Development of optimized autonomous self-healing systems for epoxy materials based on maleimide chemistry

Article abstract of DOI:10.1016/j.polymer.2012.03.061

Maleimide chemistry involving amines and thiols is presented and evaluated for the design of autonomous self-healing epoxy materials. Model reactions show that amines react rapidly with maleimide compounds at room temperature via the Michael addition reaction. Moreover, thiols and maleimides react readily in the presence of tertiary amines that are present in the epoxy material. The maleimide conjugation reaction with residual amines in the epoxy material ensures chemical bonding of the newly formed network with the original materials during crack healing, while in the crack plane, multifunctional thiols react with difunctional maleimides to fill the crack area. Healing efficiencies are evaluated using the tapered double cantilever beam (TDCB) test method with manual injection of the healing agents, revealing a maximum healing efficiency up to 121% for EPON 828 epoxy material. Furthermore, the use of maleimide chemistry has also been evaluated for self-healing applications towards a cold-curing resin that is currently used for infusion of wind turbine blades (RIM resin). While the healing efficiency is strongly dependent on the type of epoxy material, the average maximum peak load for fracture after healing is roughly the same for all tested epoxy materials.

Full text of DOI:10.1016/j.polymer.2012.03.061

Polymer 53 (2012) 2320e2326
Contents lists available at SciVerse ScienceDirect
Polymer
journal homepage: www.elsevier.com/locate/polymer
Development of optimized autonomous self-healing systems for epoxy materials
based on maleimide chemistry
,
Stijn Billiet a, Wim Van Camp a, Xander K.D. Hillewaere a, Hubert Rahier b, Filip E. Du Prez a
*
a Department of Organic Chemistry, Polymer Chemistry Research Group, Ghent University, Krijgslaan 281, S4-bis, B-9000 Ghent, Belgium
b Vrije Universiteit Brussel, Faculty of Engineering, Department of Materials and Chemistry (MACH), Physical Chemistry and Polymer Science (FYSC), Pleinlaan 2,
B-1050 Brussel, Belgium
a r t i c l e i n f o
a b s t r a c t
Article history:
Maleimide chemistry involving amines and thiols is presented and evaluated for the design of autono-
mous self-healing epoxy materials. Model reactions show that amines react rapidly with maleimide
compounds at room temperature via the Michael addition reaction. Moreover, thiols and maleimides
react readily in the presence of tertiary amines that are present in the epoxy material. The maleimide
conjugation reaction with residual amines in the epoxy material ensures chemical bonding of the newly
formed network with the original materials during crack healing, while in the crack plane, multifunc-
tional thiols react with difunctional maleimides to fill the crack area. Healing efficiencies are evaluated
using the tapered double cantilever beam (TDCB) test method with manual injection of the healing
agents, revealing a maximum healing efficiency up to 121% for EPON 828 epoxy material. Furthermore,
the use of maleimide chemistry has also been evaluated for self-healing applications towards a cold-
curing resin that is currently used for infusion of wind turbine blades (RIM resin). While the healing
efficiency is strongly dependent on the type of epoxy material, the average maximum peak load for
fracture after healing is roughly the same for all tested epoxy materials.
Received 7 February 2012
Received in revised form
23 March 2012
Accepted 29 March 2012
Available online 3 April 2012
Keywords:
Self-healing
Autonomous
Maleimide
Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction
properties of the newly formed polymer in the crack plane [17,18],
and (3) chemical compatibility and stability of the catalyst within
Throughout their lifetime, polymeric materials and coatings are
susceptible to mechanical damage such as wear, degradation, and
microcracking, all of them reducing the mechanical properties of
these materials. To overcome these limitations, a rapidly emerging
field of research has resulted in materials that can repair crack
damage in an autonomous fashion (without any stimulus or human
intervention),referred to as self-healing materials [1e12].
The microencapsulation approach, pioneered by White et al. for
the healing of epoxy materials [13], is by far the most studied self-
healing (SH) concept in recent years. Microcapsules that are
incorporated in the polymer composite material contain the self-
healing agents (monomers or network precursors) that can (co)
polymerize or crosslink upon rupture of the capsules as a result of
mechanical damage [6,7,13,14]. The healing efficiency in this
particular microcapsule-based self-healing system is dependent on
several factors including (1) monomer stability in the absence of
a catalyst and the polymerization kinetics [15,16], (2) mechanical
the matrix [19,20].
Despite increased research activities, the current research
related to self-healing epoxy materials has been mainly limited to
a few types of chemistries that will be briefly reviewed in the next
paragraph: (1) ring opening methathesis polymerization (ROMP)
based systems with dicyclopentadiene and Grubbs’ catalyst, (2)
poly(dimethylsiloxane) (PDMS) based systems, (3) solvent-based
healing systems, and (4) epoxy-based healing systems.
The healing chemistry that was originally developed by White
et al. was based on encapsulated dicyclopentadiene (DCPD) that
starts to polymerize by ring opening methathesis polymerization
(ROMP) when the DCPD comes into contact with the ruthenium/
Grubbs’ catalyst that is dispersed in the epoxy matrix [13].
