Application of alkoxyamine in self-healing of epoxy

Article abstract of DOI:10.1039/c4ta00130c

To provide epoxy with intrinsic self-healing ability, a new alkoxyamine based diol was synthesized and incorporated into an epoxy monomer, which was then compounded with traditional bisphenol A diglycidyl ether and cured by diethylenetriamine. Taking advantage of dynamic equilibrium of thermally reversible reaction of C-ON bond in alkoxyamine, cracked portions of the functionalized epoxy material can be repeatedly reconnected at certain temperature, as characterized by restoration of impact strength and visual inspection as well. Unlike the two-step self-healing approach based on reversible Diels-Alder reaction, the present one only needed one step. Owing to the steric hindrance of tertiary butyl group in the diol, onset of scission of C-ON bonds and radical recombination (i.e., healing reaction) occurred at lower temperature. Additionally, reversibility of the alkoxyamine derivatives was improved with incorporation of Si-O bonds into the epoxy chains. The present work carefully studied thermal reversibility, thermal stability, dynamic mechanical behavior and healing performance of the epoxy in relation to molecular structure. The outcomes might help to optimize the material and guide future design of novel epoxy monomers. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c4ta00130c

Journal of
Materials Chemistry A
View Article Online
View Journal | View Issue
PAPER
Application of alkoxyamine in self-healing of epoxy
a
a
Chan'e Yuan,ab Ming Qiu Zhang and Min Zhi Rong
*
*
Cite this: J. Mater. Chem. A, 2014, 2,
6558
To provide epoxy with intrinsic self-healing ability, a new alkoxyamine based diol was synthesized and
incorporated into an epoxy monomer, which was then compounded with traditional bisphenol A
diglycidyl ether and cured by diethylenetriamine. Taking advantage of dynamic equilibrium of thermally
reversible reaction of C–ON bond in alkoxyamine, cracked portions of the functionalized epoxy material
can be repeatedly reconnected at certain temperature, as characterized by restoration of impact
strength and visual inspection as well. Unlike the two-step self-healing approach based on reversible
Diels–Alder reaction, the present one only needed one step. Owing to the steric hindrance of tertiary
butyl group in the diol, onset of scission of C–ON bonds and radical recombination (i.e., healing
reaction) occurred at lower temperature. Additionally, reversibility of the alkoxyamine derivatives was
improved with incorporation of Si–O bonds into the epoxy chains. The present work carefully studied
thermal reversibility, thermal stability, dynamic mechanical behavior and healing performance of the
epoxy in relation to molecular structure. The outcomes might help to optimize the material and guide
future design of novel epoxy monomers.
Received 9th January 2014
Accepted 6th February 2014
DOI: 10.1039/c4ta00130c
www.rsc.org/MaterialsA
Epoxy is an important polymeric material widely applied in
surface coatings, structural adhesives, printed circuit boards,
Introduction
insulation materials for electronic devices, and advanced
composites, because of its high mechanical properties, chem-
ical resistance, thermal stability, ease of handling and pro-
cessing, etc. application in various elds, easy processing in a
wide temperature range, etc. With the development of self-
healing technology in recent years, the possibility of imparting
intrinsic crack remendability to epoxy has been explored.
By using thermally reversible DA reaction, which proved to
be able to reconnect broken chains in thermoset,20 Tian and co-
workers prepared self-healing epoxy composed of both ther-
mally reversible DA bonds and irreversible bonds from ring-
opening polymerization of epoxide groups.21–23 This molecular
structure allows the cured epoxy to possess similar mechanical
properties like commercial epoxies, and to eliminate cracks
through successive retro-DA and DA reaction below Tg. Unlike
the above work by Tian et al. who functionalized epoxy with
furan,21 Bai et al. synthesized a diamine carrying DA unit, which
served as a healable element and was used to crosslink
commercial epoxy monomers to achieve self-healing.24 Canadell
et al. incorporated disulde links into a rubbery epoxy network
(i.e., a commercially available disulde group-contained epoxy
cured by tetrafunctional thiol), which was able to fully restore its
mechanical property at moderate temperatures (e.g., 60 ꢀC).25
The healing mechanism was further analyzed in detail from the
viewpoint of thiol–disulde exchange.14
Polymers and polymer composites have been used in a variety of
applications. However, they are susceptible to damage induced by
mechanical, chemical, thermal, and UV irradiation factors. This
damage may bring about microcracks deep inside the materials,
which are difficult to de detected and repaired. To settle the issue,
the concept of self-healing has been considered.1 So far, the
healing approaches fall into two categories: extrinsic and
intrinsic.2 The former takes effect because of the pre-embedded
microencapsulated healing agent in the materials. Upon fracture,
the healing agent would be released to the cracked portions and
rebond the damages.3,4 However, the cracks can not be repeatedly
healed when the healing agent is exhausted. The latter relies on
the inherent physical and/or chemical interactions of macro-
molecules between the broken faces. Reversion of covalent
bonding,5,6 for example, is found to be an effective means in this
aspect owing to its practicability. Intrinsic self-healing system
does not require additional healing agents,7,8 and the reversible
bonds responsible for repeated healing cycles include Diels–Alder
(DA) bonds,9 thiuram disulde units,10 alkoxyamine moieties,11
coumarin derivatives,12 acylhydrazone bonds,13 disulde links,14
Ru-catalyzed shuffling of C–C bonds,15 hydrogen bonding,16 ion-
omers,17 p–p stacking,18 van der Waals' actions,19 etc.
