Preparation and properties of bio-based epoxy networks derived from isosorbide diglycidyl ether

Article abstract of DOI:10.1016/j.polymer.2011.06.001

Two bio-based epoxy prepolymers were synthesized from isosorbide, a carbohydrate-based C6 building block, using two different synthetic routes. The chemical structures of the bio-based epoxy prepolymers were analyzed by SEC, ESI-TOF MS, FTIR, ¹H NMR and ¹³C NMR analysis. The resulting epoxy prepolymers differ by the molar mass distribution, one consists of the pure epoxy monomer whereas the other exhibits various oligomeric species. A traditional petroleum-based epoxy prepolymer, DGEBA, which has similar epoxy equivalent weight, was also used in this study for comparison. Gelation and crosslinking reactions of the two bio-based epoxy precursors with an amino hardener, isophorone diamine, were studied using rheological measurements and differential scanning calorimetry (DSC) respectively; the effect of the stoichiometric ratio n_ah/n_e was investigated. Structures of the epoxy networks were evaluated using dynamic mechanical analysis (DMA) and thermo gravimetric analysis (TGA).

Full text of DOI:10.1016/j.polymer.2011.06.001

Polymer 52 (2011) 3611e3620
Contents lists available at ScienceDirect
Polymer
journal homepage: www.elsevier.com/locate/polymer
Preparation and properties of bio-based epoxy networks derived from isosorbide
diglycidyl ether
Marie Chrysanthos b, Jocelyne Galya b,*, Jean-Pierre Pascault a
a
,
,
,b
a
Université de Lyon, F-69003, France
INSA-Lyon, Ingénierie des Matériaux Polymères, IMP, UMR5223, F-69621 Villeurbanne, France
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 23 March 2011
Received in revised form
Two bio-based epoxy prepolymers were synthesized from isosorbide, a carbohydrate-based C6 building
block, using two different synthetic routes. The chemical structures of the bio-based epoxy prepolymers
1
13
were analyzed by SEC, ESI-TOF MS, FTIR, H NMR and C NMR analysis. The resulting epoxy prepolymers
differ by the molar mass distribution, one consists of the pure epoxy monomer whereas the other
exhibits various oligomeric species. A traditional petroleum-based epoxy prepolymer, DGEBA, which has
similar epoxy equivalent weight, was also used in this study for comparison. Gelation and crosslinking
reactions of the two bio-based epoxy precursors with an amino hardener, isophorone diamine, were
studied using rheological measurements and differential scanning calorimetry (DSC) respectively; the
27 May 2011
Accepted 1 June 2011
Available online 12 June 2011
Keywords:
Sugar-based epoxies
Isosorbide
effect of the stoichiometric ratio nah/n
uated using dynamic mechanical analysis (DMA) and thermo gravimetric analysis (TGA).
Ó 2011 Elsevier Ltd. All rights reserved.
e
was investigated. Structures of the epoxy networks were eval-
Epoxy-amine networks
1
. Introduction
Until now, numerous bio-based epoxy resins deriving from
vegetable oils, such as soybean oil, linseed oil and castor oil have
been studied [5e10]. These macromonomers have long aliphatic
chains; therefore they cannot lead to high performance networks
and compete with DGEBA based networks. They are used in epoxy
formulations as a toughening agent, that is, to reduce the inherent
brittleness of epoxy networks. Some new bio-based epoxy and/or
hardeners are prepared from cardanol which gives higher network
properties [11,12].
The objective of our work is to develop novel bio-based reactive
systems suitable for high performance composite materials.
DGEBA is derived from bisphenol A and epichlorohydrin. An easy
way to obtain a partially bio-based DGEBA is to use epichlorohydrin
from bio-based glycerol, an abundant and inexpensive polyol. Such
bio-based epichlorohydrin is commercially available. However the
molar mass of DGEBA is dominated by bisphenol A, epichlorohydrin
accounts only for 20% of the molar mass of DGEBA. So a challenge to
obtain a fully bio-based epoxy prepolymer is to replace bisphenol A
by a bio-based polyol. Another interest for replacing bisphenol A is
that bisphenol A has been known to have estrogenic properties [13].
The chemical bonds that link bisphenol A in polymer structures are
not completely stable and the polymer may release, with time,
a small amount of bisphenol A, toxic to living organisms.
In recent years, bio-based polymers derived from renewable
resources [1,2] have become increasingly important as sustainable
and eco-efficient products which can replace the products based on
petrochemical-derived stocks. Of all polymers produced, the main
part is thermoplastics while the minor part is thermosets, 82 and
18% respectively [3]. This is one important reason why research and
development has mainly concerned bio-based thermoplastics such
as linear polyesters, polyamides, polyurethanes, etc. Thermosets
prepared from renewable resources have been the subject of more
limited investigations [2].
Epoxy networks are amongst the most popular thermosetting
polymers which are used in various fields: coatings, adhesives,
laminates, electrical castings, etc. because they exhibit a broad
spectrum of properties, through selection of epoxy prepolymer and
curing agent composition [4]. The diglycidyl ether of bisphenol A,
DGEBA, is by far the most commonly used starting monomer to
formulate epoxy networks: it represents 75% of all epoxy precur-
sors [3]. Therefore there is a huge interest in developing a bio-based
epoxy oligomer able to replace the classical DGEBA and to lead to
high performance materials.
Investigations were done on the use of natural polysaccharides
in replacement of bisphenol A in the synthesis of bio-based epoxy
prepolymers. Sorbitol and maltytol were converted in multifunc-
tional epoxy monomers which were used in combination with
*
Corresponding author. INSA-Lyon, Ingénierie des Matériaux Polymères, IMP,
UMR5223, F-69621 Villeurbanne, France. Tel.: þ33 4 72 43 83 81; fax: þ33 4 72 43
8
5 27.
