Self-healing polyurethanes with shape recovery

Article abstract of DOI:10.1002/adfm.201400299

Two new thermoresponsive self-healing polyurethanes (1DA1T and 1.5DA1T) based on the Diels-Alder (DA) reaction between furan and maleimide moieties are developed that use the shape-memory effect to bring crack faces into intimate contact such that healing can take place. Unlike other self-healing polymers, these polymers do not require external forces to close cracks but rather they use the shape-memory effect to autonomously close the crack. Both polyurethanes have a stable polymer structure and comparable mechanical properties to commercial epoxies. A differential scanning calorimeter is employed to check the glass transition temperature of the polymers as well as the DA and retro-DA (rDA) reaction temperatures. These DA and rDA reactions are confirmed with variable-temperature proton nuclear magnetic resonance. Healing efficiency is calculated using a measurement of the failure load from compact tension testing. The results show that the shape-memory effect can replace external forces to close two crack surfaces and the DA reaction can be repeatedly employed to heal the cracks.

Full text of DOI:10.1002/adfm.201400299

www.afm-journal.de
www.MaterialsViews.com
Self-Healing Polyurethanes with Shape Recovery
Yunseon Heo and Henry A. Sodano*
of the DA adduct with differential scan-
ning calorimetry (DSC). The bond energy
Two new thermoresponsive self-healing polyurethanes (1DA1T and 1.5DA1T)
based on the Diels–Alder (DA) reaction between furan and maleimide moie-
ties are developed that use the shape-memory effect to bring crack faces into
intimate contact such that healing can take place. Unlike other self-healing
polymers, these polymers do not require external forces to close cracks but
rather they use the shape-memory effect to autonomously close the crack.
Both polyurethanes have a stable polymer structure and comparable mechan-
ical properties to commercial epoxies. A differential scanning calorimeter is
employed to check the glass transition temperature of the polymers as well
as the DA and retro-DA (rDA) reaction temperatures. These DA and rDA
reactions are confirmed with variable-temperature proton nuclear magnetic
resonance. Healing efficiency is calculated using a measurement of the
failure load from compact tension testing. The results show that the shape-
memory effect can replace external forces to close two crack surfaces and the
DA reaction can be repeatedly employed to heal the cracks.
of the new C–C σ bond in the DA adducts
was 96.2 kJ/mol[3] while other covalent
bond energies are 348 kJ/mol for C–C
bonding, 293 kJ/mol for C–N bonding,
and 389 kJ/mol for N–H bonding.[8] This
implies that covalent bonds are at least
3 to 4 times stronger than the intermolec-
ular C–C σ bond between the DA adducts.
Therefore, cracks are more likely to propa-
gate through new C–C σ bonds than the
other bonds. Consequently, healing is
possible on the crack surfaces through
the recovery of broken DA crosslinks and
the healing efficiency directly correlates
to the crosslink recovery. The tempera-
ture ranges for the DA and rDA reactions
are also advantageous. The DA reaction
occurs at temperatures between room
temperature and approximately 90 °C
without any other external stimulus.[9] Thus, DA polymers can
still counteract the propagation of micro-damage at ambient
temperature. Also, the range for the rDA reaction is approxi-
mately 130 °C which is low enough to avoid thermal degrada-
tion of the polymer upon heating and high enough for many
applications[9] such as a structural composite or electronic
device coating materials.
1. Introduction
The development of self-healing polymers has grown over the
past decade due to their ability to repair damage occurring
before complete failure. Current methods have allowed for the
development of self-healing polymers that are able to achieve
strong healing performance, but the healing process for these
polymers are either not inherently repeatable (i.e., a capsule
based[1] and a vascular based)[2] or require the use an external
pressure to close the crack and bring the surfaces into con-
tact.[3–5] While many promising self-healing methods are under
development, to date, polymers that utilize the Diels-Alder (DA)
reaction have been the most promising since it is highly effi-
cient, simple and can be repeatedly healed through only the
application of heat.
The Diels–Alder (DA) reaction is a thermally reversible reac-
tion between a conjugated diene and a dienophile resulting
in a cyclohexene derivative.[6] The crosslinked DA adducts can
undergo a reverse reaction at higher temperatures which is
known as a retro Diels–Alder (rDA) reaction.[7] Chen et al. deter-
mined the new σ bond energy between a furan and maleimide
When considering self-healing polymers with the DA reac-
tion, the major limitation of the current healing processes is
the need for external forces to bring the two crack surfaces
together. It is believed that the shape memory effect from
shape memory polymers (SMPs) can replace the use of external
force. SMPs are a group of polymers that have the capability of
recovering their permanent shape upon the application of an
external stimulus, even after significant deformation.[10] This is
possible due to what is known as the shape memory effect.
