A hybrid molecular dynamics (MD) simulation framework is developed to emulate mechanochemical reaction of mechanophores in epoxy-based nanocomposites. Two different force fields, a classical force field and a bond order based force field are hybridized to mimic the experimental processes from specimen preparation to mechanical loading test. Ultra-violet photodimerization for mechanophore synthesis and epoxy curing for thermoset polymer generation are successfully simulated by developing a numerical covalent bond generation method using the classical force field within the framework. Mechanical loading tests to activate mechanophores are also virtually conducted by deforming the volume of a simulation unit cell. The unit cell deformation leads to covalent bond elongation and subsequent bond breakage, which is captured using the bond order based force field. The outcome of the virtual loading test is used for local work analysis, which enables a quantitative study of mechanophore activation. Through the local work analysis, the onset and evolution of mechanophore activation indicating damage initiation and propagation are estimated; ultimately, the mechanophore sensitivity to external stress is evaluated. The virtual loading tests also provide accurate estimations of mechanical properties such as elastic, shear, bulk modulus, yield strain/strength, and Poisson’s ratio of the system. Experimental studies are performed in conjunction with the simulation work to validate the hybrid MD simulation framework. Less than 2% error in estimations of glass transition temperature (Tg) is observed with experimentally measured Tgs by use of differential scanning calorimetry. Virtual loading tests successfully reproduce the stress-strain curve capturing the effect of mechanophore inclusion on mechanical properties of epoxy polymer; comparable changes in Young’s modulus and yield strength are observed in experiments and simulations. Early damage signal detection, which is identified in experiments by observing increased intensity before the yield strain, is captured in simulations by showing that the critical strain representing the onset of the mechanophore activation occurs before the estimated yield strain. It is anticipated that the experimentally validated hybrid MD framework presented in this dissertation will provide a low-cost alternative to additional experiments that are required for optimizing material design parameters to improve damage sensing capability and mechanical properties.
In addition to the study of mechanochemical reaction analysis, an atomistic model of interphase in carbon fiber reinforced composites is developed. Physical entanglement between semi-crystalline carbon fiber surface and polymer matrix is captured by introducing voids in multiple graphene layers, which allow polymer matrix to intertwine with graphene layers. The hybrid MD framework is used with some modifications to estimate interphase properties that include the effect of the physical entanglement. The results are compared with existing carbon fiber surface models that assume that carbon fiber has a crystalline structure and hence are unable to capture the physical entanglement. Results indicate that the current model shows larger stress gradients across the material interphase. These large stress gradients increase the viscoplasticity and damage effects at the interphase. The results are important for improved prediction of the nonlinear response and damage evolution in composite materials.