COMPOSITES

SCIENCE AND
TECHNOLOGY
Composites Science and Technology 67 (2007) 2261–2270
www.elsevier.com/locate/compscitech

Effect of new epoxy matrix for T800 carbon fiber/epoxy filament

wound composites
Weiming Chen 1, Yunhua Yu 1, Peng Li, Chengzhong Wang,
Tongyue Zhou, Xiaoping Yang *
The Key Laboratory of Beijing City on the Preparation and Processing of Novel Polymers, Beijing University of Chemical Technology,
P.O. Box 34, Beijing 100029, China

Received 29 April 2006; received in revised form 30 September 2006; accepted 23 January 2007
Available online 11 February 2007

Abstract

A new epoxy resin matrix with good adherence to T800 carbon fibers (T800 CFs) in filament winding was developed by addition of
hardener and resin diluter. Interfacial behavior of the T800 CF/epoxy composites was analyzed according to the Naval Ordnance lab-
oratory (NOL) ring test, short-beam-shear test and fracture surface observation. Meanwhile, scanning electron microscopy (SEM), Fou-
rier transform infrared spectroscopy (FT-IR), X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM) were used
in analysis of the interfacial behavior. The interfacial properties of the T800 CF/epoxy filament wound composites were improved by
optimizing the matrices through increasing the toughness and reducing the viscosity, which is an important factor in influencing the wet-
tability of T800 CFs. The Interlaminar shear strength (ILSS) of the unidirectional T800 CF/epoxy composites and the tensile strength of
NOL-ring in this work reached to 123 and 2570 MPa, respectively. Also, the interfacial adhesion was much improved by the chemical
reactions between the new matrix and the sizing on the T800 CFs.
 2007 Elsevier Ltd. All rights reserved.

Keywords: A. T800 carbon fiber; E. Filament winding; A. Composites; B. Interfacial properties; A. Resin matrix

1. Introduction and meanwhile to prevent the fibers from damage through

Mechanical performance of carbon fiber reinforced
polymer (CFRP) composites depends not only on the prop-
erties of reinforcing fiber and matrix, but also on the fiber/
sizing and sizing/matrix interfacial properties [1]. There-
fore, many scientific efforts have been devoted to modify
carbon fibers by a variety of methods such as gas-phase,
liquid-phase and continuous anodic oxidation [2–6], and
then to apply a very thin (usually on nanometer scale) coat-
ing of a prepolymer or resin to the modified carbon fiber
surface [7–10] for the purpose of improving the interfacial
properties between the carbon fibers and matrix [11,12],
*
Corresponding author. Tel.: +86 10 6441 2084.
E-mail addresses: chenwm@iccas.ac.cn (W. Chen), yuyh@mail.buct.
edu.cn (Y. Yu), yangxp@mail.buct.edu.cn (X. Yang).
1
Tel.: +86 10 6442 7698.
the process of manufacture. Many investigations on the
nature of the fiber/sizing and sizing/matrix interphases
have proved that the nanometer scale sizing plays a domi-
nant role in the interfacial adhesion of the carbon fiber to
resin matrix [13,14]. The functional groups of the sizing
can react and/or interact with the matrix, which give rise
to strong adhesion between the fiber and the matrix.
The filament winding process is an efficient and viable
technique, and commonly using in the mass production
of advanced fiber reinforced composites such as pressure
vessels, pipes and shafts. Many of these kinds of filament
wound composites are also using in such applications as
aviation and spaceflight for their high strength, light weight
and high stiffness. However, owing to the extreme inertness
of surface caused by the alignment of graphitic crystallites
[15,16], high performance carbon fibers such as T800/
T1000 or M50J/M60J always failed to reach the expected

0266-3538/$ - see front matter  2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compscitech.2007.01.026
2262 W. Chen et al. / Composites Science and Technology 67 (2007) 2261–2270