An advantage of SMPs is their versatility since they can be
synthesized to have a wide range of properties depending on its
chemical structures and compositions.[11] According to Lendlein
et al., the shape memory effect arises from the combination of
the structure and morphology of the polymers. In SMPs, two
notable features exist; a soft segment (switching segment) and
a hard segment (netpoint).[12] The temporary shape of an SMP
is fixed in a soft segment that consists of the chains, while the
permanent shape is memorized in a hard segment that can
consist of covalent bonds or intermolecular interactions.[10,12,13]
Recently, Rodriguez et al.[14] and Luo and Mather[15] reported
shape memory assisted self-healing polymers (SMASH). In
both papers, the shape memory properties originated from
the covalent bonds (a hard segment); the network poly(ε-
caprolactone)[14] and the epoxy matrix[15] and the healing mech-
anism was due to the diffusion of PCL which is a thermoplastic
Y. Heo, Dr. H. A. Sodano
Materials Science and Engineering
University of Florida
Gainesville, FL 32611, USA
E-mail: hsodano@ufl.edu
Dr. H. A. Sodano
Mechanical and Aerospace Engineering
University of Florida
Gainesville, FL 32611, USA
DOI: 10.1002/adfm.201400299
©
wileyonlinelibrary.com
Adv. Funct. Mater. 2014,
DOI: 10.1002/adfm.201400299
2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1

www.afm-journal.de
www.MaterialsViews.com
(TEA) and hexamethylene diisocyanate
(HDI) were adapted for the shape memory
properties.[13] The polyurethanes that were
synthesized for testing consisted of 1 mole
of DA monomer (precrosslinked FA and
HEM), 1 mole of TEA, and 2.5 mole of HDI
for 1DA1T polymer and 1.5 mole of DA
monomer, 1 mole of TEA and 3 moles of
HDI for 1.5DA1T. The detailed syntheses
procedures for the HEM which are similar
to the method employed by Heath et al.[17]
and the polymerization method and the
polyurethane structure scheme can be found
in the Expermental Section.
2.2. Thermal Analysis of the Diels–Alder
Reaction
The mechanism for the reversible DA reac-
tion between FA and HEM is shown in
Figure 1a and the temperature ranges of
crosslinking and cleaving in polymers were
studied using DSC as shown in Figure 1b.
From the DSC data, the average transi-
tion temperatures are obtained and shown
in Table 1. The average glass transition
Figure 1. (a) The mechanism for the reversible DA reaction of FA and HEM, and (b) DSC data
for polyurethanes (1DA1T and 1.5DA1T with FA, HEM, TEA, and HDI).
polyester with a relatively low melting point below 60 °C.[16]
Thus, in both works, two different types of polymers; shape
memory type and thermoplastic PCL, were blended and PCL
were melted and flowed to heal the surfaces after closing the
cracks using the shape memory effect. One of the biggest draw-
backs of these SMASHs results due to the low melting point
of PCL, the applicable temperature ranges are thus limited to
below 60 °C. However, in this work, the synthesized polymers
are applicable up to 130 °C before the C–C σ bonds cleave and
are thermally stable up to 250 °C. Also, they combine both self-
healing and shape memory properties such that no external
forces are required to heal the crack. In order to increase sta-
bility while maintaining good mechanical properties, a ther-
moset polyurethane polymer system was chosen and the previ-
ously published shape memory polyurethane by Wilson et al.[13]
was adapted into the developed polymers with the DA self-
healing mechanism to create one complete polymer network.
This has allowed for more in depth testing and characterization
on the self-healing property in this work. All DA monomers
were designed to connect to the polyurethane’s main chain and
as a result, the DA monomers require a hydroxyl functional
group such that they can react with the isocyanate functional
groups to form polyurethane.
temperature (42.4 °C for 1DA1T and 45.6 °C for 1.5DA1T) of
the polymers are shown in the first transition in Figure 1b.
Also, the second and third transitions in Figure 1b indicate the
crosslinking temperature (the average of 97.3 °C for 1DA1T
and 96.6 °C for 1.5DA1T) and the cleaving temperature (the
average of 131.6 °C for 1DA1T and 129.3 °C for 1.5DA1T) and
these transition temperatures in two different polymers are very
similar showing for the reversible DA reaction between FA and
HEM monomers in the polyurethane network. Moreover, they
seem slightly higher when compared to the reported DA and
rDA reaction temperatures (i.e., 93 °C and 120 °C,[4] respec-
tively). This can be attributed to the requirement for additional
thermal energy to overcome the bonding energies between all
the molecules in the polymer such that the molecules become
sufficiently mobile, showing that these polymers network are
very stable.
2.3. Reversibility of the Diels–Alder and Retro Diels–Alder
Reactions
Variable temperature proton nuclear magnetic resonance
(VT NMR) was used to confirm the reversible DA reaction. VT
NMR was a better tool to analyze the actual amount of the DA
reversibility since this reaction is temperature dependent, and
the bonding information, especially of the cleaved bonding,
could be acquired at the reaction temperatures. With traditional
NMR, a specimen after cleaving would need to be immedi-
ately quenched down to a lower temperature and kept at a low
temperature to maintain the cleaved bonds. Otherwise some
crosslinking due to the sample being at an ambient tempera-
ture could occur, meaning that the cleaving NMR data are not
2. Results and Discussion
2.1. Synthesis of Polymers
Commercial furfuryl alcohol (FA) and synthesized N-(2-
hydroxyethyl)-maleimide (HEM) were employed for the self-
healing property through the DA reaction and triethanolamine
©
2
wileyonlinelibrary.com
2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Funct. Mater. 2014,
DOI: 10.1002/adfm.201400299

www.afm-journal.de
www.MaterialsViews.com
Table 1. The average glass transition temperature, crosslinking temperature, and cleaving temperature and their standard deviations for 1DA1T and
1.5DA1T.