E-mail address: jocelyne.galy@insa-lyon.fr (J. Galy).
0
032-3861/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.polymer.2011.06.001

3612
M. Chrysanthos et al. / Polymer 52 (2011) 3611e3620
DETA (diethylene triamine) to produce bio-based networks [14].
Glycerol is also an important bio-based raw material and epoxy
precursors such as glycerol polyglycidyl ether (GPE) and poly-
glycerol polyglycidyl ether (PGPE) are industrially available and
inexpensive. Such products have been used in textile and paper as
processing agents and as reactive diluents. However, in a recent
study these epoxy monomers were reacted with a bio-based curing
agent, 3-poly(L-lysine) to produce fully bio-based epoxy networks,
as well as nanocomposites based on montmorillonite [15]. Glucose
can be readily hydrogenated and yield sorbitol. Sorbitol poly-
glycidyl ether (SPE) is also a commercial multifunctional epoxy
monomer used for similar applications as GPE and PGPE; SPE and
GPE have been associated to tannic acid to obtain bio-based
networks and biocomposites with microfibrillated cellulose [16].
Synthesis of polyglycidyl ether derived from 1,4:3,6-
dianhydrohexitols is well described in patent literature [17e19].
The double dehydration of sorbitol yields isosorbide (1,4:3,6-
dianhydro-D-sorbitol (DAS)), commercially available in industrial
quantities. The present study describes bio-based epoxy networks
derived from this molecule. It was epoxidized using two different
routes. The resulting isosorbide diglycidyl ether (or Diglycidyl Ether
of DAS, DGEDAS) was characterized and bio-based epoxy networks
containing this isosorbide diglycidyl ether were prepared using one
amine curing agent, isophorone diamine (IPD). Their structures
were evaluated and compared to a conventional epoxy network
based on DGEBA cured by IPD.
Scheme 1. Synthetic route of isosorbide diglycidyl ether via epichlorohydrin.
The most convenient route is that of the reaction of the isosorbide
directly with epichlorohydrin. The synthesis is well described in the
literature [19,20] and the synthetic route is presented in Scheme 1.
It is based on the traditional and industrial route to produce DGEBA.
A flask fitted with a Dean-Stark water-separator tube was
charged with isosorbide (4.0 mole) and epichlorohydrin
(
40.0 mole). The mixture was stirred under nitrogen flow; the
ꢀ
reaction flask temperature was 115 C. When the reaction was
steadily refluxing, an aqueous solution of sodium hydroxide
(
8.0 mole) was added drop by drop to the flask. Gradually water
began to appear as an upper layer in the Dean-Stark tube and the
reaction was left to proceed steadily. The whole addition step took
about 12 h. Water and epichlorohydrin were drained off under
ꢀ
ꢀ
reduced pressure (230 mbar at 70e80 C and 96 mbar at 65 C
2
. Experimental section
respectively).
n
The diglycidyl ether of isosorbide thus obtained, DGEDAS , was
2
.1. Materials
a viscous liquid with an epoxide equivalent weight of 184 g/eq, and
a hydroxyl number IOH ¼ 105 mg/g KOH as determined by titrim-
etry. The yield of the reaction was 96e98%. We will see that this
synthesis of the diglycidyl ether of isosorbide does not give pure
diglycidyl ethers since oligomers may be formed by concomitant
reaction of the epoxy function with unreacted isosorbide.
Another synthetic route, based on the method of sucrose
epoxydation developed by Sachinvala et al. [21] was then per-
formed on isosorbide by Feng et al. [20]. As compared to the first
synthetic route discussed above, this one is not industrially
exploited. It involved a two step sequence via allylic derivative
Fig. 1 shows the structure of the reagents used in this study. The
conventional petroleum-based epoxy monomer used was a digly-
cidyl ether of bisphenol A (DGEBA) supplied by Huntsman (Araldite
5
085), with an epoxide equivalent weight (EEW) of 180 g/eq.
The cycloaliphatic diamine curing agent used was isophorone
diamine (IPD) supplied by Aldrich; the amine equivalent weight
AEW) is42 g/eq. Isosorbidewith a high purity, commercialized under
thetrade namePolysorbPhasbeenkindlysuppliedbyRoquette Frères
Lestrem, France). Epichlorohydrin and sodium hydroxide were
purchased from Aldrich. All chemicals were used as received.
(
(
(
Scheme 2). Diallyl isosorbide ether was prepared by heating the
ꢀ
isosorbide (0.4 mole) with allyl bromide (0.88 mole) at 65 C under
2.2. Synthesis of isosorbide diglycidyl ether
drop by drop addition of
a solution of sodium hydroxide
(
0.88 mole). The reaction was stopped and cooled at room
In the present work, the synthesis of the diglycidyl ether of
isosorbide (DGEDAS) was performed using two different routes.
temperature before washing it with dichloromethane. Freshly
prepared diallyl isosorbide was then added slowly to a solution of
peracid in dichloromethane. The reaction was stirred at a temper-
ꢀ
ature of 0e10 C for 94 h. The cold solution was filtrated and
washed with a solution of 10% sodium bisulfite followed by satu-
rated sodium bicarbonate and distilled water. The organic layer was
Fig. 1. Structure of the reagents used in this study.
Scheme 2. Synthetic route of isosorbide diglycidyl ether via allylic derivative.

