#2009 The Society of Polymer Science, Japan

Fabrication, Morphology and Cure Behavior

of Triethylenetetramine-Grafted Multiwalled
Carbon Nanotube/Epoxy Nanocomposites
By Kai YANG
and Mingyuan G U

Multiwalled carbon nanotubes (MWCNTs) were first treated by a 3:1 (v/v) mixture of concentrated H2 SO4 /HNO3 , and then
triethylenetetramine (TETA) grafting was carried out. X-Ray photoelectron spectroscopy and Raman spectroscopy analyses
proved the effectiveness of this kind of chemical functionalization method. The TETA thin layer was formed on the MWCNT
surface, which contributes to good dispersion of MWCNTs in epoxy matrix. Cure behaviors of MWCNT/epoxy systems
were studied with non-isothermal differential scanning calorimetry. MWCNTs have the retarding effect on the cure reaction
owing to their steric hindrance, while TETA functional groups on the MWCNT surface can accelerate the primary amine-
epoxide reaction and weaken the inhibition effect on the process of cure. A qualitative analysis of the degree of vitrification in
different epoxy systems was performed. The activation energy of the cure reactions was obtained by the Flynn-Wall-Ozawa
method.
KEY WORDS: Carbon Nanotubes / Polymer-Matrix Composites (PMCs) / Cure Behavior / Differential Scanning Calorimetry (DSC) /
Surface Treatments /

Carbon nanotubes (CNTs) have stimulated intense research Epoxy resins are especially interesting owing to their high
interest since the landmark paper by Iijima,1 largely because of modulus, low shrinkage in cure, high adhesion, good chemical
their potential applications in diverse fields of nanoscience and and corrosion resistance.21 Combining the properties of epoxy
nanotechnology,2–4 unique structure and superior physical with CNTs can supply more extensive applications. Conse-
properties such as high Young’s modulus, excellent electrical quently, it is highly worthy to study the cure behaviors of this
and thermal conductivities, low density, high surface area and type of polymer composites. Although there are some studies
high chemical stability.5–7 Recently, CNT/polymer composites on the cure reactions of CNT/polymer systems,22–24 the
have been regarded as an extremely promising research topic in influence of CNTs on the cure behaviors has not been clearly
the field of material science. Subsequently, extensive research- and in depth addressed in the literature yet. In order to obtain
es have been focused on the use of CNTs as nano-fillers to superior physical properties of CNT/polymer composites, we
produce a variety of high-performance polymer composites for need to purify CNTs to get ride of various impurities. Chemical
targeted applications. CNT/polymer composites, including modification not only contributes to the purification process,
matrices such as polyethylene,8 poly (methyl methacrylate),9 but also facilitates better dispersion of CNTs in polymer
polypropylene,10 polystyrene,11,12 nylon13,14 and polyamide,15 matrix. More specially, some functional groups can be
have been studied. The filling of CNTs brings forth very generated on the surface of CNTs through chemical modifica-
remarkable improvement of the various properties of polymer tion. These functional groups have an impact upon the
systems (such as the mechanical, electrical and thermal structures and performances of CNT/polymer composites.
properties, etc.).16–19 Thereby, it is probable to manufacture In the present work, an effective route to synthesize
novel composites by means of combining the complementary chemically functionalized MWCNT/epoxy composites is pro-
properties of CNTs and polymers. posed. In this functionalization process, grafting triethylenetetr-
In order to make an optimum use of CNT/polymer amine (TETA) on the multiwalled CNT (MWCNT) surface can
composites, we should understand the nature of their curing bridge the connection of MWCNTs to the epoxy matrix. The
process and the structure of the cured material. Especially, the presence of chemically bonded TETA transforms the smooth
study of the cure reactions of polymer composites as a function and nonreactive surface of MWCNTs into a hybrid material that
of the processing conditions is extremely critical for the has the characteristics of both MWCNTs and TETA, which will
analysis and design of processing operations. The final proper- be conducive to the performance of MWCNT/epoxy compo-
ties of CNT/polymer composites depend significantly on the sites. We used dynamic differential scanning calorimetry (DSC)
curing conditions. Accordingly, the research of the cure to investigate the cure behaviors of MWCNT/bisphenol-A
behaviors is conducive to both a better knowledge of process glycidol ether epoxy resin (DGEBA)/2-ethyl-4-methylimida-
development and an enhancement of the quality of final zole (EMI-2,4) composites and analyze the effects of MWCNTs
products related to the structures of the polymer composites.20 and chemical functionalization on the curing process.
State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai 200240, China


To whom correspondence should be addressed (Fax: +86 021 62822012, E-mail: kyang [email protected]).

