Reactive Microspheres as Active Fillers for Epoxy Resins

C. Carfagna,1 V. Ambrogi,1 G. Cicala,2 A. Pollicino,2 A. Recca,2 G. Costa3

1
Department of Materials and Production Engineering, University of Naples “Federico II”, Piazzale Tecchio, 80 - 80125
Naples, Italy
2
Department of Physical and Chemical Methodologies for Engineering, University of Catania, Viale A. Doria 6 - 95125
Catania, Italy
3
IMAG-CNR, Via De Marini, 6 - 16149 Genova, Italy

Received 21 November 2003; accepted 19 February 2004

DOI 10.1002/app.20678
Published online in Wiley InterScience (www.interscience.wiley.com).

ABSTRACT: Physical properties of epoxy resins filled analysis was applied on particles to relate surface composi-
with microparticles are presented and discussed. Micropar- tion and reactivity of microspheres. Rheological, dynamic–
ticles were synthesized in the form of micron-sized, mechanical, and mechanical properties of the cured blends
crosslinked spherical particles, with an excess of reactive were analyzed and related to the pure resin and to the same
amino groups on their outer surface, and subsequently resin modified with PES180. © 2004 Wiley Periodicals, Inc.
blended with EPON828-3,3⬘DDS in different weight per- J Appl Polym Sci 93: 2031–2044, 2004
cents (10 and 20 wt %). Differential scanning calorimetry and
scanning electronic microscopy were applied to investigate
microsphere properties such as morphology, shape, size, Key words: adhesion; crosslinking; curing of polymers; res-
and size distribution. Electron spectroscopy for chemical ins; thermosets

INTRODUCTION particles can be utilized in the field of microencapsu-

lation, which includes a wide variety of possible ap-
It is well known that physical properties of thermosets
plications among which that of drug release seems to
can be improved by inclusion of a filler, as a second
be the most promising.7 Because of the commercial
phase, to the resin batch. For example, toughness is
and scientific interest in these particles, preparation of
noticeably increased by mixing reactive rubbers with
microspheres has been an active research field in re-
epoxy resins. The same effect can be reached with the
cent years.
addition of some reactive thermoplastics.
As far as epoxy resin particles preparation is con-
In many cases, the target is viscosity modification
cerned, several protocols were reported in scientific
during processing of thermosets. For this purpose, a
literature and patents.3,8,9 In their work, Hibino and
suspension of solid particles, as well as of some ther-
moplastics, included in the resin can be very useful in Kimura8 proposed a novel synthetic approach for ep-
balancing the proper viscosity profile during process- oxy resin particles through nonaqueous dispersion
ing. polymerization by using polyethers as solvents. At the
In this article, a novel approach to change the vis- same time, Hseish and Woo9 obtained crosslinked
cosity of epoxy resins is presented. The method is epoxy microparticles of spherical shape through a
based on the inclusion of reactive microspheres pre- cure-induced phase inversion process in DGEBA/
viously prepared by reaction of the epoxy monomer DDS/PMMA blends. They explained the morphology
with the curing agent. evolution of epoxy microsphere through a phase-in-
Over the last several years, a growing interest in the version mechanism.
production of polymer particles was recorded.1– 6 In this article, curing of epoxy resins in the form of
These materials are suitable as spacers, slip properties micron-sized particles was carried out in polypro-
improvers for plastic film, paints, adhesives, sealant pylene glycol (PPG). The starting solution was com-
and binders if dispersed in a solvent, and fillers in the posed of the epoxy monomer and an amine-based
market of composite materials. Furthermore, polymer curing agent dissolved in the solvent. The as-blended
mixture was let react at constant temperature, without
stirring, until the resin particle formation occurred.
Correspondence to: C. Carfagna ([email protected]). The mechanism was very complex because of the cur-
Contract grant sponsor: MIUR; contract grant number: ing reaction between the epoxy monomer and the
COFIN 2001.
aromatic amine, which led to the crosslinked resin,
Journal of Applied Polymer Science, Vol. 93, 2031–2044 (2004) followed by a phase-separation process, influenced by
© 2004 Wiley Periodicals, Inc. the miscibility of the reacting thermoset in the PPG.
2032 CARFAGNA ET AL.