M. Chrysanthos et al. / Polymer 52 (2011) 3611e3620
3613
temperature, in order to form a homogenous system. The obtained
ꢀ
mixture was poured into a PTFE coated mould and cured at 80 C
ꢀ
for 1 h followed by 2 h at 180 C. Networks were stored in dry
conditions in order to avoid absorption of atmospheric moisture.
2.4. Measurements
Size Exclusion chromatography (SEC) was used to measure the
molar mass distribution of the epoxy prepolymers. Tetrahydrofuran
THF) was used as the eluent at a flow rate of 1 mL/min. The
(
separation was done on a set of three columns (Waters HR0.5, HR1
and HR2), with a peak detection based on the signal of a refractive
index detector (Shimadzu RID-10A).
The positive-ion electro-spray ionization time-of-flight (ESI-
TOF) mass spectra were acquired by injecting the sample (solubi-
lized in acetone and diluted 100 times in methanol) into the
ESI-TOF mass spectrometer (Waters LC-ToF). The spray tip potential
was þ3200 V and the nozzle potential was þ60 V. All the significant
peaks present in the mass spectra corresponded to species ionized
Fig. 2. SEC chromatograms of the epoxy prepolymers DGEBA, DGEDAS
n 0
and DGEDAS
(
c ¼ 4 mg/ml).
þ
then dried over anhydrous magnesium sulfate. Preparative chro-
matography was used to purify the synthesized product. The
monomer of isosorbide diglycidyl ether (DGEDAS ) is thus obtained
0
as a viscous liquid with an epoxide equivalent weight of 143 g/eq;
the yield of the reaction was 60%.
with sodium (M þ Na ).
Fourier Transform Infrared Spectroscopy (FTIR) of the samples
was operated on a Nicolet 550 infrared spectrophotometer, using
ꢁ1
KBr pellet, in the wavelength range of 500e4000 cm . The spectra
were acquired using 32 scans at a resolution of four wavenumbers.
1
13
H and C NMR were recorded at frequencies of 400 and
100.6 MHz, respectively, using a Bruker AVANCE 400 NMR spec-
trometer. Deuterated chloroform (CDCl ) was used as a solvent.
2.3. Preparation of epoxy networks
3
The formulations used in this study were based on three
Thermogravimetric analysis (TGA) was carried out on a TGA
Q500 (TA Instruments). 10 mg of the sample was loaded in an open
different epoxy monomers DGEBA, DGEDAS
with the same cycloaliphatic amine, IPD.
0 n
and DGEDAS cured
ꢀ
ꢀ
platinum pan, and heated from 25 C to 500 C under dry nitrogen
ꢀ
To prepare the different networks, the amine curing agent was
added to the epoxy prepolymer in different amounts in order to
at constant heating rates of 10 C/min.
Differential scanning analysis (DSC) measurements were carried
out on a DSC Q10 (TA Instruments). The samples were heated at
vary the stoichiometric ratio r ¼ nah/n
e
where nah is the amino
hydrogen equivalent and n the epoxy equivalent. The epoxy pre-
ꢀ
ꢀ
ꢀ
e
a rate of 10 C/min from ꢁ70 C to 200 C under a nitrogen gas
atmosphere. A second heating scan with the same conditions was
performed after cooling down the samples.
polymers and the curing agent, which are all liquid at room
temperature, were mixed vigorously and rapidly at room
Fig. 3. ESI-TOF mass spectrum (positive-ion mode).

3614
M. Chrysanthos et al. / Polymer 52 (2011) 3611e3620
Rheological properties were measured using an ARES Rheom-
eter from TA Instrument equipped with parallel plates with
0
.5e1 mm spacing and 25 mm diameter. The reactive mixture was
ꢀ
put quickly on the preheated plate (at 80 C) and a multifrequency
sweep was started when the temperature equilibrium was reached
again. The gel time was determined according to the Winter-
eChambon criterion that implies the independence of the loss
0
0
0
factor, tan
d
(¼G /G ), as a function of frequency [22].
Density of the different epoxy-amine networks was obtained
using classical Archimedes method of immersion of the networks
in water. Weight measurements were done quickly (<1 min) in
order to avoid water absorption by the networks.
Dynamic mechanical analysis (DMA) was performed with an
ARES Rheometer from TA Instrument using the torsion mode to
0
00
determine the storage (G ), and loss (G ) moduli as well as tan
d
as
3
a function of temperature. Samples (30 ꢂ 10 ꢂ 2 mm ) were heated
ꢀ
ꢀ
ꢀ
from ꢁ100 C to 200 C using a heating rate of 3 C/min in a forced
convection oven using nitrogen stream. The sample was
a
deformed sinusoidally with controlled strain amplitude of 0.5% at
a fixed frequency of 1 Hz.
n
Fig. 4. Structures of the oligomers of DGEDAS revealed by ESI-TOF.
3
3
3
. Results and discussion
segment). Linear oligomers (n ¼ 0,1, 2) but also branched oligomers
.1. Characterization of isosorbide diglycidyl ether
(
2A þ 4B, 3A þ 5B, 3A þ 6B.) are found, having the chemical
structure shown in Fig. 4. This result can be explained by
a competition during the reaction with epichlorohydrin between
the secondary hydroxy groups of isosorbide and the secondary
.1.1. SEC characterization
For first information on molar mass distribution, the epoxy
prepolymers were analyzed by SEC. The SEC chromatograms of the
n
two diglycidyl ether of isosorbide, DGEDAS and DGEDAS ,
0
synthesized via two different routes, are presented in Fig. 2 as well
as the chromatogram of DGEBA for comparison. Whereas the SEC
chromatogram of DGEDAS
route, underlines the presence of oligomers through numerous
elution peaks, only one of those peaks is present in the DGEDAS
n
, obtained via the traditional synthesis
0
chromatogram, at an elution volume of 26 ml. This peak corre-
sponds to the pure monomer of diglycidyl ether of isosorbide. The
molar mass, calculated from
a high molar mass DGEBA
(
2
eew ¼ 475e550 g/eq) calibration curve, was found equal to
35 g/mol, while the theoretical value is equal to 258 g/mol. By
and
chromatograms, the proportion of monomer in the
sample is equal to 30%. The other part of DGEDAS is
comparing the area of the monomer peak present in DGEDAS
DGEDAS
DGEDAS
0
n
n
n
composed of oligomers of higher molar mass, as it can be observed
in more condensed DGEBA.