properties in their reinforced composites. Therefore, mod-
ification of the interfacial behaviors between carbon fiber
and matrix, and between matrix and sizing is a feasible
way to improve the mechanical properties of the high per-
formance CFRP.
In this research, development of new epoxy resins is
tried to obtain high performance T800 CF/epoxy compos-
ites. For this purpose, a combination of hardeners was car-
ried out and a diluter was added to the combined matrix,
and then the effect of the new epoxy matrix was investi-
gated according to the NOL-ring test and other mechanical
property test methods. SEM, FT-IR, XPS and AFM were
used to analyze the interfacial behaviors. Also, sizing effect
was studied in consideration of chemical reactions between
sizing and matrix, which could improve the performance of
composites.
2. Experimental
2.1. Materials
Three types of epoxy resins were used in this work. The
diglycidyl ether of bisphenol A (DGEBA) type epoxy resin
was supplied by Yueyang resin factory China (epoxy value,
0.51), the diglycidyl ester of aliphatic cyclo (DGEAC) type
epoxy resin was provided by Tianjin Jindong chemical fac-
tory (epoxy value, 0.85), and the tetraglycidyl diaminodi-
phenyl methane (TGDDM) type epoxy resin was
supplied by Shanghai Institute of Synthetic Resins (epoxy
value, 0.80).
MeTHPA, DDM and 2,4-EMI, bought from Tianjin
Synthetic Material Research Institute, China, were used
as hardeners, 1 phr of 2,4-EMI was added into matrix
when the hardener was MeTHPA. The unit phr is an
abbreviation of part per one hundred base resins. A mix-
ture hardener of two diethyltoluene diamine (DETDA)
isomers (74–80% 2,4-isomer and 18–24% 2,6-isomer)
was purchased from Lonza, Switzerland. The chemical
structures of the resins and hardeners are shown in
Table 1.
The liquid aromatic diamine (DETDA) was selected to
be a less reactive hardener with an epoxy resin [17]. So in
this work, it was mixed with the hardener DDM in a spe-
cific ratio to prolong the shelf life of the matrix. A diglyc-
idyl ether type diluter with a low molecular weight was
synthesized by authors to reduce the viscosity of resins.
Both T300 and T800 CFs were obtained from Toray Co.,
Japan. T800 CFs possess a diameter of 5 lm, a tensile
strength of 5.5 GPa, a modulus of 294 GPa and an elonga-
tion of 1.9%.

Table 1
Chemical structures of the materials
Materials Chemical structures
DGEBA CH3 CH 3
CH2 CH CH2 O C O CH2 CH CH 2 O C O CH2 CH CH2
O n O
CH3 OH CH3
DGEAC O O
C O CH2 CH CH2
O
C O CH2 CH CH2
O O
TGDDM O O
CH2 CH CH2 CH2 CH CH2
N CH2 N
CH2 CH CH2 CH2 CH CH2
O O
MeTHPA O
H3C
O

O
DDM
H2 N CH 2 NH 2

DETDA H2N C2 H5 C2H5

H3C (2,6-isomer) + H3C NH2 (2,4-isomer)

H2N C2H5 H 2N C2H5

2-methyl-4-eshyl-EMI H
N
C2H5

H 3C N
Diluter CH2-CH-CH2O-(CH2) 4-O-CH2-CH-CH2
O O
 W. Chen et al. / Composites Science and Technology 67 (2007) 2261–2270 2263

A According to the ASTM D2344, short-beam-shear test

of the unidirectional composites was carried out on an
Instron-1196 universal testing machine at a span-to-depth
ratio of 6/1. More than six composite specimens with
dimensions of 20 mm · 6 mm · 2 mm were selected for
each ILSS test.
The commercially sized fibers were desized according to
150 mm
the method as reported [21,22]. Firstly, the carbon fibers
were immersed in dichloromethane (CH2Cl2) for 1 week,
then washed with distilled water for 2 days, and then dried
under vacuum at 110 C for 24 h. Finally, the CH2Cl2 was
evaporated for residue analysis.

2.3. Analysis and characterization

Section A A

Fig. 1. Schematic of NOL-ring specimens.

2.2. Specimen preparation and testing
The resin casts and unidirectional composite laminates
with 60% volume content of carbon fibers were cured at
80 C/2 h + 120 C/2 h + 150 C/4 h when the hardener
was MeTHPA and at 80 C/1 h + 120 C/2 h + 150 C/
3 h + 180 C/1 h when the hardener was amine.
A NOL-ring, as shown in Fig. 1, is one kind of filament
wound composite sample. Not only can it reflect the capa-
bility of the composite to transfer load, but also assess the
interfacial adhesion of the composite [18,19]. In this work,
NOL-rings were produced on a filament winding machine
(MAW20-LS1-6, Mikrosam Co., Macedonia) with a wind-
ing tension of 25 N. The cure cycles were the same as that
of the unidirectional composite laminates. The tensile
strength of the NOL-ring was tested on an Instron-1196
universal testing machine at a rate of 5 mm/min [20].
It should be noted that the tensile strength (r) can be
determined as
r ¼ p=2tw;
where p is the ultimate burst force recorded in Newton (N),
t and w are the thickness and width of the NOL-ring in mil-
limeter (mm), respectively. Six specimens were tested for
each experiment to get an average value.
FT-IR spectra were taken with a Magna IR TM Spec-
trometer 550 (Nicolet) in the reflectance mode. Data acqui-
sition was performed automatically using an inter-faced
computer and a standard software package. The samples
were dried under vacuum at 110 C. The sample chamber
was kept purged with nitrogen during the whole experi-
ment. The surface of carbon fibers was analyzed using a
VG Scientific LAB MK-II X-ray photoelectron spectrome-
ter (XPS). The spectra were collected using a Mg Ka X-ray
source (1253.6 eV). The pressure in the chamber was held
below 5 · 108 Torr during analysis. All XPS spectra were
recorded at a 45 take-off angle.
The samples were observed with Cambridge Stereoscan
250MK3 scanning electron microscope (SEM). All samples
were coated with gold to avoid the electric charge. AFM
measurement was performed on a Nanoscope III instru-
ment by fastening a carbon fiber filament to a steel sample
mount using double sided tape to observe the surficial mor-
phology of the carbon fiber. All images were collected in air
using the tapping mode with a silicon nitride probe.
3. Results and discussion
3.1. Mechanical properties of resin casts
The mechanical properties of the three kinds of resin
casts (DGEBA, DGEAC and TGDDM) are shown in
Table 2. It was seen that the three kinds of resins with

