December 1996

ELSEWIER Materials Letters 29 (1996) 235-240

Quantitative study of correlation between fracture surface

roughness and fatigue properties of Sic/Al composites
X.W. Li *, J.F. Tian, N.L. Han, Y. Kang, Z.G. Wang
State Key Labor-atop for Fatigue and Fracture of Materials, Institute of Metal Research, Academia Sinica, Shenyang 110015, China

Received 19 April 1996; revised 5 June 1996; accepted 10 June 1996

Abstract

Quantitative measurements were carried out on the fatigue fracture surface of the SIC/AI composite by a sectioning
method. It was shown that the cyclic plastic strain amplitude, Sic volume fraction and particle size have effects on fracture
surface roughness R,. The measured fracture surface roughness R, is closely related to the fatigue-crack propagation path
and may corresponds to the fatigue life. Moreover, it was found that there is an obvious difference in the R, values for
fatigue fractures which are due to different fracture mechanisms. These results show that it is possible to reflect the fracture
mechanism using fracture surface roughness and relate it to the fracture properties of materials.

Keywords: Composite; SIC/Al; Fatigue fracture; Roughness: Fractal dimension; Sectioning method

1. Introduction profile structure factor. This sectioning method is

based on classical stereological principles and veri-
In recent years, there has been extensive research fied with the aid of computer simulations for some
on quantitative characterization of the fracture sur- ruled surfaces. In recent works [7,8], we have experi-
face roughness which is now a major object of study mentally confirmed its effectiveness.
in “Quantitative Fractography” [l-4]. Gokhale and The main quantitative parameter describing the
Underwood [5,6] have proposed an assumption-free surface roughness is fracture surface roughness R,,
method for estimating the fracture surface roughness which is equal to the true area S of a fracture surface
by vertically sectioning the fracture surface into three divided by its apparent projection area A. It may be
parts at an angle of 120” with each other, which expressed as:
could be expressed as follows:
R, = S/A. (2)
R, = RL$, (‘1
Generally, the desired parameter R, is very diffi-
where R, and R, are the fracture surface and profile
cult to measure directly, and is often estimated
roughness parameters respectively, and $ is the
through parametric relations between the fracture
surface roughness R, and the profile roughness R,.
The fracture profile roughness R, is defined as:

* Corresponding author R, = L/L,, (3)

00167-577X/96/$12.00 Copyright 0 1996 Elsevier Science B.V. All rights reserved.

PII SOl67-577X(96)00150-4
236 X. W. Li et al. /Materials Letters 29 (lYY6i 235-240

where L is the true length of a fracture profile and were analyzed using a IPS-500 image analyzer and
LO is its apparent projected length. the requisite profiles were selected and then digi-
Most of the fractal analyses on fracture surfaces tized. By utilizing the technology of image process
have shown that the fracture surface or profile does and analysis and a self-designed program, the values
show a fractal character [9-l I]. The corresponding of R, and D, can be obtained. The detailed proce-
fractal equations are: dure can be found in Ref. [7].

R,(q) = cr+)-‘), (4)

R,(n) = c2++2), (5) 3. Experimental results and discussion

where C, and C, are constants without dimension 3.1. Effect of plastic strain amplitude on fracture
and 77 is the size of the measuring unit. sulfate roughness
In the present study, the sectioning method was
applied to obtain a quantitative analysis of some The fatigue tests were performed on pure alu-
fatigue fractures of Sic particulate reinforced alu- minum reinforced with 20% Sic particles of 10 p,m
minum composites. The correlation between the frac- average size using three plastic strain amplitudes, i.e.
ture surface roughness and fatigue properties of the 0.15%, 0.2% and 0.3%. The three corresponding
SIC/Al composite was explored tentatively. fatigue fractures are shown in Fig. 1. It can be seen

2. Experimental procedures

The SIC/Al composite used in the present study

was made by powder metallurgy. The Sic particu-
lates with a size of 10 km or 32 &rn were mechani-
cally mixed with 14 km aluminum powder. The
mixture was then hot compacted at 893 K and
subsequently extruded to bars. The unreinforced alu-
minum matrix was produced in the same process
except for mixing. The test specimens were ma-
chined from the as-extruded bars with the loading
axis parallel to the extrusion direction. The gauge
length and diameter of the specimens were 14 mm
and 6 mm, respectively. The low cycle fatigue tests
were performed under fully reversed plastic strain
control using a mechanical servo-controlled testing
system. The tests were conducted in air at 298 K and
the cycling frequency was 0.125 Hz.
Electroless plating of the fatigue fracture surface
provided a Ni-P protective layer to avoid distortion
during sectioning and metallographic polishing. Then
metallographic sectioning was done along three ver-
tical sectioning planes mutually at an angle of 120”
and subsequent metallographic polishing was per-
formed to clearly reveal the fracture profile, Three
representative fields of view from each profile were Fig. 1. SEM images of the morphology of fatigue fractures for
taken by backscattered electron imaging using a different strain amplitudes. (a) Aep /2 = O.lS%, (b) Aep /2 =
scanning electron microscope &EM). The images 0.2%. (c) A .zp /2 = 0.3%.
 X. W. Li et al. / Materials Letters 29 (19961 235-240 231