Although DCPD is a cheap and readily available healing agent, the
Grubbs’ catalyst is not, and is toxic. Moreover, the catalyst shows
a limited stability and reactivity due to prolonged exposure to
oxygen [21], moisture and the amine curing agent, while the
monomer is prone to autopolymerization [20]. To overcome some
of these issues, Cho et al. reported on poly(dimethylsiloxane)
(PDMS) based self-healing materials [22]. The polycondensation
reaction of hydroxyl-terminated siloxanes and alkoxysilanes is
* Corresponding author. Tel.: þ32 9 264 45 03/44 89 (secr.); fax: þ32 9 264 49 72.
E-mail address: filip.duprez@ugent.be (F.E. Du Prez).
0032-3861/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.polymer.2012.03.061

S. Billiet et al. / Polymer 53 (2012) 2320e2326
2321
catalyzed by organotin compounds that are less expensive, less
toxic, and proceed in the presence of moisture, which makes this
approach more suitable for practical applications. However, as the
polymerized healing agent is not structurally similar to the polymer
matrix, improvement of the interfacial compatibility between the
healing agent and the substrate (crack surface) was desired. As
a result, a diglycidyl ether bisphenol A based epoxy resin was
encapsulated, along with an imidazoleemetal complex as a latent
hardener [23]. As the main drawback, the system requires external
heat to promote the imidazole complex to dissociate into its reac-
tive species, and can thus not be classified as an autonomous one.
In order to reach a more autonomic epoxy-based self-healing
harsh conditions. As far as we know, a healing chemistry using
maleimide conjugation reactions has not been reported yet.
In this contribution, we have evaluated the use of maleimide
chemistry for two different types of epoxy materials. First, the
proposed strategy was applied to EPON 828, which is considered as
a reference material for cold-curing epoxy materials in self-healing
applications and it still remains the most investigated matrix in the
field. Then we have extended the same methodology to an epoxy
material, RIM 135, that is currently widely used for high-end
applications. For example, Owens Corning and PPG use it as
a benchmark resin for glass fibre properties testing [39,40]. Due to
its extremely low viscosity, it is currently widely used in the
European market as a cold-curing resin for infusion of large wind
turbine blades. Those blades are typically made of low temperature
curing thermoset resins with glass fibre reinforcement. In the past,
polyester and vinylester were the standard resins, but epoxy resins
are increasingly used because of their superior mechanical prop-
erties and of the environmental problems like stryrene emission
styrene emission problems with other materials. This application is
particularly interesting for self-healing composites, because wind
turbine blades have to withstand biaxial variable amplitude fatigue
loadings for 20 or more years. In an early stage, the fatigue damage
typically leads to fibre/matrix debonding and small matrix cracks in
the composite [41e44].
system,
a polythiol curing agent (pentaerythritol tetrakis(3-
mercaptopropionate)) was separately encapsulated, allowing
curing at room temperature [24]. Microcapsules containing amine
hardeners were also investigated, but the encapsulation of amines
although possible [25] remains challenging [26].
Microcapsules containing solvents have also been explored for
use in self-healing polymer composites [27]. The release of solvents
upon damage locally swells the matrix and promotes the mobility
of functional groups to entangle and react across the crack plane.
The proposed strategy was further improved by mixing a part of
epoxy resin with the solvent, which combines the solvent healing
effect and additional crosslinking reactions [28].
‘Click’ chemistry, as introduced by Sharpless [29], was also
proposed as healing mechanism. Binder et al. [30e32] made use of
azide-alkyne chemistry to heal a poly(isobutylene) matrix. In this
process, a liquid polymer was successfully used as a reactive
compound.
Another important topic in the self-healing field is to expand the
variety of polymer materials and composites that are capable of
healing. Although most of the self-healing approaches have been
focussing on the epoxy systems, a few recent reports show the
interest in the exploration of other systems and chemistries
[45e49]. The improvement of self-healing strategies and the
extension to other materials requires to broaden our approach and
to move away from the currently applied types of chemistry. From
this point of view, our current research aims to broaden the specific
chemistries that are currently applied in the microencapsulation
approach for epoxy materials. The encapsulation process itself is
currently being optimized in our research group.
In a recent contribution, Zhang et al. have explored the use of
glycidyl methacrylate (GMA)eloaded microcapsules for making
a self-healing epoxy material [33]. As GMA is a low viscosity liquid
chemical, it combines two healing actions: solvent effect on one
hand, and chemical reactions with unreacted functionalities in the
matrix via the reactive epoxide groups and double bonds on the
other hand. This promising concept was shown to be effective for
cold-curing epoxy materials (EPON 828). At the same time, it
broadened the spectrum of possible healing agents as the nucleo-
philic addition reaction of double bonds as a healing chemistry was
reported for the first time. However, as a result of the reactive
nature of the applied methacrylate group, the shelf-life of the
healing agent is limited at relatively high temperature or the
monomer may polymerize during the high curing temperatures
when the synthesis of high-end composite materials, which require
a curing temperature between 120 and 180 ꢀC, is targeted. For the
same reason, one may look for compounds that have an even higher
boiling point than GMA (189 ꢀC) to prevent high pressures in the
microcapsules during these curing steps and thus prevent micro-
capsule breakage or leakage.
