A thermally remendable epoxy resin

Article abstract of DOI:10.1039/b811938d

To provide epoxy resin with crack healing capability, an epoxy containing both furan and epoxide groups, N,N-diglycidyl-furfurylamine (DGFA), was synthesized through a two-step approach. When it reacted with N,N′-(4,4′-diphenylmethane) bismaleimide (DPMBM

Full text of DOI:10.1039/b811938d

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/materials | Journal of Materials Chemistry
A thermally remendable epoxy resin†
a
a
b
Qiao Tian, Yan Chao Yuan, Min Zhi Rong and Ming Qiu Zhangb
*
Received 14th July 2008, Accepted 2nd December 2008
First published as an Advance Article on the web 22nd January 2009
DOI: 10.1039/b811938d
To provide epoxy resin with crack healing capability, an epoxy containing both furan and epoxide
groups, N,N-diglycidyl-furfurylamine (DGFA), was synthesized through a two-step approach. When it
reacted with N,N0-(4,40-diphenylmethane) bismaleimide (DPMBMI) and methylhexahydrophthalic
anhydride (MHHPA), respectively, a crosslinked polymer with two types of intermonomer linkage was
yielded. That is, thermally reversible Diels–Alder (DA) bonds from the reaction between furan and
maleimide groups, and thermally stable bonds from the reaction between epoxide and anhydride
groups. In this way, cured DGFA possessed not only similar mechanical properties as commercial
epoxy, but also thermal remendability that enabled elimination of cracks. The latter function took
effect as a result of successive retro-DA and DA reactions, which led to crack healing in a controlled
manner through chain reconnection.
thermally reversible crosslinked polyamides from maleimide-
Introduction
containing polyamides and a tri-functional furan compound.11
On the other hand, Ober and co-workers showed a series of
reworkable epoxy compounds that incorporated thermally
cleavable groups, such as secondary or tertiary esters, into the
networks.12,13 These groups were chosen to decompose between
200 and 300 ꢀC so that the cured epoxy can be removed without
damaging the underlying structure. Also, Small et al. made
In recent years, there have been continuous efforts to prepare
self-repairing polymers and polymer composites.1–4 It is desired
that microcracks deep inside structures can be healed in an
autonomic way.
So far, the achievements in this aspect fall into two categories:
extrinsic and intrinsic self-healing. In the case of extrinsic self-
healing, the healing agent has to be encapsulated and embedded
into the materials in advance. As soon as cracks destroy the
fragile capsules, the healing agent would be released into the
crack planes due to the capillary effect, re-binding the cracks.5–7
For intrinsic self-mending, cracks can be healed by polymers
themselves as a result of physical and/or chemical interactions at
the broken faces. The work by Wudl and co-workers is a repre-
sentative in this aspect.8,9 They synthesized highly crosslinked
polymeric materials with multi-furan and multi-maleimide via
a
thermally-removable encapsulant from a bis(maleimide)
compound and tris(furan) or tetrakis(furan), which can be easily
ꢀ
removed by heating to temperatures higher than 90 C, prefer-
ably in a polar solvent.14
Epoxy has been extensively employed in many applications
because of its excellent properties.15,16 It is worth noting that curing
reactions of epoxy with hardeners are generally irreversible.
Consequently, conventional cured epoxy can hardly exhibit
remendable behavior due to lack of the ability of recombination of
broken molecules. Since epoxy materials mostly work under severe
circumstances where long-term service is required, it would be ideal
if they were integrated with multiple self-healing capability.
In this work, a novel epoxy resin, N,N-diglycidyl-furfuryl-
amine (DGFA, Scheme 1, IUPAC name: N-(2-furylmethyl)-
N,N-bis(oxirane-2-ylmethyl)amine) was synthesized to combine
ꢀ
Diels–Alder (DA) reaction. At temperatures above 120 C, the
‘‘intermonomer’’ linkages disconnect (corresponding to retro-
DA reaction) but then reconnect upon cooling (i.e. DA reaction).
This process is fully reversible and can be used to restore frac-
tured parts of the polymers. In principle, an infinite of crack
healing is available without the aid of additional catalysts,
monomers and special surface treatment.
Following the approach of Wudl and co-workers, Liu and
Hseih prepared a thermally remendable and removable cross-
linked polymer through DA reactions between multifunctional
furan and maleimide monomers, by using epoxy compounds as
precursors.10 In a subsequent paper, Liu and Chen reported
aKey Laboratory for Polymeric Composite and Functional Materials of
Ministry of Education, OFCM Institute, School of Chemistry and
Chemical Engineering, Zhongshan University, Guangzhou 510275, P. R.