752 Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009 doi:10.1295/polymj.PJ2009064
Preparation and Cure Behavior of Modified MWCNT/Epoxy Composites

EXPERIMENTAL

Materials
Epoxy resin utilized in this work was a nominally difunc-
tional epoxy resin, Epon 828 supplied by Shanghai Resin Co.
Epon 828 is, basically, DGEBA with the epoxy value of 0.48–
0.52 mol/100 g and its structure is described by Zhou.25 The
curing agent, EMI-2,4, was provided by Beijing Chemical
Reagent Co. Imidazole is a type of nucleophilic hardener. The
net epoxy cured by it possesses better heat endurance, higher
modulus and wider curing temperature range than that cured by
amine curing agents. In this work, EMI-2,4 has excellent
compatibility with DGEBA. The cure temperature of DGEBA/
EMI-2,4 system is relatively low and its gel time is long. In the
meantime, the final cured product has high heat deformation
temperature, good heat durability, superior chemical stability
and mechanical properties. Therefore, we used EMI-2,4 to
achieve the cure of epoxy. MWCNTs were obtained from the
Nanotech Port Company, Shenzhen, China. These MWCNTs
were produced by chemical vapor deposition (CVD). This kind
of catalytic production is simple and has a high productivity.
The external diameter and the length of MWCNTs are 60–
100 nm and 5–15 mm, respectively.
Figure 1. Scheme for the process of grafting TETA onto the MWCNT
surface.
Chemical Functionalization of MWCNTs
The synthetic procedure is revealed in Figure 1. MWCNTs
were first treated by a 3:1 (v/v) mixture of concentrated Instrumental Analysis
H2 SO4 /HNO3 , with sonication at 40  C for 10 h. After acid X-Ray photoelectron spectroscopy (XPS) analysis was
treatment, MWCNTs were washed using deionized water, performed in an ultra high vacuum system equipped with a
filtered until the pH value reached 7 and dried at 80  C for 24 h. Kratos AXIS Ultra hemispherical electron analyzer, using a
The acid-treated MWCNT (MWCNT-COOH) was reacted with monochromated Al K
source (1486.6 eV), at a base pressure
excess SOCl2 for 24 h under reflux to yield acyl chloride- of 2  1010 mbar. Charging effects were compensated by a
functionalized MWCNT (MWCNT-COCl). The MWCNT- flood gun, and binding energies were calibrated using the main
COCl was further reacted with TETA at 120  C under magnetic C 1s peak at 284.6 eV as internal standard. Raman spectra were
stirring for 96 h to produce MWCNT-TETA. obtained using a Renishaw RM3000 Raman microscope
equipped with a 1800 lines/mm grating and a motorized
Preparation of MWCNT/Epoxy Nanocomposites three-axis XYZ mapping stage. A 514.5 nm excitation line of
In order to check the wetting of MWCNT-TETA in organic an argon laser was used, while the laser power, measured
solvent, such as chloroform, it was dispersed in chloroform and directly on the specimen, was 10 mW. High-resolution
wait for 72 h. MWCNT/epoxy composites were prepared in transmission electron microscopy (HRTEM) images were
solution blended method, namely, using chloroform as the co- taken on JEM 2100F to characterize the microstructures of
solvent. Unmodified or modified MWCNTs were dispersed MWCNTs. The morphological observations were carried out
separately in chloroform and sonicated for 1 h. The dispersion with a field emission scanning electron microscope (FESEM)
was mixed with DGEBA. The mixture was sonicated and stirred (FEI SIRION 200, USA). The specimens were prepared by
for homogenization at 60  C. This treatment increased the mixing DGEBA, EMI-2,4 and unmodified or chemically
viscosity of the dispersion, which in turn limited the MWCNT modified MWCNTs (their weight ratios are displayed in
aggregation. Subsequently, the mixture was treated by a post Table I). Then, the mixtures were sonicated and subsequently
heating at 90  C in order to completely remove the solvent. And homogenized by magnetic stirring for 2 h to break down the
then, the mixture was kept in a vacuum oven for 24 h to get rid agglomerations of MWCNTs. Approximately 5 mg sample of
of air bubbles. Subsequently, the curing agent was added to the the mixture was weighed accurately into an aluminum DSC
mixture, which was stirred by using a magnetic bar for 30 min sample pan and then covered with an aluminum lid. Thermal
under sonication. Finally, the mixture was poured into a mould analysis was carried out by non-isothermal DSC using a
and the whole system was placed in an oven. The MWCNT/ Perkin-Elmer DSC-7 system. The specimens were scanned at
epoxy compound was precured at 70  C for 1 h, cured at 110  C three different heating rates of 10, 15 and 20 Kmin1 under a
for 1.5 h and postcured at 140  C for 1.5 h. constant flow of nitrogen at 50 mLmin1 .

Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009 #2009 The Society of Polymer Science, Japan 753
 K. YANG and M. GU

Figure 2. XPS spectra of (a,b) as-received MWCNTs and (c,d) TETA-grafted MWCNTs.

Table I. Compositions of the specimens (by weight) Table II. Atomic percentage of XPS spectra