In view of increasing the industrial feasibility for the widespread
use of microcapsule-based self-healing thermosets and composites,
we have explored and evaluated the use of maleimide chemistry for
the application in autonomous self-healing epoxy materials. Mal-
eimides were already extensively used in self-healing as a member
of the DielseAlder reaction with furan compounds [34e37]. On the
other hand, maleimides are known as well to react at room
temperature with amines via the Michael addition reaction.
Moreover, maleimides also react with thiols, while the presence of
a tertiary amine catalyst tremendously speeds up the reaction [38].
As tertiary amines as well as unreacted amines are present in epoxy
materials, maleimide chemistry has been considered as a prom-
ising chemical strategy for self-healing materials. Moreover, mal-
eimides show a low tendency to homopolymerize and have very
high boiling points, which is important in view of surviving the
high curing temperatures or the use of self-healing materials in
2. Results and discussion
2.1. Model study for the kinetics of maleimides and amines/thiols at
room temperature
In order to check the feasibility of the proposed maleimide
chemistry for self-healing applications, we have first performed
a detailed study of the reaction kinetics at room temperature of
monofunctional maleimides with amines and thiols. As the lifetime
of a material is dependent on the result of crack growth vs. crack
repair, reaction kinetics are of utmost importance. At the same
time, we have compared the maleimide e amine/thiol conjugation
reaction kinetics with those of the glycidyl methacrylate e amine
reactions that were proposed by Zhang et al. [33].
Glycidyl methacrylate reacts with amines via the reactive
epoxide group as well as via the methacrylate functionality (C]C
bond). In order to assess the contribution of each reactive group, we
have selected two model compounds, each containing the respec-
tive reactive group. Butyl glycidyl ether (BGE) was chosen as
a model compound for the epoxide group, while methyl meth-
acyrlate (MMA) was chosen as a model compound to assess the
reactivity of the C]C bond (see Scheme 1, reactions (1) and (2)).
Both compounds were reacted with a primary amine and the
conversion was monitored by 1H NMR. For the maleimides, the
aromatic phenylmaleimide (PM) was monitored as
a model
compound (Scheme 1, reaction (3)). In the case of MMA and PM, the
aliphatic, monofunctional propylamine (PA) was chosen as the
amine to mimic the reactivity of the primary amine groups of

2322
S. Billiet et al. / Polymer 53 (2012) 2320e2326
Scheme 1. Model reactions in m-cresol for a comparison of the reactivity of amines with (1) epoxide, (2) methacrylate and (3) maleimide functionality, respectively. (4) Model
reaction for a comparison of the reactivity of thiols with maleimide functionality.
diethylene amine (DETA), while in the case of BGE, benzyl amine
(BA) was chosen to prevent overlap of the 1H NMR signals. All
reactions were performed in m-cresol to ensure similar reaction
conditions as with the aromatic maleimides used later on.
As can be noticed from Fig. 1, maleimides are more reactive
towards amines than the epoxide or methacrylate groups. The
reaction of PM with PA (Scheme 1, reaction (3)) reaches full
conversion after 10 min, while the reaction of BGE and BA (Scheme
1, reaction (1)) reaches only 14% conversion after 120 min, and only
10% of the C]C bonds react with PA after 120 min (Scheme 1,
reaction (2)). The high reactivity of maleimides may show faster
healing rates and hence increase the rate of crack healing vs. crack
propagation.
In a similar way, we have also investigated the reactivity of
thiols with maleimides, using PM and dodecanethiol (DT) to mimic
the reactivity of the used multifunctional thiols (Scheme 1, reaction
(4)) [50]. As can be seen in Fig. 1, the reaction proceeds rather
slowly, reaching 15% of maleimide conversion in 120 min. However,
the use of a tertiary amine catalyst (NEt3) dramatically increases the
reaction rate [51] and 100% conversion is reached in less than
10 min. An identical reaction with grinded epoxy material added as
the catalyst (average size of the particles ꢁ 100
mm) reached 90% of
conversion of the maleimide functionalities in 120 min, clearly
demonstrating the ability of the tertiary amines in the epoxy
material to catalyze the thiolemaleimide reaction.
2.2. Healing strategy
Since the above described model reactions have demonstrated
that the maleimideeamine reaction proceeds faster than the
epoxide-amine or methacrylateeamine reaction, the former reac-
tion shows a promising chemical strategy for application in self-
healing materials. Consequently, we propose a healing strategy
based on maleimide and thiol compounds on one hand, and amine
species that are present in the epoxy material on the other hand.
Actually, one can make a distinction between the bulk area of
the crack and the crack interface (Fig. 2) in a damaged epoxy
material. At the crack interface, residual (unreacted) amines, that
arise from the incomplete curing reaction of the matrix, can react
with maleimide compounds. The covalent bond will ensure
Fig. 2. Schematic depiction of a crack in the epoxy material. At the crack interface,
a reaction can occur between unreacted amine functionalities at the crack interface
and maleimide compounds as the healing agents, ensuring covalent bonding of the
newly formed material and the original material. In the crack area, multifunctional
thiols and maleimides form a network.
Fig. 1. Reaction kinetics at room temperature of the reaction of BGE and BA (-), MMA
and PA (:), PM and PA (A), PM and DT (C), PM and DT catalyzed by TEA ( ), and PM
=
and DT catalyzed by grinded EPON 828 epoxy material ( ).