TABLE I
Concentrations, Reaction Temperature, and Times Chosen in the Preparation of Epoxy Resin Particles
and Glass Transition Temperatures (Tg’s) of Crosslinked Particles
Concentration Reaction Reaction
Solute Solvent (wt %)a Symbol temperature (°C) time (h) Tg (°C)b

9.0 I 130 15 150

PPG1000 33.3 II 130 15 143
EPON825⫹DAT
47.4 III 130 15 153
PPG4000 33.3 IV 130 15 150
a
%wt is referred to as the weight ratio between EPON825⫹DAT (solute) and total weight (solute ⫹ solvent). The
percentage values correspond to solute/solvent ratios of 10%, 50%, 90%, respectively;
b
Tgs values were obtained from DSC second heating scans.

To relate experimental parameters, such as nature of respectively, with the purpose of obtaining two homo-
solvent and composition of the solution, with the geneous solutions of EPON825 in PPG and DAT in
properties of microspheres, four systems were pre- PPG to be mixed.
pared, which differed in the epoxy ⫹ amine content, DAT was taken in a 35% molar excess with respect
in the solvent, or in viscosity of the PPG. Electron to the stoichiometric ratio (stoichiometric ratio be-
spectroscopy for chemical analysis (ESCA) was ap- tween EPON825 and DAT is 2 : 1) to be sure that all
plied on particles with the aim of relating the surface the epoxy monomer introduced was reacted and to
composition to the reactivity of microspheres. guarantee an excess of curing agent on the surface of
One group of microspheres was selected as fillers the microspheres, as confirmed by ESCA, to react with
for EPON828/3,3⬘DDS matrices. Two systems were the epoxy monomer of the matrix.
analyzed; the first composed by 10 wt % and the
second composed by 20 wt % of microspheres in the
epoxy matrix. Rheological, dynamic–mechanical, and
Materials used for rheological analysis
mechanical properties were studied in comparison
with the neat resin and with EPON828/3,3⬘DDS/15 Rheological analyses were carried out on two systems
wt % PES180, where PES180 is a modifier. composed by a Bisphenol-A-type resin (EPON828威,
supplied by-Shell Co., EEW 185–192), 3,3⬘-diaminodi-
phenylsulphone (3,3⬘DDS; Aldrich, Milan, Italy), and
EXPERIMENTAL
two different weight percentages of epoxy micro-
Materials spheres, 10 and 20 wt %, respectively. In particular,
the microspheres obtained from system II (see Table I)
Epoxy resin microparticles
were used for these studies. The comparison of the
A commercial available Bisphenol-A-type resin two modified systems was made with the EPON828/
(EPON825威, supplied by Shell Co.) and an aromatic 3,3⬘DDS neat resin. We will refer to the three systems
amine, the 2,4-diaminotoluene (DAT; purchased from as 10% MICs, 20% MICs, and 0% MICs, respectively,
Aldrich Milan, Italy) were used for the preparation of later on.
particles. PPG with different molecular weights In all the cases, the epoxy monomer, the curing
(PPG4000 with Mn ⫽ 40,000 and PPG1000 with Mn ⫽ agent, and the microspheres (when present) were
1000) were selected as solvents and purchased from mixed at the temperature of 110°C without solvent.
Aldrich. All these products were used without further Mixtures of EPON828 and 3,3⬘DDS modified by
purification. addition of 15 wt % of a thermoplastic copolymer
The different experimental conditions selected for were used as comparison to the systems containing
the preparation of epoxy microspheres are reported in the microspheres.
Table I. Glass transition temperatures (Tg’s) of the The thermoplastic polymer was an amine-termi-
obtained particles are also indicated. nated copolymer 40 : 60 PES : PEES (polyethersul-
In a typical procedure, EPON825 and DAT were phone : polyetherethersulphone) prepared in our lab-
dissolved in PPG. The mixture was allowed to heat at oratories10 with a molecular weight of 9000 Mn. Resin
a constant temperature (see Table I) for 15 h without mixtures were prepared by mixing the thermoplastic
stirring, until a precipitate appeared. The solid resid- with the EPON828 and stirring the resulting mixture
ual product was washed with hot methanol, filtered, for 1 h at 120°C; then, the curing agent was added to
and vacuum dried at T ⫽ 60°C for 2 h. the mixture and stirred for 1 h at 80°C.
Prior to being mixed, EPON825 and DAT were dis- We will refer to the 40 : 60 PES : PEES copolymers as
solved in PPG separately at T ⫽ 40°C and T ⫽ 80°C, PES180 later on.
REACTIVE MICROSPHERES AS ACTIVE FILLERS 2033