Table 2
Tensile properties of resin casts
Resin/Hardener Tensile strength (MPa) Elongation (%) Tensile modulus (GPa)
DGEBA MeTHPA 60 ± 5 2.1 ± 0.2 2.2 ± 0.2
DDM 70 ± 5 2.3 ± 0.1 2.2 ± 0.1
DDM/DETDA 72 ± 6 2.5 ± 0.2 2.0 ± 0.2
DGEAC MeTHPA 62 ± 5 2.5 ± 0.1 3.2 ± 0.2
DDM 88 ± 4 3.1 ± 0.2 3.6 ± 0.3
DDM/DETDA 98 ± 5 3.7 ± 0.2 3.2 ± 0.2
TGDDM MeTHPA 65 ± 6 3.0 ± 0.3 2.4 ± 0.2
DDM 71 ± 4 2.9 ± 0.3 2.7 ± 0.5
DDM/DETDA 75 ± 6 3.1 ± 0.4 2.8 ± 0.4
2264 W. Chen et al. / Composites Science and Technology 67 (2007) 2261–2270

160 3.2. The ILSS and NOL-ring tensile strength of T800

MeTHPA
CF/epoxy composites
DDM
DDM/DETDA
Fig. 2 shows the ILSS of the unidirectional T800 CF
120
109
104 composites with the three kinds of resin matrices. For the
102 101
DGEBA resin, no matter what kind of hardener was used,
ILSS (MPa)

91
80 77
69 72
67
40
0 over 100 MPa. However the T800 CF reinforced DGEAC/
DGEBA DGEAC TGDDM
Fig. 2. ILSS of T800 CF composites with different matrices.
the hardener of DDM or DDM/DETDA showed higher
tensile properties than the resins with the hardener of
MeTHPA. For the DGEBA resin casts, no matter what
hardener was, all of their tensile properties were lower than
those of the other two resin casts. With regard to the
TGDDM resin casts, they showed tensile strengths close
to those of the DGEBA resin casts but higher elongations.
While the DGEAC resin casts with the mixture hardeners
of DDM and DETDA showed the highest tensile strength
and modulus. Therefore, DGEAC/DDM/DETDA system
is recommended to a promising matrix in this work.
the T800 CF composites showed the lowest ILSS. The main
reason is that T800 CF, like other high toughness carbon
fibers [23], has an elongation of 1.9%, which needs higher
toughness of matrix. However, as shown in Table 2, the
DGEBA resin casts only had an elongation of the order
of 2.1–2.5%, which will result in premature failure in the
interphase of the T800 CF composites. In contrast, all of
the T800 CF reinforced TGDDM composites have a ILSS
DDM/DETDA composites showed the highest ILSS.
Combining the data of Table 2 and Fig. 2, a good relation-
ship between the toughness of resin casts and the ILSS of
T800 CF reinforced composites is found, that is, the higher
toughness of matrix is, the higher ILSS of composites will
be.
SEM observations of the interphase regions of the T800
CF composites after ILSS test are shown in Fig. 3(a)–(c).
Compared with the T800 CF/DGEBA/MeTHPA compos-
ites [Fig. 3(a)], both T800 CF/DGEAC/DDM/DETDA
and T800 CF/TGDDM/MeTHPA composites had stron-
ger interfacial adhesion, with carbon fibers well covered
by matrices.
However, it is interesting to see that the changes in the
tensile strength of T800 CF NOL-ring composites shown
in Fig. 4 are different from the changes in the ILSS of

Fig. 3. SEM photographs of (a) T800 CF/DGEBA/MeTHPA, (b) T800 CF/DGEAC/DDM/DETDA and (c) T800 CF/TGDDM/MeTHPA.
 W. Chen et al. / Composites Science and Technology 67 (2007) 2261–2270 2265

4000
MeTHPA 3000 1
DDM/DETDA
Viscosity
3200 Tensile strength of NOL rings 2570 0
2510

Viscosity (Pa⋅s/25ºC)
ILSS 2430

Strength (MPa)
2500
Tensile strength (MPa)

2355
2220 2230 2245
-1
2400 2220
2030 2119 2021 2000
1889 1994 125
123 122 -2
118 117
116
1600 109 110
-3
100

800 -4
0 5 10 15 20 25 30
Diluter (phr)

0 Fig. 5. Change in mechanical properties of T800 CF/DGEAC/DDM/

DGEBA DGEAC TGDDM DETDA composites with matrix viscosity.