;

S. Billiet et al. / Polymer 53 (2012) 2320e2326
2323
a chemical bonding of the newly formed network and the original
epoxy material. Moreover, it has been shown that the use of an
excess of amine hardener during the synthesis of the epoxy
material can result in an important contribution to the healing
reaction [33]. On the other hand, in the bulk area of the crack,
a healing reaction can occur by reaction of thiols with the mal-
eimide compounds. These reactions are catalyzed by the tertiary
amines that are present at the crack interface, as has been shown in
the model reaction of PM and DT catalyzed by grinded EPON 828
epoxy material, which obviously contains tertiary amines at the
surface (see Fig. 2). Moreover, polythiols have been encapsulated
for the use in self-healing materials before [24,52,53], mostly as an
alternative to polyamines that are more difficult to encapsulate
[26,54].
As we aim to form a crosslinked material in the crack plane, we
propose the use of multifunctional compounds. A polythiol with
four functionalities (pentaerythritol(3-mercaptopropionate)) was
chosen for its high functionality and flexibility, high boiling point
and earlier demonstration of its successful encapsulation for self-
healing epoxy materials. For the multifunctional maleimides, we
have chosen two difunctional maleimides: 1,1(methylenedi-4,1-
phenyl)bismaleimide (MBM) that has a central bisphenyl moiety,
and N,N0-(1,3 phenylenedi)maleimide (PDM) that contains a phenyl
group (see Scheme 2). The structural similarities should ensure
a good compatibility of the newly formed material with the original
material. As both maleimide compounds are solid compounds, they
are solubilized in m-cresol that was chosen because of the lack of
solubility of the bismaleimide in other solvents. Moreover, m-cresol
has some advantages such as its high boiling point (203 ꢀC), its
hydrophobic nature that will facilitate the microcapsule production
via an ‘oil-in-water’ microencapsulation process and the ability to
act as an inhibitor for the homopolymerization of bismaleimides,
a possible side reaction that would limit the shelf-life of the healing
agent. All of the mentioned healing agents are stable for a long
period when stored in a dry and cool place. The maximal solubility
of MBM and PDM was determined as 30 and 15 wt%, respectively.
Tapered Double Cantilever Beam (TDCB) test method. TDCB
samples with a long moulded groove of 47 mm were used to
evaluate the self-healing ability, according to the protocol
described by White et al. [13]. With this test, TDCB samples were
pulled until crack formation occurred, which results in
a maximum load that the virgin material can resist. Note that at
this stage of the research, the various healing agents are manually
injected in the crack plane. 30
mL of a combination of healing
agents (equimolar amounts of MBM and tetrathiol, MBM and
DETA, PDM and tetrathiol, PDM and DETA, respectively, see
experimental part for details) was then injected in the crack plane,
the sample unloaded, and allowed to heal at 25 ꢀC for 3 or 5 days.
The same test was then repeated to obtain the maximum load that
is necessary to break the healed sample. The ratio of these two
values is defined as the healing efficiency [55]. The TDCB tests are
applied to two types of epoxy materials: EPON 828 and RIM 135
resin (vide supra). For the latter material, we have also evaluated
the effect of the use of excess hardener during the epoxy material
preparation.
2.3.1. EPON 828 epoxy material
In the case of the EPON 828 resin, an equimolar distribution of
resin and diethylene amine (DETA) hardener is used (100/12 w/w
ratio of resin/hardener). Using the usual temperature program for
curing the matrix material (24 h at room temperature, followed by
24 h at 40 ꢀC), a DSC study (see Fig. 3) revealed a conversion of the
epoxide groups of 85% [56]. As equimolar amounts of the amine
hardener were used, 15% of the initial amine functionalities (most
likely secondary and some primary amines) have not reacted and
are still available to take part in the healing reaction. During the
curing process, the vitrification of the mixture will hamper higher
conversions because of diffusional constraints. After the curing
process used in this study, the Tg of the epoxy material is 56 ꢀC,
while after a heating cycle up to 200 ꢀC to ensure complete curing,
the Tg rises to about 130 ꢀC. This can best be seen in the reversing
heat capacity curve where only the heat capacity is retrieved while
the reaction exotherm is retrieved in the non-reversing curve (not
shown).
2.3. Quantification of the healing efficiency: tapered double
cantilever beam (TDCB) tests
Fig. 4 shows the healing efficiency for the healing of the EPON
828 epoxy material with 4 mixtures of healing agent for 3 and 5
days at 25 ꢀC. The healing efficiency shows the average value of
a minimum of 10e15 specimens, with the corresponding standard
deviation. This standard deviation can be assigned to small
In a next step, the effectiveness of the proposed chemistry for
self healing applications was determined by the quantitative
Fig. 3. MDSC thermograms (amplitude ¼ 0.5 ꢀC and period 60 s) of the cured EPON
828 resin. 1st heating showing the Tg (55.85 ꢀC) after the standard cure procedure
followed by the residual reaction. 2nd heating(128.55 ꢀC) showing the final Tg after full
cure. Curves are shifted vertically for clarity.
Scheme 2. Healing agents pentaerythritol(3-mercaptopropionate) (tetrathiol),
1,1(methylenedi-4,1-phenyl)bismaleimide (MBM), and N,N0-(1,3 phenylenedi)mal-
eimide (PDM).