Materials used for DMTA and LEFM testing 12 ⫻ 40 ⫻ 3 mm. Storage modulus and tan ␦ were
recorded by a torsion mode.
Cured samples were prepared by mixing the micro-
Differential scanning calorimetry (DSC) (TA Instru-
pheres with the EPON828 and stirring the resulting
ments model 2920) was used to evaluate the Tg’s of
mixture for 30 min at 120°C; then, the curing agent
microspheres. For each sample, a double heating scan
was added to the mixture and stirred for 1 h at 80°C.
from room temperature to 280°C at a rate of 10°C/
The mixture was poured in a preheated steel mold and
min, under nitrogen atmosphere, was carried out.
degassed for 30 min at 145°C. Then, the temperature
Scanning electronic microscopy (SEM) was used to
was increased at 2°C/min up to 180°C and held at that
study morphology of particles and their sizes and to inves-
value for 3 h. At the end of the curing cycle, the panels
tigate the characteristics of fracture surfaces of composites,
were left to cool down slowly at room temperature.
using a Hitachi microscope model S-2300. Prior to being
observed, samples were metallized with a gold layer.
Morphological characterization was also carried out
Instrumental
by using a transmission electron microscope (TEM;
ESCA spectra were recorded by using a VG Instru- Zeiss model EM900) applying an acceleration voltage
ment equipped with a X-ray source having a Mg an- of 80 kV. The specimen support grids, lying on filter
ode (MgK␣1,2 1253.6 eV). In standard conditions, X-ray paper, were coated with carbon black. Five milligrams
source worked at 100 W, 10 kV, and 10 mA. The of the polymer was ultrasonically dispersed for 4 min
instrument base pressure was 5 ⫻ 10⫺10 Pa and spec- in 10 ml isopropanol and 2 drops of the polymer
tra were recorded at a pressure of 2 ⫻ 10⫺8 Pa. Pass suspensions were placed onto the coated grids; the
energy was 100 eV for widescans, 50 eV for narrows- solvent was then evaporated. The grids were prepared
cans, and 20 eV for O1s regions recorded for curve immediately before the observation to avoid the ag-
fitting. To get information on depth profiling, three gregation of polymer particles during storage.
spectra of each sample were recorded at different take-
off angles of analyzed electrons (at 25°, 45°, and 90°
with respect to sample surface, respectively). All data RESULTS AND DISCUSSION
analyses (curve fitting, linear background subtraction,
Reaction mechanism of microspheres formation
and area determination) were carried by using version
6 of VGX900⫻ software of VG Instrument. Formation of epoxy resin microparticles was achieved in
Rheological properties of the matrix resin and ther- a stationary reaction system, using a selected solvent. In
moplastic modified systems were recorded on an Ares a typical procedure, epoxy and curing agent (in a fixed
rheometer (Rheometric Scientific). Dynamic-viscosity molar ratio) were introduced in the dispersant and let
␩*, elastic modulus G⬘, and loss modulus G⬙ were react at a selected temperature, without stirring. As soon
obtained by oscillatory shear measurements as a func- as the crosslinking reaction started, the material sepa-
tion of temperature and by a series of isothermal rated, as a result of its reduced solubility in the solvent,
experiments employing parallel plates (40 mm diam- and was deposited as microsized particles.
eter) separated by a 1.2-mm gap. Isothermal tests were In the early stage of the reaction, the system was
performed by preheating the sample chamber to the found to be in the physical state of miscible liquid,
fixed temperature; then, the sample was quickly in- which macroscopically corresponded to a situation of
serted and the analysis was started. The analysis was homogeneity and transparency of the solution. This
done with optimized instrumental parameters of 35% state was held until a mechanism took place, which
strain and angular frequency of 30 rad/s. was able to force the energetic barrier (related in some
Panels (80 ⫻ 16 mm) were produced for the linear way to the viscosity of the medium, the affinity with
elastic fracture mechanics (LEFM) and dynamical me- the solvent, the solvent/resin ratio, etc.) and induce
chanical thermal analysis (DMTA). All the tests were phase separation. This mechanism was represented by
conducted by using a sample size of 65 ⫻ 12.5 ⫻ 5 mm the formation of the first nuclei of the cured resin, as a
with a single-edge notch in a flexural mode at a con- consequence of the crosslinking reaction between the
stant speed of 1 mm/min. The precrack was applied epoxy monomer and the amine. In fact, as soon as the
by the razor tapping technique. The crack opening reaction between epoxy and amine occurred at a given
stress intensity factor, Kc, and the energy of fracture time t, at random points the first nuclei were assumed
factor, Gc, were calculated according to the protocol [a] to appear in the solution, similarly to the formation of
established by the European Structural Integrity Soci- the crystallization nuclei from a molten polymer. Be-
ety (ESIS). All measurements were made at 23°C. cause the overall system was in a metastable state,
DMTA tests were carried out for cured samples these bodies could dissolve into the solvent or sepa-
with an Ares rheometer (Rheometric Scientific) at a rate from it as solid particles. As the curing reaction
fixed frequency of 10 rad/s and 0.1 strain with a went on, the nuclei formed reached a critical degree of
2°C/min heating rate by using samples of sizes crosslinking (i.e., a critical dimension), so that it was
2034 CARFAGNA ET AL.