1DA1T
42.4
Standard Deviation
1.5DA1T
45.6
Standard Deviation
2.06
2.45
2.66
3.77
1.53
2.48
Glass Transition Temperature [°C]
Crosslinking Temperature [°C]
Cleaving Temperature [°C]
97.3
96.6
131.6
129.3
representative of what actually occurs at cleaving temperatures.
Thus, VT NMR is better for investigating the amount of the
cleaved bonds and their reversibility. Moreover, the crosslinking
and cleaving times can be estimated by repeatedly scanning the
peaks until the reactions are complete. These recorded reaction
times and temperatures were then used to design the healing
process.
through the DA reaction. Then, as shown in Figure 2a and c,
the two peaks disappear as the temperature increases to 140 °C
since the crosslinked HEM and FA monomers become discon-
nected. These four NMR spectra indicate that the DA reaction
occurs repeatedly between the HEM and FA monomers. The
residue peaks of the new σ bonds still appear for the cleaved
cases (shown in Figure 2a and c) even though the specimen
and solvent in the NMR tube was kept at 140 °C until the peaks
did not show any changes. This indicates that the rDA reaction
is not 100% efficient, meaning that the DA and rDA revers-
ibility was not 100% effective.
For the VT NMR testing, the precrosslinked DA monomers
were tested as a model system to evaluate the feasibility of the
crosslinking/cleaving reactions between HEM and FA mono-
mers that have the healing property in the polymers. First, the
sample was kept at 140 °C for 30 min for the cleaving process
(Figure 2a) and then cooled down to 80 °C for 120 min for the
crosslinking DA reaction (Figure 2b). To confirm the repeat-
ability of the reactions, the sample was re-heated to 140 °C and
kept at this temperature for 30 min (Figure 2c) and finally re-
cooled down to 80 °C for 120 min for crosslinking DA reaction
(Figure 2d). As clearly shown in Figure 2b and d, two peaks at
6.52 ppm and 5.08 ppm appear as HEM and FA monomers are
connected at 80 °C. These peaks were identified from the NMR
data of the HEM monomer synthesis comparing the product
(2) and HEM in the Experimental Section and the actual NMR
peaks are also attached in Figure S3 and Figure S5 in the Sup-
porting Information. As furan is removed from product (2),
the 6.52 ppm and 5.08 ppm peaks disappear in the NMR data,
meaning that these peaks indicate the new σ bond was created
2.4. Self-Healing Efficiency
The healing performance shape memory polymers were evalu-
ated through fracture testing. Compact tension (CT) tests
were performed on an Instron 5969 to measure the peak load
to induce fracture in the glassy state of the as cast and healed
specimens. CT test specimens with a pin-pin configuration and
slightly modified dimensions from the ASTM D 5045 standard
were constructed using a computer numerical control (CNC)
mill with the dimensions shown in Figure 3a. The first modi-
fication of the ASTM dimensions was the addition of a resting
hole along the crack path to terminate the propagation in the
middle of the sample. This was needed since the shape memory
effect will arise from the portion of the sample that is left in tact
and the polymers exhibit uncontrolled crack growth in its glassy
state. In addition, the notch length was shortened to enhance the
shape memory effect. Using the initial and post healing max-
imum loads that were acquired through the use of CT testing,
the healing efficiency was calculated using Equation (1).[18]
Max.Loadhealed
Healing Efficiency (%) =
× 100
(1)
Max.Loadinitial
Figure 3c and d show the CT test results for 1DA1T and
1.5DA1T polymer samples. The average healing efficiencies
were calculated with the fracture loads from 6 specimens for
1DA1T and 8 specimens for 1.5DA1T. More 1.5DA1T speci-
mens were tested for the statistical stability since 1.5DA1T
polymer data had larger variations. For the first, second, and
third healing processes, the efficiencies of 79.76%, 69.30%,
and 59.26% for 1DA1T and 84.08%, 84.34%, and 75.89% for
1.5DA1T, respectively, were calculated and the error bars as
shown in Figure 3b indicate the standard deviation of multiple
specimens. The 1.5DA1T polymer efficiencies show the higher
and more stable healing efficiency compared to them of 1DA1T
due to the amount of available furan and maleimide molecules
to heal the crack surfaces.
Figure 2. VT NMR data for HEM and FA precrosslinked monomer to
see the Diels–Alder reversibility; the precrosslinked monomer was (a)
kept at 140 °C for 30 min (cleaving, rDA reaction), (b) cooled down to
80 °C for 120 min (crosslinking, DA reaction), (c) heated up to 140 °C
for 20 min (cleaving, rDA reaction), and (d) cooled to 80 °C for 120 min
(crosslinking, DA reaction)
©
wileyonlinelibrary.com
Adv. Funct. Mater. 2014,
DOI: 10.1002/adfm.201400299
2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
3

www.afm-journal.de
www.MaterialsViews.com
Figure 3. (a) Dimension of a compact tension test specimen (unit in inches) with a thickness of 0.35”, (b) the average healing efficiencies of 1DA1T
and 1.5DA1T for the first, second and third healing (79.76%, 69.30%, and 59.26% for 1DA1T and 84.08%, 84.34%, and 75.89% for 1.5DA1T), (c) 1DA1T
example plot of failure loads as a function of extension (showing healing efficiencies of 79.15%, 73.99%, and 58.77% for the first, second and third
healing, respectively), and (d) 1.5DA1T example plot of failure loads as a function of extension (showing healing efficiencies of 85.59%, 86.12%, and
69.60% for the first, second, and third healing, respectively).