2324
S. Billiet et al. / Polymer 53 (2012) 2320e2326
Fig. 4. Healing efficiency for the healing of EPON 828 epoxy material with various
mixtures of healing agents at 25 ꢀC for 3 days (white bars) and 5 days (grey bars),
respectively. All the specimens were cured at an equimolar ratio of DETA hardener to
epoxy resin (12/100, w/w).
variations in the precrack and deviations of the crack plane. These
values are in accordance to previous reports in which TDCB has
been used to test self-healing materials [27,55]. The results clearly
show a difference between the various mixtures of healing agents,
while also the time allowed for healing shows to be an important
parameter. The MBM compound results in a higher healing effi-
ciency compared to PDM (maximum of 121% compared to 51% after
5 days, respectively). The combination of maleimides with tetra-
thiol results in a higher healing efficiency than the combination
with DETA (121% vs. 70%). These results show effective healing,
with the combination of tetrathiol with MBM as the most effective
healing agent combination.
Fig. 5. Healing efficiency for the healing of RIM 135 epoxy material with various
mixtures of healing agents at 25 ꢀC for 3 days (white bars) and 5 days (grey bars),
respectively. (A) specimens were cured at an equimolar ratio of epoxy resin to RIMH
137 hardener (100/30, w/w). (B) specimens were cured at an excess ratio of epoxy resin
tot RIMH 137 hardener (100/50, w/w).
2.3.2. RIM 135 epoxy material
The RIM 135 epoxy resin is a combination of bisphenol A
diglycidyl ether (DGEBA) and 1,6-hexanediol diglycidyl ether. The
hardener RIMH 137 is mainly composed of alkyl ether amine and
isophorone diamine.
When using an equimolar ratio of resin to hardener (100/30 w/w
resin/hardener) in the case of RIM 135 epoxy material, and using the
usual temperature program for curing (24 h at 40 ꢀC, followed by
16 h at 80 ꢀC), a DSC study revealed a complete conversion of the
epoxides.
Indeed, the Tg of the epoxy material after the applied curing
process and after a heating cycle up to 200 ꢀC was found to be
identical, i.e. 76 ꢀC. As equimolar amounts of the amine hardener
were used, the amine functionalities have reacted during the curing
program. Thus, the RIM epoxy material that results from curing of
an equimolar ratio of resin to hardener (100/30 w/w) does not
contain a significant amount of unreacted amines, and hence no
residual amine functionalities at the crack surfaces that can help in
chemically linking the epoxy surface to the self healing agent.
Fig. 5A shows that also in this case, the MBM curing agent
performs better. The difference resulting from the use of tetrathiol
or DETA is less clear in this case. Healing efficiencies after 5 days are
slightly higher than after 3 days. When compared to the healing
efficiencies of EPON 828 material, the healing efficiency for the RIM
material is generally lower with a maximum healing efficiency of
about 85%.
resin/hardener), as this material will bear unreacted amine func-
tionalities at the crack surface, which will be able to react with the
maleimide healing agents. In this case, the Tg of the material after the
curing process is 48 ꢀC; this value remains identical after a heating
cycle up to 200 ꢀC, revealing that all epoxy functionalities have
reacted. Fig. 5B shows the healing efficiency for various mixtures of
healing agents at 25 ꢀC for 3 days (white bars) and 5 days (grey bars),
respectively. Also in this case, MBM reveals to be a more efficient
healing agent than PDM. However, the healing efficiencies are lower
than in the case of EPON 828 and RIM with an equimolar ratio of
hardener to resin.
2.3.3. Average peak load vs. healing efficiency
Although the healing efficiencies for the epoxy material EPON
828 are higher than the values for the corresponding RIM (100:30
and 100:50) epoxy material, one should note that the healing
efficiency is a relative value, and is influenced by the strength of the
original material. For this reason, it is of interest to evaluate the
maximal force that is needed to break the healed material. The
white bars in Fig. 6 represent the maximal peak load to break the
various virgin epoxy materials. The average maximal peak load to
break the virgin sample is 58 N in the case of EPON 828, 80 N in the
case of RIM (100:30), and 120 N in the case of RIM (100:50). It is
clear that EPON 828 is an intrinsically weaker material than RIM
We also have investigated the effect of using an excess ratio of
hardener for the preparation of RIM 135 epoxy material(100/50 w/w

S. Billiet et al. / Polymer 53 (2012) 2320e2326
2325
we are exploring other chemical strategies to further optimize
the healing chemistry and investigate the possibility for the
application of these concepts in state-of-the-art materials for daily
applications.
4. Experimental
4.1. Materials
Butyl glycidyl ether (BGE, Aldrich, 95%), benzyl amine (BA, Acros,
99,5%), diethylene triamine (DETA, Aldrich, 99%), phenylmaleimide
(PM, Aldrich, 97%), propylamine (PA, Acros, 98%), 1-dodecanethiol
(DT, Fluka, 97%), pentaerythritol(3-mercaptopropionate) (Tetra-
thiol, Aldrich, 95%), triethylamine (TEA, Acros, 99%),1,1(methylenedi-
4,1-phenyl)bismaleimide (MBM, Aldrich, 95%), N,N0-(1,3 phenyl-
enedi)maleimide (PDM, Acros, 97%), EPON 828 epoxy resin (Low
viscosity, Momentive Specialty Chemicals GmbH), RIM 135 epoxy
resin (Momentive Specialty Chemicals GmbH) and RIMH 137 hard-
ener (Momentive Specialty Chemicals GmbH) were used as received.
