Axial Compressive Strength of Carbon Fiber

TADASHI OHSAWA, MINORU MIWA,* and MASATO KAWADE,

Faculty of Engineering, Gifu University, Yanagido, Gifu 501-1 1,
Japan, and EIKI TSUSHIMA, Corporate Research and
Development Laboratory, Toa Nenryo Kogyo Co., Ltd.,
Nishi-tsurugaoka 1 -chorne, Ohi-machi, Iruma-gun,
Saitama 354, Japan

Synopsis
Efforts were made to estimate the axial compressive strengths of carbon fibers from the fiber
fragment lengths produced by subjecting to a strain greater than the fiber ultimate strain for
PAN-based and pitch-based carbon fibers. The estimated compressive strength of carbon fibers
decreases with increasing temperature in a temperature range from room temperature to 100OC.
This decrease in compressive strength may be accounted for by a decrease in the radial
compressing force. The real compressive strength, determined by extrapolating a linear relation-
ship between the estimated compressive strength and the radial compressing force, is approxi-
mately 25-60% of tensile strength for PAN-based fibers, while it is approximately 10-35% for
pitch-based fibers.

INTRODUCTION
Relationship between the mechanical properties (strength, fracture strain,
and modulus) of composites in tension and those of the fibers used as
reinforcement have been studied both experimentally and theoretically in
fairly great detail. However, in spite of the fact that the compressive charac-
teristics of composites depend on the characteristics of reinforcing fibers, there
is almost no study on the mechanical properties of composites in compression
because it is difficult to determine accurately the axial compressive properties
of reinforcing fibers.
Efforts were made to determine the axial compressive properties of reinforc-
ing fibers. Sinclair and Chamis' estimated the compressive modulus and
strength of carbon fibers from those of unidirectional composites according to
a rule of mixtures. Jones and Johnson' estimated the compressive fracture
strain of carbon fibers from the radius of curvature of the fiber loop a t the
point of maximum strain using the elastica method. Moreover, Hawthorne
and Teghtsoonian3 studied the axial compression fracture of carbon fibers by
embedding a single fiber in the epoxy resin and compressing the specimens
parallel to the fiber axis. They identified the earliest stages of fracture by
careful optical monitoring of the fiber surface and estimated the fiber axial
compression failure strengths from both the earliest fracture strain and the
tensile modulus.

*To whom correspondence should be addressed.

Journal of Applied Polymer Science, Vol. 39, 1733-1743 (1990)

0 1990 John Wiley & Sons, Inc. CCC 0021-8995/90/081733-11$04.00
1734 OHSAWA E T AL.

There are some questions in the above-mentioned studies. In the report by

Sinclair and Chamis’ a rule of mixtures doesn’t always hold in compression. In
the fiber loop,2 the fiber doesn’t always fail in the compression side of fiber.
Hawthorne and Teghtsoonian3 used the tensile modulus to estimate the fiber
axial compression failure strengths. However, it is doubtful that compressive
modulus is equal to the tensile modulus and the linearity holds in the
compressive stress-strain curves.
I n this paper, efforts were made to determine accurately the fiber axial
compressive strengths using the yield shear strength a t the fiber-matrix
interphase. Namely, if a sufficiently long fiber is embedded in the resin matrix
and the system is compressed, the fiber eventually breaks into many pieces, on
the assumption that a theory according to Kelly and T y ~ o n ,while
~ . ~ assuming
that the yield shear strength a t the fiber-matrix interphase is constant, may
be applied to the system in compression. By measuring the lengths of the
broken pieces, we estimated the fiber axial compressive strengths.

MEASUREMENTS

Principle
If a sufficiently long fiber is embedded in the resin matrix and the system is
subjected to a tensile strain greater than the fiber ultimate strain, the fiber
eventually breaks into many pieces. Assuming that the fiber strength is
uniform along the length of the fiber, the critical fiber length (Zc)ten in tension
is given as follows6:

where is the mean fragment length.