aKey Laboratory for Polymeric Composite and Functional Materials of Ministry of
Education, GD HPPC Lab, School of Chemistry and Chemical Engineering, Sun
Yat-sen University, Guangzhou 510275, P. R. China. E-mail: ceszmq@mail.sysu.edu.
cn; cesrmz@mail.sysu.edu.cn; Fax: +86-20-84114008; Tel: +86-20-84114008
bNational Engineering Research Center of Electronic Circuits Base Materials, Shengyi
Technology Co. Ltd., Dongguan 523808, P. R. China
Leibler's group reported that with proper catalyst (e.g., zinc
acetate), the topologies of crosslinked networks from epoxy-acid
and epoxy-anhydride reactions can be rearranged by trans-
esterication exchange reactions without modication of the
6558 | J. Mater. Chem. A, 2014, 2, 6558–6566
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry A
numbers of links and average functionality.26 At lower temper- hoped that aer curing, an epoxy with balanced performance
ature, the exchange reactions are sluggish, and the materials would be yielded (Fig. 1), in which the C–ON bonds help to heal
have properties of classical epoxy thermosets. At higher cracks via their reversible homolysis/recombination habit.
temperature, however, the polymers behaved like viscoelastic
liquids. Lap-shear tests indicated that these networks can be
welded aer treatment at 150 ꢀC for 1 h. Peterson et al. built up
Experimental
reversible reinforcement/matrix interfaces in glass ber-rein- Materials and reagents
forced composites taking advantage of DA reaction between a
Benzoyl peroxide (BPO), chlorodimethylsilane, and ally glycidyl
ether were purchased from Alfa Aesar GmbH, Germany. 4-
Hydroxy-2,2,6,6-tetramethylpiperidinyloxy (4-OH-TEMPO),
furan-functionalized epoxy-amine thermosetting matrix with a
maleimide-functionalized glass ber.27,28 At room temperature
the DA adduct is formed spontaneously and above 90 ꢀC the
adduct breaks apart to reform the original furan and maleimide
moieties. Signicant healing of the interface characterized by
single ber microdroplet pull-out testing was observed.
diethylenetriamine (DETA), triethylamine, and diethylene glycol
diglycidyl ether (DGDE) were supplied by Aldrich. Karstedt's
catalyst solution (divinyltetramethyldisiloxane–Pt(0), 2–3% w/w
in xylene) was provided by Gelest, and bisphenol A diglycidyl
ether (EPIKOTE 828) by Shell Chemical Company. Dibutyltin
dilaurate (DBTL) was supplied by Aldrich. All the above chem-
icals were used as received. a-Methyl styrene (Alfa Aesar GmbH,
Germany) was puried by vacuum distillation aer elimination
of inhibitor and pre-dried with sodium sulfate. Tetrahydrofuran
(THF) and triethylamine were purchased from Aldrich, pre-
dried and distilled over Na in advance.
In this work, a new healing chemistry, reversible covalent
bonding through dynamic exchange reaction of alkoxyamines,
is introduced to make self-healing epoxy. Under certain
homolysis temperature, C–ON bonds in alkoxyamines undergo
a reversible radical reaction, producing transient carbon-
centered and persistent nitroxide radicals in equal amounts.29–34
There exists a fast dynamic equilibrium between dissociation
and association with a very high equilibrium constant of 1.7 ꢁ
1015 [Mꢂ1] for bulk living polymerization of styrene at 135 ꢀC.30
Our early work demonstrated that crosslinked polystyrene
containing C–ON bonds as crosslinkers can indeed rebind
fractured surfaces repeatedly.11 Since covalent bond homolysis
and radical recombination synchronously take place in this
case, the aforesaid two-step healing strategy based on DA bonds
is simplied into a single-step one, and possible material
deformation due to collapse of crosslinked networks in the
course of crack remending would thus be avoided even when
the healing is carried out above glass transition temperature.