The RIM 135 epoxy resin is a combination of bisphenol A diglycidyl
ether (DGEBA) and 1,6-hexanediol diglycidyl ether [57]. The RIMH
137 hardener is mainly composed of low MW poly(propylether
amine) (MW z 230) and isophorone diamine. Solvents were
purchased from Acros and used without purification.
Fig. 6. Overview of the maximal peak load to break various epoxy samples. The white
bars represent the maximal peak load to break the original material. The grey bars
represent the maximum peak load to break the healed material. The horizontal line
shows the average maximum peak load for all healed specimens (irrespective of the
original material). The specimens were healed with a healing agent combination of
MBM and tetrathiol for 5 days at 25 ꢀC.
(100:30) and RIM (100:50). However, note that the high apparent
strength of the RIM (100:50) arises from plastic deformation of the
RIM material due to its relatively low Tg (48 ꢀC). The grey bars in
Fig. 6 represent the maximal peak load to break the healed mate-
rial. The various epoxy materials were healed for 5 days at 25 ꢀC
with a combination of tetrathiol and MBM, as this showed to be the
most effective healing agent combination. It should be noted that
the average maximum peak load to break the healed specimens is
not very different (around 62 N), irrespective of the virgin material.
Taking into account this average value, the healing efficiency for
EPON 828 is 121%, 90% for RIM (100:30) and 48% for RIM (100:50).
Thus, although the maximum peak load is nearly the same, the
healing efficiency decreases dramatically as a result of the high
maximum load to break the original material.
4.2. Model study for the kinetics of the reaction of amines and
epoxide/CeC double bonds
PA (0.44 mle5.28 mmol) is added to MMA (0.56 mle5.26 mmol)
in m-cresol (10 wt%) in a glass (10 ml) vial as reference for the
reactivity of amines on the double bond of GMA. As a reference for
the epoxide group in GMA, BA (0.43 mle3.96 mmol) is added to
BGE (0.57 mle3.99 mmol) in m-cresol (10 wt%). 1H NMR is used to
follow conversions of the mentioned reactions.
4.3. Model study for the kinetics of the reaction of maleimides and
amines/thiols
PM(0.208ge1.20mmol)isdissolvedinm-cresol(1.72mle10wt%),at
roomtemperature,inaglass(2ml)vial.Thevialisputintheultrasonic
bathforashorttimetoensurethatahomogeneoussolutionisobtained.
InthissolutionPA(0.1mle1.20mmol)orDT(0.285mle1.20mmol)is
added. For the reaction with DT, a reaction was conducted without
3. Conclusions
An alternative chemistry based on the Michael addition
between bismaleimides and amines or thiols was developed for the
self-healing of epoxy materials. The proposed healing reaction was
characterized via an extensive kinetic study of monofunctional
model compounds. This study lead to the selection of multifunc-
tional healing agents that were tested for their healing efficiency of
various epoxy materials via the ‘tapered double cantilever beam’
(TDCB) test method, with manual injection of the healing agents
catalyst, with TEA (17
mle0.122 mmol) as a catalyst, or EPON 828 as a
catalyst(0.125ge100:12wt%egrinded,averagesizeofgrainsꢁ100
m
m),
respectively.Conversionofthemaleimidefunctionalitywasmonitored
by1HNMR.
4.4. Stock solutions of healing agents
(30 mL). For EPON 828, an average healing efficiency of 121% was
reached with the healing agent combination MBM þ tetrathiol.
Moreover, we have extended the self-healing testing towards resins
that are currently used as a cold-curing resin for infusion of large
wind turbine blades. In this case, we have also investigated the
effect of unreacted amine functionalities in the epoxy material by
using an excess of hardener to resin. Also in the case of the RIM
resin, the combination of MBM þ tetrathiol was found to be the
most promising healing agent. It reached healing efficiencies going
from 80% (equimolar) to 50% (excess amine hardener) after a heal-
ing period of 5 days at room temperature. However, the healing
efficiency was found to be strongly dependent on the type of epoxy
material, while the absolute average maximum peak load for
fracture after healing was found to be roughly the same for all
epoxy materials. Further research is focusing on the encapsulation
of maleimide compounds in microcapsules. At the same time,
MBM (0.385ge1 mmol)isdissolvedin m-cresol (1.24 mle25wt%)
in
a
(190
glass vial (2 ml), followed by addition of tetrathiol
le0.5 mmol) or DETA (108 le1 mmol), respectively. Alterna-
m
m
tively, PDM (0.268 ge1 mmol) is dissolved in m-cresol
(1.60 mle15 wt%) in a glass vial (2 ml), followed by addition of tet-
rathiol (190
mle0.5 mmol) or DETA (108 mle1 mmol), respectively.