Figure 1 TEM image of microspheres obtained from system II.

not possible for the particles to dissolve in the solvent Figures 1 and 2 report the images obtained from TEM
anymore. Hence, as a result of their growth, the par- and SEM techniques, respectively, on sample II (see
ticles separated from the solvent and were deposited Table I). The particles are regular in shape and size, the
by virtue of gravity. latter ranging from 1 to 5 ␮m. The same was observed in
All the samples of extracted spheres exhibited a the case of samples I and III, which also exhibit highly
distinctly single Tg, indicating that the spheres were smooth and homogeneous surfaces, with regular dimen-
likely to be formed by an homogeneous structure. Tg sions.
values are reported in Table I. Furthermore, the value On the contrary, sample IV appears irregular in its
of Tg was similar to that of the in bulk cured epoxy (Tg morphology and dimension (Fig. 3): this may be attrib-
⫽ 150°C). In addition, the microspheres were insolu- uted to the high viscosity of PPG4000 in comparison
ble in all the common organic solvents, confirming the with PPG1000. It seems that, due to the high viscosity of
crosslinked nature of the materials. the solvent, the forming network segregates and grows
To relate the nature of solvent and the monomer con- at a lower rate, leading to bigger sized particles.
centrations with particle morphology, four samples were
prepared according to same procedure, at the same re-
action temperature (T ⫽ 130°C). In particular, sample IV
ESCA analysis
was prepared by using PPG4000 as solvent (see Table I),
while samples I, II, and III were obtained from PPG1000 Microsphere action on crosslinking kinetics has to be
solutions, with different epoxy ⫹ amine to solvent related to their surface composition. Among surface
weight ratio. characterization techniques, ESCA gives the opportu-
REACTIVE MICROSPHERES AS ACTIVE FILLERS 2035

Figure 2 SEM image of microspheres obtained from system II.

nity to get semiquantitative information on the mean In the interpretation of these data, it must be kept in
composition of the outermost layers (⬃ 100 Å). As is mind that on the surface both DAT and PPG can be
known from the photoelectronic peak areas (obtained present at the same time. Actually, it is PPG presence
multiplying the experimental values by the right sen- that, determining a decreasing of the N1s/C1s ratio
sitivity factors), it is possible to get data on the atomic and the contemporaneous increasing of the O1s/C1s,
abundance. Data obtained by this technique are re- makes necessary a further data elaboration before
ported in Table II. their correct interpretation. Aim of the elaboration is

Figure 3 SEM image of microspheres obtained from system IV.

2036 CARFAGNA ET AL.

TABLE II
Atomic Ratios at Several Take-Off Angles
(t.o.a.) in Samples I–IV
Sample O1s/C1s N1s/C1s

I (t.o.a.25°) 0.163 0.0367

I (t.o.a.45°) 0.161 0.0329
I (t.o.a.90°) 0.160 0.0342
II (t.o.a.25°) 0.210 0.0343
II (t.o.a.45°) 0.220 0.0360
II (t.o.a.90°) 0.207 0.0353
III (t.o.a.25°) 0.205 0.0412
III (t.o.a.45°) 0.204 0.0373
III (t.o.a.90°) 0.199 0.0363
IV (t.o.a.25°) 0.204 0.0238
IV (t.o.a.45°) 0.202 0.0371
IV (t.o.a.90°) 0.206 0.0351 Figure 5 ESCA C1s envelope of sample IIbis.