Fig. 4. Tensile strength of T800 CF NOL-ring.

T800 CF composites shown in Fig. 2. Because of high vis-
cosity and short shelf life, the epoxy/DDM system was not
used anymore as the matrix to make NOL-rings in filament
winding. As shown in Fig. 4, the T800 CF NOL-ring com-
posite with the DGEAC/DDM/DETDA system as matrix
showed the highest tensile strength. On the other hand, the
T800 CF/TGDDM NOL-ring composites did not show
high enough tensile strength in spite of high ILSSs due to
the high viscosity of resin, which has adversely influenced
the wettability to carbon fibers in the filament winding
[24]. Therefore, unlike the static process of unidirectional
composites, a matrix with low viscosity and good wettabil-
ity to carbon fibers is more necessarily recommended for
the dynamic filament winding process.
3.3. The ILSS and NOL-ring tensile strength of T800
CF/toughened-epoxy composites
In order to satisfy the requirements of the viscosity of
the resin matrix and the wettability to carbon fibers in fila-
ment winding process, a viscosity diluter, which was syn-
thesized by authors, was added to the DGEAC/DDM/
DETDA system. Table 3 shows the mechanical properties
of the DGEAC/DDM/DETDA system with different vis-
cosity. The addition of the diluter to the DGEAC/DDM/
DETDA system has reduced the viscosity of the matrix
obviously. In addition, both tensile strength and elongation
were improved while the modulus was reduced with addi-
tion of the diluter.
Fig. 5 shows the effect of matrix viscosity on the mechan-
ical properties of T800 CF reinforced DGEAC/DDM/
DETDA composite. Both the ILSS and tensile strengths
of NOL-rings were improved when the diluter was added.
With 25 phr diluter, the ILSS of the T800 CF composite
has reached to 122 MPa and the tensile strength of NOL-
ring increased to 2570 MPa. In consideration of the com-
mercial T300 or T700 CF composites with only about
90 MPa of ILSS [19,21–23], T800 CF composites show
about 36% higher in ILSS value with small amount of
diluter addition. These results suggested that the optimum
viscosity of matrix is necessary to increase the interfacial
properties of T800 CF/epoxy filament wound composites.
Fig. 6 is the fracture pictures of T800 CF NOL-rings for
DGEBA/MeTHPA, DGEAC/DDM/DETDA/diluter and
TGDDM/MeTHPA after burst tests. Each picture on the
right is the amplified section of the NOL-ring shown in
the left one. As shown in Fig. 6(a), the T800 CF/
DGEBA/MeTHPA NOL-ring composite showed a frac-
ture of delamination, while fractures of the other two
NOL-rings showed that all the T800 CFs carried the load
synchronously and ruptured together. These kinds of bro-
ken modes have been reported to identify whether the
interfacial adhesion of NOL-ring is good or not [20]. The
results confirm that DGEAC/DDM/DETDA/diluter or
TGDDM/MeTHPA matrix has a better interfacial adhe-
sion to T800 CFs. The results can also be convinced by

Table 3
Mechanical properties of the DGEAC/DDM/DETDA system with different viscosity
Diluter (Phr) Viscosity (Pa s/25 C) Tensile strength (MPa) Elongation (%) Tensile modulus (GPa)
0 1.40 98 ± 5 3.7 ± 0.2 3.2 ± 0.2
5 1.11 100 ± 3 3.8 ± 0.2 2.9 ± 0.3
10 0.68 102 ± 4 3.8 ± 0.3 2.6 ± 0.2
15 0.61 101 ± 3 3.9 ± 0.2 2.5 ± 0.2
20 0.57 103 ± 5 4.1 ± 0.1 2.3 ± 0.4
25 0.51 102 ± 4 4.4 ± 0.1 2.2 ± 0.2
30 0.44 101 ± 5 5.0 ± 0.2 1.9 ± 0.4
2266 W. Chen et al. / Composites Science and Technology 67 (2007) 2261–2270

Fig. 6. Failure sections of NOL specimens of (a) T800 CF/DGEBA/MeTHPA, (b) T800 CF/DGEAC/DDM/DETDA/Diluter and (c) T800
CF/TGDDM/MeTHPA.