that the three fatigue fractures are basically alike in Table 1

The measured values of fatigue life, N,, fracture surface rough-
character so that it is very difficult to determine
ness, R,, and fractal dimension, D,, for different strain ampli-
which roughness is qualitatively larger. The rough- tudes (A eP /2 = strain amplitude)
ness of these fracture surfaces was quantitatively
A+ /2 (“/cl Nf Rs DS
measured by the sectioning method. The results are
El 0.15 38000 1.x74 2.093
listed in Table 1 and the quantitative relationship
E2 0.2 13500 1.947 2.097
among the fracture surface roughness R,, the plastic E3 0.3 2616 2.188 2.099
strain amplitude be,/2 and the fatigue life Nt
(cycles) are shown in Fig. 2.
From Table I and Fig. 2, it can be seen that the
measured value of R, increases with increasing creases. Thus, when he,/2 is larger, the measured
plastic strain amplitude A~r/2. In general, the fa- value of R, increases.
tigue fracture surface roughness is closely related to Moreover, Fig. 2 shows that the relationship be-
the fatigue-crack propagation path. The more tortu- tween the plastic strain amplitude her,/2 and the
ous the fatigue-crack propagation path, the larger the fatigue life Nf of the SIC/Al composite fits the
roughness R,. To adequately understand the increase Coffin-Manson relationship. The plot of R, versus
of R, with the strain amplitude A~r/2, it is neces- A ~r/2 appears to be approximately a straight line. It
sary to consider the highly localized stresses that seems that there might be a corresponding relation
develop in the matrix around the reinforcement parti- between R, and Nf. Table I also lists the values of
cles. This concentration of stresses results from the the measured fractal dimension D, for fracture sur-
constrained deformation in the matrix that occurs face. The value of D, does not change essentially
because of the significant difference in the elastic with the plastic strain amplitude. Thus, the slight
modules of the reinforcement and the matrix. During difference in character does not enable one to distin-
cyclic loading these highly localized stresses lead guish among the three fracture surfaces using the
naturally to highly localized plastic deformation [ 121. fractal dimension Ds. Therefore, it may be more
Besides, Fig. 1 shows that there are much more efficient to describe the fracture surface roughness
secondary cracks on the fracture surfaces, which using R, rather than D,.
mostly nucleate at the particle-matrix interfaces.
With increasing the plastic strain amplitude, the plas- 3.2. Eflect of Sic volume ,fruction on ,fructure surjke
tic area at the crack tip enlarges and the number of roughness
SIC particles participating in nucleating cracks in-
The materials used in this study were the unrein-
fored aluminum matrix and Sic/Al composites with
2.22
two SIC volume fractions, i.e. 6% and 25%. The
a^ 2.17
fatigue tests were performed under the same plastic
2 12 strain amplitude of 0.5%. The corresponding three
;
fatigue fractures are shown in Fig. 3. Table 2 lists
5 207

k
the measured results. From Fig. 3 it can be seen
a 202
qualitatively that there is quite a difference in the
9
E 1.97 features of the three fatigue fracture surfaces. So,
83 192
their surface roughness should be evidently distinc-
tive, which can also be seen from the different
1.87
profiles (see Fig. 4) obtained by sectioning the frac-
1.82 ture surfaces. From the measured results shown in
0 10 0.15 (1.20 0.25 0.30 0.35 0 40
Fig. 5 it was found that with increasing SIC volume
Strain Amplitude A& (9k) fraction, the fatigue life Nr decreases and the frac-
Fig. 2. Roughness parameter R, and fatigue life Nf versus strain ture surface roughness R, increases. For the com-
amplitude AC, /2. posite that has less ductility than the unreinforced
238 X. W. Li et al. /Materials

Fig. 4. Fracture profiles of pure Al and composites with different

volume fractions of Sic. The plastic strain amplitude was 0.5%.
Fig. 3. SEM images of the morphology of fatigue fractures of pure
(a) Pure Al CO%), (b) 6%~~(c) 25%.
Al and composites with different volume fractions of Sic. The
plastic strain amplitude was 0.5%. (a) Pure Al (0%). (b) 6%. (c)
25%.
surfaces and the fatigue-crack propagation path is
comparatively even and smooth. Therefore, the cor-
responding fracture surface roughness R, is small.
matrix it is relatively easy to initiate microcracks
under high fatigue strain either by interface cracking
or by particle fracturing. Furthermore, the ductility
of the composite is reduced more with increasing
Lz40
I ‘O’
5? 2.20
Sic volume fraction so that the fatigue life A$ de- d