Moreover, according to Kelly and Tyson4* the relationship between critical
fiber length (Zc)ten and the tensile strength of fiber (af),,, results in the
following equation:

where r is the yield shear strength a t the fiber-matrix interphase and d is the
diameter of the fiber. Accordingly, r can be determined from eqs. (1) and (2)
by measuring mean fragment length in tension.
When the above-mentioned specimen is compressed in the direction of the
fiber axis, we assumed that the yield shear strength of the same value as that
in tension, but the working direction is the reverse. Therefore, if the specimen
is compressed, the fiber eventually breaks into many pieces in the same
manner as in the case of tension. Measuring the mean fragment in compres-
sion ( the critical fiber length in compression ( Zc)comp is given as follows:
 AXIAL COMPRESSIVE STRENGTH OF CARBON FIBER 1735

Moreover, according to the above-mentioned assumption, the following equa-

tion holds :

where ( uf )camp is the fiber axial compressive strength. As 7 and d in eq. (4) are
the same values as those in eq. (2), the fiber axial compressive strength
( uf )camp can finally be estimated from the following equation:

(Zc)ten and in eq. (5) are characteristic values for a given composite
system under the same environmental condition. Therefore, when critical fiber
lengths will be determined under the same environmental condition, eq. (5)
shows that the ratio of fiber compressive strength to tensile strength is given
by the ratio of critical fiber length in compression to that in tension.
It is impossible to obtain the length dependence of compressive strength of
carbon fiber for the present. Accordingly, to rough approximation, assuming
that the ratio compressive strength to tensile strength is the same value
regardless of the fiber length, it is conceivable that the compressive strength
( uf )camp obtained by substituting the tensile strength a t a given length into eq.
(5) is the estimated compressive strength corresponding to the tensile strength.
In this experiment, to accurately give the fiber the tensile strain, a rectangu-
lar (beam) in which a long fiber was embedded at a constant tension in the
neighborhood of the surface prepared [Fig. 1 (a)] and the four-point bending
method was adopted. If the side having a fiber of the specimen is bent outside,
the fiber is elongated. As the fiber is evenly strained within the distance
between the loading heads C and D [Fig. 1 (b)], the fragment lengths of the
fiber can be measured over the section C-D.
Second, if the side having a fiber of the specimen is bent inside, the fiber is
compressed. However, when the compressive fracture strain of carbon fibers

,matrix

(a)
9

fiber ‘ n e u t r a l axis

A\ vD B
/
a (b)

-/
Fig. 1. Schematic diagram of four-point bending test in tension.
1736 OHSAWA ET AL.

Steel Phosphor bronze plate

Fig. 2. Schematic diagram of four-point bending test in compression.

is larger than the tensile ultimate strain, the fracture of rectangular beam is
often preceded by the tensile fracture of the resin matrix and the experiments
becomes a failure in spite of all the effort. Therefore, an equipment was
devised to compress the specimen [Fig. 2 (a)]. The specimen embedding a long
fiber was set on the phosphor bronze plate between the loading heads C and D
of the equipment. Accordingly, a laminated beam consisting of both the
phosphor bronze plate and the specimen was composed. The thickness of
phosphor bronze plate is 1 nun. Young’s modulus of phosphor bronze plate is
nearly two figures greater than that of the resin matrix. When the laminated
beam is bent, it always has the neutral plane in the neighborhood of the upper
surface of phosphor bronze plate. Consequently, the specimen is compressed
only.
Using the method shown in Figure l(b), the drop (Y),,, of heads C and D
required to a constant tensile strain ( c )ten is given by the following equation
according to the bending theory:

( C f )ten
(Y)t,n = 7 . s . (3L - 4 s )
6h

where h‘ is the distance between the fiber and the neutral axis, L is the span
length, and S is the distance between the lower supporting point A (or B) and
the upper loading head C (or D). Similarly, using the method shown in Figure
2(b), the drop ( Y )camp required to give a constant compressive strain ( E )camp
is given by

(€1 )camp . s . ( L - 2 s )
(Y)comp = (7)
2h’

In compressive experiments, thinner specimens were preferred to be prof-

itable the observations of fiber fragments under the optical microscope. In
order to secure the required value of h’, a spacer made of the same material as
the resin matrix was inserted between the specimen and the phosphor bronze
plate.
 AXIAL COMPRESSIVE STRENGTH OF CARBON FIBER 1737

TABLE I
Mechanical Properties of Materials (at 20"C).'