the individuation of the PPG amount eventually Comparing the different shape of the C1s envelope of
present on the microsphere surface and the calculation samples II and IIbis (Figs. 4 and 5), we concluded that
of new values of the atomic ratios reported in Table II on microsphere surface there is a high concentration of
obtained by subtracting the contributions deriving PPG.
from PPG. To do this, we used another property of The peak fitting elaboration of C1s envelope was
ESCA spectra. In fact, each photoelectronic peak is the carried out by taking into account the component
result of the sum of all the contributions deriving from position reported in literature for the carbon atoms of
the presence, in the analyzed layer, of different oxida- PPG11 and EPON82512 and DAT11 centered at 285.0 eV
tion states for the same atomic species (chemical shift). for hydrocarbons, at about 286.4 eV for carbons single
Then, it is possible, by a peak-fitting process, to fit into bonded to oxygen and at about 287.1 eV for carbon
the photoelectronic peak all these contribution and in present in the oxirane ring, respectively. The results
our case it can be worth analyzing the components of obtained in the case of the sample II spectrum re-
the C1s peak. corded at a take-off angle of 25° (Fig. 4) show a relative
The ESCA spectra of sample II (Fig 4) showed three abundance of these three peaks of 54.7, 33.8, and
photoelectronic peaks centred at about 285.1 eV (C1s 10.5%, respectively. These values are in good agree-
photoelectrons), 532.7 eV (O1s photoelectrons), and ment with a composition of the outermost layers of
399.1 eV (N1s photoelectrons), respectively. To get in- 38.2 wt % of PPG, 49.8 wt % of EPON825, and 12.0 wt
formation on the peak positions, we prepared a sam- % of DAT (molar ratio ⫽ 9:2:1.35). Such a composition
ple (IIbis) of crosslinked EPON825 with DAT in the would give a O1s/C1s area ratio of 0.217 and a N1s/C1s
absence of PPG by using the same conditions of tem- area ratio of 0.0344 very close to the experimental
perature and stoichiometry of sample II. The resulting values. Through the same elaboration, it has been
spectra showed C1s, O1s, and N1s envelopes centered possible to calculate the amounts of PPG present in all
at 285.0, 533.0, and 399.5 eV, respectively (Fig. 5). the samples (Table III). A first analysis of the data
reported in Table III shows that, in each sample, the
abundance of the components varies as a function of
the take-off angle. Remembering that the higher the
take-off angle is, the deeper the analyzed layer, it is
clear that at the surface of all samples there is a con-
centration gradient and that the PPG concentration is
higher in the outermost layers. Moreover, analyzing
the four sets of output for the different microspheres,
it can be concluded that the gradient of concentration
of PPG is present independently from solute concen-
tration as well as molecular weight of solvent.
Having evaluated by this method the amount of
PPG present at the sample surface, the following step
was the elimination of the PPG contribution from the
C1s and O1s area values, knowing that in EPON825
there is a proportion of 21 carbon atoms for every 4
Figure 4 ESCA C1s envelope of sample II. oxygen atoms, whereas in the PPG repeating unit, the
REACTIVE MICROSPHERES AS ACTIVE FILLERS 2037