Systems DGEBA EMI-2,4 MWCNTs Signal As-received MWCNTs TETA-grafted MWCNTs

S1 100 6 0 C 1s (%) 98.82 85.26
S2 100 6 1 (as-received MWCNTs) O 1s (%) 1.18 8.02
S3 100 6 1 (TETA-grafted MWCNTs) N 1s (%) 0 6.72

spectrum of as-received MWCNTs shows a dominant peak at

RESULTS AND DISCUSSION 284.6 eV attributed to the sp2 carbon atoms of the grapheme
sheets, a peak at 285.6 eV attributed to defects on the nanotube
Characterization of TETA-grafted MWCNTs structure, a peak at 287.9 eV due to –C=O bonds26 and the -
Figure 2 shows the XPS survey spectra of as-received 
transition loss peak at 291.5 eV.27 With respect to TETA-
MWCNTs and TETA-grafted MWCNTs and the deconvolu- grafted MWCNTs, the two peaks appear at 286.8 eV and
tions of the corresponding C 1s spectra. All the spectra exhibit 289.0 eV, which can be assigned to the C–OH bonds of
C 1s and O 1s peaks, whereas the XPS spectrum of TETA- MWCNTs and the COO bonds in carboxylic groups, respec-
grafted MWCNTs exhibits the remarkable intensification of tively. The peaks at 287.8 eV and 292.7 eV are respectively
O 1s peak. The reason for this phenomenon could be due to the attributed to N–C=O bonds and C=C bonds. The above
great effect of H2 SO4 /HNO3 treatment on the MWCNT analysis presents the effectiveness of chemical functionaliza-
surface. In addition, a N 1s peak is also found in this spectrum, tion of MWCNTs.
which indicates that TETA has been successfully grafted onto The Raman spectra exhibited in Figure 3 reveal the D and G
the MWCNT surface (shown in Figure 2(a) and Figure 2(c)). bands at 1352 cm1 and 1580 cm1 for as-received MWCNTs
The surface atomic concentration of the two samples is and TETA-grafted MWCNTs, which are derived from disor-
summarized in Table II. The C 1s XPS spectra of the two dered graphite structures and concentric cylinders of graphite
samples were deconvoluted into separated peaks (presented layers, respectively. Figure 3b depicts the D band intensity of
in Figure 2(b) and Figure 2(d)). Deconvolution of the C 1s TETA-grafted MWCNTs, suggesting that chemical function-

754 #2009 The Society of Polymer Science, Japan Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009
Preparation and Cure Behavior of Modified MWCNT/Epoxy Composites

Figure 3. Raman spectra of (a) as-received MWCNTs and (b) TETA-grafted

MWCNTs.

alization will increase the D band intensity. Moreover, a

shoulder band located at 1622 cm1 can be observed for
Figure 4. (a) HRTEM image of an as-received MWCNT; (b) Magnification of
TETA-grafted MWCNTs. This peak has been related to the the position marked with the white square in (a); (c) HRTEM image
sp2 hybridization of carbon and is regarded as an evidence of of a TETA-grafted MWCNT; (d) Magnification of the position
marked with the white square in (c).
the disruption of the aromatic system of p electrons by the
attached molecules.28,29 The obvious shoulder peak at 1622
cm1 of TETA-grafted MWCNTs as compared to as-received that as-received MWCNTs are curled and entangled; however,
MWCNTs demonstrates a significant content of organic there are few big agglomerates in them. After acid treatment,
molecules covalently attached. MWCNTs are severely entangled and form big agglomeration.
The microstructures of as-received MWCNT and TETA- The interspaces formed in acid-treated MWCNTs are extreme-
grafted MWCNT are shown in Figure 4. It can be seen from ly small. While further TETA grafting breaks those big
Figure 4(a) that the structure of the as-received MWCNT is agglomerates and recovers the loose state of MWCNTs. The
clear. However, the structure of the TETA-grafted MWCNT is reasons for the changes in morphology of MWCNTs are
indistinct (presented in Figure 4(c)), which may be due to analyzed in Figure 6. H2 SO4 /HNO3 treatment would introduce
TETA grafting procedure. In Figure 4(b), the surface of the as- a number of carboxyl and hydroxyl groups onto the MWCNT
received MWCNT is smooth. However, the surface of the walls. Acid-treated MWCNTs have large polar forces derived
TETA-grafted MWCNT becomes rough, which is attributed to from the strong interactions among these polar groups on their
a TETA layer formed on its tube wall. The clearer structures of surface. So the interspaces formed in H2 SO4 /HNO3 -treated
the TETA layer and the tube wall of the TETA-grafted MWCNTs get very small. TETA grafting varies the forces
MWCNT are revealed in Figure 4(d). Thus HRTEM analysis among MWCNTs from polar to nonpolar; In the meantime,
supplies the direct evidence that TETA is effectively grafted TETA molecules may increase the tube-to-tube distances.
onto the MWCNT wall to form a thin layer. Accordingly, the interspaces formed in TETA-grafted
Figure 5 displays FESEM microphotographs of as-received MWCNTs become large again. Moreover, TETA-grafted
MWCNTs (Figure 5(a)), acid-treated MWCNTs (Figure 5(b)) MWCNTs are looser than as-received MWCNTs, which is
and TETA-grafted MWCNTs (Figure 5(c)). It can be observed conducive to their better dispersion in epoxy matrix.

Figure 5. FESEM images of (a) as-received MWCNTs, (b) acid-treated MWCNTs and (c) TETA-grafted MWCNTs.

Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009 #2009 The Society of Polymer Science, Japan 755
 K. YANG and M. GU

Figure 6. Reasons for varying sized interspaces formed in unmodified and chemically modified MWCNTs.