3.1.2. Electro-spray mass spectroscopy characterization
The epoxy molecule DGEDAS was analyzed by Electro-Spray
n
mass spectroscopy, to obtain more precise information on molar
mass. ESI-TOF mass spectrum is presented in Fig. 3; it shows a high
number of peaks which indicates the presence of numerous olig-
omers. The formulas corresponding to some of these species were
confirmed by exact molar mass analysis and are given in Table 1, (A
represented the isosorbide segment and B the epichlorohydrin
Table 1
Molar mass values of some DGEDAS oligomers revealed by ESI-TOF.
Molar mass (g/mol)
Formula
Structure
258
404
460
516
662
718
774
C
12
C
18
C
21
C
24
C
30
C
33
C
36
H
18
H
28
H
32
H
36
H
46
H
50
H
54
O
O
O
O
O
O
O
6
A þ 2B (n ¼ 0)
2A þ 2B
10
11
12
16
17
18
2A þ 3B (n ¼ 1)
2A þ 4B
3A þ 4B (n ¼ 2)
3A þ 5B
3A þ 6B
o n
Fig. 5. FTIR spectra of (a) DGEDAS , (b) DGEDAS .

M. Chrysanthos et al. / Polymer 52 (2011) 3611e3620
3615
Fig. 6. 1H NMR spectra of DGEDAS
.
0
hydroxy groups formed after a first reaction of isosorbide diglycidyl
ether with isosorbide. The molar mass of the linear diglycidyl
oligomers is given by M ¼ 258 þ n.202. Electro-spray analysis gives
main difference between DGEDAS
intensity of the eOH peak in DGEDAS
which indicates the absence of eOH group in DGEDAS
shown here) mainly differs from DGEDAS by the presence of
0
and DGEDAS
n
is the very weak
0
(due to residual moisture)
. DGEBA (not
0
information on the molar mass of the oligomer species of DGEDAS
n
ꢁ
1
but, in the conditions used, the magnitude of the peaks is not
quantitative and therefore cannot be linked to the amount of each
species.
aromatic structures: eC]C bond (1607, 1579 and 1508 cm ) and
ꢁ
1
aromatic CeO (1243 cm ).
3
.1.4. NMR characterization
3.1.3. FTIR characterization
Figs. 6 and 7 show the 1H NMR and 13C NMR spectra of the
Fourier transform infrared (FTIR) spectroscopy was used to
0
diglycidyl ether of isosorbide monomer (DGEDAS ); the proton and
identify the molecular structure and the chemical bonds of the
three different epoxy prepolymers. Referring to the FTIR spectra of
carbon assignments are indicated. The epoxy protons H1a, H1b and
H2, give the signals between 2.5 and 3.1 ppm. The peaks between
4.4 and 4.6 ppm are assigned to the protons common to the two
isosorbide cycles, H . Because of the conformation of the cycle and
thus the non-equivalence of the protons of the isosorbide cycle [23]
and the H3 protons, the peaks between 3.3 and 4.1 ppm
DGEDAS
0 n
and DGEDAS shown in Fig. 5, the different peaks indicate
ꢁ1
*
the presence of eOH group (3470 cm ), eCH bond (2930,
ꢁ
1
ꢁ1
2
870 cm and 1460, 1370 cm ), eCOe bond form aliphatic ether
ꢁ1 ꢁ1
(
1090 and 1020 cm ) as well as epoxyde group (910 cm ). The
Fig. 7. 13C NMR spectra of DGEDAS
.
0

3616
M. Chrysanthos et al. / Polymer 52 (2011) 3611e3620
Fig. 8. 1H NMR spectra of DGEDAS
.
n
corresponding to these protons are difficult to assign precisely. The
of polymerization, n, cannot be calculated due to the presence of
integrated values for the different signals are in good agreement
branched oligomers and as a consequence DGEDAS
functionality higher than 2.
The ratio R of epoxy cycle to isosorbide cycle is given by the
following equation:
n
has a mean
13
with the structure shown in Scheme 1. The C NMR spectrum
shows very well-defined peaks: the carbons of the epoxy ring, C1
and C3, show signals at 44 and 50.5 ppm respectively. The carbon,
C3, belonging to the glycidyl ether unit has a resonance at 70 ppm.