Tensile Young's Breaking

strength modulus strain Diameter
Materials @Pa) @Pa) (Pm)

Carbon fiber
Pitch-based
Carbonized fiber H T X 3.34 230 1.46 10.1
Graphitized fiber HMX 4.33 482 0.90 10.1
PAN-based
Carbonized fiber T-300 3.50 206 1.70 7.1
Graphitized fiber M-40 2.88 388 0.80 7.1
Matrix
Epoxy resin 0.068 1.9 8.8 -

"Gauge length for fiber, 5 mm; H T X and HMX are Tonen's pitch based high performance
carbon fibers. They are the materials under developing.

Preparation of Specimen
The fibers used in this study were PAN-based carbon fibers (Toray, Torayca
7.1 pm in diameter) and pitch-based carbon fibers (Toa Nenryo Kogyo, Tonen
10.1 pm in diameter). Moreover, carbonized (higher strength type) and graphi-
tized fibers (higher modulus type) were respectively used (Table I). Matrix
material was an epoxy resin (Epikote 828, Yuka Shell). Epoxy resin, 100 parts,
was mixed with 10 parts of an amine hardening agent (S-Cure 661, Nihon
Kayaku). The mixture was agitated thoroughly and then defoamed under
vaccum for about 30 min. This mixture was poured into a mold holding a fiber
a t a constant tension and subjected to curing a t 65°C for 17 h and postcuring
a t 140°C for 5 h. The specimen was then allowed to cool to room temperature
a t a cooling rate of about 0.5"C/min. The dimension of the mold resulted in a
specimen which measured 13 mm in thickness, 10 mm in width, and 210 mm
in length for the tensile experiments, and 4, 10, and 100 mm for the compres-
sion, respectively.
The specimens prepared in this manner were submitted to measurement of
fragment length; each specimen was subjected to a tensile (or compressive)
strain of 4% a t a drop rate of the upper heads of 10 mm/min. Span length L
was 180 mm. The distance S between points A (or B) and C (or D) is 50 mm.
In order to investigate the effect of radially compressing force on the fiber
axial compressive strength, measurements were made a t intervals of 20°C
from 20 to 140°C. Over 1000 fragments were examined for each experimental
condition.

RESULTS AND DISCUSSION

The photographs of fiber fragments of the specimen compressed by the
above-mentioned method are shown in Figure 3. The fiber is broken into a
number of fragments. In the fiber breaking a t 20°C [Fig. 3 (a)], the fiber
fractures without the buckling. At 140°C [Fig. 3 (b)], the fiber buckles near
the fracture point and the fracture appearances are remarkably different from
those a t 20°C. Namely, it is conceivable that the fiber is buckled by the
1738 OHSAWA ET AL.

U
1 OOprn

-
IOOprn
Fig. 3. Photographs of fiber breaking points in compression test (pitch-based carbonized fiber,
HTX): (a) at 20°C; (b) at 140°C.

decrease in the radial force compressing the fiber owing to both the decrease
in residual thermal stress and Young’s modulus of the resin matrix. Figure 4
illustrates typical distribution of fiber fragment length. Similar results were
obtained a t various temperatures. The relationship between the critical fiber
length (Zc)ten in tension obtained from the mean fragment length
according to eq. (1) and the temperature is shown in Figure 5. Moreover, the
relationship between the critical fiber length ( Zc)comp in compression obtained
from the mean fragment length (~),o,,,p in compression according to eq. (3) and
the temperature is shown in Figure 6. Figure 7 shows the relationship between
the value of fiber axial compressive strength estimated from the critical fiber
length ( Zc)ten (Fig. 5 ) in tension and the critical fiber length ( Zc)comp (Fig. 6) in
compression according to eq. ( 5 ) and the temperature. With these systems, the
estimated compressive strength decreases with increasing temperature a t a
temperature range lower than 100°C. As the real axial compressive strength of
carbonized and graphited fiber ought not to depend on the temperature a t a

20 -

I
I
-

1
;15-
v

g 10
5

0.5
0 1.0 1. 0 0.5 1 .o 1.5 2.0
Fragment length (mrn)
Fig. 4. Distribution of fragment length (pitch-based carbonized fiber, HTX): (a) a t 40°C; (b)
a t 100°C.
 AXIAL COMPRESSIVE STRENGTH OF CARBON FIBER 1739

3.0
(a)

I I I 1 6 I t

0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140

Temperature ("C)
Fig. 5. Relation between temperature and critical fiber length in tension: (a) pitch-ba 1
carbon fiber; (b) PAN-based carbon fiber; (0)carbonized fiber, HTX; (0) graphitized fiber,
HMX; (A) carbonized fiber, T-300; (A) graphitized fiber, M-40.