The functionalized MWCNTs usually present a much better

dispersion or higher solubility in solvents. A stable dispersion
was obtained in the case of TETA-grafted MWCNTs when
equal amounts of as-received MWCNTs and TETA-grafted
MWCNTs were respectively added to the same volume of
chloroform and mechanically mixed (shown in Figure 7). As
shown in Figure 7a, all of nanotubes sank in the case of
raw MWCNTs. This indicates that TETA-grafted MWCNTs
(Figure 7b) have a higher degree of miscibility than as-
received MWCNTs owing to the presence of TETA functional
groups on the MWCNT surface.

Morphology and Cure Behavior of MWCNT/Epoxy Nano-

composites
Figure 8 reveals FESEM images of the fracture surfaces of Figure 7. Photograph of (a) as-received MWCNTs and (b) TETA-grafted
as-received MWCNT/epoxy composite and TETA-grafted MWCNTs dispersed in chloroform and wait for 72 h.

MWCNT/epoxy composite. The white regions in Figure 8(a)

represent the aggregated MWCNTs and the region indicated On the other hand, Figure 8(c) shows that TETA-grafted
with an arrow is magnified as shown in Figure 8(b) in order to MWCNTs are uniformly dispersed in epoxy matrix. This
verify existence of the MWCNTs. So as-received MWCNTs means that TETA grafting would be beneficial to good
dispersed in epoxy matrix exist in the form of agglomerates. dispersion of MWCNTs in epoxy resin.

756 #2009 The Society of Polymer Science, Japan Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009
Preparation and Cure Behavior of Modified MWCNT/Epoxy Composites

Figure 8. (a) FESEM image of the fracture surface of as-received MWCNT/epoxy composite; (b) Magnification of the position marked with the white arrow in (a);
(c) FESEM image of the fracture surface of TETA-grafted MWCNT/epoxy composite. The content of the MWCNTs was 1% by weight.

Figure 9. Dynamic DSC thermograms of the different systems. (a) S1 ; (b) S2 ; (c) S3 .

The curing mechanism of DGEBA/EMI-2,4 system in- (gram mass of EMI-2,4 per 100 g mass of DGEBA), the
cludes two stages consisting of the adduct and etherification variation amplitude of
H is small. However, the cure reaction
reactions.30 The adduct reactions were considered to be the will be incomplete if the content of EMI-2,4 is below 3 wt.-%.
catalyst that initiated the etherification reactions which cross- In this work, the concentration of EMI-2,4 is 6 wt.-% which is
link the epoxy resins and determine their final performance. in the range of 3–10 wt.-%. The dynamic DSC thermograms of
Previous work presented that the adduct formation was S1 , S2 and S3 are illustrated in Figure 9(a)–(c), respectively. As
necessary prior to the etherification reactions. Consequently, is shown, only a main exothermic peak of the etherification
low EMI-2,4 concentrations can be used to analyze the reaction appears. So the heat of reaction for the etherification
etherification reactions through suppressing the adduct forma- reaction peak is approximately the total heat of the curing
tion.30 In an attempt to objectively meet this specification, the reaction. Additionally, the functional groups on the chemically
molar heat of the reaction
H should approach a constant. modified MWCNT surface can act as curing agents to react
When the concentration of EMI-2,4 is between 3 and 10 wt.-% with the epoxy resin. However, the released heat is much

Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009 #2009 The Society of Polymer Science, Japan 757
 K. YANG and M. GU

Table III. The onset temperature, the peak temperature and the heat
of cure in S1 –S3 systems at the different heating rate
1
Heating rate (K
min ) 10 15 20
S1 Ti (K) 393.6 401.4 404.8
Tp (K) 403.9 412.1 418.0

H (J
g 1 ) 415.50 417.97 420.48
S2 Ti (K) 404.6 412.3 415.8
Tp (K) 414.6 422.4 427.8

H (J
g 1 ) 403.06 406.18 409.32
S3 Ti (K) 399.7 407.4 411.5
Tp (K) 409.8 417.9 423.3

H (J
g 1 ) 407.83 411.28 413.11

smaller than the heat of the curing reaction, which means that it
can be negligible. Accordingly,
H can be regarded as a Figure 10. Advancement of the cure reactions for non-isothermal cure at
constant over the whole cure reaction. From a dynamic DSC 20 K
min 1 heating rate of S1 , S2 and S3 systems. Dotted lines
are given only for showing the tendency.
run, the total area A of the exothermal peak (the region between
the exotherm and the baseline) is in direct proportion to the
molar heat of the reaction
H released during the whole cure This fact arose from the contributive effect of TETA functional
reaction. The fraction extent of conversion
at any temper- groups on the cure reaction.
ature T can be expressed as:
¼
HT =
H ¼ AT =A (0  Figure 10 presents a comparison of the curing kinetics, at the