Then the different carbons, from the isosorbide cycle, are observed
between 71 and 86 ppm.
number of epoxy cycle
R ¼
number of isosorbide cycle
Figs. 8 and 9 show the 1H NMR and 13C NMR of the diglycidyl
ðI protons from epoxy cycleÞ=3
¼
(1)
ꢃ
ether of isosorbide oligomer (DGEDAS
molar mass distribution of the oligomers, the proton NMR inter-
pretation of the DGEDAS spectrum becomes complex. Based on
n
). Because of the broad
ðI protons H from isosorbide cycleÞ=2
Using the proton integration values (Table 2) a value of R ¼ 1.24 is
n
the previous assignments of peaks, the integration (I) of signals
found.
between 2.6 and 3.2 ppm (protons from epoxy cycle and hydro
ether unit) and the peaks at 4.5 and 4.7 ppm (H from the protons
common to the two isosorbide cycle) can be used to calculate R, the
ratio of epoxy cycle to isosorbide cycle. However the average degree
n
The carbon NMR spectrum of DGEDAS is also more compli-
*
cated. In addition to the resonances assigned previously, new peaks
appear at 69 ppm (Ca) and 79 ppm (Cb). These peaks give indica-
tion on the structure of some oligomers included in the product:
Fig. 9. 13C NMR spectra of DGEDAS
.
n

M. Chrysanthos et al. / Polymer 52 (2011) 3611e3620
3617
Table 2
Determination of R using 1H NMR.
Protons
I (Integration value)
H1a þ H1b
1.063 þ 1.000
1.008
H2
H* (common to the isosorbide cycle)
Ratio R
0.842 þ 0.806
1.24
a
the peak at 69 ppm is assigned to C eOH peak from the hydroxy
ether segment that links two isosorbide cycles. This peak shifts to
13
7
9 ppm with an epichlorohydrin recombination. So the C NMR
confirms the presence of branched oligomers, previously observed
by ESI-TOF mass spectroscopy.
A quantitative carbon NMR was also carried out using an
“inverse gate decoupling” sequence in order to calculate R the ratio
of epoxy cycle to isosorbide cycle. Using conditions leading to
quantitative carbon NMR signals, integration values of the carbons
are available (Table 3). The ratio of epoxy cycle to isosorbide cycle R
is given by the following equation:
number of epoxy cycle
R ¼
number of isosorbide cycle
ðI carbons from epoxy cycleÞ=2
¼
À
Áꢀ
(2)
ꢃ
I carbons C from isosorbide cycle
2
1
The value obtained for R, 1.27, is very close to the one given by H
NMR. Integration values of the signals corresponding to Ca and Cb
are 1.052 and 0.803 respectively; it means that in the oligomers
(n > 0) the number of branched units represents 43% and the
number of linear units (hydroxy ether) represents 57%
3.1.5. TG analysis
Fig. 10. (a) TGA and (b) DTGA curves of DGEBA, DGEDAS0 and DGEDASn.
The mass loss as a function of temperature for DGEBA, DGEDAS
n
and DGEDAS
0
prepolymers and their derivatives are shown in
Fig. 10(a) and (b).
considered, i.e. epoxy prepolymer þ IPD at nah/n ¼ 1 calculated
e
The reference DGEBA thermally degrades mainly through
from the amino hydrogen and epoxy equivalent experimental
values.
a simple step process with an initial degradation (T5%) close to
ꢀ
ꢀ
2
50 C and a maximum rate at 320 C. The TGA curve of the
monomer of diglycidyl ether of isosorbide (DGEDAS ) shows
a continuous single step degradation process well before DGEBA. It
As an example, Fig.11 shows the evolution of the loss factor tan
as a function of time, in a multifrequency mode (from 1 rad/s to
0 rad/s) for the system DGEDAS eIPD. The gel time, tgel, is deter-
d
0
5
0
ꢀ
ꢀ
begins after 130 C and with a maximum rate equal to 250 C. In the
mined by the crossover of the loss factor curves at various
case of diglycidyl ether of isosorbide oligomer (DGEDAS
dation begins like DGEDAS
stages of degradation with maximum rate at 250 and 410 C. In the
first step, the sample lost 30% of its weight, and in the second step it
lost the remaining weight. It revealed that the first step decom-
n
), degra-
ꢀ
0
around 130 C but with two distinct
ꢀ
position of the DGEDAS
DGEDAS contains 30% of monomer DGEDAS
previous results of SEC analysis. Higher molar mass compounds
n ¼ 1, 2 and branched oligomers) are more stable or not so volatile.
n
is the monomer decomposition. Therefore,
n
0
, which confirms the
(
3.2. Curing behavior
Gel time was determined by rheological measurements, at
ꢀ
a temperature equal to 80 C, for the three reactive systems
Table 3
Determination of R using 13C NMR.
Carbons
I (Integration value)
C1
2.000
C2
2.020
C* (common to the isosorbide cycle)
Ratio R
1.558 þ 1.590
ꢀ
Fig. 11. Loss factor tan
DGEDAS eIPD.
d
,
isotherm at 80 C, example of the reactive system
1.27
0

3618
M. Chrysanthos et al. / Polymer 52 (2011) 3611e3620
Table 4
Table 5
Gel time data for the reactive systems DGEBA/IPD, DGEDAS
n
0
/IPD and DGEDAS /IPD.
n 0
DSC results for DGEBA/IPD, DGEDAS /IPD and DGEDAS /IPD.
ꢀ
ꢀ
Tpeak ( C)
System
tgel (min)
tan
d
gel
D
System
T
g0 ( C)
ΔH (J/g)
ΔH (kJ/ee)
DGEBAeIPD
16
7
18
1.5
1.0
1.6
0.63
0.50
0.64
DGEBAeIPD
ꢁ35
ꢁ36
ꢁ60
415
404
547
92
91
89
114
104
108
DGEDAS
DGEDAS
n
eIPD
eIPD
DGEDAS
n
eIPD
eIPD
0
DGEDAS
0
frequencies. Table 4 summarizes the gel time for the three different
reactive systems. It appears that the new bio-based reactive system
to the two others which can be explained also by the fact that due
to branching there are two different types of epoxy groups; cyclo-
aliphatic and aliphatic epoxy groups, as shown in Fig. 4. The values
of the enthalpy of reaction are the same for the three systems and in
the range of the classical value obtained for epoxy-amine reaction
at a stoichiometric ratio of 1.