China
bMaterials Science Institute, Zhongshan University, Guangzhou 510275,
P. R. China. E-mail: cesrmz@mail.sysu.edu.cn; Fax: +86-020-84114008;
Tel: +86-20-84114008
† Electronic supplementary information (ESI) available: Synthesis and
characterization of DGFA, isothermal curing kinetics, and time
dependences of DA conversion. See DOI: 10.1039/b811938d
Scheme 1 DGFA synthesized by a two-step route.
J. Mater. Chem., 2009, 19, 1289–1296 | 1289
This journal is ª The Royal Society of Chemistry 2009

View Article Online
epoxide with furan groups in one molecule. The epoxide groups
can react with a traditional curing agent like anhydride to form
an epoxy network, providing the material with outstanding
mechanical properties and thermal resistance as usual. Mean-
while, the furan groups can react with maleimide to introduce
thermally reversible DA bonds into the epoxy network. Even-
tually, the molecular networks in the cured material are
comprised of two types of intermonomer linkage. In this context,
the advantages of the epoxy and an intrinsic self-healing ability
join together.
interpretation of the characteristic spectra can be found in the
ESI.†
FTIR max(KBr)/cmꢁ1: 3144; 3117; 3051; 1503; 1346; 1250;
1148; 1013; 918; 851; 752. dH (300 MHz; CDCl3, ppm): 7.3 (1 H);
6.3 (1 H); 6.2 (1 H); 3.8 (2 H); 3.1–2.4 (10 H). dC (300 MHz;
CDCl3, ppm): 151; 141; 110; 109; 66.6; 66.1; 61.1; 60.9; 60.6; 44.8;
44.5. Elemental analysis: (Found: C, 62.49; H, 7.15; N, 6.47.
C11H15NO3 requires C, 63.14; H, 7.23; N, 6.69%). Viscosity of
ꢀ
DGFA at 25 C is 0.02 Pa$s.
This paper discussed the synthesis and characterization of
DGFA and the related substances. Thermal reversibility and
remendability of the resultant were also examined for having
a deeper understanding of the structure-property relationship.
In addition, the newly developed epoxy is intended to possess
thermal remendability below its glass transition temperature.
To attain this objective, for such a DGFA molecule, the curing
feature of its epoxy groups with anhydride, and the recycla-
bility of retro-DA and DA reactions between furan and
maleimide groups, were carefully studied. The kinetic inhibition
effect of the epoxy network on the reversible DA reaction was
revealed from retro-DA and DA reactions of the cured DGFA
resin.
DA adduct of DGFA and DPMBMI
To achieve the retro-DA reaction feature of the adduct of DGFA
and DPMBMI, a solution method was chosen to obtain the
adduct as entirely as possible. Firstly, 5.71 g (0.016 mol)
DPMBMI was dissolved in 50 mL anhydrous tetrahydrofuran
(THF). The solution was charged into a 100 mL three-necked
round-bottom flask equipped with magnetic stirring. Then,
6.68 g (0.032 mol) DGFA was slowly added into the solution.
Afterwards, the solution was refluxed at about 66 ꢀC for 24 h
under N2 atmosphere, and cooled down to room temperature.
The reaction solution was poured into a large excess of diethyl
ether. The precipitate was filtered, washed with methanol, and
then dried under vacuum (<1.3 ꢂ 103 Pa) at room temperature.
Experimental
Preparation of DGFA/MHHPA/DPMBMI crosslinked polymer
Materials and reagents
First, 17.13 g (0.048 mol) DPMBMI was dissolved in 20 g
(0.096 mol) DGFA liquid under stirring at 90 ꢀC for 10 min
without the help of solvent. Then, 25.86 g (0.1536 mol)
MHHPA was mixed with the above mixture at 80 ꢀC for an
additional 10 min. The resultant homogeneous liquid was
degassed, poured into a closed silicone rubber mold, and cured
at 70 ꢀC for 24 h. The curing temperature was selected
according to the optimal temperature for DA bond formation
(see the sub-section in the Results and discussion section: DA
and retro-DA reactions between DGFA and DPMBMI), so
that most furan and maleimide groups could take part in the
reaction prior to the solidification of the system. Furthmore,
a non-stoichiometric ratio of epoxy ring/anhydride of 1 : 0.8 was
used to slow down curing of the resin.
Reagents and solvents used in the syntheses, including furfural
amine, epichlorohydrine, ethyl acetate, hexane and sodium
hydroxide, were obtained from Alfa Aesar GmbH, Germany.
The curing agents, methylhexahydrophthalic anhydride
(MHHPA) and N,N0-(4,40-diphenylmethane) bismaleimide
(DPMBMI), were supplied by Aldrich Chemical Co., USA. All
the chemicals were used as received.
Measurements
Fourier transform infrared (FTIR) spectra were recorded with
a Bruker EQUINOX55 Fourier transformation infrared spec-
trometer coupled with an infrared microscope spectrometer.