Fig. 7. SEM photographs of failure section for NOL-ring of (a) T800 CF/DGEBA/MeTHPA, (b) T800 CF/DGEAC/DDM/DETDA/Diluter and (c)
T800 CF/TGDDM/MeTHPA.
 W. Chen et al. / Composites Science and Technology 67 (2007) 2261–2270
the SEM photographs of the failure sections of NOL-ring
composites after burst test as shown in Fig. 7. It can be
seen from Fig. 7(b) and (c) that carbon fibers are well
adhered by the matrix, while in Fig. 7(a) some carbon fibers
are separated from matrix and pulled out with an indicator
of poor adhesion.
3.4. The interfacial behavior analysis of T800 CF/epoxy
composites
As well-known, the interface/interphase plays a vital
role in stress transfer from one fiber to another in the
matrix of a composite. An optimal interfacial adhesion is
required for taking advantage of the excellent mechanical
properties of T800 CFs in composites. For good bonding
with matrix and facilitating handling during composite
manufacture, carbon fibers are generally sized with differ-
ent polymers after surface-treatment [25–28]. Among the
different commercial sizings of CFs, DGEBA is the most
common [21,22,29,30]. It has been also reported that the
presence of sizing may improve the wetting of the fiber
by the matrix resin and protect its reactivity [31,32]. There-
fore, to analyze the interfacial behavior of the T800/
DGEAC/DDM/DETDA/diluter composite with good
interfacial adhesion, sizing was analyzed first.
Fig. 8 shows the FT-IR results of DGEBA and the siz-
ings on T800 CF and T300 CF, respectively. The spectra of
the sizings on T800 CF and T300 CF are almost same and
also similar with that of DGEBA. As reported [21,22], the
sizing obtained from T300 CF was DGEBA, with an ali-
phatic ether added. Therefore, it can be deduced that the
sizing on the T800 CFs is also the matter of DGEBA, with
an aliphatic ether added.
However, it should be noted that the bands of T300 and
T800 CFs in the 3700–3100 and 1740 cm1 are obvious dif-
DGEBA sizing of T800
sizing of T300
3500 3000 2500 2000
Wavenumber (cm-1)
Fig. 8. FT-IR spectra of DGEBA resin and sizings on carbon fibers.
2267
ferent from the bands of DGEBA. The downward convex
bands in 3700–3100 and 1740 cm1 are due to the vibration
of O–H and C@O which may come from the curing reac-
tion of epoxy resin and carboxylic acid anhydride [33,34].
Therefore, the sizing on T800 CFs may be the matter of
DGEBA with anhydride added to form certain cross-link-
ing. That is to say, the sizing may contain high molecular
weight of DGEBA.
Since all the DGEAC, the diluter and the sizing contain
epoxy groups, a three-dimensional network can be formed
between the sizing and the DGEAC/DDM/DETDA/
diluter matrix system by taking epoxy–amine curing reac-
tion. The mechanism of epoxy–amine curing reactions
has been widely analyzed [35–37].
The complex epoxy–amine curing reactions and the typ-
ical autocatalytic behavior make it very difficult to quanti-
tatively characterize the structures of the interphase and
the bulk resin in the T800 CF/DGEAC/DDM/DETDA/
diluter system. However, some qualitative analysis can be
used to investigate the compatibility of the T800 CF with
the DGEAC/DDM/DETDA/diluter matrix.
The elemental compositions for three kinds of T800
CFs, which can be obtained from XPS spectrum are shown
in Fig. 9. It is generally accepted that the active functional
groups (C–OH; C@O; COOH) on the carbon fiber surface
contribute an important role in fiber/matrix adhesion
because of the possibility of forming a chemical bond
between fiber and matrix. Therefore, O1s/C1s composition
ratio was used to predict the adhesion between fiber and
matrix [8,38]. It was seen in Fig. 9 that the unsized T800
CF showed the lowest O1s/C1s ratio, while the commer-
cially sized T800 CF showed a higher O1s/C1s composition
ratio than the unsized T800 CF, indicating that sizing is
helpful to improve the interfacial adhesion between T800
CFs and matrix.
1500 1000 500
2268
C1S
120
O 1S
Elemental composition (%)
O 1S /C 1S
90
74.40
69.67
60
30 20.01
0
unsized T800 sized T800
Fig. 9. Elemental composition and O1s/C1s ratio of carbon fibers.
In addition, compared with the commercially sized T800
CF, the T800 CF in the T800 CF/DGEAC/DDM/