4.5. Synthesis of epoxy materials
EPON 828 resin is mixed with DETA in a 100/12 (equimolar)
weight/weight ratio, RIM 135 resin is mixed with RIMH 137 in
a 100/30 (equimolar) or 100/50 (excess hardener) weight/weight
ratio. After homogeneous mixing, air bubbles are removed by
applying high vacuum for 5 min. The solution is then poured in
a silicon mould with specific dimensions to produce the Tapered

2326
S. Billiet et al. / Polymer 53 (2012) 2320e2326
[13] White SR, Sottos NR, Geubelle PH, Moore JS, Kessler MR, Sriram SR, et al.
Nature 2001;409(6822):794e7.
[14] Wu DY, Meure S, Solomon D. Progress in Polymer Science 2008;33(5):479e522.
[15] Mauldin TC, Rule JD, Sottos NR, White SR, Moore JS. Journal of the Royal
Society Interface 2007;4(13):389e93.
Double Cantilever Beam (TDCB) samples [55]. The filled moulds are
cured at different temperatures e EPON 828 (1 day at 25 ꢀC and
16 h at 40 ꢀC) and RIM 135 (1 day at 40 ꢀC and 16 h at 80 ꢀC).
[16] Rule JD, Moore JS. Macromolecules 2002;35(21):7878e82.
[17] Lee JK, Liu X, Yoon SH, Kessler MR. Journal of Polymer Science Part B-Polymer
Physics 2007;45(14):1771e80.
4.6. TDCB testing
A precrack was made in the TDCB sample with the aid of a razor
blade, which was inserted into the groove to the TDCB specimen
and tapping it into the moulded notch starter. The samples were
then clamped on the tensile testing machine to be stretched at
[18] Liu X, Lee JK, Yoon SH, Kessler MR. Journal of Applied Polymer Science 2006;
101(3):1266e72.
[19] Wilson GO, Caruso MM, Reimer NT, White SR, Sottos NR, Moore JS. Chemistry
of Materials 2008;20(10):3288e97.
[20] Jones AS, Rule JD, Moore JS, White SR, Sottos NR. Chemistry of Materials 2006;
18(5):1312e7.
[21] Coope TS, Mayer UFJ, Wass DF, Trask RS, Bond IP. Advanced Functional
Materials 2011;21(24):4624e31.
[22] Cho SH, Andersson HM, White SR, Sottos NR, Braun PV. Advanced Materials
2006;18(8):997.
[23] Rong MZ, Zhang MQ, Zhang W. Advanced Composites Letters 2007;16(5):
167e72.
[24] Yuan YC, Rong MZ, Zhang MQ, Chen B, Yang GC, Li XM. Macromolecules 2008;
41(14):5197e202.
[25] Yuan YC, Ye XJ, Rong MZ, Zhang MQ, Yang GC, Zhao JQ. ACS Applied Materials
and Interfaces 2011;3(11):4487e95.
a speed of 5
occurs, resulting in the peak load to break the virgin sample. For
healing of the samples, 30 l of healing agent was manually
mm/s [55]. The specimens were loaded until failure
m
injected in the crack plane and the samples were clamped together
to ensure a good contact of the healing agent to the crack surface.
The healed samples were stored at 25 ꢀC for 3 or 5 days and then
broken at the same initial opening speed to calculate the healing
efficiency, which is defined as the ratio of the maximum peak load
of the healed sample to the maximum peak load of the virgin
sample. The average value of minimum 5 samples was taken to get
accurate results.
[26] McIlroy DA, Blaiszik BJ, Caruso MM, White SR, Moore JS, Sottos NR. Macro-
molecules 43(4):1855e1859.
[27] Caruso MM, Delafuente DA, Ho V, Sottos NR, Moore JS, White SR. Macro-
molecules 2007;40(25):8830e2.
[28] Caruso MM, Blaiszik BJ, White SR, Sottos NR, Moore JS. Advanced Functional
Materials 2008;18(13):1898e904.
[29] Kolb HC, Finn MG, Sharpless KB. Angewandte Chemie International Edition
2001;40(11):2004e21.
4.7. Instrumentation
1H NMR spectra were recorded with a Bruker AVANCE 300
(300 MHz) FT-NMR spectrometer. Differential Scanning Calo-
rimetry (DSC) thermograms were recorded using a TA Instru-
ments Q2000 DSC with autosampler option and Refrigerated
Cooling System (RCS). Nitrogen gas was used as purge gas. The
samples were studied in TAi Tzero Hermetic Aluminium sample
pans and a scan rate of 10 K/min. TDCB tests were conducted
on a Tinius Olsen H10KT with a load cell of 5000 N. The load-
displacement curves were analyzed with the aid of the Test
Navigator software.
[30] Gragert M, Schunack M, Binder WH. Macromolecular Rapid Communications
2011;32(5):419e25.
[31] Schunack M, Gragert M, Döhler D, Michael P, Binder WH. Macromolecular
Chemistry and Physics 2012;213(2):205e14.
[32] Sheng X, Mauldin TC, Kessler MR. Journal of Polymer Science Part A: Polymer
Chemistry 2010;48(18):4093e102.
[33] Meng LM, Yuan YC, Rong MZ, Zhang MQ. Journal of Materials Chemistry 2010;
20(29):6030e8.
[34] Chen X, Dam MA, Ono K, Mal A, Shen H, Nutt SR, et al. Science 2002;
295(5560):1698e702.