1H-NMR and 13C-NMR spectra were measured on a VARIAN
Mercury-Plus 300 (300 Hz). Elemental analysis was performed
with a Vario EL elemental analyzer. Dynamic viscosity of the
Remendability assessment
ꢀ
resin at 25 C was measured by an advanced rheometric expan-
The crosslinked polymer plates (5 mm thick) were impacted with
an iron ball to generate visible cracks. The samples were then
thermally treatꢀed at different temperatures for 20 min and
annealed at 80 C for different times. Variations in the damage
patterns during annealing were monitored with a camera.
sion system (ARES Rheometer, TA). Differential scanning
calorimetry (DSC) was performed on a TA Instruments DSC
Q10 using nitrogen purge and an empty aluminium pan as
a reference.
Dynamic mechanical analysis (DMA) was conꢀducted on a TA
Instruments DMA 2980 at a heating rate of 5 C/min. Tensile
and flexural properties were tested at room temperature by an
Instron 4505 universal testing apparatus under a constant
crosshead rate of 2 mm/min. Each batch included ten specimens
to yield an averaged value.
Results and discussion
Synthesis and characterization of DGFA
The novel epoxy resin, DGFA, was synthesized according to
a two-step reaction mechanism (Scheme 1). In the first step,
a ring-opening addition reaction between the oxirane ring of
epichlorohydrine and the amine group of 2-furfurylamine
occurred to produce a chlorohydrine end-group compound. In
the second step, DGFA formed through the ring-closing reaction
Synthesis of DGFA
Synthesis of DGFA followed the routine processes of producing
a glycidyl amine-type epoxy resin. The experimental details and
1290 | J. Mater. Chem., 2009, 19, 1289–1296
This journal is ª The Royal Society of Chemistry 2009

View Article Online
of chlorohydrine groups owing to the dehydrochlorination
action of NaOH solution. After purification by chromatography
on a silica gel column, the product of DGFA was characterized
1
with FTIR (ESI, Fig. S1†), H-NMR (ESI, Fig. S2), 13C-NMR
(ESI, Fig. S3) and elemental analysis.
The data shown in the Experimental section provide sufficient
evidences for the chemical structure of DGFA. Further analysis
of the spectra is available in the ESI.
To highlight the effect of purification during the preparation
of DGFA, the FTIR spectrum of the raw DGFA product is
compared to that of its purified version (cf. ESI, Fig. S1A and
S1B). A broad absorption at 3471 cmꢁ1 appears on the former,
which is ascribed to the stretching mode of the chlorohydrine
groups (O–H). It means that the ring-closing reactions of the
chlorohydrine groups in the second step (Scheme 1) were not
completely carried out, which explains the necessity of purifica-
tion by chromatography to obtain high purity DGFA.
Since the structure of DGFA has been confirmed, the reac-
tivity of the attached epoxide groups has to be checked.
Accordingly, the isothermal curing kinetics of DGFA with
MHHPA at a stoichiometric ratio were studied by DSC using
a modified Avrami equation (see ESI, Table S1†).17 It is found
that DGFA can perform the curing reaction with MHHPA like
a conventional epoxy resin, implying that the synthesized
DGFA might be applicable for the field in which conventional
epoxy resin is employed. In addition, the activation energy,
61.5 kJ/mol, agrees well with that of a similar epoxy-anhydride
system (71.6 kJ/mol).18 On the other hand, DGFA can be cured
by MHHPA at a temperature as low as ꢃ50 ꢀC (see ESI,
Fig. S4), which contrasts sharply with the case of conventional
epoxy cured by anhydride even in the presence of an accelerant
(i.e. 100ꢃ150 ꢀC). The catalytic effect of the tertiary amine
group on the DGFA molecules (see Scheme 1) should take the
responsibility.
Scheme 2 DA and retro-DA reactions between DGFA and DPMBMI.
Fig. 1 FTIR spectra of (A) a mixture of DGFA and DPMBMI, and (B)
a mixture of DGFA and DPMBMI treated at 66 ꢀC for 24 h.
DA and retro-DA reactions between DGFA and DPMBMI
The objective of this work is to develop a thermally remendable
epoxy resin by introducing thermally reversible groups into the
cured networks. Therefore, the reversibility of the DA reaction
between DGFA (containing furan ring, electron-rich diene) and
DPMBMI (involving maleimide, electron-poor dienophile)
should be verified in addition to the above-proved reactivity of
the epoxide groups of DGFA.