The measured healing efficiency is approximately 80–85%
of the virgin fracture load. There are two primary reasons for
the efficiency falling below 100%, the first is a result of the
efficiency of the re-crosslinking reaction of the DA molecules
and the second being related to the sharpening of the crack tip
due to fracture is the virgin sample. For the crack to completely
heal the polymer network needs to be completely reformed;
however, it is probable that following healing disconnected mol-
ecule pairs exist in the specimen. As shown in the VT NMR
data (Figure 2), the cleaving process (rDA reaction) is not
100% effective in solution and is expected to be lower in the
solid state due to steric effects, which may result in diminished
healing efficiency. A second explanation is that the healing
mechanism is also difficult to accurately measure since the
crack in the healed specimens will always be sharper than the
pre-crack produced in the virgin sample and thus higher stress
intensity should exist following healing. This higher stress
intensity would produce an apparent rather than true reduc-
tion in healing efficiency. The influence of this factor could be
reduced through the formation of a starter crack by creating a
stable crack growth process, however this was challenging with
the glassy polymer and efforts to accomplish this were unsuc-
cessful. Each of these reasons for the incomplete efficiency
gradually reduces the healing efficiency with increasing healing
cycles. However, both the 1DA1T and 1.5DA1T polymers pos-
sess superior healing efficiencies when compared (using
fracture strength) to similar polymers calculated the and seen
in the literature, including 2MEP4F[4] and polymer 400[19] with
reported average first healing efficiencies of 80% and 46%,
respectively.
2.5. Healing Process
The temperature profile for the healing process was designed to
have enough time to crosslink and cleave based on the VT NMR
data. After a crack was formed in the specimens and without the
use of an external force to close the crack, the CT specimens were
placed face-down in a vacuum oven under a nitrogen environ-
ment such that no gravitational forces contributed to the crack
closure. The oven was slowly heated up to 135 °C over 2.5 h and
kept at this temperature for 2 h. Then, the specimens were cooled
down to 90 °C over 1 h and kept at this temperature for 2 h and
finally cooled to 70 °C over 1 h and kept again at this temperature
for another 2 h. The specimens were heated up to a temperature
above the cleaving temperature so that the unbroken bonds could
separate into two individual molecules. This allowed for a greater
chance to heal the cracks when undergoing the DA reaction
since polymer cannot return the crack faces into its exact prior
condition to allow the formation of identical bonds. Also, 2 h
rest steps were taken at 90 °C and 70 °C since the gradual tem-
perature changes ensured that the specimens were cooled
©
4
wileyonlinelibrary.com
2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Funct. Mater. 2014,
DOI: 10.1002/adfm.201400299

www.afm-journal.de
www.MaterialsViews.com
Figure 4. Image of a specimen after cracking (left) and after the thermal
healing process without the use of external forces (right).
Figure 5. Images of 1.5DA1T films show a) permanent shape (48.6 mm
× 8.73 mm × 1.08 mm), b) temporary shape after deforming and fixing,
c) permanent shape after recovering, d) permanent shape (76.5
mm × 8.6 mm × 1.06 mm), e) spiral temporary shape, and f) recovered
permanent shape
uniformly. Figure 4 shows a cracked specimen before (left) and
the same specimen after the thermal healing process (right).
2.6. Shape Memory Property of the Polymer
and the Young’s modulus from 12 specimens were 76.8 MPa
and 2.86 GPa for 1DA1T and 87.3 MPa and 2.48 GPa for
1.5DA1T, respectively. They are comparable to composites
grade epoxy but the ultimate tensile strain of 1DA1T is much
larger than epoxy. This indicates that 1DA1T and 1.5DA1T
polymers still exhibit elastomeric behavior as a result of the
TEA and HDI polyurethanes even after adding DA crosslinking
to the polymer. For fracture toughness, a total of 6 specimens
of each polymer were tested and 1.43 MPa for 1DA1T and
1.06 MPa were obtained. Both average fracture toughness are
more than that of an epoxy. In Table 2, all data are listed and
the standard deviations of each property are also shown. These
data were compared to composite grade epoxy since the devel-
oped polyurethane was studied as a possible substitute to epoxy
fractions in structural composites, also generally known as
polymer matrix composites.
Shape memory polyurethanes were adapted from the previ-
ously published work by Wilson et al. and their shape memory
properties were confirmed by the cyclic tensile testing.[13] How-
ever, in order to evaluate the shape memory property of 1DA1T
and 1.5DA1T, the recovery capability tests in two different
directions under free-strain conditions were conducted and
the representative results for 1.5DA1T are shown in Figure 5.