creases. Concerning the unreinforced aluminum ma-

trix, many fatigue striations were seen on the fracture

2 1.60

Table 2 E
9 1.40
The measured values of fatigue life, Nr, fracture surface rough-
P
ness, R,, and fractal dimension, D,, for composites with different 1.20 o Rs l N
volume fractions, V, of SIC

v (%) Nt
1.00 ’ J 102
Rs Ds 0 5 10 15 20 25 30
VI 0 2260 1.382 2.053
SC Volume Fraction (%)
v2 6 1340 1.768 2.082
v3 25 788 2.222 2.101 Fig. 5. Roughness parameter R, and fatigue life Nr versus SIC
volume fraction.
 X. W. Li et al. /Materials Letters 29 (1996) 235-240 239

different surface features, the measured fractal di-

mension D, for the fracture surface is quite different
(see Table 2). In this case, the description of fracture
surface roughness using D, is sufficient. However,
in a word, R, appears to be superior to D, in
characterizing the fracture surface roughness.

3.3. Effect of SC particle size on fracture su$ace

roughness
Fig. 6. SEM morphology of the fatigue fracture of a composite
with large SIC particles.
The fatigue test was performed with the compos-
ite containing Sic particles of 32 km in size using a
In the case of composites it was hard to see fatigue plastic strain amplitude of 0.5%. Comparing Fig. 6
striations, but a population of unequiaxed dimples with Fig. 1, it can be noted that the fracture surface
appears on the fracture surfaces. In the center of of the composite with large particulate is quite dif-
most dimples there were broken or detached Sic ferent in character from that with small particulates.
particles. On propagating, the crack encounters SIC It can also be seen from the fracture profiles shown
particles and will pass by or through the particles in Fig. 7 that the fracture profiles in samples with
and its propagation path becomes more tortuous. small particulates are more tortuous while in samples
This trend is more evident with increasing SIC vol- with large particulates the profiles are more even and
ume fraction. Thus, the measured value of R, in- straight. The measured values of R, and D, for
creases. samples with large particulates are equal to 1.827
It was found from Fig. 5 that there is also a and 2.074, respectively, which are smaller than that
corresponding relation between R, and Nf analo- with small particulates (see Table 1). The results
gous to that in Fig. 2. The results show that it is show quantitatively that there exist two quite differ-
quite possible to relate the fracture surface roughness ent kinds of fracture in the composite, depending on
R, with the fatigue life Nt. However, to be exact, the reinforcement sizes.
the fatigue life should be divided into the initiation
life and propagation life. Here, R, was only related
with the total fatigue life, which might be incom- 4. Conclusions
plete. So, it remains to be studied further. In addi-
tion, Table 2 and Fig. 5 show that the measured (1) The plastic strain amplitude, Sic volume frac-
value of R, of pure aluminum differs significantly tion and particle size have effects on the fracture
from that obtained for the composite due to different surface roughness R,. The measured R, is closely
fracture mechanisms. It is thus possible to reflect the related to the fatigue-crack propagation path and
fracture mechanism to some extent by using R,. may correspond to the fatigue life.
In regard to the fatigue fractures with obviously (2) It is possible to reflect the fracture mecha-

Fig. 7. Fracture profiles of composites with two particle sizes: (a) 32 km, (b) 10 pm.
nisms to some extent using the fracture surface [61 A.M. Gokhale and E.E. Underwood, Metall. Trans. 21 A
(1990) 1201.
roughness R, .
[71Xiaowu Li, Jifeng Tian, Yan Kang and Zhongguang Wang.
(3) It may be more efficient to characterize the Scripta Metall. Mater. 33 (1995) 803.
fracture surface roughness and relate it to the frac- Bl Xiaowu Li, Master Thesis, Institute of Metal Research.
ture properties of materials using R, rather than D,. Academia Sinica. Shenyang (1995).
[91B.B. Mandelbrot, D.E. Passoja and A.J. Panllay, Nature 308
(1984) 721.
1101D.J. Alexander, Quantitative methods in fractography, ASTM
References STP 1085, eds. B.M. Stauss and SK. Putaturda (ASTM.
Philadelphia, 1990) p. 39.
[l] E.E. Underwood, JOM 42 (1990) 10. Ill1 E.E. Underwood and K. Banerji, Metals handbook, 9th Ed..
[2] S.M. El-Soudani, JOM 42 (1990) 20. Vol. I2 (ASM, Metals Park, OH, 1987) p. 2 11.
[3] L. Wojnar, Acta Stereol. 10 (1991) 37. 1121J. Llorca, S. Suresh and A. Needleman, Metall. Trans. 23 A
[4] J.M. Bennett, Meas Sci. Technol. 3 (1992) 1119. (1992) 919.
[5] A.M. Gokhale and E.E. Underwood, Metall. Trans. 21 A
(1990) 1193.