According to the literature,19 DA and retro-DA reaction
temperatures of the current system were set at 70 and 110 ꢀC
(Scheme 2), respectively. First, DA reaction of the furan and
maleimide groups in an equivalent molar mixture of DGFA and
1
DPMBMI was monitored by FTIR and H-NMR spectroscopy
as well. The mixture solution in acetone (50% w/v) was coated on
the surface of a KBr tablet, and then the FTIR spectrum was
collected after the tablet had been dried at room temperature. As
Fig. 2 1H-NMR spectra of a mixture of DGFA and DPMBMI kept at
70 ꢀC for (A) 0 h, (B) 2 h, and (C) 6 h.
shown in Fig. 1A, there is only a negligible peak at 1771 cmꢁ1
,
which is specific to DA adducts of maleimides.20 This peak
appears obviously after keeping the mixture solution in THF at
about 66 ꢀC for 24 h (see details in the Experimental section)
(Fig. 1B). It manifests that the DA reaction can take place very
slowly at room temperature, while is accelerated at around 70 ꢀC.
The DA reaction between DGFA and DPMBMI was further
evaluated by in-situ 1H-NMR spectroscopy. An equivalent molar
mixture of DGFA and DPMBMI was dissolved in DMSO-d6
(76 mmol/L) and then sealed in a glass tube under N2
This journal is ª The Royal Society of Chemistry 2009
J. Mater. Chem., 2009, 19, 1289–1296 | 1291

View Article Online
1
atmosphere. Then, H-NMR spectra were collected at specific
at 90 ꢀC for 5 min, and then quickly cooling down to room
temperature. Upon being heated in a DSC cell, the mixture
time intervals to monitor the DA reaction proceeding in the
mixture solution at 70 ꢀC. The results show that at the beginning
of the reaction, no characteristic peaks between 5.1 and 5.3 ppm
corresponding to the DA adduct are observed (Fig. 2A). After
2 h, these peaks appear (Fig. 2B), and become more evident when
the reaction has proceeded for 6 h (Fig. 2C). The feasibility of the
DA reaction between DGFA and DPMBMI is thus confirmed.
On the other hand, the content of the DA adduct and
ꢀ
performs a DA reaction from 27 to 86 C with an exothermal
peak at 65 ꢀC. Afterward, an endothermic peak appears at
110 ꢀC representing the retro-DA reaction, followed by two
ꢀ
evident exothermal peaks at 160 and 214 C due to the epoxide
ring-opening reaction and polymerization of maleimide moieties,
respectively. This DSC analysis proves that the DA reaction can
take place at 70 ꢀC. Therefore, an isothermal DSC scanning was
0
ꢀ
unreacted DGFA can be characterized by the integrals of H2
0
subsequently carried out at 70 C to evaluate the extent of the
and H3 peaks, and those of H2 and H3 peaks, respectively
(Fig. 2). This is because H20 and H30 peaks at 6.6 and 6.5 ppm are
due to the H2 and H3 protons of the DA adduct,21,22 while the
peaks at 6.3 and 6.2 ppm are assigned to the H2 and H3 protons
of unreacted DGFA. As a result, time-dependent reaction
conversion of the DA adduct, x, is known:
DA reaction. After 200 min of reaction, the exothermal enthalpy
is found to be 216 J/g, corresponding to 83.8 kJ/mol (per DA
bond). The value agrees with that reported in the literature.19
In addition, the retro-DA reaction of the DA adduct between
DGFA and DPMBMI in solution (see details in the Experi-
mental section) was evaluated by 1H-NMR and 13C-NMR
spectroscopy. The DA adduct was dissolved in DMSO-d6 in
a glass tube and heated at 110 ꢀC for 20 min to disconnect the DA
adduct. After the retro-DA reaction, the product was
0
0
Integral area of ðH02 þ H03Þ
x ¼
(1)
Integral area of ðH02 þ H03 þ H2 þ H3Þ
Also, first- and second-order kinetics of the reaction can be
estimated from:
ln1/(1 ꢁ x) ¼ kt
1/(1 ꢁ x) ¼ kt
(2)
(3)
where k stands for the rate constant. Quantification of the DA
reaction between DGFA and DPMBMI based on in-situ
1H-NMR spectroscopy (see ESI, Fig. S5†) indicates that a linear
ꢀ
fit of the data at 70 C gives a rate constant with a regression
coefficient of 0.9882 for the first-order plot and 0.9960 for the
second-order one, suggesting that the second-order kinetics
reaction should be more reasonable. Similarly, the reaction at
60 ꢀC also exhibits a greater regression coefficient for the second-
order plot than that of the first-order one. The activation energy
derived from the second-order kinetics is 42.9 kJ/mol, which falls
in the similar ranges reported for the other furan-maleimide DA
reactions.10,23–26
Fig. 4 1H-NMR spectra of (A) the adduct of DGFA and DPMBMI,
and (B) the retro-DA product of the adduct of DGFA and DPMBMI by
heating the adduct at 110 ꢀC for 20 min.