Figure 5a shows the permanent shape (dimensions of 48.6 mm ×
8.73 mm × 1.08 mm) and Figure 5b presents the temporary
shape after heating the film for 100 °C for 5 min, stretching
the film, and then cooling it down to room temperature in
order to fix the shape. Finally, Figure 5c displays the recovered
permanent shape after heating the sample to 100 °C, which is
above the glass transition temperature. An additional test was
performed to show the spiral shape memory property of the
polymer. Figure 5d shows the permanent shape (dimensions
of 76.5 mm × 8.6 mm × 1.06 mm) and Figure 5e exhibits the
coiled temporary shape at room temperature after heating the
film at 100 °C for 5 mi and fixing it. Lastly, Figure 5f illustrates
the permanent shape of the polymer after recovery. Through
these tests, it can be concluded that the developed self-healing
polyurethanes contain shape recovery and this shape recovery
force can be used to close the crack with the application of heat
during the healing process.
3. Conclusion
This work has developed polymers combining two functionali-
ties, self-healing polymers and shape memory polymers. This
eliminates the need to apply an external force to close a crack
during the healing process. The polyurethane backbone cre-
ated excellent mechanical properties comparable to epoxy while
allowing the DA monomers to link to the main polymer chains
and achieve healing as well as its shape memory ability. The
healing efficiencies calculated using maximum loads from CT
testing confirmed that repeated healing without external pres-
sure was possible with only slight loss efficiency.
2.7. Mechanical Properties of the 1DA1T and 1.5DA1T Polymers
Finally, the mechanical properties were investigated for the self-
healing polyurethanes. The average ultimate tensile stresses
Table 2. Comparison of mechanical properties between 1DA1T, 1.5DA1T, and epoxy.
Property
1DA1T
76.8
Standard Deviation
1.5DA1T
87.8
Standard Deviation
Epoxy
35–13020
2–521
Ultimate tensile strength [MPa]
Ultimate tensile strain [%]
Young’s modulus [GPa]
Glass transition temperature [°C]
Fracture toughness [MPa·√m]
8.2
7.6
2.8
30.5
23.5
0.25
2.06
0.14
5.4
2.37
2.57
0.17
3.77
0.04
2–620
42.4
45.6
50–25020
0.6522a)
1.43
1.06
a)Epoxy resin (L135i) with an amine hardener (H137i).
©
wileyonlinelibrary.com
Adv. Funct. Mater. 2014,
DOI: 10.1002/adfm.201400299
2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
5

www.afm-journal.de
www.MaterialsViews.com
Scheme 1. Three steps synthesis of N-(2-hydroxyethyl) maleimide.
DMSO-d6) δ 6.97 (s, 2H), 4.80 (br, 1H), 3.44 (br, 4H) ppm; 13C NMR
(300 Hz, DMSO-d6) δ 171.53, 134.88, 58.37, 40.38 ppm.
4. Experimental Section
Pre-crosslinking DA Adducts (DA): To make sure that all DA adducts are
completely crosslinked before cracking, HEM and furfuryl alcohol (FA)
are precrosslinked by using the DA reaction before the polymerization
step as shown in Scheme 2. For HEM and FA, each monomer is added
into toluene with a 1:1 molar ratio and then the mixtures are heated
to 75 °C under stirring for 12 h, then a light yellowish precipitation is
observed. The product is then vacuum filtered and washed twice with
ether (yield 90.09%). mp 109 °C; 1H NMR (300 Hz, DMSO-d6) δ 6.52
(br, 2H), 5.07 (s, 1H), 4.92 (br, 1H), 4.75 (br, 1H), 4.01 (d, 1H), 3.71 (d,
1H), 3.41 (s, 4H), 3.03 (d, 1H), 2.89 (d, 1H) ppm; 13C NMR (300 Hz,
DMSO-d6) δ 176.85, 175.40, 138.53, 136.91, 92.07, 80.63, 59.40, 57.74,
50.39, 48.21, 41.00 ppm.
Polymers: Maleimide moieties usually are solid at room temperature
and have high melting points compared to the rest of the monomers
(i.e., FA, TEA, and HDI). A bulk or film polymerization method that
used by other researchers[3,4,23] is to mix and dissolve the monomers
into a solvent and then let the solvent evaporate slowly to acquire the
polymer products. However, if a solvent is not completely removed
during this process, the material properties of a polymer could end up
changing. In order to avoid this potential problem, in this work, the
solid DA monomers are heated and melt at 105 °C and then mixed the
hot resin monomer with the rest of monomers (TEA and HDI) at the
same temperature. Then, the mixtures are poured into the Teflon mold
and place the mold into the preheated oven to 100 °C under nitrogen
condition. The temperature is increased to 125 °C and stayed for 2 h and
then deceased to 90 °C over 1 h and kept for 2 h. After that, the oven is
cooled down to 70 °C over 1 h and kept for 2 h and finally the polymers
are cooled down in oven slowly to room temperature. The polymerization
and cooling temperature profile is similar to the healing process that is
designed to maximize the number of crosslinked condition in polymer
when the reaction is done. This polymerization process with the HEM
monomer is shown in detail (see Scheme 3 ).