Here, for purposes of obtaining alkoxyamine-containing
epoxy (diEP, see Scheme 1), a new alkoxyamine-based diol is
designed and synthesized, which should possess lower homol-
ysis-recombination temperature in comparison with the
alkoxyamine-based diol used in our previous work.11 Mean-
while, Si–O bonds are introduced through reaction between the
diol and chlorosilane to improve mobility of molecular chains
connected with alkoxyamine groups and facilitate the subse-
quent healing reaction across crack faces. By hydrosilylation
between silicon hydrogen bond and allyl glycidyl ether, the
alkoxyamine-containing epoxy (diEP) is yielded. Considering
that the cured version of the alkoxyamine-containing epoxy
might be too so as a result of its exible long segment struc-
ture, traditional bisphenol A diglycidyl ether is incorporated as
a rigid component to strengthen the ultimate material. It is
Synthesis of alkoxyamine-based diol
The alkoxyamine-based diol (4-hydroxy-1-(20-hydroxy-10-phenyl-
10-methyl) ethyl-TEMPO) was synthesized as follows. Benzoyl
peroxide (14.52 g, 0.06 mol) and 4-OH-TEMPO (20.64 g, 0.12
mol) were dissolved in a-methyl styrene (250 mL), heated at
ꢀ
100 C under argon for 3 h, and then cooled to room temper-
ature. The excess a-methyl styrene was removed under reduced
pressure. Aerwards, ethanol (300 mL) and KOH aqueous
solution (3 N, 120 mL) were added into the system and the
mixture was stirred at room temperature for 12 h, followed by
evaporation
to
dryness
and
partitioning
between
Scheme 1 Synthesis of alkoxyamine-containing epoxy (diEP).
Fig. 1 Thermally remendable epoxy containing alkoxyamines.
J. Mater. Chem. A, 2014, 2, 6558–6566 | 6559
This journal is © The Royal Society of Chemistry 2014

View Article Online
Journal of Materials Chemistry A
Paper
dichloromethane and deionized water. The aqueous phase was resin cured by DETA (labeled as N-epoxy) was also prepared.
extracted with dichloromethane three times and the organic Additionally, another control specimen made from the
phases were gathered, dried with anhydrous sodium sulfate for compounds of bisphenol A epoxy resin and DGDE (denoted by
24 h. Aer the solvent was removed, the crude product was 828/DGDE), in which DEDE replaced diEP in 828/diEP, was
puried by column chromatography using ethyl acetate and designed to reveal the effect of entanglement of so molecular
petroleum ether as the eluent, offering the diol as a pale yellow segments on self-healing performance. In this case, the molar
1
solid (yield ¼ 34.7%). H NMR (CDCl3): d/ppm 0.89 (3H, CH3), ratio of 828/DGDE was still 1.5 : 1 and DETA served as the
0.99 (3H, CH3), 1.19 (3H, CH3), 1.20 (3H, CH3), 1.58 (3H, CH3), hardener again.
1.53–1.84 (4H, 2CH2), 3.84 (1H, CH), 3.96 (1H, CH), 4.05
(1H, CH), 7.21–7.54 (5H, aromatic); 13C NMR (CDCl3): d/ppm
21.96, 22.06, 23.93, 34.19, 49.49, 49.75, 61.01, 61.10, 63.04,
Characterization
1H NMR and 13C NMR spectra were collected by a VARIAN
70.12, 81.91, 126.88, 127.19, 128.14, 146.20; mass spectrum for
18H29NO3 [M + 1]+ found 308.2.
Mercury-Plus 300 (300 MHz) with chloroform (CDCl3) as
C
solvent. Mass spectrometry (MS) measurement was conducted
on a Thermo Finnigan LCQDECA P LC-MS spectrometer. Elec-
tron spin resonance (ESR) spectra were recorded by JEOL JES-
FA200 to investigate the habit of thermal reversibility of the
resultant epoxy materials. Thermogravimetric analysis (TGA)
was performeꢂd1on a TA Instruments TGA Q50 at a heating rate
of 10 ꢀC min using a nitrogen purge. Dynamic mechanical
analysis (DMA) was carried out with a 01Db-metravib-DMA25
apparatus under nitrogen atmosphere using tensile mode.
