Multiscale Modeling of Epoxies and Epoxy-Based Composites
285
Glass- and carbon-fiber reinforced epoxy composites have been widely studied
both experimentally and computationally [8, 105]. One example is single and
multiwall carbon nanotube (CNT)-reinforced nanocomposites, which are of interest
because of their wide potential applications [23, 106, 107]. Although there is no
chemical bonding between the CNT and epoxy, the electrostatic and van der Waals
forces are strong enough to create a tightly bonded interface [23].
Putz et al. conducted dynamic scanning calorimetry (DSC) testing and showed
that introducing multiwall CNTs in an epoxy matrix alters the T g depending
on the degrees of cross-linking of the matrix [8]. In particular, the inclusion of
nanoparticles results in an increase/decrease in T g at low/high degrees of crosslinking. From their observation, the authors proposed three mechanisms that can
be used to explain the interphase phenomenon (see Fig. 9): (1) incomplete curing
of the network near the interphase region induced by the limited mobility of the
prepolymers and cross-linker leads to the formation of dangling unreacted end
groups of epoxide chains; (2) the epoxy curing process is disrupted at the interphase
region, and the formed network is different than the bulk epoxy; and (3) there is
a retarded dynamic at the interphase so that epoxy forms a distinguished structure
involving physical and/or chemical interactions with the fiber surfaces. All three
mechanisms can happen at the interface region, and various simulation efforts have
been conducted to understand the exact mechanisms, not only for T g but also for
structural and mechanical properties as well. Note that fiber/epoxy interface is
different from fiber/plastic interface, as the latter has relatively long un-cross-linked
polymer molecules and physical adsorption and wrapping are most likely to happen
at the interface [108].
All-atom MD and CG-MD simulations of nanocomposites typically include
one nano-sized fiber/particle and a surrounding epoxy matrix to represent a small
region of the nanocomposites. The resulting material properties obtained from these
simulations can be considered as local properties of the nanocomposites. Langeloth
et al. [52] studied a DGEBA/DETA epoxy system as a bulk material and with
a solid surface using CG-MD models. They found that T g is a function of the
conversion degree and it is a local material property that changes at a 3 nm region
around the fiber (the interphase region). A multiscale study from Choi et al. [81]
showed that a soft and slippery layer of a polymer at the CNT/DGEBA interface
reduces the mechanical strength in the transverse and shear directions compared
to neat epoxies. They have also bridged between the MD and FEM models using
an iterative matching process to obtain key parameters (i.e., interphase thickness
of 1.14–1.92 nm and strain energy density) for the FEM model (see Fig. 10a).
Kim et al. also used a matching process between MD and a continuum model to
create an effective interphase region between silica particle and epoxy matrix [22].
They found the interface structural conformation to depend on the degree of crosslinking of the epoxy. Liu et al. proposed a novel multiscale simulation approach
that combined MD and CG-MD simulations to create a cured epoxy network with
a carbon fiber substrate and investigated the diffusion effect on the cured structure
[109]. Based on Koo et al.’s work of neat epoxy [50], Subramanian et al. created an
MD simulation of CNT/epoxy nanocomposites and modeled its fracture behaviors
285
Glass- and carbon-fiber reinforced epoxy composites have been widely studied
both experimentally and computationally [8, 105]. One example is single and
multiwall carbon nanotube (CNT)-reinforced nanocomposites, which are of interest
because of their wide potential applications [23, 106, 107]. Although there is no
chemical bonding between the CNT and epoxy, the electrostatic and van der Waals
forces are strong enough to create a tightly bonded interface [23].
Putz et al. conducted dynamic scanning calorimetry (DSC) testing and showed
that introducing multiwall CNTs in an epoxy matrix alters the T g depending
on the degrees of cross-linking of the matrix [8]. In particular, the inclusion of
nanoparticles results in an increase/decrease in T g at low/high degrees of crosslinking. From their observation, the authors proposed three mechanisms that can
be used to explain the interphase phenomenon (see Fig. 9): (1) incomplete curing
of the network near the interphase region induced by the limited mobility of the
prepolymers and cross-linker leads to the formation of dangling unreacted end
groups of epoxide chains; (2) the epoxy curing process is disrupted at the interphase
region, and the formed network is different than the bulk epoxy; and (3) there is
a retarded dynamic at the interphase so that epoxy forms a distinguished structure
involving physical and/or chemical interactions with the fiber surfaces. All three
mechanisms can happen at the interface region, and various simulation efforts have
been conducted to understand the exact mechanisms, not only for T g but also for
structural and mechanical properties as well. Note that fiber/epoxy interface is
different from fiber/plastic interface, as the latter has relatively long un-cross-linked
polymer molecules and physical adsorption and wrapping are most likely to happen
at the interface [108].
All-atom MD and CG-MD simulations of nanocomposites typically include
one nano-sized fiber/particle and a surrounding epoxy matrix to represent a small
region of the nanocomposites. The resulting material properties obtained from these
simulations can be considered as local properties of the nanocomposites. Langeloth
et al. [52] studied a DGEBA/DETA epoxy system as a bulk material and with
a solid surface using CG-MD models. They found that T g is a function of the
conversion degree and it is a local material property that changes at a 3 nm region
around the fiber (the interphase region). A multiscale study from Choi et al. [81]
showed that a soft and slippery layer of a polymer at the CNT/DGEBA interface
reduces the mechanical strength in the transverse and shear directions compared
to neat epoxies. They have also bridged between the MD and FEM models using
an iterative matching process to obtain key parameters (i.e., interphase thickness
of 1.14–1.92 nm and strain energy density) for the FEM model (see Fig. 10a).
Kim et al. also used a matching process between MD and a continuum model to
create an effective interphase region between silica particle and epoxy matrix [22].
They found the interface structural conformation to depend on the degree of crosslinking of the epoxy. Liu et al. proposed a novel multiscale simulation approach
that combined MD and CG-MD simulations to create a cured epoxy network with
a carbon fiber substrate and investigated the diffusion effect on the cured structure
[109]. Based on Koo et al.’s work of neat epoxy [50], Subramanian et al. created an
MD simulation of CNT/epoxy nanocomposites and modeled its fracture behaviors