TABLE III respond to a N1s/C1s ratio of 0.0408, the results clearly

Weight % of PPG Present on the Analyzed Layer and show that in the outermost layers there is an excess of
N1s/C1s Atomic Ratios Corrected from PPG
DAT in each of the four samples. The amount of this
Contribution for Samples I–IV
excess, valuable from the value of the N1s/C1s ratio at
Recalculated a take angle of 25°, in the case of sample II reaches the
Sample Wt % of PPG N1s/C1s
35 wt %.
I (t.o.a.25°) 28 0.0488
I (t.o.a.45°) 25 0.0422
I (t.o.a.90°) 22 0.0424
II (t.o.a.25°) 38 0.0522 Rheological analysis results
II (t.o.a.45°) 20 0.0432
II (t.o.a.90°) 17 0.0357 Figure 6 reports the three curves of viscosity versus
III (t.o.a.25°) 33 0.0584 temperature obtained for the 0% MICs, 10% MICs, and
III (t.o.a.45°) 21 0.0457 20% MICs systems, respectively, at a heating rate of
III (t.o.a.90°) 22 0.0450 2°C/min. The temperature value for the minimum
IV (t.o.a.25°) 28 0.0317
IV (t.o.a.45°) 22 0.0460
viscosity decreases with increasing the percentage of
IV (t.o.a.90°) 19 0.0421 microspheres in the sample, thus indicating that an
increase of the microsphere percentage is responsible
for a higher reactivity of the system. This effect may be
related to the presence of reactive amino groups on the
atomic ratio carbon/oxygen is 3 to 1. In the same way, microsphere surface, as determined by ESCA. Further-
the contribution deriving from DAT was calculated more, the figure shows that viscosity increases over
considering that each nitrogen atom present at the the entire temperature range with the amount of mi-
surface theoretically carries with itself 3.5 carbon at- crospheres. This result is promising for the reduction
oms. These calculations allowed us to remove the of the bleeding phenomena and thus makes the mi-
contribution because of PPG from the C1s area values crospheres particularly attractive as rheology control
and to recalculate the N1s/C1s area ratio reported in modifiers for fiber-reinforced composites and adhe-
Table III. sives.
A first analysis of these results shows the presence Figure 7 compares the EPON828/3,3⬘DDS/20%
of a DAT concentration gradient with an increase of MIC with the EPON828/3,3⬘DDS/15% PES180 sys-
the curing agent moving toward the surface. More- tem. The viscosity increase is of the same order of
over, knowing that the presence of a stoichiometric magnitude in both cases. Moreover, even if the copol-
amount of DAT with respect to EPON825 would cor- ymer is aminoterminated, as reported by literature,

Figure 6 Viscosity versus temperature curves for the 0% MICs, 10% MICs, 20% MICs systems.
2038 CARFAGNA ET AL.

Figure 7 Viscosity versus temperature curves for the 20% MICs, 15% PES180 0% MICs, 0% PES180 0% MICs systems.

the influence on the reactivity is higher in the case of order to relate the amount of microparticles to the
microspheres compared with the thermoplastic. This curing reaction, as well as to build up a model to use
makes microspheres useful modifiers for medium as a useful tool for predicting the gel times, as the
temperature cycles. point where the dynamic storage G⬘ and loss G⬙ mod-
Figures 8, 9, and 10 report the isothermal test per- uli crossover (i.e., tan ␦ ⫽ 1), according to the criterion
formed on 0% MICs, 10% MICs, and 20% MICs sys- proposed by Tung and Dynes,13 at different tempera-
tems, respectively. These tests were carried out in tures. This model turns out to be very useful for the

Figure 8 Viscosity versus time curves for the 0% MICs, 10% MICs, 20% MICs systems (T ⫽ 120°C).
REACTIVE MICROSPHERES AS ACTIVE FILLERS 2039

Figure 9 Viscosity versus time curves for the 0% MICs, 10% MICs, 20% MICs systems (T ⫽ 140°C).

selection of proper processing conditions. Rheological that cure reactions may be described by equation
tests performed at different temperatures confirmed containing one apparent activation energy,14,15 as
that with increasing the percentage of microspheres, follows:
viscosity and reactivity become higher. In fact, gel
times decrease as the amount of microspheres is
raised. An overall activation energy for polymeriza-
tion can be calculated from gelation times, assuming
dx
dt
⫽ A exp 冉 冊
⫺ Ea
RT
f共x兲 (1)

Figure 10 Viscosity versus time curves for the 0% MICs, 10% MICs, 20% MICs systems (T ⫽ 150°C).
2040 CARFAGNA ET AL.

Figure 11 In Tgel versus 1/T curves for the 10% MICs and 20% MICs systems.

where A is a constant, Ea is the apparent activation the experiments is therefore based on a linear interpo-
energy for the overall reaction, R is the gas constant, T lation of eq. (3).
is the absolute isothermal cure temperature, and f(x) is Figure 11 shows the interpolations obtained for the
a function of the reaction mechanism and the extent of 10% MICs and 20% MICs systems. The calculation of
reaction x. Integration of this equation from x ⫽ 0 to x parameters, intersection, and slope of eq. (3) enables
⫽ xgel by taking natural logarithms leads to us to draw out the gel_time versus temperature curves
shown in Figure 12. Their trend represents a further

ln 冕0
xgel
dx
f共x兲
⫽ ln A ⫹ ln 共Tgel兲 ⫺ 冉 冊
Ea
RT
(2)
confirmation of the higher reactivity related to the
presence of microspheres.