[35] Chen X, Wudl F, Mal AK, Shen H, Nutt SR. Macromolecules 2003;36(6):1802e7.
[36] Liu Y-L, Hsieh C-Y, Chen Y-W. Polymer 2006;47(8):2581e6.
[37] Toncelli C, De Reus DC, Picchioni F, Broekhuis AA. Macromolecular Chemistry
and Physics 2012;213(2):157e65.
[38] Mather BD, Viswanathan K, Miller KM, Long TE. Progress in Polymer Science
2006;31(5):487e531.
Acknowledgements
[39] Corning Owens. Industrial customer presentation:“High performance rein-
forcements”; 2011.
Stijn Billiet thanks the Agency for Innovation by Science and
Technology in Flanders (IWT) for a PhD scholarship. W. Van Pae-
peghem (Ghent University) and D. Van Hemelrijck (Vrije uni-
versiteit Brussel) are acknowledged for the scientific discussions.
W. V. C. thanks the Research Foundation e Flanders (FWO) for
a postdoctoral research fellowship. F. D. P. acknowledges the
Belgian Program on Interuniversity Attraction Poles initiated by the
Belgian State, Prime Minister’s office (Program P6/27), and the
European Science Foundation e Precision Polymer Materials (P2M)
program for financial support.
[40] PPG Industries Inc. Composite materials for wind blades: current performance
and future directions. Presented by Juan Camilo Serrano at the 2010 Wind
Turbine Blade Workshop. Sandia National Laboratories; July 20e21, 2010.
[41] Reifsnider KL, Introduction. Fatigue of composite materials: composite
material series 4. Elsevier; 1990. pp. 1e9.
[42] Reifsnider KL. Fatigue behaviour of composite laminates. Fatigue of composite
materials. Composite material series 4. Elsevier; 1990. pp. 105e180.
[43] Reifsnider KL. Life prediction for resin-matrix composite materials. Fatigue of
composite materials. Composite material series 4. Elsevier; 1990. pp.
431e483.
[44] Van Paepegem W, Degrieck J. Composites Part A 2001;32(10):1433e41.
[45] Wang HP, Yuan YC, Rong MZ, Zhang MQ. Colloid and Polymer Science 2009;
287(9):1089e97.
[46] Yang JL, Keller MW, Moore JS, White SR, Sottos NR. Macromolecules 2008;
41(24):9650e5.
References
[47] Wang X, Li G, Wei J, Guan W. Journal of Applied Polymer Science 2009;113(2):
1008e16.
[48] Sheng X, Mauldin TC, Kessler MR. Journal of Polymer Science Part A-Polymer
Chemistry 48(18):4093e4102.
[49] Montarnal D, Tournilhac F, Hidalgo M, Leibler L. Journal of Polymer Science
Part A-Polymer Chemistry 48(5):1133e1141.
[50] Chan JW, Hoyle CE, Lowe AB, Bowman M. Macromolecules 43(15):6381e6388.
[51] Li GZ, Randev RK, Soeriyadi AH, Rees G, Boyer C, Tong Z, et al. Polymer
Chemistry 1(8):1196e1204.
[1] Norris BC, Bielawski CW. Macromolecules 2010;43(8):3591e3.
[2] Williams KA, Dreyer DR, Bielawski CW. MRS Bulletin 2008;33(08):759e65.
[3] Caruso MM, Davis DA, Shen Q, Odom SA, Sottos NR, White SR, et al. Chemical
Reviews 2009;109(11):5755e98.
[4] Wudl F, Murphy EB. Progress in Polymer Science 2010;35:223e51.
[5] Mauldin TC, Kessler MR. International Materials Reviews 55(6):317e346.
[6] Syrett JA, Becer CR, Haddleton DM. Polymer Chemistry 2010;1(7):978e87.
[7] Hager MD, Greil P, Leyens C, van der Zwaag S, Schubert US. Advanced
Materials 22(47):5424e5430.
[52] Yuan YC, Rong MZ, Zhang MQ. Polymer 2008;49(10):2531e41.
[53] van den Dungen ETA, Klumperman B. Journal of Polymer Science Part
A-Polymer Chemistry 48(22):5215e5230.
[8] Blaiszik BJ, Kramer SLB, Olugebefola SC, Moore JS, Sottos NR, White SR. Annual
Review of Materials Research 40(40):179e211.
[54] Yuan L, Liang GZ, Xie JQ, He SB. Colloid and Polymer Science 2007;285(7):
781e91.
[9] Wool RP. Soft Matter 2008;4(3):400e18.
[10] Burattini S, Greenland BW, Chappell D, Colquhoun HM, Hayes W. Chemical
Society Reviews 1973-1985;39(6).
[11] van der Zwaag S. Self healing materials. Dordrecht, The Netherlands:
Springer; 2007.
[12] Williams KA, Boydston AJ, Bielawski CW. Journal of The Royal Society Inter-
face 2007;4(13):359e62.
[55] Brown EN, Sottos NR, White SR. Experimental Mechanics 2002;42(4):372e9.
[56] Menczel JD, Prime RB. Thermal analysis of polymers: fundamentals and
applications. Hoboken, New Jersey: John Wiley & Sons; 2009.
[57] Momentive Specialty Chemicals. Technical data sheet EPIKOTE Resin MGS RIM
135.