temperature ranges of 20-100°C, it is conceivable that the radial force

compressing the fiber decreased by both the decreases in residual thermal
stress and Young's modulus of the resin matrix with increasing temperature
may cause a temperature dependence of the estimated value of the fiber axial
compressive strength. The thermal stress was produced by the difference in
thermal expansion coefficient between fiber and resin. The estimated compres-
sive strength increases greatly with increasing temperature above 120°C. It is

Fig. 6. Relation between temperature and critical fiber length in compression: (a) pitch-based
carbon fiber; (b) PAN-based carbon fiber; (0)carbonized fiber, HTX; (0) graphitized fiber,
HMX; (A) carbonized fiber, T-300; (A) graphitized fiber, M-40.
1740 OHSAWA ET AL.

8
I

% 0 20 40 60 80 100 120 140

W Temperature (OC)
Fig. 7. Relation between temperature and estimated compressive strength: (0) pitch-based
carbonized fiber, HTX; (0)pitch-based graphitized fiber, HMX; (A) PAN-based carbonized fiber,
T-300; (A) PAN-based graphitized fiber, M-40.

conceivable that this result may be caused by not satisfying the conditions for
application of the above-mentioned eq. (5). As shown in Figure 3, when the
specimen is compressed, the fiber buckles owing to the greater decrease in the
Young's modulus of resin matrix. Accordingly, further details are discussed for
the results obtained at temperature range lower than 100°C.
When a fiber is embedded in the resin and the system is allowed to cure, the
thermal stress ( P)T working perpendicularly on the fiber-resin interface is
approximately given by the following e q ~ a t i o n ~ . ~ :

where a is the thermal expansion coefficient, E is the Young's modulus, v is

the Poisson's ratio, AT is the difference in temperature from molding temper-
ature, and the subscript m denotes matrix. Figure 8 shows the relationship
between the thermal stress ( P ) T obtained by eq. (8). The thermal expansion
coefficient ( a m ) ,Young's modulus ( E m ) ,and Poisson's ratio (v,) required to
calculate the thermal stress ( P ) T are shown in Table 11. Figure 9 shows the
relationship between the estimated compressive strength and the thermal
stress. It is obtained from the estimated compressive strength of the
fiber-temperature relationship (Fig. 7) and the thermal stress-temperature
relationship (Fig. 8). The estimated compressive strength of carbonized and
graphitized fibers decreases linearly with decreasing thermal stress.
It is conceivable that the real compressive strength of carbonized and
graphitized fiber is the strength when the radial compressing force; i.e., the
residual thermal stress in this experiment is zero. Accordingly, we considered
the value obtained by extrapolating the straight line in Figure 9 to ( P ) T= 0
as the real compressive strength of carbonized and graphitized fiber and
 AXIAL COMPRESSIVE STRENGTH OF CARBON FIBER 1741

6 1 'b,

Temperature ( " C )
Fig. 8. Relation between temperature and thermal stress.

TABLE I1
Properties of Epoxy Resin at Various Temperatures

Temperature ("C) 20 40 60 80 100 120 140

Thermal expansion
coefficient a, ( x ~ O - ' / ~ C ) 6.1 7.2 8.1 14.3 14.7 15.2 15.2
Young's modulus
Em(GPa) 1.90 1.80 1.74 0.11 0.02 0.01 0.01
Poisson's ratio
'rn 0.30 0.33 0.35 0.38 0.40 0.43 0.45

.-111a , -
1 ,-. -
?'-
-
Q

E '';
A A

--
A
s ,p
XI
al
c

.-2 1 1
1742 OHSAWA ET AL.

TABLE I11
Tensile and Estimated Compressive Strength of Carbon Fiber

Pitch-based PAN-based
Carbonized Graphitized Carbonized Graphitized
Carbon fiber fiber HTX fiber HMX fiber T-300 fiber M-40

