Molecular Dynamics (MD) simulations were used in investigating interface adhesion in glass fibre reinforced polymer composites. Novel contributions were made for; (i ) determining the solubility limit of water in a crosslinked epoxy system, (ii ) generating slab composite interfaces, (ii ) determining the solubility limit of water in at composite interfaces, and (iv ) quantifying interfacial adhesion in a slab composite for both dry and saturated systems.

Procedures were developed for uses in modelling crosslinked DGEBA/IPD epoxy systems. A new term was proposed, the reduced chemical potential µ˜, and was applied in assessing the solubility of water in a polymeric system. DGEBA/IPD was found to have a moisture solubility of 3.50-3.75 wt.%. Experimentally, an average maximum water content was found as 2.66 wt.%. The average simulated elastic modulus for DGEBA/IPD saturated to 3.6 wt.% water was found as 4.11 GPa, a decrease of 16.5% from the dry system. An experimental value of 3.74 GPa was found, a decrease of 13.8%. The onset of the glass transition temperature was also considered, results indicate a significant decrease of 17.9% with an average value of 96.48 C, while experimentally a decrease of 15.4% to a value of 91.54 C was seen. The simulated system overestimates the maximum water content, and slightly overestimates the relative loss of properties due to the presence of absorbed moisture.

Ten thermodynamically equivalent slab composite interface structures (containing glass fibre, sizing, and epoxy) were generated. Novel procedures for saturating the system based in Grand Canonical Monte Carlo exchanges of noble water molecules were developed. An average solubility limit of 2.07 wt.% water was found, and calculated to correspond to 0.99 wt.% water in an equivalent full composite, an experimental value of 1.15 wt.% was found.

The work of adhesion was calculated for replicates of dry, moist, and fully saturated slab composites and compared to experimental tensile strength in unidirectional composites. The work of adhesion for the saturated SiO2-sizing interface was found as 267.88 mJ m−2, a 32% decrease. For the sizing-epoxy interface, a value of 191.01 mJ m−2 was found, a decrease of 24%. The SiO2-epoxy interface (representing an unsized fibre) a value of 148.61 mJ m−2 was found, a 29% decrease. Results for both dry and saturated conditions indicated SiO2-sizing as the critical interface for failure. Non-bonded and torsional terms make up the largest portion of adhesion, and that they consistently see the most significant relative decrease with increasing water content. This suggested that the loss of interface adhesion could be minimised through altering the fibre sizing system in a way such that the non-bonded and torsional terms are retained with increasing moisture content.

SEM analysis was undertaken on both dry and saturated samples after tensile failure. The results suggest more cohesive failures in the epoxy for the dry samples, while relatively clean fibre surfaces in the saturated samples indicate adhesive failure at the interface.