DETDA/diluter composites showed higher content of O1s
and higher ratio of O1s/C1s which possibly arose from the
carbon fiber surface and the epoxy rings of matrix. This
is due to the strong chemical reaction between matrix
and sizing which resulted in good interfacial adhesion of
T800 CF/DGEAC/DDM/DETDA/diluter composite.
Therefore, we suggest that the interface/interphase
adhesion of T800 CF composites is one model of ‘‘double
interface/interphase’’. One is the adhesion between carbon
fiber and sizing, another is that between the sizing and
matrix. Because T800 carbon fibers like T300 or T700
CFs is one of commercial carbon fibers, they have been
already surface treated. These treatments increase the sur-
Fig. 10. Three-dimensional AFM topographical images of (a) unsized T800 CF, (b) commercial sized T800 CF, (c) T800 CF in DGEAC/DDM/DETDA/
diluter matrix composite.
W. Chen et al. / Composites Science and Technology 67 (2007) 2261–2270
0.4 face-active sites and improve bonding between the fibers
and the resin matrix [11–14]. However, strong treatment
may damage the carbon fiber surface, and decrease its
0.3 properties. Some researchers reported that the treatment
on the commercial carbon fiber surface did not improve
71.02
O1S /C1S
the fiber/matrix adhesion well [21,22]. So, it indicates that
0.2 the interface/interphase adhesion formed through the
chemical reactions between the sizing and the matrix is
more important in the interfacial behaviors of T800 CF
21.03 22.11 0.1 composites than the adhesion formed by the T800 CFs
and the sizing. In order to improve the interfacial adhesion
of T800 CF composites, emphasis should be placed on the
T800 in composites
0.0 chemical interface/interphase between the sizing and the
matrix rather between the T800 CFs and the sizing by
developing functional groups.
AFM was used to characterize the roughness of the
three T800 CF surfaces. Resultant AFM images of
2 lm · 2 lm are shown in Fig. 10. Like other PAN-based
CFs [38–40], all the three T800 CF surfaces have clear
ridges and striations running along the axises of the fibers.
As seen in Fig. 10a and b, the longitudinal ridges and the
striations of unsized T800 CF are less well defined than
commercial sized T800 CF. This difference presented that
the original features of the surface topography were chan-
ged by the sizing [8,10]. In addition, Fig. 10c reveals that
the longitudinal ridges and striations of T800 CF are cov-
ered by matrix on a nanometer scale. This is presumably
due to the chemical reactions between the sizing and the
matrix to form a coating along the carbon fiber [41]. These
assertions obtained from Fig. 10 were corroborated with
the roughness parameters in Table 4.
 W. Chen et al. / Composites Science and Technology 67 (2007) 2261–2270
Table 4
Main roughness parameters of the T800 CFs measured by AFM on
images of 2 lm · 2 lm
Carbon fibers Ra (nm) Rmax (nm)
Unsized T800 CF 12.66 69.83
Commercial sized T800 CF 25.84 115.40
T800 CF in composites 42.49 188.30
Table 4 summarizes the results of the roughness analysis
of T800 CF as obtained from AFM, where Ra is the mean
surface roughness value and Rmax is the maximum height
roughness value. Like some other carbon fibers [8,10], the
result in this work revealed that the roughness of the com-
mercial sized T800 CF was twice higher than that of the
unsized T800 CF. Comparison the roughness values of
the commercial sized T800 CF with T800 CF in
DGEAC/DDM/DETDA/diluter/T800 composites, the
interfacial reaction between the sizing and the matrix
changes the surface topography very much. The Ra and
the Rmax values increased from 25.84 to 42.49 nm and from
115.40 to 188.30 nm, respectively. It was reported that the
improved interfacial adhesion was strongly related to the
surface structure, while the chemical component of the sur-
face had little influences on the interfacial adhesion [42,43].
In contrast, many investigators have recognized that the
surface structure does not play significant role in the adhe-
sion and the chemical component is underlined [44,45].
Therefore, the results obtained from AFM analysis in this
study suggest that the chemical interface/interphase
between the sizing and the matrix plays an important role