Molecular weight between crosslinks of the cured epoxy, Mc,
was calculated from:36
Synthesis of diEP
The synthesis of diEP referred to the route reported in ref. 35
(Scheme 1). The aforesaid alkoxyamine-based diol (3.07 g, 0.01
mol) and triethylamine (3.03 g, 0.03 mol) were dissolved in
tetrahydrofuran (100 mL), and then chlorodimethylsilane
(2.835 g, 0.03 mol) was added dropwise into the system under
argon gas ow to ensure an anhydrous reaction environment,
followed by being stirred at room temperature for 5 h. Aer the
generated triethylamine hydrochloride was removed, the
mixture was evaporated to dryness under reduced pressure and
partitioned between toluene and deionized water. The organic
phase was washed with water three times and dried with
anhydrous sodium sulfate for 24 h. Aerwards, Karstedt's
catalyst (0.018 g, 7.8 ꢁ 10ꢂ5 mol) and allyl glycidyl ether (2.394
g, 0.021 mol) were added into the obtained toluene solutionꢀof
intermediate product, and the compound was heated at 60 C
under argon atmosphere for 2.5 h. The solvent and excess allyl
glycidyl ether were removed by vacuum distillation to give the
product as a pale yellow oil (yield ¼ 63.2%). The obtained
alkoxyamine derivative with epoxy group at both ends is labeled
ꢀ
ꢁ
r
2
E0 ¼ 3
RT 1ꢂ
Mc
f
where E0 stands for storage modulus at rubbery plateau zone, r
density, R gas constant, T absolute temperature, and f func-
tionality of epoxy, respectively. In this work, E0 values at T ¼ Tg +
ꢀ
30 C were used for the calculation.
Healing property of the epoxy was characterized by Izod
notched impact tests on the basis of GBT 1843–1996 standard
according to the method proposed by Jones et al.37 Firstly, the
specimen (52.8 mm ꢁ 12.3 mm ꢁ 10.1 mm) was impacted to
failure using a JJ-20 impact tester (Changchun Research Insti-
tute for Testing Machines Co. Ltd., China). Then, the broken
halves of the specimen were recombined, with a pressure of
about 0.2 MPa to make sure that their faces were kept in
alignment and intimate contact. Healing was conducted at
90 ꢀC under argon atmosphere for 1.5 h. Having been cooled to
room temperature, the xed specimen was tested to failure
again. The ratio of impact strength of the healed specimen to
that of the virgin one gave a measure of healing efficiency. Each
batch included six specimens to yield average value.
1
as diEP. H NMR (CDCl3): d/ppm ꢂ0.16 (2H, CH2), ꢂ0.21 (2H,
CH2), 0.04–0.07 (12H, 4 Si–CH3), 0.325–1.695 (23H, 4CH3, 1 CH3,
2CH2, 2CH2), 2.55 (2H, CH), 2.73 (2H, CH), 3.08 (2H, CH), 3.31–
3.51 (8H, 4CH2), 3.64 (1H, CH), 3.80 (1H, CH), 3.95 (1H, CH),
7.12–7.39 (5H, aromatic); 13C NMR (CDCl3): d/ppm ꢂ2.06,
ꢂ1.01, 0.73, 1.41, 12.58, 13.27, 14.65, 18.66, 22.09, 23.58, 23.86,
33.65, 34.54, 44.53, 50.04, 50.28, 51.11, 60.23, 30.80, 63.78,
71.47, 71.70, 74.36, 74.49, 82.99, 126.62, 126.88, 127.51, 147.34;
mass spectrum for C34H61NO7Si2 [M + 1]+ found 652.4.
Preparation of epoxy specimens
The reason why we used impact test rather than tensile test
to evaluate healing performance of the material lies in the fact
that material loss would take place at the fracture surface when
the specimen was failed due to tensile test. As a result, it was
hard to ensure intimate contact between the uneven fracture
surface of the broken halves of the specimen.
The synthesized diEP was blended with bisphenol A epoxy resin
(denoted by 828) at the molar ratio of 1 : 1.5, and then the
hardener DETA was added according to the stoichiometric ratio
of the reactive epoxide group of each ingredient. The epoxy
specimens (called 828/diEP) used for further characterization
were cured in a silicone rubber mold under an argon atmo-
sphere step by step at 30, 80 and 100 ꢀC for 12, 5 and 1 h,
respectively.
Results and discussions
Thermal reversibility
To study the relationship between homolysis of alkoxyamine
moieties and thermal stability of the material when being Since the C–ON bonds in alkoxyamines should play the leading
heated, diEP specimen in the absence of bisphenol A epoxy role in self-healing and homolysis-rebonding of C–ON bonds is
6560 | J. Mater. Chem. A, 2014, 2, 6558–6566
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry A
related to free radicals, ESR was employed to understand the
To further investigate the reversibility of C–ON bond in
thermally reversible behavior of the materials at the beginning cured epoxy containing alkoxyamines, ESR spectra were
of the discussion. Fig. 2a shows typical ESR spectra of N-epoxy collected during repeated heating–cooling cycles between 20
ꢀ
measured at various temperatures. In addition, Fig. 2b gives the and 90 C (Fig. 3), as the change in radical relative intensity
normalized absorption lines corresponding to the ESR curves in reects homolysis-recombination habit of C–ON bond.