In order to reveal the temperature range of the DA reaction
between DGFA and DPMBMI, a DSC heating experiment was
conducted (Fig. 3). The sample mixture of DGFA and DPMBMI
was obtained by dissolving stoichiometric DPMBMI into DGFA
Fig. 5 13C-NMR spectra of (A) the adduct of DGFA and DPMBMI,
and (B) the retro-DA product of the adduct of DGFA and DPMBMI by
heating the adduct at 110 ꢀC for 20 min.
Fig. 3 DSC heating trace of a mixture of DGFA and DPMBMI.
Heating rate: 2 ꢀC/min.
1292 | J. Mater. Chem., 2009, 19, 1289–1296
This journal is ª The Royal Society of Chemistry 2009

View Article Online
immediately cooled down to room temperature and analyzed.
From the 1H-NMR spectra in Fig. 4, it is seen that the peaks at
5.1–5.3 ppm ascribed to the DA adduct (Fig. 4A) entirely
disappear in the spectrum of the retro-DA product (Fig. 4B).
Also, the characteristic peaks at 82 and 93 ppm,8 which are
clearly found in the 13C-NMR spectrum of the DA product
(Fig. 5A), are no longer perceived in the spectrum of the retro-
DA product (Fig. 5B). These data demonstrate that the retro-DA
ꢀ
reaction can be completed at 110 C for 20 min.
Thermal reversibility of DGFA/MHHPA/DPMBMI crosslinked
polymer
The curing reaction of DGFA with MHHPA and DPMBMI was
carried out at 70 ꢀC for 24 h. In the FTIR spectrum of cured
DGFA (Fig. 6), the peaks attributed to the epoxide groups at
918 cmꢁ1 (oxirane ring breathing) and 851 cmꢁ1 (C–O–C)
disappear, indicating that all epoxide groups have reacted with
anhydride groups of MHHPA to form epoxy networks. An
evident peak of hydroxyl groups emerges at 3500 cmꢁ1 as a result
of this reaction. Besides, the DA reaction of furan and maleimide
groups in the cured sample is also proved by the existence of the
Fig. 7 DSC heating traces of DGFA/MHHPA/DPMBMI crosslinked
polymer (heating rate: 5 ꢀC/min). The sample was first heated for
recording curve (A), and then naturally cooled down to room tempera-
ture in the DSC cell, followed by a second heating to record curve (B).
reaction but the glass transition of the cured epoxy is detected by
the second DSC heating scan. In comparison to the retro-DA
reaction at 110 ꢀC in solution (Fig. 4 and 5), the higher retro-DA
reaction temperature in the solid (126 ꢀC) is indicative of the
restraining effect of the cured epoxy. It can also be concluded
that reconnection of the furan and maleic groups in the cured
DGFA has to proceed at a rate much slower than that in solu-
tion, due to the same restriction of the epoxy network. Never-
theless, the DSC results demonstrate that the retro-DA reaction
is an accessible reaction pathway under heating circumstances
that is preferred over the bond-breaking degradation reaction in
the cured epoxy network.
peak at 1771 cmꢁ1
.
As mentioned in the Introduction, the cured DGFA polymer
should contain two types of crosslinedk covalent bonds: (i)
thermally stable bonds from the reaction between the epoxide
and anhydride groups, and (ii) thermally reversible bonds from
the DA reaction of the furan and maleic groups. Although the
DA and retro-DA reactions between DGFA and DPMBMI have
been proved by the simulation tests (see Fig. 1–5), it is still
unknown whether these thermally reversible reactions would be
inhibited by the stable crosslinked epoxy networks. Therefore,
the thermal reversibility of the cured DGFA polymer should be
studied, as it is directly related to thermal remendability of the
material.
The reversibility of the crosslinked DGFA polymer was
further studied by applying cyclic retro-DA (heat treatment at
ꢀ
140 C for 20 min to attain a retro-DA sample) and DA (heat
treatment at 80 ꢀC for 72 h to attain a DA sample) reactions up to
Fig. 7 shows the heating DSC curves of the cured DGFA
polymer. The first heating curve has an endothermic peak at
126 ꢀC, whileꢀthe second one shows no peak except for a transi-
tion at 128.5 C. It is thus reasonable to deduce that the endo-
therm appearing on the first curve must result from the retro-DA
reaction. Because there is not enough time for the recovered
furan and maleimide moieties to be reconnected during the
subsequent cooling and reheating processes, no more retro-DA
Fig. 8 DSC heating traces of DGFA/MHHPA/DPMBMI crosslinked
polymer (heating rate: 5 ꢀC/min). DA0: as-manufactured sample; rDA1:
DA0 treated at 140 ꢀC for 20 min, and then quenched to room temper-
ature; DA1: rDA1 treated at 80 ꢀC for 72 h, and then cooled inside oven
by switching off the electricity supply; rDA2: DA1 treated at 140 ꢀC for
20 min, and then quenched to room temperature; DA2: rDA2 treated at
80 ꢀC for 72 h, and then cooled inside the oven by switching off the
electricity supply; rDA3: DA2 treated at 140 ꢀC for 20 min, and then
quenched to room temperature; DA3: rDA3 treated at 80 ꢀC for 72 h, and
then cooled inside the oven by switching off the electricity supply.