in the interfacial adhesion and composite mechanical prop-
erties. This result is in good agreement with the previous
XPS and FT-IR analysis.
4. Conclusion
The new epoxy matrix, DGEAC/DDM/DETDA/
diluter, has a good interfacial adhesion to T800 carbon
fibers. The ILSS and tensile strength of NOL-ring of this
T800 CFs reinforced composites could reach up to
123 MPa which is about 36% higher than commercial
T300 or T700 composites and 2570 MPa, respectively.
The chemical reaction of T800 CF/DGEAC/DDM/DET-
EDA/diluter was proved as an epoxy–amine reaction.
Also, sizing the carbon fiber is very important to increase
the performance of composite because the sizing and the
matrix makes another chemical reaction and resulting in
the ‘‘double interface/interphase’’ adhesion in the
composites.
References
[1] Donnet JB, Bansal RC. Carbon fiber. 3rd ed. New York: Marcel
Dekker; 1998.
[2] Fitzer E, Weiss R. Effect of surface treatment and sizing of carbon
fibers on the mechanical properties of CFR thermosetting and
thermoplastic polymers. Carbon 1987;25(4):455–67.
2269
[3] Severini F, Formaro L, Pegoraro M, Posca L. Chemical modification
of carbon fiber surfaces. Carbon 2002;40(5):735–41.
[4] Pittman Jr CU, Jiang W, He GR, Gardner SD. Oxygen plasma and
isobutylene plasma treatment of carbon fibers: determination of
surface functionality and effects on composite properties. Carbon
1998;36(1):25–37.
[5] Pamula E, Rouxhet PG. Bulk and surface chemical functionalities of
type III PAN-based carbon fiber. Carbon 2003;41(10):1905–15.
[6] Pittman Jr CU, Jiang W, Yue ZR, Gardner S, Wang L, Toghiani H,
et al. Surface properties of electrochemically oxidized carbon fiber.
Carbon 1999;37(11):1797–807.
[7] Montes-Moran MA, Young RJ. Raman spectroscopy study of high-
modulus carbon fibers: effect of plasma-treatment on the interfacial
properties of single-fiber-epoxy composite, part II: characterization of
the fiber–matrix interface. Carbon 2002;40(6):857–75.
[8] Dilsiz N, Wightman JP. Surface analysis of unsized and sized carbon
fibers. Carbon 1999;37(7):1105–14.
[9] Iron JO, Yuan W. Surface properties of carbon fiber modified by
electrodeposition of polyamic acid. Polymer 1996;37(18):4197–203.
[10] Hideko T, Oyama, Wightman JP. Surface characterization of PVP-
sized and oxygen plasma-treated carbon fiber. Surf Interf Anal
1998;26:39–55.
[11] Ivens J, Albertsen H, Wevers M, Verpoest I, Peters P. Interlaminar
fracture toughness of CFRP influenced by fiber surface treatment.
Part 2: Modeling of the interface effect. Compos Sci Technol
1995;54(2):147–59.
[12] Montes-Morán MA, Matı́nez-Alonso A, Tascón JMD, Paiva MC,
Bernardo CA. Effect of plasma oxidation on the surface and
interfacial properties of carbon fiber/polycarbonate composites.
Carbon 2001;39(7):1057–68.
[13] Yumitori S, Wang D, Jones FR. The role of sizing resins in carbon
fibre-reinforced polyethersulfone (PES). Composites
1994;25(7):698–705.
[14] Gao SL, Mäder E, Zhandarov SF. Carbon fiber and composites with
epoxy resins: topography, fractography and interphases. Carbon
2004;42(3):515–29.
[15] Bogoeva-Gaceva G, Mäder E, Häussler L, Dekanski A. Character-
ization of the surface and interphase of plasma-treated HM carbon
fibers. Composites A 1997;28(5):445–52.
[16] Hughes JDH. The carbon fiber/epoxy interface – a review. Compos
Sci Technol 1991;41(1):13–45.
[17] Naffakh M, Dumon M, Dupuy J, Gérard JF. Cure kinetics of an
epoxy/liquid aromatic diamine modified with poly(ether imide). J
Appl Polym Sci 2005;96(3):660–72.
[18] Zhang JS, Karbhari VM, Wu L, Reynaud D. Field exposure based
durability assessment of FRP column wrap systems. Composites B
2003;34(1):41–50.
[19] Ping C, Xigao J, Hui C, Julong G, Bing H, Xingsong Z. Study of
toughened epoxy resins matrix for carbon fiber composite motor case.
Acta Mater Comp Sinica 2002;19(2):24–7.
[20] Liu L, Huang YD, Zhang ZQ, Jiang B, Nie J. Ultrasonic modification
of aramid fiber–epoxy interface. J Appl Polym Sci
2001;81(11):2764–8.
[21] Marieta C, Schulz E, Irusta L, Gabilondo N, Tercjak A, Mondragon
I. Evaluation of fiber surface treatment and toughening of thermoset
matrix on the interfacial behaviour of carbon fiber-reinforced cyanate