DGEDAS
n
reacted with IPD has shorter gel time than the traditional
eIPD has almost the
one DGEBA-IPD, whereas the system DGEDAS
0
same gel time as DGEBAeIPD. Different hypotheses can explain the
shorter gel time of the system based on the oligomeric bio-based
prepolymer: first the epoxy-amine reaction can be catalyzed by
the hydroxyl groups of this oligomer, secondly the functionality of
this compound is probably higher than 2 due to branched mole-
cules which means a conversion at the gel point lower than
3.3. Influence of stoichiometry on the value of T
g
The effect of variation of the amino hydrogen-to-epoxy ratio,
is examined in this study as an indicator of possible
0
.58e0.60 [24]. At the gelation point, the storage and loss modulus
r ¼ nah/ne, on T
g
can be described by a power law as a function of the pulsation:
0
00
D
side reactions. Indeed it has been shown that assuming the only
mechanism is epoxy-amine addition without side reaction, the
G (u) f G (u) f u , where D is the relaxation exponent that can be
predicted by Rouse’s percolation theory. So at gelation tan
d is
maximum attainable T
sition [25]. For each epoxy prepolymer the variation of the ratio
was studied between 0.5 and 1.5.
For the three systems, the variation of the glass transition
g
coincides with the stoichiometric compo-
independent on frequency and its value is:
nah/n
e
d
¼
p
$
D
=2
(3)
gel
The values obtained for the relaxation component
D
are similar
temperature, measured during a second DSC run, is plotted versus r
for DGEBAeIPD and DGEDAS
are in agreement with data reported in the literature on die-
poxyediamine systems [4]. The value obtained for DGEDAS eIPD
system is lower and equal to 0.5; for this system gelation occurs
0
eIPD networks (Table 4), these values
in Fig. 13. In each case, the curves exhibit a maximum T for r equal
to 1, which means that the only mechanism occurring during the
crosslinking of the reactive systems is the amine-epoxy addition.
No side reaction, such as etherification, occurs during the
crosslinking.
g
n
0
00
when G ¼ G .
Fig. 12 shows the non-isothermal DSC thermograms of the three
The T obtained for both bio-based reactive systems are lower
g
different epoxy precursors reacted with IPD at nah/n
transition temperature of the systems before reaction, Tg0, is the
lowest for the system based on DGEDAS , this monomer has a lower
molar mass as compared to the two others. The peak maximum
temperature, Tpeak, and the total heat of reaction (ΔH) obtained
from the DSC analysis are summarized in Table 5. There are no large
differences between the three systems. DGEBAeIPD and DGE-
e
¼ 1. The glass
than the T observed for the conventional DGEBAeIPD system. This
observation cannot be attributed to incomplete curing as no
g
0
residual heat was observed by DSC after the curing cycle and FTIR
ꢁ1
showed no peak at 910 cm . A similar result was found by Feng
ꢀ
et al. who noticed a decrease of Tg about 40 C on networks
synthesized from diglycidyl ether of isosorbide and Jeffamine T403,
an aliphatic curing agent [20]. Because of the short and cyclic
structure of the monomer present in the bio-based epoxy resin,
DAS
0
eIPD thermograms have similar shape, with a main peak at
ꢀ
114 and 108 C respectively, and a shoulder on the high tempera-
higher T was expected for those systems. It was shown for aliphatic
g
ture side. This shoulder is explained by the different reactivity of
the amino groups of IPD. In the third system, DGEDAS eIPD, the
exothermic peak of reaction is more symmetric, Tpeak is slightly
lower, confirming the highest reactivity of this system as compared
and aromatic polyesters that isosorbide structure is much more
rigid than usual diols like ethane diol or butane diol [26]. In our case
we have to compare the effect of bisphenol A and isosorbide,
n
Fig. 12. DSC thermograms of dynamic curing of DGEBA, DGEDAS
IPD (heating rate 10 C/min), r ¼ 1.
n
, and DGEDAS
0
with
Fig. 13. Glass transition temperature versus r for networks based on DGEBA, DGEDAS
and DGEDAS in combination with IPD.
n
ꢀ
0

M. Chrysanthos et al. / Polymer 52 (2011) 3611e3620
3619
a comparison could be done from linear polycarbonate synthesized
from these two diols [27]. It seems that in this case the isosorbide
structural unit brings a lower contribution to T , with a decrease of
5 C as compared to bisphenol A-based polycarbonate. Besides, for
epoxy networks T depends on the chain stiffness, through the
g
ꢀ
1
g
structure of the epoxy prepolymer and hardener, and also on the
crosslink density, n, given by their respective concentrations [4,28].
Many methods based on physical or empirical approaches have
been developed for the prediction of T . The relation established by
g
DiMarzio is well adapted to epoxy-amine networks:
TgL
T
g
¼ 1
(4)
ꢁ KDMFn
where KDM is the DiMarzio universal constant, F is a flex parameter
and TgL is the glass transition of a hypothetical linear polymer; TgL
represents the copolymer effect and can be calculated using an
additivity law, for example the Fox approach. The low molar mass
of the isosorbide based monomer and thus the increase of the
proportion of hardener, IPD, can explain the lower T
the DGEDAS eIPD networks compared to the one of DGEBAeIPD
network. For 100 g of epoxy monomer we need 24 g of IPD in the
case of DGEBA (r ¼ 1), and 30 g of IPD in the case of DGEDAS . The
mass fraction of IPD increased from 19.3% in DGEBAeIPD to 23.1% in
DGEDAS eIPD, so the contribution of the hardener to T is greater
in the case of DGEDAS eIPD than DGEBAeIPD.