Fig. 6 FTIR spectra of DGFA/MHHPA/DPMBMI crosslinked polymer.
This journal is ª The Royal Society of Chemistry 2009
J. Mater. Chem., 2009, 19, 1289–1296 | 1293

View Article Online
four times. The higher heating temperatures for both retro-DA
and DA reactions were chosen to increase the reaction efficiency.
Again, the reactions were monitored by DSC (Fig. 8). The
original cured DGFA (DA0) gives an endothermic peak of the
retro-DA reaction at 126 ꢀC, while the rDA1 only shows a glass
retain the integity of the material when the retro-DA or DA
reaction occurs.
Fig. 8 also exhibits that there is an exothermic peak at around
225 ꢀC on the heating curves of DA0–DA3 samples. It should be
ascribed to the polymerization of the free maleimide moieties
resulting from the retro-DA reaction. This reaction is detri-
mental to the thermal reversibility of the cured DGFA polymer
because it converts maleimide to irreversible bonds. To avoid the
negative effect, therefore, the material should not be heated to an
ꢀ
transition at 128.5 C. This implies that the retro-DA reaction
has already been completed during the heat treatment at 140 ꢀC
for 20 min. Subsequently, the rDA1 is allowed to reconnect the
cleavage groups by the DA reaction, and then an endothermic
peak appears again at 131 ꢀC for DA1. It is evident that the DA
reaction between the disconnected furan and maleimide moieties
has definitely taken place in the thermal treatment process. The
recovery efficiency of the second DA reaction in comparison to
the original cured DGFA polymer is about 77% as given by the
ratio of DH1 (enthalpy of DA1) over DHo (enthalpy of DA0). It
is noted that DHo is 122.9 J/g, corresponding to 80.2 kJ/mol (per
DA bond), which is quite close to the aforesaid exothermal
enthalpy per mole of DA bond formation. This means that all of
the DA bonds can be cleaved during the first retro-DA reaction,
and the crosslinked epoxy networks do not block the retro-DA
reaction.
ꢀ
elevated temperature, e.g. 180 C.
DMA spectra of the cured DGFA polymer are shown in
Fig. 9. For the original sample, its storage modulus exhibits
a two-stage behavior with a rise in temperature. At around
110 ꢀC, a drastic decrease in the modulus starts, and then
a plateau appears at 118 ꢀC. When temperature is further raised,
the modulus drops again at 121 ꢀC. Meanwhile, the temperature
dependence of tan d shows a knee point at 119 ꢀC and a peak at
128 ꢀꢀC, respectively. In contrast, after retro-DA reaction at
140 C for 20 min, the sample only shows a single transition on
the curve of storage modulus versus temperature. The corre-
sponding tan d plot also has only one peak and the aforesaid knee
point is absent. Clearly, the two-stage variation of storage
modulus of the original cured DGFA polymer reflects the retro-
DA reaction and glass transition, respectively. The retro-DA
reaction happens below the glass transition temperature of the
material, but there is still certain overlap between the two
processes. Furthermore, the DMA analysis coincides with
previous DSC data, which also show a retro-DA reaction peak at
around 120 ꢀC and a glass transition at around 128 ꢀC.
In this manner, the recyclability of retro-DA and DA reactions
in the crosslinked DGFA polymer has been verified by the
repeated endotherms and Tg transitions for the DA and retro-DA
samples. The characteristic parameters of the reactions inspected
by DSC (Fig. 8) are summarized in Table 1. It is noted that the
endothermic peak temperature of the DA sample slightly
increases with heat treatment time, implying that the retro-DA
reaction becomes more difficult with the heating cycles. This
phenomenon may be related with the increase in crosslink density
of the epoxy resin after heat treatment, as evidenced by the
gradual increase in the Tg of rDA1 (128.5 ꢀC), rDA2 (132.5 ꢀC)
and rDA3 (136 ꢀC). Accordingly, the enthalpy of each endo-
thermic peak, DH, of the retro-DA reaction gradually decreases
as of the second heating cycle, indicating that the increased
crosslink density of the epoxy part somewhat hinders recovery of
the DA bonds. Nevertheless, the decrease of each retro-DA
endothermic enthalpy is marginal. The DA and retro-DA reac-
tions in the cured DGFA can still be considered as reversible.
