.
9) N. E. Dowling: Mechanical Behavior of Materials 2nd edition,
(Prentice-Hall International, Inc., London, 1999) pp. 239–280.
10) J. G. Choi and S. B. Lee: Analysis material science, (Cheong Moon
Gak Publishers, Seoul, 2000) pp. 196–198.