 1) where
HT is the heat of reaction of partially cured heating rate of 20 Kmin1 , among S1 , S2 and S3 systems. All
samples heated up to the temperature T. Table III represents curves show a sigmoidal form, which denotes that MWCNT/
the onset temperature Ti , the peak temperature Tp , and the heat epoxy systems have an autocatalytic kinetics. It is known that
of cure
H of several systems at the different heating rates. DGEBA/EMI-2,4 system has autocatalytic kinetics. That is
With increasing the heating rate, the values of Ti and Tp to say, the filling of unmodified and chemically modified
increase in each system. The reason for Ti and Tp following the MWCNTs does not change the cure mechanism of epoxy resin.
sequence S1 < S2 and S1 < S3 is the introduction of MWCNTs In addition, it is obvious that S2 and S3 systems exhibit the
in the systems, which indicates that MWCNTs delay the cure characteristic of delayed cure reactions, which indicates that
reaction due to their steric hindrance.23 The possible mecha- MWCNTs have the retarding effect on the cure reaction of
nism for MWCNTs to inhibit the cure reaction of epoxy resin epoxy. The reason for the degree of cure retardation effect
system may be the decrease of the propagation reaction speed. following the sequence S3 < S2 may be directly attributed to the
With regard to MWCNT/epoxy system, the contents of role of TETA functional groups on the chemically functional-
DGEBA and EMI-2,4 are lower than those in the correspond- ized MWCNT surface, namely, accelerating the curing reaction
ing unfilled epoxy and viscosity may increase,31 which causes of epoxy. For the development of the composites, the cure
decelerating the propagation reaction and extends the gel time inhibition effect caused by MWCNTs could have some negative
of epoxy. Furthermore, the reason for Ti and Tp following the effects on the composite manufacture since it needs longer post-
sequence S3 < S2 may be directly attributed to the role of cure time. In addtion, combining Figure 10 with Figure 8, we
TETA functional groups on the surface of chemically func- need to explain a phenomenon. In Figure 8(a), the dispersion of
tionalized MWCNTs. In terms of the curing mechanism for the MWCNTs is very poor. However, the curing property of S2 is not
DGEBA/EMI-2,4 system,30 TETA functional groups on the close to that of S1 (shown in Figure 10). It would be natural to
MWCNT surface can play the role of curing agents and think that steric hindrance by MWCNTs during curing in the
facilitate the primary amine-epoxide reaction. Therefore, it can matrix of S2 should be small due to poor dispersion. This seems
be inferred from the mechanism that these TETA functional to be the contradiction. The logical explanation of this phe-
groups can enervate the retardation effect on the cure reaction nomenon can be made from the following points: (i) There are
caused by MWCNTs, that is to say, the values of Ti and Tp many MWCNT agglomerates distributing over the epoxy matrix
show a descending tendency. In our systems, the changes of the of S2 , so the dispersion of MWCNTs is poor. These micro-size
heating rate have no effect on the total curing reaction heat MWCNT agglomerates may still generate the retarding effect on
estimated from the areas under the exothermic peaks of S1 , S2 the curing of S2 . (ii) There might be some interaction between
and S3 . Namely, the heat of cure is almost independent of the EMI-2,4 and MWCNT surface. EMI-2,4 concentration in and
heating rate. The phenomenon for
H following the sequence around MWCNT agglomerates might be higher, resulting in
S1 > S2 and S1 > S3 may be ascribed to the two aspects: the that curing property of S2 is not similar to that of S1 .
proportional reduction of epoxy in the composites and the The fractional extent of conversion
vs. temperature T at
retarding effect caused by MWCNTs on the cure reaction. different heating rates of S1 , S2 and S3 systems is separately
Moreover,
H of S3 system was a little higher than that of S2 reported in Figure 11(a)–(c). It can be seen that for each figure
system, which means that the extent of cure was larger in S3 . all the curves have the same functional form and only shift a

758 #2009 The Society of Polymer Science, Japan Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009
Preparation and Cure Behavior of Modified MWCNT/Epoxy Composites

Figure 11. Advancement of the cure reactions for non-isothermal cure at different heating rates of S1 , S2 and S3 systems. Dotted lines are given only for showing
the tendency. (a) S1 ; (b) S2 ; (c) S3 .

distance along the temperature axis. So all the
vs. T curves at and eq (2), d
=dT can be expressed as:
different heating rates should be superposable through shifting
each curve along the temperature axis relative to a curve at an d
f ð
Þ
¼ AeE
=RT ð3Þ
arbitrary reference heating rate by a shift factor, ’ðÞ ¼ dT 
Tref  T ð
¼ 0Þ, where  is the heating rate. In this work, the then
curve of 20 Kmin1 heating rate was considered to be the

d
E

master curve. The master curve represents the progression of ln  ln½A f ð
Þ ¼   ln  ð4Þ
dT RT
the cure reactions of the materials cured at a reference heating
rate. Figure 12(a)–(c) display the superposed curves of
vs. T In respect of some arbitrary fractional extent of conversion

for S1 , S2 and S3 systems. (0 
 1), the values of ln½Afð
Þ in the curves at the
Kinetic analysis of non-isothermal resin-cured system is different heating rates are equal, while the values of d
=dT ,
based upon the following rate equation:32 namely, the tangent slope of the point corresponding to the
fractional extent conversion
in the curves at the different
d
d

 ¼ kðTÞ f ð
Þ ð1Þ heating rates, are not always equal to those of d
=dTref in
dt dT
the reference curve. So there are three possibilities: d
=dT ¼
where  ¼ dT=dt is a constant heating rate, kðTÞ is a temper- d
=dTref , d
=dT < d
=dTref or d
=dT > d
=dTref . And then,
ature-dependent reaction rate constant, and f ð
Þ is the different we can separately analyze the three cases. If the case is
conversion function depending on the reaction mechanism. In d
=dT ¼ d
=dTref , from eq (4) we can obtain:
terms of Arrhenius law, kðTÞ is shown in eq (2): E
E

  ln  ¼   ln ref ð5Þ
kðTÞ ¼ AeE
=RT ð2Þ RT RTref
 
1 1 lnðref =Þ
where A is the pre-exponential factor, E
is the activation  ¼R ð6Þ
energy, and T is the absolute temperature. According to eq (1) T Tref E

Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009 #2009 The Society of Polymer Science, Japan 759
 K. YANG and M. GU

Figure 12. Superposed curves of  vs. T for S1 , S2 and S3 systems. Dotted lines are given only for showing the tendency. (a) S1 ; (b) S2 ; (c) S3 .

then Based upon eq (10), it can be inferred that Tref  T will

  augment with the increase of
. According to the foregoing
lnðref =Þ
Tref  T ¼ T Tref R ð7Þ discussion, the increase of Tref  T means that d
=dT >
E

d
=dTref will happen. This point contradicts with the studied
The curve of 20 K/min heating rate was thought of as the second case, namely, d
=dT < d
=dTref . So the possibility
reference curve, thus ref = > 1 and Tref  T > 0. From of d
=dT < d
=dTref is still untenable. Finally, we check
Figure 11(a)–(c), it can be observed that T Tref will increase the third case: d
=dT > d
=dTref . From eq (4) we can
with increasing
. According to eq (7), Tref  T will increase, obtain:
which indicates that d
=dT > d
=dTref will happen. However, E
E

this case is in contradiction with the previous assumption, that   ln  >   ln ref ð11Þ
RT RTref
is, d
=dT ¼ d
=dTref . Accordingly, the case of d
=dT ¼  
d
=dTref is not tenable. Furthermore, we analyze the second 1 1 lnðref =Þ
 <R ð12Þ
case, namely, d
=dT < d
=dTref . Based on it, eq (4) can be T Tref E

changed as follows: then
E
E
 
  ln  <   ln ref ð8Þ lnðref =Þ
RT RTref Tref  T < T Tref R ð13Þ
E

 
1 1 lnðref =Þ Under the condition of d
=dT > d
=dTref , Tref  T will
 >R ð9Þ
T Tref E
increase. As the fractional extent of conversion
increases,
then T Tref R lnðref =Þ=E
will continuously increase. It can be
 
lnðref =Þ seen from eq (13) that Tref  T is not larger than
Tref  T > T Tref R ð10Þ T Tref R lnðref =Þ=E
. Therefore, eq (13) is logically reason-
E

able.

760 #2009 The Society of Polymer Science, Japan Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009
Preparation and Cure Behavior of Modified MWCNT/Epoxy Composites

Based on the above discussion, it can be obtained that the Values of E
of S1 , S2 and S3 systems obtained for each value
reference offset Tref  T is not a constant. If the reaction is of
are shown in Figure 14. The dependence of the activation
only kinetically controlled, the curves of
vs. T should branch energy of S1 , S2 and S3 systems as a function of degree of
away from the master curve and the reference offset Tref  T curing was observed. Trying various multiple regression
will increase with the increase of
. Additionally, in respect of equations to fit the values of E
, we found that cubic
some arbitrary
(0 <
< 1), the reference offset Tref  T is polynomial can have enough goodness of fit, that is, if high-
not equal in S1 , S2 and S3 systems. The reason for Tref  T order moments are used to fit the values of E
, the coefficient
following the sequence S1 > S2 > S3 may be ascribed to the of the items whose numbers of order are greater than or equal
degree of vitrification following the sequence S3 > S2 > S1 at to 4 in the polynomial expressions will be smaller than 105 .
any of the heating rates considered. The filling of MWCNTs Consequently, cubic polynomial is good enough to fit the
resulted in the increase of the degree of vitrification in the values of E
. The fitted multiple regression equations for
epoxy, especially at lower heating rate. Moreover, the cure values of E
of S1 , S2 and S3 systems can be separately
reaction is more easily controlled by diffusion at larger
, then expressed in eqs (16), (18) and (20).
the values of d
=dT decrease. Furthermore, TETA functional S1 system:
groups on the surface of chemically functionalized MWCNTs Eð
Þ ¼ 69:80026  35:79507

caused the increase of the degree of vitrification in the epoxy to
þ 74:85288
2  72:30322
3
2 ð0; 1Þ ð16Þ
some extent. This kind of influence of the TETA functional
groups on the vitrification phenomena is complex. Conse- dE=d
¼ 35:79507 þ 149:70576
 216:90966
2 < 0 ð17Þ
quently, a qualitative analysis of the degree of vitrification in S2 system:
S1 , S2 and S3 systems can be made through comparing the Eð
Þ ¼ 70:33147  10:78226

values of Tref  T from the curves in Figure 12(a)–(c). The
þ 6:99794
2  15:68283
3
2 ð0; 1Þ ð18Þ
filling of MWCNTs results in the increase of the degree of
2 3
vitrification in the epoxy and the corresponding decrease of dE=d
¼ 10:78226 þ 13:99588
 47:04849
< 0 ð19Þ
d
=dT , which denotes that the cure reaction is inclined to be S3 system:
diffusion controlled at lower heating rate and larger degree of Eð
Þ ¼ 70:29673  13:10851