Fig. 2a, from which integrated areas of the relative signal Moreover, comparison between the relative signal intensity
intensities are calculated and plotted as a function of temper- aer thermocycling and that of the original state at 20 ꢀC
ꢀ
ature (see Fig. 2c). Clearly, radical signal is detectable at 20 C would help to evaluate thermal reversibility of the epoxy. It is
and its intensity increases with a rise in temperature. The seen from Fig. 3 that when temperature alternates between
phenomenon resembles the case of crosslinked polystyrene 20 and 90 ꢀC, the amount of free radicals uctuates as a result
containing alkoxyamine,11 but the initial temperature for of scission of alkoxyamine moieties and recombination of the
appearance of radical signal is much lower than that of the radicals. These processes are fully reversible, but the valley
latter (ꢃ100 ꢀC). This is because tertiary butyl groups rather values of the relative intensity slightly increases with the
than ethyl groups link up with the N–O bonds in the diol thermal cycles especially in the case of 828/diEP. The latter
(Scheme 1), which produce larger steric hindrance, reducing the behavior might be attributed to permanent dissociation of a
bond energy of C–ON bond and lowering its homolysis few alkoxyamines as the combination of tertiary butyl radi-
temperature. On the other hand, the results in Fig. 2 and ref. 11 cals, and then nitroxide radicals is somehow retained in the
conrm that alkoxyamines derivatives attached to different solid state.38 The relatively higher thermal reversibility of
polymers exhibit similar thermal reversibility, which provides N-epoxy than 828/diEP should originate from the larger
us with ample room to develop different thermally reversible quantity of so diEP in the former. On one hand, exibility of
polymers for different requirements.
the abundant so diEP chains increases probability of colli-
sion between radicals. On the other hand, the molecular
chain of diEP (651 g molꢂ1) is longer than that of 828 (330 g
molꢂ1), thus making a lower crosslinking density for N-epoxy.
It is also conducive to molecular motion and contact with
each other. Nevertheless, 828/diEP possesses much better
reversible performance than the crosslinked polystyrene with
alkoxyamines mentioned above.11 Apparently the higher
stability of tertiary butyl radicals together with so segments
in epoxy take the responsibility, which can hinder the
combination between carbon radicals but benet reconnec-
tion of the carbon and nitroxide radicals attached to the
macromolecular chains.
Fig.
2 (a) Typical ESR spectra of N-epoxymeasured at various
temperatures. (b) Normalized absorption lines obtained from the data of
(a). (c) Temperature dependence of relative ESR signal intensities Fig. 3 Relative ESR signal intensities of (a) N-epoxy, and (b) 828/diEP
calculated from integral areas of the normalized absorption curves in (b). measured during heating–cooling cycles between 20 and 90 ꢀC.
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. A, 2014, 2, 6558–6566 | 6561

View Article Online
Journal of Materials Chemistry A
Paper
Thermal stability and dynamic mechanical behavior
Considering the fact that the C–ON bonds in alkoxyamines
frequently cleave but immediately recombine under certain
homolysis temperature, thermal stability of the epoxy materials
containing alkoxyamines must be affected accordingly. The
TGA curves in Fig. 4a show that the presence of alkoxyamines
signicantly decreases the onset temperature of epoxy, which is
in proportion to the concentration of alkoxyamines. The
homolysis of C–ON bonds and the resultant radicals should
account for this. In addition, the onset temperature of 828/
ꢀ
DGDE is 30 C lower than that of 828 (see the inset of Fig. 4a),
implying not only alkoxyamines but also so segments in the
polymer network lower the thermal stability of epoxy.
When looking at the curves of rate of weight loss (Fig. 4b),
two peaks are found for N-epoxy. By reference to the peak
position of 828 that does not contain any alkoxyamines, it is
known that the peak at higher temperature results from
destruction of irreversible covalent bonds, and the other reects
the weight loss caused by cleavage of C–ON bonds.
In addition to TGA that provides temperature dependent
structural information under quiescent condition, DMA
enables study of the issue in a dynamic way. As shown in Fig. 5,
the N-epoxy specimen can be repeatedly tested, demonstrating
the existence of reversible equilibrium between C–ON bonds
homolysis and recombination; otherwise, even the second test
could not be conducted if the radicals failed to be immediately
re-bonded and the material behaved like thermoplastics due to
the destruction of the networks. During the repeated heating–
cooling cycles, the temperature dependence of storage modulus
of N-epoxy keeps almost the same, but the proles of its tan d
peaks are obviously changed. In fact, there are two peaks on the
tan d curves of N-epoxy. Compared to the higher temperature
one, both position and height of the lower temperature one
Fig. 5 Temperature dependence of (a) storage modulus and (b) tan d
of N-epoxy measured by repeated DMA scans.
change much more remarkably with the repeated scans. This
should be the result of the dynamic equilibrium feature of the
reversible C–ON bonding. That is, the lower temperature peak
on the tan d curve of N-epoxy is contributed by segment motion
of the dangling chains generated by disconnection of C–ON
bonds, while the higher temperature one represents the
undamaged networks. Owing to the random nature of the
homolysis/recombination behavior of C–ON bond, the dangling
chain lengths change from time to time so that the position and
height of the lower temperature peak are different for different
tests. In contrast, the higher temperature peak is basically
unchanged due to the relatively denite molecular weight
between cross-links of the undamaged networks.