In fact, the crosslink density of the epoxy network would not
ceaselessly change after a couple of heating-cooling cycles, so
that the reversibility of the DA and retro-DA reactions in the
cured DGFA can eventually be fixed. Moreover, the epoxy
network should favor the disconnected furan and maleimide
moieties remaining at their original places during cooling. As
a result, they can be reconnected very efficiently upon being
heated at 80 ꢀC. In other words, the epoxy network would help to
It is interesting to see from the DMA measurements (Fig. 9)
that after retro-DA reaction, the cured DGFA polymer possesses
a higher Tg and significantly lower room temperature storage
modulus. The former phenomenon should result from the
increased crosslinking density of the epoxy network, which has
been discussed when analyzing Fig. 8. The latter can be explained
by the disconnection of DA bonds, which may lead to an
increased amount of chain ends or low molecular weight
substances, and hence reduces the stiffness of the material.
When the temperature is higher than Tg, all the DA bonds are
broken. Consequently, the storage modulus is only related with
the crosslinking density of epoxy network. In other words, the
Table 1 Quantification of endothermic peaks of cured DGFA polymer
measured by DSC heating scan
Onset
temperature (ꢀC)
Peak
temperature (ꢀC)
Sample IDa
Enthalpy (J/g)
DA0
DA1
DA2
DA3
120.7
126.2
128.4
129.2
126.2
131.6
133.7
136.5
122.9
94.7
92.6
90.0
Fig. 9 DMA spectra of DGFA/MHHPA/DPMBMI crosslinked poly-
mer. (A) as-manufactured sample; (B) as-manufactured sample treated at
140 ꢀC for 20 min, and then quenched to room temperature.
a
Descriptions of the samples are listed in the caption of Fig. 8.
1294 | J. Mater. Chem., 2009, 19, 1289–1296
This journal is ª The Royal Society of Chemistry 2009

View Article Online
effect of the crosslinking density variation on the storage
modulus of the polymer can be perceived in the rubbery region. It
is seen that the sample that experienced the retro-DA reaction
exhibits a higher modulus in the rubbery region than the
as-manufactured one. It again reveals that the crosslinking
density of the epoxy network can be further enhanced by
a heating treatment above Tg.
to those of commercial epoxy resins. It means that DGFA
possesses the potential ability to replace traditional epoxy resin
in future applications.
Thermally remendable processes of artificial damage on the
cured DGFA are illustrated in Fig. 10. Since the interfaces of the
cracks have a higher refractive index, they are easily visually
monitored. All the samples were thermally treated at different
temperatures (100ꢃ125 ꢀC) for 20 min for disconnecting DA
covalent bonds via retro-DA reaction, before being annealed at
80 ꢀC for a while to repair the cracks by DA reaction between the
disconnected furan and maleimide moieties.
Mechanical properties and thermal remendability of the cured
DGFA polymer
As shown in Fig. 10a, no crack repair is observed when the
ꢀ
temperature of the retro-DA reaction (110 ꢀC). When heat
treatment is carried out at a temperature (119 ꢀC) higher than the
onset temperature of the retro-DA reaction, most of the cracks
The tensile and flexural properties of the DGFA/MHHPA/
DPMBMI crosslinked polymer were measured at room
temperature, in comparison with those of conventional bisphe-
nol-A epoxy cured by MHHPA (Table 2). The cured version of
the newly synthesized epoxy has a higher modulus but a moder-
ately lower strength as compared to the cured bisphenol-A
epoxy.27 On the whole, the data of the cured DGFA are still close
heat treatment temperature (100 C) is lower than the starting
ꢀ
can be healed after DA reaction treatment at 80 C (Fig. 10b).
Some larger cracks finally remain unhealed, probably due to the
relatively low retro-DA treatment temperature, and/or material
loss during damaging, and/or misalignment of the cracked
portions. In the case of the higher retro-DA reaction temperature
Table 2 Mechanical properties of the cured epoxy
ꢀ
(125 C), which is close to the Tg of the cured DGFA polymer
DGFA/MHHPA/
DPMBMI crosslinked
polymer
Bisphenol-A
epoxy cured
by MHHPA27
(128 ꢀC), larger cracks can also be repaired owing to the
enhanced segment mobility (Fig. 10c). These results clearly
demonstrate that the synthesized epoxy has been provided with
thermal remendability as expected.
Properties
Young’s modulus (GPa)
Tensile strength (MPa)
Elongation at break (%)
Flexural modulus (GPa)
Flexural strength (MPa)
2.5
53
1.8
65
1.5
4.6
110
—
2.6
128
Conclusions
A novel epoxy resin DGFA was synthesized by a two-step route,
which combined epoxide with furan groups into one molecule.