curing.
þ 4:76735
2  18:77349
3
2 ð0; 1Þ ð20Þ
It is known that the activation energy represents the 2
potential barrier and the mechanism of the cure reaction. dE=d
¼ 13:10851 þ 9:5347
þ 56:32047
< 0 ð21Þ
Accordingly, the research on the activation energy is conducive It can be seen from Figure 14 that the values of E
of S1 , S2
to better understanding of the cure behaviors of MWCNT/ and S3 systems decrease with increasing
, which can be
epoxy systems. The isoconversional method assumes that the further proved by the fact that with regard to S1 , S2 and S3
activation energy is the function of the degree of curing. systems, the differential equation of Eð
Þ, that is, dE=d
, is
Two different isoconversional methods can be applied: the smaller than zero, which indicates that Eð
Þ is a decreasing
Friedman method (differential isoconversional method) and function. The corresponding expressions of dE=d
are pre-
the Flynn-Wall-Ozawa method (integral isoconversional meth- sented in eqs (17), (19) and (21). In terms of the above
od). The activation energy obtained from the Flynn-Wall- discussion, the filling of unmodified and chemically modified
Ozawa method has been reported to be more reliable than that MWCNTs does not change the cure reaction mechanism of
from the Friedman method. Moreover, the Flynn-Wall-Ozawa epoxy; namely, the process of cure is still autocatalytic. The
method reveals less sensitivity to noise than the Friedman first step in the cure reaction of DGEBA/EMI-2,4 system is the
method.33 In this work, we apply the Flynn-Wall-Ozawa formation of both 1:1 and 2:1 adducts.30 The 2:1 adduct, which
method to calculate the activation energy of the cure reactions contains the nucleophilic alkoxide ion, was regarded as the
of S1 , S2 and S3 systems. This method is based upon catalyst that activated the etherification reaction. Moreover, the
eqs (14)–(15). number of the 2:1 adduct does not reduce because the alkoxide


 ion is continuously regenerated during the etherification
AE
E

ln  ¼ ln  ln gð
Þ  5:331  1:052 ð14Þ reaction. Based on the fact that the heat of cure is mainly
R RT
Z
generated from the etherification reaction, it is reasonable to
d
expect a decrease in E
of MWCNT/DGEBA/EMI-2,4
gð
Þ ¼ ð15Þ
0 f ð
Þ system. In addition, the phenomena for E
following the
where g(
) is the integral conversion function. sequence S2 > S1 and S3 > S1 may be attributed to the
As a result, the plot of ln  vs. 1=T at the same
from a retardation effect caused by MWCNTs on the cure reactions of
series of dynamic DSC thermograms at different heating rates epoxy. This kind of the inhibition effect enhances potential
would render a straight line where the slope allows the barrier of the cure reaction, so the activation energy is
determination of the activation energy. Figure 13(a)–(c) exhibit increased. Furthermore, the reason for E
following the
Flynn-Wall-Ozawa plots of S1 , S2 and S3 systems for various sequence S3 < S2 may be ascribed to the role of TETA
values of
(
¼ 0:05; 0:10; 0:15; 0:20; . . . ; 0:90; 0:95). A good functional groups on the chemically modified MWCNT sur-
linear relationship can be seen from Flynn-Wall-Ozawa plots. face. These TETA functional groups promote the primary

Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009 #2009 The Society of Polymer Science, Japan 761
 K. YANG and M. GU

Figure 13. Flynn-Wall-Ozawa plots at various degrees of curing of S1 , S2 and S3 systems. (a) S1 ; (b) S2 ; (c) S3 .

CONCLUSIONS

In this work, TETA grafting of MWCNTs was carried out.

XPS and Raman analyses show the effectiveness of chemical
functionalization of MWCNTs. HRTEM analysis indicates that
TETA is grafted onto the MWCNT surface to form a thin layer.
FESEM images denote that TETA grafting varies the forces
among MWCNTs from polar to nonpolar, thus big agglomer-
ates of acid-treated MWCNTs are broken and the loose state is
recovered. The interspaces formed in TETA-grafted MWCNTs
are larger, which is beneficial to their better dispersion in epoxy
matrix. Cure behavior of MWCNT/epoxy system and the
effect of chemical functionalization of MWCNTs on the cure
reaction were studied. In the dynamic DSC experiments, the
heat of cure is nearly independent of the heating rate. The
Figure 14. Values of the activation energy of S1 , S2 and S3 systems filling of unmodified and chemically modified MWCNTs does
obtained from Flynn-Wall-Ozawa plots at different .
not change the cure reaction mechanism of epoxy, while
MWCNTs delay the cure reaction because of their steric
hindrance. TETA functional groups on the chemically func-
amine-epoxide reaction and efficiently weaken the retarding tionalized MWCNT surface can act as curing agents and
effect caused by MWCNTs on the cure reaction, which facilitate the primary amine-epoxide reaction, which means
contributes to the lowering of potential barrier of cure and that they weaken the retardation effect caused by MWCNTs on
the corresponding decrease of the activation energy. the cure reaction of epoxy. The phenomena for Ti , Tp and E