4.1. Syntheses
Materials: Furfuryl alcohol (Acros Organics), hexamethylene
diisocyanate (Acros Organics), and triethanolamine (Sigma-Aldrich) are
used as received. Maleimide monomers, N-(2-hydroxyethyl)-maleimide
(HEM), is prepared with Diels–Alder and retro Diels–Alder cycloaddition
through a three steps process as shown in Scheme 1 . Heath et al.’s
method[17] is mainly used but modified slightly. For the HEM synthesis,
maleic anhydride (Alfa Aesar), furan (Acros Organics), ethanolamine
(Acros Organics), methanol (Fisher Chemical) and toluene (Fisher
Chemical) were used as received.
4,10-Dioxatricyclo[5.2.1.02,6]dec-8-ene-3,5-dione (1) : Maleic anhydride
(30 g, 305.94 mmol) and furan (21 g, 308.51 mmol) are mixed and
stirred in toluene (300 mL) at room temperature for 24 h. Furan is added
slightly more than maleic anhydride to make sure all maleic anhydride
are crosslinked with furan. After 24 hours, a fine white solid powder
is precipitated and the mixture is vacuum filtered and dried to obtain
the product (1). Unreacted furan is filtered during the vacuum filtration
process and a further purification step is not necessary (yield 85.6%).
mp 123 °C; 1H NMR (300 Hz, DMSO-d6) δ 6.59 (s, 2H), 5.36 (s, 2H),
3.32 (s, 2H) ppm; 13C NMR (300 Hz, DMSO-d6) δ 171.98, 137.30, 82.11,
49.53 ppm.
4-(2-Hydroxy-ethyl)-10-oxa-4-aza-tricyclo[5.2.1.02,6]dec-8-ene-3,5-
dione (2) : 4,10-Dioxatricyclo[5.2.1.02,6]dec-8-ene-3,5-dione (1) (30 g,
180.58 mmol) and methanol (60 mL) are added to a flask and the mixture
is kept in an ice bath for 15 min. Ethanolamine (11.04 g, 180.75 mmol)
is mixed well with methanol (5 mL) and this mixture is slowly added
over 30 min to the flask in the ice bath under stirring. The final mixture
is then kept in the ice bath for 30 min and then at room temperature
for an additional 30 min. The product (1) is dissolved completely into
methanol while adding ethanolamine. Then, the solution is refluxed for
24 h. Afterwards, the solution is cooled down and a slightly yellowish
solid powder is crystallized. This powder is then vacuum filtered and
dried (yield 46.6%). mp 142 °C; 1H NMR (300 Hz, DMSO-d6) δ 6.53
(s, 2H), 5.10 (s, 2H), 4.74 (br, 1H), 3.40 (br, 4H), 2.90 (s, 2H) ppm;
13C NMR (300 Hz, DMSO-d6) δ 176.93, 136.89, 80.73, 57.73, 47.57,
41.06 ppm.
N-(2-Hydroxyethyl)-maleimide (HEM): 4-(2-Hydroxyethyl)-10-oxa-
4-aza-tricyclo[5.2.1.02,6]dec-8-ene-3,5-dione (2) (30 g) is refluxed in
toluene (180 mL) for over 24 h until it is completely cleaved. To confirm
the cleaving, NMR was used and 6.520 ppm and 5.104 ppm should
disappear once furan is removed from 4-(2-Hydroxyethyl)-10-oxa-4-
aza-tricyclo[5.2.1.02,6]dec-8-ene-3,5-dione (2). The solution is cooled
down and white powders are crystallized and they can be collected
by the vacuum filtration. (yield 83.02%); mp 75 °C; 1H NMR (300 Hz,
4.2. Thermal Analysis of the Polymers
A thermogarvimeteric anaylsis (TGA), TA Q50, was utilized to check the
polymer decomposition temperature. The TGA data also can be used to
anticipate the maximum temperature for the following DSC test. Each
sample with a mass of approximately 15 mg was placed in a platinum
pan first. The temperature was increased from 20 °C up to 600 °C with
a heating rate of 20 °C/min under a nitrogen atmosphere. The balance
purge flow and sample purge flow of nitrogen were set at 40 mL/min
and 60 mL/min, respectively. TGA data can be seen in Figure S9 in the
Supporting Information.
Scheme 2. A schematic of the precrosslinked FA and HEM.
©
6
wileyonlinelibrary.com
2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Funct. Mater. 2014,
DOI: 10.1002/adfm.201400299

www.afm-journal.de
www.MaterialsViews.com
Scheme 3. A schematic of the polymerization reaction with DA, TEA, and HDI monomers.