The resultant was a liquid with low viscosity and good process-
ability, and can be cured by using anhydride and maleimide
without the help of solvent. Cured DGFA contained two types of
intermonomer linkage: thermally reversible DA bonds from the
reaction between furan and maleimide groups, and thermally
stable bonds from the reaction between epoxide and anhydride
groups. It possessed mechanical properties similar to those of
commercial epoxy resin. The thermal reversibility of the DA
bonds in cured DGFA proved to be workable below the glass
transition temperature of the material. Taking advantage of the
above feature, cracks in cured DGFA can be mended in
a controlled manner through chain reconnection resulting from
successive retro-DA and DA reactions.
Acknowledgements
The authors thank the support of the Natural Science Founda-
tion of China (Grants: 50573093, U0634001 and 20874117).
References
1 R. P. Wool, Soft Matter, 2008, 4, 400.
2 D. S. Bergman and F. Wudl, J. Mater. Chem., 2008, 18, 41.
3 Y. C. Yuan, T. Yin, M. Z. Rong and M. Q. Zhang, Express Polym.
Lett., 2008, 2, 238.
4 C. Dry and N. R. Sottos, in Smart Structures and Structures 1993:
Smart Materials, ed. V. K. Varadian, Proc. SPIE, vol. 1916, 1993,
p. 438.
5 S. M. Bleay, C. B. Loader, V. J. Hawyes, L. Humberstone and
P. T. Curtis, Compos. Part A-Appl. Sci., 2001, 32, 1767.
Fig. 10 Visual inspection of thermal remendability of the cured DGFA
polymer. The damaged samples were firstly treated at (a) 100 ꢀC, (b) 119
ꢀC and (c) 125 ꢀC for 20 min, respectively. Next, they were moved to an
oven preset at 80 ꢀC for (1) 0 h, (2) 12 h and (3) 72 h, respectively.
This journal is ª The Royal Society of Chemistry 2009
J. Mater. Chem., 2009, 19, 1289–1296 | 1295

View Article Online
6 R. S. Trask, G. J. Williams and I. P. Bond, J. R. Soc. Interface, 2007,
4, 363.
7 N. Sottos, S. White and I. Bond, J. R. Soc. Interface, 2007, 4, 347.
8 X. X. Chen, M. A. Dam, K. Ono, A. Mal, H. Shen, S. R. Nutt,
K. Sheran and F. Wudl, Science, 2002, 295, 1698.
17 W. B. Xu, S. P. Bao, S. J. Shen, W. Wang, G. P. Hang and P. S. He,
J. Polym. Sci. Polym. Phys., 2003, 41, 378.
18 F. Y. C. Boey and W. Qiang, Polymer, 2000, 41, 2081.
19 X. X. Chen, Ph.D Thesis, University of California, Los Angeles,
2003.
9 X. X. Chen, F. Wudl, A. K. Mal, H. B. Shen and S. R. Nutt,
Macromolecules, 2003, 36, 1802.
20 Y. Imai, H. Itoh, K. Naka and Y. Chujo, Macromolecules, 2000, 33,
4343.
10 Y. L. Liu and C. Y. Hsieh, J. Polym. Sci. Polym. Chem., 2006, 44, 905.
11 Y. L. Liu and Y. W. Chen, Macromol. Chem. Phys., 2007, 208, 224.
€
12 S. Yang, J.-S. Chen, H. Korner, T. Breiner and C. K. Ober, Chem.
Mater., 1998, 10, 1475.
21 C. Gousse and A. Gandini, Polym. Int., 1999, 48, 723.
22 M. Watanabe and N. Yoshie, Polymer, 2006, 47, 4946.
23 M. J. S. Dewar and A. B. Pierini, J. Am. Chem. Soc., 1984, 106,
203.
13 J.-S. Chen, C. K. Ober and M. D. Poliks, Polymer, 2002, 43, 131.
14 J. H. Small, D. A. Loy, D. R. Wheeler, J. R. McElhanon and
R. S. Saunders, US Pat., 6271335, 2001.
15 H. Lee and K. Neville, Handbook of Epoxy Resins, McGraw-Hill,
New York, 1972.
24 N. S. Isaacs and N. Keating, J. Chem. Soc., Chem. Commun., 1992, 12,
876.
25 E. Goiti, F. Heatley, M. B. Huglin and J. M. Rego, Eur. Polym. J.,
2004, 40, 1451.
26 Y. L Liu, C. Y. Hsieh and Y. W. Chen, Polymer, 2006, 47, 2581.
27 Z. Q. Tao, S. Y. Yang, Z. Y. Ge, J. S. Chen and L. Fan, Eur. Polym.
J., 2007, 43, 550.
16 S. H. Goodman, Handbook of Thermoset Plastics, Noyes
Publications, Berkshire, 2nd edn, 1998.
1296 | J. Mater. Chem., 2009, 19, 1289–1296
This journal is ª The Royal Society of Chemistry 2009