Based on the Flory theory,16 the extent of reaction at DMTA analysis results
gelation is constant, so the above equation can be Figure 13 reports the storage modulus (G⬘) behavior
expressed as versus temperature for the systems with 0, 10, and 20
wt % of microspheres. The diagram shows a clear flex
Ea point associated with the glass transition temperature
ln共Tgel兲 ⫽ C ⫹ (3)
RT for all the systems. The Tg increases with the amount
of the microspheres. This effect can be attributed to an
where C is a constant. The model utilized to predict increase of crosslink density associated with the mul-
gel times at temperatures different from those used in tifunctional nature of the microspheres that, as it has

Figure 12 Gel time versus temperature curves for the 0% MICs, 10% MICs, 20 % MICs systems.
REACTIVE MICROSPHERES AS ACTIVE FILLERS 2041

Figure 13 Storage modulus (G⬘) versus temperature behavior for cured samples.

been showed previously by ESCA, are characterized the systems modified by the addition of the micro-
by the presence of reactive amino groups on the sur- spheres and reports the behavior of the PES180 mod-
face. ified systems for comparison.
The storage modulus G⬘ (Fig. 13) in the rubbery region The increase of crosslink densities due to the addi-
is shown to increase with increasing percentages of mi- tion of the microspheres causes a considerable in-
crospheres. Therefore, if we consider the relationship (4) crease of Tg. Figure 14 reports the curves for the sys-
drawn from the theory of the rubber elasticity, that links tems modified by the addition of the thermoplastic
the molecular weight between the crosslink points (Mc) copolymer PES180. The comparison shows that the
to the storage modulus G⬘, we can conclude that the 20% MICs blend has a Tg of the same order to the
network formed by the addition of the microspheres system 30% PES180. Therefore, it can be concluded
results in higher crosslink densities that the reactive synthesized microspheres constitute a
valid and economic alternative to the use of thermo-
3 ␳ RT plastics in those applications where the Tg increase is
Mc ⫽ (4) a primary issue.
G

where ␳ is the density of the resin, R is the universal

LEFM analysis results
constant, T is the temperature, and G⬘ is the storage
modulus in the rubbery region. Fracture mechanic tests were carried out in bending
Figure 14 reports the tan_delta (G⬙/G⬘) curves for mode and quasistatic speed to characterize the micro-

Figure 14 Tan_Delta versus temperature behavior for cured samples.

2042 CARFAGNA ET AL.

Figure 15 Energy of fracture (Gc) and stress intensity factor (Kc) for samples modified with microspheres and thermoplastic
copolymer PES180.

mechanical behavior of modified samples. Figure 15 sistance can be considered an advantageous feature
reports the Gc and Kc values obtained. The Gc and Kc for future application of the microspheres.
showed slightly increased at higher content of the
microspheres. The fracture energy factor at 20% load-
SEM analysis
ing of microspheres has a value comparable to the
systems modified by the addition of 15% of PES180. SEM analysis provided useful information about mi-
Due to the crosslinked nature of the microspheres, it crosphere segregation, degree of adhesion with the
can be attributed to a crack pinning mechanism, as matrix, and fracture behavior.
reported for rigid particles,17 but further studies are Figure 16 shows the fracture surface of a sample
required to better understand the fracture mechanical fractured in N2. By a careful observation of the resin
behavior. However, the slight increase in fracture re- surface, it can be seen that in some part of the sample

Figure 16 SEM image showing the morphology of 10% MICs sample fractured in N2 (magnification, ⫻724K).
REACTIVE MICROSPHERES AS ACTIVE FILLERS 2043

Figure 17 SEM image showing the morphology of 10% MICs sample after LEFM tests (magnification, ⫻724K).