With the incorporation of 828 that contains no reversible C–
ON bonds and increases crosslinking density of the system, the
tan d curves of 828/diEP become unimodal as the dangling
chains lengths of the disconnected chains produced by
homolysis of alkoxyamine moieties have to be much shorter due
to the existence of irreversible networks, which cannot afford to
form to a new damping peak (Fig. 6). Nevertheless, it is worth
noting that peak temperature and intensity on the tan d curves
decrease aer each test cycle. This should be interpreted as the
appearance of a few permanently dissociated alkoxyamines
originating from the difficulty of recombination of tertiary butyl
radicals and nitroxide radicals in solid (Fig. 3b). Accordingly,
crosslinking density of the epoxy network is reduced with the
test cycles (Table 1). The analysis is supported by the tempera-
ture dependence of storage modulus, which moves downward
during repeated DMA tests (Fig. 6a). The slight decline of the
stiffness should be attributed to the permanently destructed C–
ON bonds that could no longer bear load. The decay is marginal
on the whole, which is also proved by ESR data, meaning the
Fig. 4 Pyrolytic behaviors of cured epoxies.
6562 | J. Mater. Chem. A, 2014, 2, 6558–6566
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry A
Fig. 6 Temperature dependence of (a) storage modulus and (b) tan d of 828/diEP measured by repeated DMA scans.
thermal reversibility of cured 828/diEP would effectively work
Table 2 lists that the average healing efficiency of the rst
for crack healing. On the other hand, a comparison of Fig. 5 and healing as 62.0%, while those of the second and third healing are
6 indicates that the peak temperature of tan d of 828/diEP is 59.5 and 55.1%, respectively. The decline in healing efficiency
about 20 ꢀC higher than that of N-epoxy. That is due to the with healing/re-fracture cycles might result from incomplete
reduced concentration of exible –Si–O– bonds in 828/diEP.
alignment of the specimens and/or permanently irreversible C–
ON bond caused by radical deactivation at the fractured surfaces.
Compared to the crosslinked polystyrene system,19 the healing
efficiency of 828/diEP decreases much more slowly. It is mainly
because the long so segments of diEP in epoxy not only increase
the collision probability of radicals to reform covalent C–ON
bonds, but also benet interdiffusion of the dangling chains
across the fractured surfaces. Fig. 8 depicts the healing efficiency
of 828/diEP as a function of healing time. With a rise in healing
time, the healing efficiency increases and then levels off when the
time exceeds 2 h. It implies that no more reversible C–ON bonds
are involved at the interface with prolonged time. A case should
be pointed out that although the C–ON healing chemical equi-
librium is fairly fast, 2 h is still needed to reach steady state
healing. It reects the importance of interdiffusion of macro-
molecules and collision probability of radicals between the
fractured surfaces, which act as the control step.
To investigate the effects of C–ON bonds and long so
segments on healing performance of epoxy, controlled experi-
ments were carried out as follows. Firstly, 828/DGDE was prepared
by replacing diEP with DGDE that does not contain any reversible
bond but has long so structure like diEP. It is found that only
ꢃ7% of impact strength can be restored under the same condi-
tions (Table 2), suggesting that the reversible C–ON bonding rather
than the so chains entanglement or reaction of latent functional
groups at the interface makes the decisive contribution to the crack
healing. Secondly, the impact ruptured specimens of 828/diEP
were exposed to air at 90 ꢀC for 1.5 h, and then they were cooled
down to room temperature, recombined together and healed in air
at 90 ꢀC for 1.5 h. Due to the oxygen induced deactivation of
radicals, the reversible reaction of C–ON bonds cannot proceed in
air. As a result, ꢃ6% healing efficiency is detected (Table 2). The
result proves the importance of C–ON bonding once again.
It is worth noting that the self-healing 828/diEP has not yet
achieved full recovery of impact strength owing to the presence of
unmendable 828 structure that acts as the stiff backbone of the
networks. To further improve healing efficiency of the epoxy while
keeping high strength, novel epoxy monomer should be synthe-
sized instead of blending. The current work provides the knowl-
edge of structure–properties relationship for molecular design.
Healing performance
As mentioned in the Experimental, healing capability of the
materials was evaluated by impact test. Fig. 7 shows three
successive macro-repairing processes of an 828/diEP specimen.
The virgin specimen was rstly impacted apart, and then the
broken parts were recombined. Having been healed in argon at
90 ꢀC for 1.5 h, no shape change was observed for the specimen
and the crack line was absent under visual inspection. It can
bear a weight of 200 g (i.e., 15.8 kPa under tensile condition).
Aerwards, the healed specimen was subjected to impact
loading again, and repaired under the same conditions. The
healed version still possessed load bearing ability, and the
cracking line was invisible. The third impact-healing tests also
gave the same results as before. Clearly, the epoxy has acquired
the ability to repeatedly remend as expected. It is worth noting
that the healing temperature is higher than Tg of the material,
but the specimen still keeps its original shape aer rehabilita-
tion, which evidently reects one of the advantages of employ-
ing dynamic reversible C–ON bond as healing group.