There is no large variation in T for the two bio-based systems:
g
obtained for
0
0
0
g
0
g
on the one hand the molar mass distribution is very different as
well as the epoxy equivalents, but on the other hand the highest
functionality of DGEDAS
prepolymer shows only a slight decrease as compared to DGEDAS
moreover the variation of T versus a/e is less pronounced.
n g
may explain why the T based on this
0
;
g
3.4. Thermo-mechanical properties of the networks
The dynamic mechanical properties of the networks were
investigated using DMA. In Figs. 14a), b) the storage modulus (G )
0
Fig. 14. Storage modulus versus temperature (a) and tan
DGEBA, DGEDAS and DGEDAS cured with IPD.
d
versus temperature (b) for
and the loss factor (tan
compared with the DGEBA based one.
A typical plot of the storage modulus G and tan
temperature exhibits two relaxations. The main transition,
d) of the two bio-epoxy based networks are
n
0
0
d
as a function of
in the
a
shortest chains between crosslinks. This behavior is linked to the
branched structure of the DGEDAS oligomer and to its high func-
tionality as compared to DGEDAS
According to the rubber elasticity theory, the following equation
is employed to describe the relationship between M (average
high temperature region, is associated with the glass transition; the
ꢀ
n
.
secondary relaxation
b
, below 0 C is assigned to short molecular
0
segment motion, hydroxyl ether groups in the particular case of
0
epoxy-amine networks [29]. The values of T
b
, T
a
, G
R
are reported in
c
Table 6.
molar mass of the segment between crosslinking points) and
There is no significant difference in the sub-T
g
behavior between
0
storage modulus (G ) of a thermoset above the T
g
[30]:
the three types of networks: T
b
is in the same range of tempera-
ꢀ
tures, around ꢁ50 C, but the magnitude of the relaxation is higher
for the isosorbide based networks (Table 6). It is linked to the
concentration of the relaxing species per unit volume, hydroxy
ether units, that is for the same reason explained previously higher
’
G ¼ dRT=M
c
(5)
0
where G , d, R, T are the storage modulus at T þ 30 K, the density
g
of the polymer, the gas constant (8.314 J/mol K) and the tempera-
in DGEDAS
For DGEDAS
0
eIPD networks as compared to DGEBAeIPD network.
eIPD networks, the hydroxy ether groups can come
ture (K) respectively. The values of molar mass between cross-
linking, calculated as cited above, are summarized in Table 6. They
n
initially from the oligomers and from the epoxy-amine reaction,
their concentration in the final network should be more or less the
same as in DGEDAS
0
eIPD network.
Table 6
T
a
follows the same trend as the T
g
measured by DSC: the
ꢀ
Dynamic mechanical analysis of the networks: DGEBA/IPD, DGEDAS
DGEDAS
/IPD for r ¼ 1.
n
/IPD and
highest value, 155 C, is obtained for the DGEBAeIPD network, then
0
a decrease of T
a
is observed for the bio-based networks which reach
eIPD and DGEDAS eIPD respectively.
eIPD network has the highest rubbery modulus,
eIPD and the DGEBAeIPD networks with
ꢀ
System
Tb
(
ꢀ
C) Ta
(
ꢀ
C)
G (MPa)
0
d
Mc
Mcth
112 and 96 C, for DGEDAS
0
n
R
ꢀ
3
(
at T
g
þ 30 C) (g/cm ) (g/mol) (g/mol)
The DGEDAS
n
DGEBAeIPD
DGEDAS
DGEDAS
ꢁ46
155
96
112
17.1
24.7
16.6
1.13
1.24
1.25
250
170
260
300
302
247
followed by the DGEDAS
0
n
eIPD ꢁ42
eIPD ꢁ40
equivalent rubbery moduli. This means that the network in which
the molecular motion occurs at the lowest temperature has also the
0

3620
M. Chrysanthos et al. / Polymer 52 (2011) 3611e3620
based on the oligomeric resin is more reactive than the two others,
which behaves similarly. In all cases the maximum glass transition
temperature was obtained using a stoichiometric ratio equal to 1
which means that only epoxy-amine reactions occur. Networks
synthesized from the bio-based epoxy prepolymers have high glass
transition temperature, yet lower than the DGEBA based network.
However they have higher rubbery modulus due either to the lower
molar mass or to the higher functionality of the prepolymer. These
results suggest that isosorbide based epoxy precursors could be
good candidates to replace bisphenol A-based epoxy prepolymers.
However storage of these bio-based prepolymers must be done in
dry conditions because these compounds are hygroscopic and the
presence of water may deteriorate the properties of the networks,
depending on the type of hardener used [20].
Acknowledgment
0 n
Fig. 15. TGA curves of networks based on DGEBA, DGEDAS and DGEDAS in combi-
nation with IPD.
The authors acknowledge the financial support of FEDER (Fonds
Européens pour le Développement Economique Régional). The
authors are also thankful to Roquette (Lestrem, F) and Huntsman
Advanced Materials (Bâle, CH) for supplying samples and for the
synthesis of isosorbide based epoxies, and to the team of NMR
Polymer Center, Institut de Chimie de Lyon (ICL) for their help.
are compared to the theoretical value calculated from the molar
mass of the epoxy prepolymer and curing agent, Mcth. The molar
mass of the epoxy prepolymer was calculated from the value of the
epoxy equivalent and assuming a functionality of 2. Theoretical and
experimental results concerning M
networks based on pure monomers (DGEBA and DGEDAS
have a functionality of 2. For the network based on DGEDAS
c
are in agreement for the
References
0
) which
n
[1] Kaplan DL. Biopolymers from renewable resources. Berlin: Springer-Verlag;
oligomer a significant difference is noticed between the experi-
1998.
mental value and the theoretical value. This difference is due to the
[2] Gandini A. In: Pascault JP, Williams RJJ, editors. Epoxy polymers: new mate-
rials and innovations. Wiley-VCH; 2010.
wrong functionality used to calculate the theoretical M
confirms indirectly the complex chemical structure of DGEDAS
There are two types of crosslink points in DGEDAS based networks
Fig. 4):
c
and thus
[3] Shen L, Haufe J, Patel MK. Product overview and market projection of
n
.
emerging biobased plastics. PROBIP; 2009.