A differential scanning calorimetry (DSC), TA Q20, was used for
thermal analyses such as the glass transition, crosslinking, and cleaving
temperatures of each polymer. For testing, each polymer sample was
preheated at 130 °C for 2 h and then quenched in an ice bath. This
procedure was necessary to promote cleaved bonds in the polymer
structures so that the crosslinking transition which is observed at lower
temperature can be shown clearly. Then, samples that weighted between
10 mg and 15 mg were placed in standard Tzero aluminium pan. Two
aluminium pans with and without samples were put on each sensor
platform. The temperature ranges for DSC testing were from −10 °C
to 160 °C under the nitrogen atmosphere and the sample purge flow
rate of nitrogen was set at 50 mL/min. In more detail, a sample was
cooled down to an equilibration temperature of −10 °C for 2 utes and
linearly heated up to 160 °C with a ramping rate of 10 °C/min. After
data acquisition, all data were analyzed by using “TA Universal Analysis”
software offered by the TA Company. The average glass transition
temperatures, crosslinking temperatures and cleaving temperatures of
6 specimens of 1DA1T and 1.5DA1T polymers are reported in Table 1.
at the resting hole. Under the ASTM D 5045 standard, at least five
specimens are required and averaged to obtain the values. Thus, 6
specimens were tested for 1DA1T and 8 specimens were examined
for 1.5DA1T. More 1.5DA1T specimens were test to achieve better
statistical stability since the initial 6 values were varied. The standard
deviations of each polymer set were also calculated and shown as error
bars in Figure 3b.
4.5. Shape Memory Characterization
In addition to the self-healing properties of the polyurethane, the
shape memory properties were also confirmed. The tests to verify
the recovery capability in two different directions under free-strain
were undertaken to evaluate the shape memory property. One newly
synthesized film with dimensions of 48.6 mm × 8.73 mm × 1.08 mm
was heated at 100 °C for 5 min, stretched in longitudinal direction, and
cooled down to room temperature in order to fix the temporary shape.
Then, the film was re-heated to 100 °C for 5 min to let the permanent
shape recover. Another film with dimensions of 76.5 mm × 8.6 mm ×
1.06 mm was used to show the spiral shape memory property. Similar
to the other sample, it was heated to 100 °C for 5 min, coiled, and
cooled down to room temperature to fix the temporary shape. Then,
to recover the permanent shape the film was re-heated to 100 °C
for 5 min.
4.3. Reversibility of Self-Healing Mechanism
In order to study the reversibility of the DA reaction and retro DA
reaction, variable temperature proton nuclear magnetic resonance
(VT NMR) was utilized to observe new σ bond peaks (i.e., 6.52 ppm
and 5.18 ppm) in the chemical shifts. An Inova 2 500 MHz NMR
instrument was used and the deuterated dimethyl sulfoxide (DMSO-
d6) was used for all NMR specimens as a solvent. A specimen was
heated up with an internal heating function of the Inova 2 instrument
and cooled down with an external liquid nitrogen cooling system in air.
Precrosslinked HEM and FA monomers in dimethyl sulfoxide-d6 solvent
were used and all spectra were acquired at different temperatures.
The samples were kept at 140 °C for 30 min for the cleaving process
and then cooled down to 80 °C for 120 min for the crosslinking DA
reaction. Then, the same temperature and time profiles were used again
to confirm the reversibility of crosslinking and cleaving reactions. The
temperatures, 80 °C and 140 °C, were chosen since it was believed that
these temperatures were in the range of the crosslinking and cleaving
reactions, and the peaks were scanned every 10 min to study the
reaction kinetics. These time and temperature data were used later to
design the healing process.
4.6. Mechanical Properties
The dimensions of the ASTM 638 Type V sample were tested and a
total of 12 flat specimens of each 1DA1T and 1.5DA1T were prepared
to measure the ultimate tensile strength, strain, and Young’s modulus.
All tensile tests were done at room temperature and the opening rate of
5 mm/min was used. In addition, the maximum loads at fracture from a
compact tension test were obtained to calculate the fracture toughness.
The test conditions, procedures and dimensions of the specimen
for the CT test were satisfied by the ASTM D 5045 standards. Each of
6 specimens was tested at room temperature on an Instron 5969 at an
opening rate of 5 mm/min.
Supporting Information
4.4. Healing Efficiency
Supporting Information is available from the Wiley Online Library or
from the author.
In order to calculate the healing efficiency, the fracture load was
measured with compact tension (CT) testing. This was conducted using
an Instron 5969 at room temperature. The ASTM D 5045 standard was
followed with slight modifications in dimensions as shown in Figure 3a
and the constant opening rates for all tests before and after each healing
process were used (3 mm/min for 1DA1T and 1 mm/min for 1.5DA1T).
The opening rates were adjusted to initiate crack propagation without
deforming the samples, avoid crazing and to ensure the crack stops
Acknowledgements
The authors gratefully acknowledge the support from the Army
Research Office (Award #: W911NF-12-1-0014) and would also like to
©
wileyonlinelibrary.com
Adv. Funct. Mater. 2014,
DOI: 10.1002/adfm.201400299
2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
7

www.afm-journal.de
www.MaterialsViews.com
thank Mr. M. H. Malakooti for all discussion and help related to the
mechanical testing.
[8] D. L. Reger, S. R. Goode, D. W. Ball, in Chemistry: Principles and
Practice, Thomson Brooks/Cole, Belmont, USA 2009.
[9] T. C. Mauldin, M. R. Kessler, Int. Mater. Rev. 2010, 55, 317.
[10] A. Lendlein, S. Kelch, Angew. Chem. Int. Edit. 2002, 41, 2034.