In this work, 90 ꢀC is chosen as the healing temperature
because of the following reasons. First of all, the materials should
be heated above Tg to ensure higher molecular chain mobility for
occurrence of the reversible reaction across the crack interface.
Secondly, higher homolysis rate or radicals concentration at
elevated temperature is important for effective healing.
Table
1 Molecular weight between crosslinks, Mc, of the cured
epoxies estimated from repeated DMA scans in Fig. 5 and 6 with data
of the epoxies without C–ON bonds as references
Mc (g molꢂ1
)
Materials
Cycle 1
Cycle 2
Cycle 3
Cycle 4
N-Epoxy
828/diEP
828/DGDE
828
1685
350
558
161
1664
371
—
1729
384
—
1790
—
—
—
—
—
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. A, 2014, 2, 6558–6566 | 6563

View Article Online
Journal of Materials Chemistry A
Paper
Fig. 7 Repeated impact damaging–healing of an 828/diEP specimen.
Table 2 Healing efficiencies of 828/diEP measured during three
successive impact damaging–healing tests with the data of control
specimens as references
Healing efficiency (%)
Materials
1
st repaired
2nd repaired
3rd repaired
828/diEP
828/DGDE
828/diEP (healed in air)
62.0 ꢄ 3.5
6.8 ꢄ 4.2
6.9 ꢄ 3.8
59.5 ꢄ 5.7
7.2 ꢄ 2.5
5.4 ꢄ 3.6
55.1 ꢄ 3.2
6.4 ꢄ 2.7
6.0 ꢄ 2.9
Mechanical properties of the blend of 828/diEP (at 1.5/1
molar ratio) and 828 are listed in Table 3. The relatively poorer
properties of 828/diEP should result from the lower crosslink
Fig. 8 Effect of healing time on healing efficiency of 828/diEP. Healing
was conducted in argon at 90 ꢀC.
6564 | J. Mater. Chem. A, 2014, 2, 6558–6566
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry A
Table 3 Mechanical properties of the cured epoxies
4 T. C. Mauldin and M. R. Kessler, Int. Mater. Rev., 2010, 55,
317–346.
5 M. Q. Zhang and M. Z. Rong, Polym. Chem., 2013, 4, 4878–
4884.
Materials
Properties
828/diEP
828
6 M. Q. Zhang and M. Z. Rong, Acta Polym. Sin., 2012, 11, 1183–
1199.
Flexural strength (MPa)
Flexural modulus (MPa)
Tensile strength (MPa)
Young's modulus (MPa)
71.4 ꢄ 11.7
124.8 ꢄ 8.1
2409.0 ꢄ 62.2
38.8 ꢄ 8.7
2996.6 ꢄ 88.9
73.7 ꢄ 5.9
7 S. Billiet, X. K. D. Hillewaere, R. F. A. Teixeira and F. E. Du
Prez, Macromol. Rapid Commun., 2013, 34, 290–309.
8 M. Q. Zhang, eXPRESS Polym. Lett., 2012, 6, 95.
9 X. Chen, M. A. Dam, K. Ono, A. Mal, H. Shen, S. R. Nutt,
K. Sheran and F. Wudl, Science, 2002, 295, 1698–1702.
10 Y. Amamoto, H. Otsuka, A. Takahara and K. Matyjaszewski,
Adv. Mater., 2012, 24, 3975–3980.
11 C. Yuan, M. Z. Rong, M. Q. Zhang, Z. P. Zhang and
Y. C. Yuan, Chem. Mater., 2011, 23, 5076–5081.
12 J. Ling, M. Z. Rong and M. Q. Zhang, Polymer, 2012, 53, 2691–
2698.
991.8 ꢄ 69.7
380.7 ꢄ 5.9
1347.9 ꢄ 76.5
1103.2 ꢄ 55.5
Impact strength (J mꢂ2
)
density in diEP. Therefore, mechanical properties including
impact strength of 828/diEP are adversely affected despite the
fact that Si–O bonds are known for their exibility. Further
efforts should be made to improve overall properties of this
system, as it indeed has potential to become an upgraded
version of conventional epoxy without self-healing capability.
13 G. Deng, C. Tang, F. Li, H. Jiang and Y. Chen,
Macromolecules, 2010, 43, 1191–1194.
14 M. Pepels, I. Filot, B. Klumperman and H. Goossens, Polym.
Chem., 2013, 4, 4955–4965.
Conclusions
Repeated self-healing ability was successfully imparted to epoxy
by introducing alkoxyamines into the networked structure.
Owing to the synchronous C–ON bond homolysis and radical
recombination of alkoxyamines, damages in the cured version
of the functionalized epoxy can be healed at certain homolysis
temperature. Compared with the remendable polymers based
on DA bonds that healed cracks in a two-step fashion, the
present one completed the rehabilitation by only one step.