762 #2009 The Society of Polymer Science, Japan Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009
Preparation and Cure Behavior of Modified MWCNT/Epoxy Composites

following the sequence S2 > S3 > S1 may be attributed to the 13. W. D. Zhang, L. Shen, I. Y. Phan, and T. Liu, Macromolecules, 37,
256 (2004).
combined effects of MWCNTs and TETA functional groups on
14. T. Liu, I. Y. Phang, L. Shen, S. Y. Chow, and W.-D. Zhang,
the cure reactions. Additionally, the filling of MWCNTs causes Macromolecules, 37, 7214 (2004).
the continuous increase of the degree of vitrification in the 15. D. Gan and L. A. Lyon, Macromolecules, 35, 9634 (2002).
epoxy system and the corresponding decrease of d
=dT , 16. A. Moisala, Q. Li, I. A. Kinloch, and A. H. Windle, Compos. Sci.
which denotes that the cure reaction is prone to be diffusion Technol., 66, 1285 (2006).
controlled at lower heating rate and larger degree of curing. 17. A. Bagchi and S. Nomura, Compos. Sci. Technol., 66, 1703 (2006).
18. Y. Ando, X. Zhao, H. Shimoyama, G. Sakai, and K. Kaneto, Int. J.
Inorg. Mater., 1, 77 (1999).
Acknowledgment. The authors are grateful to Instrumental 19. F. H. Gojny, M. H. G. Wichmann, B. Fiedler, I. A. Kinloch, W.
Analysis Centre of Shanghai Jiao Tong University for provid- Bauhofer, and A. H. Windle, Polymer, 47, 2036 (2006).
ing the support for this work. 20. D. Rosu, A. Mititelu, and C. N. Cascaval, Polym. Test., 23, 209
(2004).
Received: March 18, 2009 21. D. Rosu, C. N. Cascaval, F. Mustata, and C. Ciobanu, Thermochim.
Accepted: May 28, 2009 Acta, 383, 119 (2002).
22. D. Puglia, L. Valentini, I. Armentano, and J. M. Kenny, Diamond
Published: July 15, 2009
Relat. Mater., 12, 827 (2003).
23. J. Bae, J. Jang, and S.-H. Yoon, Macromol. Chem. Phys., 203, 2196
(2002).
REFERENCES
24. D. Puglia, L. Valentini, and J. M. Kenny, J. Appl. Polym. Sci., 88, 452
1. S. Iijima, Nature, 354, 56 (1991). (2003).
2. A. K. T. Lau and D. Hui, Composites Part B, 33, 263 (2002). 25. T. Zhou, M. Gu, Y. Jin, and J. Wang, J. Polym. Sci., Part A: Polym.
3. Z. Ounaies, C. Park, K. E. Wise, E. J. Siochi, and J. S. Harrison, Chem., 44, 371 (2006).
Compos. Sci. Technol., 63, 1637 (2003). 26. G. Zhang, S. Sun, D. Yang, J.-P. Dodelet, and E. Sacher, Carbon, 46,
4. M. J. Biercuk, M. C. Llaguno, M. Radosvljevic, J. K. Hyun, and A. T. 196 (2008).
Johnson, Appl. Phys. Lett., 80, 2767 (2002). 27. K. Papagelis, M. Kalyva, D. Tasis, J. Parthenios, A. Siokou, and
5. H. Geng, R. Rosen, B. Zheng, H. Shimoda, L. Fleming, and J. Liu, C. Galiotis, Phys. Status Solidi B, 244, 4046 (2007).
Adv. Mater., 14, 1387 (2002). 28. H.-X. Wu, R. Tong, X.-Q. Qiu, H.-F. Yang, Y.-H. Lin, and R.-F. Cai,
6. Y. J. Kim, T. S. Shin, H. D. Choi, J. H. Kwon, Y.-C. Chung, and Carbon, 45, 152 (2007).
H. G. Yoon, Carbon, 43, 23 (2005). 29. H.-X. Wu, X.-Q. Qiu, W.-M. Cao, Y.-H. Lin, R.-F. Cai, and S.-X.
7. Y. S. Song and J. R. Youn, Carbon, 44, 710 (2006). Qian, Carbon, 45, 2866 (2007).
8. W. Tang, M. H. Santare, and S. G. Advani, Carbon, 41, 2779 (2003). 30. T. Zhou, M. Gu, Y. Jin, and J. Wang, Polym. J., 37, 833 (2005).
9. S. J. Park, M. S. Cho, S. T. Lim, H. J. Cho, and M. S. Jhon, 31. Z. Fang, J. Wang, A. Gu, and L. Tong, Front. Mater. Sci. Chin., 1,
Macromol. Rapid Commun., 24, 1070 (2003). 415 (2007).
10. S. Mondal, Man-Made Text. India, 46, 329 (2003). 32. N. Sbirrazzuoli and S. Vyazovkin, Thermochim. Acta, 388, 289
11. K. Liao and S. Li, Appl. Phys. Lett., 79, 4225 (2001). (2002).
12. C. A. Mitchell, J. L. Bahr, S. Arepalli, J. M. Tour, and R. 33. C. Jubsilp, S. Damrongsakkul, T. Takeichi, and S. Rimdusit,
Krishnamoorti, Macromolecules, 35, 8825 (2002). Thermochim. Acta, 447, 131 (2006).

Polymer Journal, Vol. 41, No. 9, pp. 752–763, 2009 #2009 The Society of Polymer Science, Japan 763