the epoxy microspheres appear to be partially covered the microsphere surfaces. This explains the high reac-
by the resin, indicating that the degree of adhesion be- tivity of microspheres towards epoxy resins.
tween them is high. In other parts, spherical voids are This reactivity was confirmed by the rheological
well detectable, indicating that fracture carried out in results. Experiments performed on EPON828 resin
liquid nitrogen removes microspheres from the matrix charged with 10 and 20 wt % microspheres led to the
as a consequence of lack of adhesion at the interface. conclusion that an increase of the microsphere per-
For samples fractured after LEFM test at room tem- centage was responsible for a higher reactivity of the
perature, no microspheres are detectable, indicating system, because the temperature value for the mini-
that, upon fracture, the matrix plastically deforms, mum viscosity decreased as the percentage of micro-
covering the inclusions (Fig. 17). spheres in the sample increased. This effect may be
related to the presence of reactive amino groups on the
microsphere surfaces.
CONCLUSION
The influence of microspheres on the reactivity is
In this article, the influence of microspheres based on higher compared with the thermoplastic usually uti-
epoxy resins on the viscosity, dynamic–mechanical lized in these cases and this makes microspheres use-
properties, and fracture toughness of an epoxy systems ful modifiers for medium temperature cycles.
was investigated. For this purpose, a three-step study The addition of the microspheres has proved to be
was followed, consisting of the preparation of epoxy very effective in increasing the Tg of the blends with-
microparticles, the characterization of their morphology out sacrificing the fracture resistance but slighting in-
and reactivity of the outermost layer, and the rheological creasing the Gc and Kc values. Moreover, due to the
and mechanical characteristics of epoxy resin charged presence of reactive amino groups, the adhesion be-
with different amounts of microparticles. tween microspheres and matrix is improved, as in-
The preparation of microparticles was carried out in ferred from SEM analysis.
four different experimental conditions, in which the Therefore, it can be concluded that the microspheres
composition of the starting solutions (i.e., the epoxy constituted a valid and economically advantageous
⫹ amine to solvent ratio) and the viscosity of solvent alternative for high engineering thermoplastic in those
were changed. The temperature was kept constant in applications where thermal resistance is a primary
all cases. This was done with the aim of understanding requisite and toughness is not a critical issue.
whether the change in experimental parameters was
responsible for different characteristics of the recov- The financial support from MIUR (COFIN 2001) is gratefully
ered microparticles. acknowledged.
Actually, from ESCA analysis, it was inferred that
the PPG content was constant in the four samples (it References
only varied with the take-off angles), and that an 1. Pascault, J. P.; Valette, L.; Barbeau, P.; Magny, B. PCT Pat.
excess of DAT (used as curing agent) was present on WO0059953, 2000.
2044 CARFAGNA ET AL.

2. Geisler, J. P.; Petri, S. U.S. Pat. 5,358,982, 1994. 11. Beamson, G.; Briggs, D. in High Resolution XPS of Organic
3. Okamatsu, T.; Ochi, M. Polymer 2002, 43, 721. Polymers—The Scientia ESCA300 Database; Wiley; Chichester, J
4. Hattori, M.; Sudol, E. D.; El-Asser, M. S., J Appl Polym Sci. 1993, 1992.
50, 2027. 12. Watts, J. F.; Abel, M. L.; Perruchot, C.; Lowe, C.; Maxted, J. T.;
5. Saenz, J. M.; Asua, J. M. J Polym Sci, Part A: Polym Chem. 1995, White, R. G.; J Electr Spectros Rel Phen. 2001, 121, 233.
33, 1511 13. Tung, C. M.; Dynes, J. P. J Appl Polym Sci. 1982, 27, 569.
6. Chai, Z.; Zheng, X.; Sun, X. J Polym Sci, Part B: Polym Phys. 14. Barton, J. M.; Greenfield, D. C. L.; Hodd, K. Polymer 1992, 33, 1177.
2003, 41, 159. 15. Oyanguren, P. A.; Williams, R. J. J. J Appl Polym Sci. 1993, 47, 1361.
7. Kawaguchi, H. Progr Polym Sci 2000, 25, 1171. 16. Flory, J. P. in Principles of Polymer Chemistry; Cornell Univ.
8. Hibino, K.; Kimura, Y.; Colloid Polym Sci. 2000, 278, 565. Press; Ithaca, NY, 1953; (a) Moore, D. R.; Pavan, A.; Williams,
9. Hseih, H. K.; Woo, E. M.; J Polym Sci, Part B: Polym Phys. 1996, J. G. Fracture Mechanics Testing Methods for Polymers, Adhe-
34, 2591. sives and Composites. ESIS publication 28; Elsevier; Oxford,
10. Mamo, A.; Orestano, A.; Pollicino, A.; Recca, A. J Polym Eng. 2001.
2002, 22, 209. 17. Lange, F. F. Philos Mag 1970, 22, 11.