[4] Pascault JP, Sautereau H, Verdu J, Williams RJJ. In: Thermosetting polymers.
Taylor & Francis Books, Inc; 2002.
n
(
[
5] Miyagawa H, Mohanty AK, Misra M, Drzal L. Macromol Mater Eng 2008;289:
29e35.
[6] Miyagawa H, Mohanty AK, Misra M, Drzal L. Polym Eng Sci 2005;45:487e95.
6
4
. Thermal stability of the networks
[
[
[
7] Park SJ, Jin FL, Lee JR. Macromol Chem Phys 2004;205:2048e54.
8] Park SJ, Jin FL. Polym Int 2008;57:577e83.
9] Park SJ, Jin FL, Lee JR, Shin JS. Eur Polym J 2005;41:231e7.
The mass loss as a function of temperature for DGEBA, DGEDAS
and DGEDAS cured with IPD is shown in Fig. 15. The network
0
n
[10] Shabeer A, Sundararaman S, Chandrashekhara K, Dharani LR. J Appl Polym Sci
2007;105:656e63.
[11] Patel MB, Patel RG, Patel VS. J Therm Anal 1989;35:47e57.
12] SATO S. WO Patent 2009/080209 A1; 2009.
13] Chapin R. NTP-CERHR expert panel report on the reproductive and develop-
mental toxicity of bisphenol A. Birth Defects Res, Part B 2008;3:157e395.
[14] Acierno D, Russo P, Savarese R. PPS-24 2008 Salerno (Italy).
15] Takada Y, Shinbo K, Someya Y, Shibata MJ. Appl Polym Sci 2009;113:479e84.
16] Shibata M, Nakai K. J Polym Sci, Part B: Polym Phys 2010;48:425e33.
DGEBAeIPD thermally degrades through a simple step process with
ꢀ
an initial degradation (T5%) close to 330 C and a maximum rate at
[
[
ꢀ
3
70 C. The TGA curve of the network obtained with DGEDAS
n
combined with IPD shows also a simple step process with an initial
ꢀ
ꢀ
degradation and a maximum rate, respectively at 330 C and 380 C,
[
[
very close to the values observed for DGEBAeIPD network. The
DGEDAS
0
eIPD network exhibits poorer thermal stability compared
[17] Morrison G. US Patent 3,041,300; 1962.
[18] Zech J. US Patent 3,272,845; 1966.
ꢀ
tothe other networks, with an initial degradation (T5%) at 305 C and
a maximum rate at 350 C. It is probably related to the low thermal
[
19] East A, Jaffe M, Zhang Y, Catalani LH.
a) US Patent 2008/0021209A1; 2008 (New Jersey Institute of Technology);
(b) US Patent 2008/0009599A1; 2008 (New Jersey Institute of Technology);
c) US Patent 7,619,056 B2; 2009 (New Jersey Institute of Technology);
20] Feng X, East AJ, Hammond WB, Zhang Y, Jaffe A. Polym Adv Technol 2011;22:
39e50.
ꢀ
(
stability observed on the DGEDAS
. Conclusions
In this work, two novel bio-based epoxy monomers were
0
prepolymer (Fig. 10).
(
[
[
5
1
21] Sachinvala ND, Winsor DL, Menescal RK, Ganjian I, Niemczura WP, Litt MH.
J Polym Sci, Part A: Polym Chem 1998;36:2397.
[22] Winter HH, Chambon F. J Rheol 1986;30:367e82.
[23] Hopton FJ, Thomas GHS. Can J Chem 1969;47:2395.
[24] Miller DR, Macosko CW. Macromol 1976;9:206e11.
successfully synthesized from renewable resources (isosorbide) via
two different routes. The chemical structures were studied in detail
1
13
using SEC, ESI-TOF MS, H and C NMR, and FTIR analysis. Digly-
cidyl ether of dianhydro-sorbitol (DGEDAS) was obtained either as
a pure monomer or as an oligomeric resin of higher functionality.
Reactivity and physical properties of these new epoxy prepolymers
cured with isophorone diamine (IPD) were compared with a tradi-
tional petroleum-derived epoxy prepolymer (DGEBA) cured with
the same hardener and under the same conditions. The system
[25] Galy J, Sabra A, Pascault JP. Polym Eng Sci 1986;26:1514e23.
[26] Fenouillot F, Rousseau A, Colomines G, Saint-Loup R, Pascault JP. Prog Polym
Sci 2010;35:578.
[27] Fuji M, Akita M, Tanaka T, EP Patent 2,033,981 A1; 2009, (Mitsubishi Chem.)
[28] Won YG, Galy J, Pascault JP, VerduPolym JJ. Sci, Part B: Polym Phys 1991;29:
81e7.
29] Won YG, Galy J, Gérard JF, Pascault JP, Bellenger V, Verdu J. Polymer 1990;31:
787e92.
[30] Nielsen LE. J Macromol Sci Rev Macromol Chem 1969;C3:69.
9
[
1