matrix composites. Compos Sci Technol 2005;65(14):2189–97.
[22] Marieta C, Schulz E, Mondragon I. Characterization of interfacial
behaviour in carbon-fibre/cyanate composites. Compos Sci Technol
2002;62(2):299–309.
[23] Baoyan Z, Xjanao C, Ping L, Xiaosu Y. Toughness and hot/wet
properties of a novel modified BMI/carbon fiber composite. J Mater
Sci Technol 2001;17(1):17–8.
[24] Morgan RJ, Walkup CM. Epoxy matrices for filament-wound carbon
fiber composites. J Appl Polym Sci 1987;32(1):37–46.
[25] Drzal LT, Micheal RJ, Koenig MF, Lloyd PF. Adhesion of graphite
fibers to epoxy matrices. II. The effect of fiber finish. J Adhes
1983;16:133–52.
2270 W. Chen et al. / Composites Science and Technology 67 (2007) 2261–2270
[26] Lee JS, Kang TJ. Changes in physico-chemical and morphological
properties of carbon fiber by surface treatment. Carbon
1997;35(2):209–16.
[27] Piggott MR. The interface in carbon fibre composites. Carbon
1989;27(5):657–62.
[28] Blackketter DM, Vpadhyaya D, King TR, King JA. Evaluation of
fiber surfaces treatment and sizing on the shear and transverse tensile
strengths of carbon fiber-reinforced thermoset and thermoplastic
matrix composites. Polym Compos 1993;14(5):430–6.
[29] Dilsiz N, Wightman JP. Effect of acid–base properties of unsized and
sized carbon fibers on fiber: epoxy matrix adhesion. Colloid Surf A
2000;164(2–3):325–36.
[30] Fernández B, Arbelaiz A, Valea A, Mujika F, Mondragon I. A
comparative study on the influence of epoxy sizings on the mechanical
performance of woven carbon fiber–epoxy composites. Polym Com-
pos 2004;25(3):319–30.
[31] Huttinger KJ, Krekel G. Polydimethylsiloxane coated carbon fibres
for the production of carbon-fibre reinforced carbon. Carbon
1991;29(8):1065–70.
[32] Chang TH, Zhang J, Yumitori S, Jones FR, Anderson CW. Sizing
resin structure and interphase formation in carbon fibre composites.
Composites 1994;25(7):661–70.
[33] Stevens GC. Cure kinetics of a low epoxide/hydroxyl group-ratio
bisphenol a epoxy resin–anhydride system by infrared absorption
spectroscopy. J Appl Polym Sci 1981;26(12):4279–97.
[34] Stevens GC. Cure kinetics of a high epoxide/hydroxyl group-ratio
bisphenol a epoxy resin–anhydride system by infrared absorption
spectroscopy. J Appl Polym Sci 1981;26(12):4259–78.
[35] Ramos JA, Paganib N, Riccardi CC, Borrajo J, Goyanes SN,
Mondragon I. Cure kinetics and shrinkage model for epoxy–amine
systems. Polymer 2005;46(10):3323–8.
[36] Cedeňo AJ, Vázquez-Torres H. Kinetic study of the effect of
poly(phenyl sulfone) on the curing of an epoxy/amine resin by
conventional and by temperature modulated differential scanning
calorimetry. Polym Int 2005;54(8):1141–52.
[37] Tanaka Y, Mika TF. In epoxy resins, chemistry and technology. In:
May CA, Tanaka Y, editors. New York: Marcel Dekker; 1973.
[38] Zhang Z, Liu Y, Huang Y, Liu L, Bao J. The effect of carbon-fiber
surface properties on the electron-beam curing of epoxy–resin
composites. Compos Sci Technol 2002;62(3):331–7.
[39] Brasquet C, Rousseau B, Estrade-Szwarckopf H, Cloirec PL.
Observation of activated carbon fibres with SEM and AFM corre-
lation with adsorption data in aqueous solution. Carbon
2000;38(3):407–22.
[40] Figueiredo JL, Serp P, Nysten B, Issi JP. Surface treatments of vapor-
grown carbon fibers produced on a substrate. Part II: Atomic force
microscopy. Carbon 1999;37(11):1809–16.
[41] Cazeneuve C, Castle JE, Watts JF. The structure of the interface in
carbon fibre composites by scanning auger microscopy. J Mater Sci
1990;25:1902–8.
[42] Yuan LY, Shyu SS, Lai JY. Plasma surface treatments on carbon
fibers. II. Mechanical property and interfacial shear strength. J Appl
Polym Sci 1991;42(9):2525–34.
[43] Krekel G, Huttinger KJ, Hoffman WP. Enhanced interfacial adhesion
of carbon bers to vinyl ester resin using poly(arylene ether phosphine
oxide). J Mater Sci 1994;29:3984–92.
[44] Scola DA, Brooks CS. Surface aspects of new fibers, boron, silicon
carbide, and graphite. J Adhes 1970;2:213–37.
[45] Farrow GJ, Atkinson KE, Fluck N, Jpnes C. Effect of low-power air
plasma treatment on the mechanical properties of carbon fibres and
the interfacial shear strength of carbon fibre–epoxy composites. Surf
Interf Anal 1995;23:313–8.