Blending alkoxyamine-containing epoxy with conventional
epoxy proved to be an effective way to produce self-healing
epoxy with balanced performance. In the long run, new epoxy
monomer coupled with both healability and mechanical
strength should be developed according to the molecular
structure–properties relationships revealed before.
15 Y. Lu and Z. Guan, J. Am. Chem. Soc., 2012, 134, 14226–
14231.
´
16 P. Cordier, F. Tournilhac, C. Soulie-Ziakovic and L. Leibler,
Nature, 2008, 451, 977–980.
17 R. J. Varley and S. van der Zwaag, Polym. Int., 2010, 59, 1031–
1038.
18 S. Burattini, H. M. Colquhoun, J. D. Fox, D. Friedmann,
B. W. Greenland, P. J. F. Harris, W. Hayes, M. E. Mackay
and S. J. Rowan, Chem. Commun., 2009, 6717–6719.
19 M. Yamaguchi, S. Ono and M. Terano, Mater. Lett., 2007, 61,
1396–1399.
20 Y. Liu and T. Chuo, Polym. Chem., 2013, 4, 2194–2205.
21 Q. Tian, Y. C. Yuan, M. Z. Rong and M. Q. Zhang, J. Mater.
Chem., 2009, 19, 1289–1296.
22 Q. Tian, M. Z. Rong, M. Q. Zhang and Y. C. Yuan, Polym. Int.,
2010, 59, 1339–1345.
23 Q. Tian, M. Z. Rong, M. Q. Zhang and Y. C. Yuan, Polymer,
2010, 51, 1779–1785.
24 N. Bai, K. Saito and G. P. Simon, Polym. Chem., 2013, 4, 724–
730.
25 J. Canadell, H. Goossens and B. Klumperman,
Macromolecules, 2011, 44, 2536–2541.
26 M. Capelot, D. Montarnal, F. Tournilhac and L. Leibler,
J. Am. Chem. Soc., 2012, 134, 7664–7667.
27 A. M. Peterson, R. E. Jensen and G. R. Palmese, Compos. Sci.
Technol., 2011, 71, 586–592.
28 A. M. Peterson, R. E. Jensen and G. R. Palmese, ACS Appl.
Mater. Interfaces, 2013, 5, 815–821.
Acknowledgements
The authors thank the Natural Science Foundation of China
(Grants: 51333008, 51273214, 51073176 and U0634001),
Doctoral Fund of Ministry of Education of China (Grant:
20090171110026), the project of key technological break-
through for emerging industries of strategic importance
(Grants: 2011A091102001, and 2011A091102003), the Science
and Technology Program of Guangdong Province (Grants:
2010B010800021 and 2010A011300004) and the Basic Scientic
Research Foundation in Colleges and Universities of Ministry of
Education of China (Grant: 12lgjc08) for support.
29 R. P. N. Veregin, M. K. Georges, P. M. Kazmaier and
G. K. Hamer, Macromolecules, 1993, 26, 5316–5320.
Notes and references
1 M. Q. Zhang and M. Z. Rong, Self-Healing Polymers and 30 R. P. N. Veregin, M. K. Georges, G. K. Hamer and
Polymer Composites, John Wiley & Sons, Inc., Hoboken, 2011.
2 R. P. Wool, So Matter, 2008, 4, 400–418.
3 S. R. White, N. R. Sottos, P. H. Geubelle, J. S. Moore,
P. M. Kazmaier, Macromolecules, 1995, 28, 4391–4398.
31 S. J. Rowan, S. J. Cantrill, G. R. L. Cousins, J. K. M. Sanders
and J. F. Stoddart, Angew. Chem., Int. Ed., 2002, 114, 938–993.
M. R. Kessler, S. R. Sriram, E. N. Brown and 32 S. J. Rowan, S. J. Cantrill, G. R. L. Cousins, J. K. M. Sanders
S. Viswanathan, Nature, 2001, 409, 794–797.
and J. F. Stoddart, Angew. Chem., Int. Ed., 2002, 41, 898–952.
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. A, 2014, 2, 6558–6566 | 6565

View Article Online
Journal of Materials Chemistry A
Paper
33 M. A. Mannan, A. Ichikawa and Y. Miura, Polymer, 2007, 48, 36 M. He, Polymer Physics, Fudan University Press, Shanghai,
743–749. 1990, p. 334.
34 C. J. Hawker, A. W. Bosman and E. Harth, Chem. Rev., 2001, 37 S. A. Hayes, W. Zhang and F. R. Jones, J. R. Soc. Interface,
101, 3661–3688. 2007, 4, 381–387.
35 N. A. A. Rossi, Q. Wang, K. Amine and R. West, Silicon, 2010, 38 F. Wang, M. Z. Rong and M. Q. Zhang, J. Mater. Chem., 2012,
2, 201–208.
22, 13076–13084.
6566 | J. Mater. Chem. A, 2014, 2, 6558–6566
This journal is © The Royal Society of Chemistry 2014