Tensile strength
(at Iten (GPa) 3.34 4.33 3.50 2.88
Estimated compressive
strength (q Icomp (GPa) 1.25 0.54 2.06 0.78
(a/ )comw’(uf )ten (W) 37.4 12.5 58.9 27.1

showed those values in Table 111. For comparison, the tensile strength and the
ratio of compressive strength to the tensile strength are also shown in Table
111. The estimated real compressive strength of PAN-based carbonized fiber
(higher strength type) is the highest and pitch-based carbon fiber follows. For
both PAN-based and pitch-based fibers, the estimated compressive strength of
graphitized fiber (higher modulus type) is always less than half of that of
carbonized fiber (higher strength type). Furthermore, the ratio of compressive
strength to the tensile strength is approximately 35-60% for carbonized fibers,
while its ratio is approximately 10-25% for graphitized fibers. It is noted that
the estimated real compressive strength of graphitized fiber is remarkably low.
Hawthorne and Teghtsoonian3 reported that the compressive strength ( ur ),
of pitch-based carbon fiber depends on its Young’s modulus. According to
their study, the compressive strength (or)=is constant, about 2.3 GPa, in
Young’s modulus ranges of 130-330 GPa. Furthermore, when Young’s modu-
lus of the fiber is 450 GPa (or), decreases to 1.2 GPa. On the other hand, the
compressive strength (or), of PAN-based carbon fiber with Young’s modulus
of 340 GPa is 2.1 GPa. These values are approximately double what we obtain.
It is conceivable that these differences may be caused by their assumption
that the compressive modulus is the same value as Young’s modulus and the
carbon fibers in compression show Hooken’s behaviour.
Kitano et a1.8 obtained the compressive strength ( o f ) , of about 4.4 GPa for
PAN-based carbonized fiber (T-300) and about 2.8 GPa for PAN-based graphi-
tized fiber (M-40) by using the loop method.’ Moreover, (ur ), of about 1.3 GPa
was obtained for pitch-based fiber. In their report, the tensile strength of
carbonized fiber (T-300) is 3.53 GPa and that of graphitized fiber (M-40) is
2.74 GPa. The obtained compressive strength is higher than the tensile
strength. We consider that the compressive strength of the carbon fibers is
lower than the tensile strength owing to the fiber structure. The results
obtained by Kitano et al. are completely in contrast to our experimental
results.

CONCLUSIONS
If a sufficiently long fiber is embedded in the neighborhood of the surface of
rectangular beam and the system is subjected to a tensile (or compressive)
strain greater than the fiber ultimate strain according to the bending method,
the fiber eventually breaks into many pieces. Measuring the lengths of the
 AXIAL COMPRESSIVE STRENGTH OF CARBON FIBER 1743

broken pieces, attempts were made to estimate the axial compressive strength
of carbon fibers.
The estimated compressive strength of carbon fibers decreases with increas-
ing temperature range from room temperature to 100°C. This decrease in
compressive strength may be accounted for by a decrease in the radial
compressing force owing to (1) decrease in residual thermal stress and (2) a
decrease in Young’s modulus of the resin matrix. On the other hand, the
estimated compressive strength increases greatly with increasing temperature
above 120°C. This increase in compressive strength may be accounted for by
unsubjecting to accurately a compressive strain owing to the fiber buckling
occurring with more of a decrease in Young’s modulus of the resin matrix.
There is a linear relationship between the estimated compressive strength
and radial compressing force in a temperature range from room temperature
to 100°C. The real compressive strength of the fibers, determined by extra-
polating this straight line until the radial compressing force is zero,
approximately 25-60% of tensile strength for PAN-based fibers, while it is
approximately 10-35% for pitch-based fibers.

References
1. J. H. Sinclair and C. C. Chamis, Compression Testing of Homogeneous Materials and
Composites, ASTM S T P 808, ASTM, Philadelphia, 1983, p. 155.
2. W. R. Jones and J. W. Johnson, Carbon, 9, 645 (1971).
3. H. M. Hawthorne and E. Teghtsoonian, J . Muter. Sci., 10, 41 (1975).
4. A. Kelly and W. R. Tyson, J . Mech. Phys. Sol&, 13, 329 (1965).
5. A. Kelly and W. R. Tyson, J . Mech. Phys. Sol&, 14, 177 (1966).
6. T. Ohsawa, A. Nakayama, M. Miwa, and A. Hasegawa, J . Appl. Polym. Sci., 22, 3203
(1978).
7. G. Gerard and A. C. Gilbert, J . Appl. Mech. ( A S M E ) , 24, 355 (1957).
8. A. Kitano, T. Norita, and K. Noguchi, 12th Fukugo Zuiryo Symp. Kohen-Yoshishu, 125
(1987).

Received August 31, 1989

Accepted September 6, 1989