[11] A. Lendlein, A. M. Schmidt, R. Langer, Proc. Natl. Acad. Sci. USA
2001, 98, 842.
Received: January 27, 2014
Revised: March 27, 2014
Published online:
[12] M. Behl, A. Lendlein, Mater. Today 2007, 10, 20.
[13] T. S. Wilson, J. P. Bearinger, J. L. Herberg, I. J. E. Marion,
W. J. Wright, C. L. Evans, D. J. Maitland, J. Appl. Polym. Sci. 2007,
106, 540.
[14] E. D. Rodriguez, X. Luo, P. T. Mather, ACS Appl. Mater. Interfaces
2011, 3, 152.
[15] X. Luo, P. T. Mather, ACS Macro Lett. 2013, 2, 152.
[16] C. Liu, H. Qin, P. T. Mather, J. Mater. Chem. 2007, 17, 1543.
[17] W. H. Heath, F. Palmieri, J. R. Adams, B. K. Long, J. Chute,
T. W. Holcombe, S. Zieren, M. J. Truitt, J. L. White, C. G. Willson,
Macromolecules 2008, 41, 719.
[1] a) S. R. White, N. Sottos, P. Geubelle, J. Moore, M. R. Kessler,
S. Sriram, E. Brown, S. Viswanathan, Nature 2001, 409, 794;
b) E. N. Brown, N. Sottos, S. R. White, Exp. Mech. 2002, 42, 372;
c) B. Blaiszik, N. Sottos, S. White, Compos. Sci. Technol. 2008,
68, 978; d) S. H. Cho, S. R. White, P. V. Braun, Adv. Mater. 2009,
21, 645; e) B. Blaiszik, S. Kramer, S. Olugebefola, J. S. Moore,
N. R. Sottos, S. R. White, Ann. Rev. Mater. Res. 2010, 40, 179.
[2] a) C. Dry, Compos. Struct. 1996, 35, 263; b) R. Trask, G. Williams,
I. Bond, J. R. Soc. Interface 2007, 4, 363; c) G. Williams, R. Trask,
I. Bond, Compos. Part A: Appl. S. 2007, 38, 1525; d) J. W. Pang,
I. P. Bond, Compos. Sci. Technol. 2005, 65, 1791; e) M. Keller,
N. Sottos, Exp. Mech. 2006, 46, 725; f) S. Kim, S. Lorente, A. Bejan,
Jpn. J. Appl. Phys. 2006, 100, 063525; g) K. Wang, S. Lorente,
A. Bejan, J. Phys. D. Appl. Phys. 2006, 39, 3086.
[18] S. Burattini, B. W. Greenland, D. Chappell, H. M. Colquhoun,
W. Hayes, Chem. Soc. Rev. 2010, 39, 1973.
[19] E. B. Murphy, E. Bolanos, C. Schaffner-Hamann, F. Wudl,
S. R. Nutt, M. L. Auad, Macromolecules 2008, 41, 5203.
[20] M. W. Hyer, S. R. White, in Stress Analysis of Fiber-Reinforced Com-
posite Materials, McGraw-Hill Companies, Inc., New York, USA
1998, p.26.
[3] X. Chen, M. A. Dam, K. Ono, A. Mal, H. Shen, S. R. Nutt, K. Sheran,
F. Wudl, Science 2002, 295, 1698.
[21] I. M. Daniel, O. Ishai, I. M. Daniel, I. Daniel, in Engineering
Mechanics of Composite Materials, 1st Ed, Oxford University Press,
New York, USA 2006.
[22] F. H. Gojny, M. H. G. Wichmann, B. Fiedler, K. Schulte, Compos.
Sci. Technol. 2005, 65, 2300.
[23] a) T. A. Plasited, S. Nemat-Nasser, Acta Mater. 2007, 55, 5684;
b) J. Zhang, Y. Niu, C. Huang, L. Xiao, Z. Chen, K. Yang, Y. Wang,
Polym. Chem. 2012, 3, 1390; c) H. Weizman, C. Nielsen,
O. S. Weizman, S. Nemat-Nasser, J. Chem. Educ. 2011, 88, 1137;
d) N. Yoshie, S. Saito, N. Oya, Polymer 2011, 52, 6074.
[4] X. Chen, F. Wudl, A. K. Mal, H. Shen, S. R. Nutt, Macromolecules
2003, 36, 1802.
[5] a) R. Gheneim, C. Perez-Berumen, A. Gandini, Macromolecules
2002, 35, 7246; b) A. Amalin Kavitha, N. K. Singha, J. Polym. Sci. A1.
2007, 45, 4441; c) Y. Zhang, A. A. Broekhuis, F. Picchioni, Macro-
molecules 2009, 42, 1906; d) C. Toncelli, D. C. De Reus, F. Picchioni,
A. A. Broekhuis, Macromol. Chem. Phys. 2012, 213, 157.
[6] O. Diels, K. Alder, Liebigs. Ann. Chem. 1928, 460, 98.
[7] B. Rickborn, in Organic Reactions, John Wiley
New York, USA 1998.
& Sons, Inc.,
©
8
wileyonlinelibrary.com
2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Funct. Mater. 2014,
DOI: 10.1002/adfm.201400299