Experimental studies on epoxy resins and their composites show that water has two effects on their dynamic moduli. First, their glass (or α-) transition point shifts to lower temperatures as the water concentration increases in the sample. Second, an unexplained ω-transition appears at a temperature below the glass transition temperature. The magnitude of this transition increases with the water contend in the sample. The magnitude of the ω-transition also increases in composites with poor interfaces that absorb more water than composites with good interfaces. The temperature of the ω-transition varies little with the water content and is not strongly dependent on the magnitude of the transition. An increase in the ω-transition loss modulus peak accompanies a decrease in the α-transition loss modulus peak.
The problem is to identify a molecular-level or microstructural mechanism that can create the various phenomena observed in the form of the ω-transition.
The experimental evidence from composites with poor interfaces suggests that the ω-transition is most likely a microstructural phenomenon and supports the following hypothesis: the ω-transition is simply the α-transition of a new phase formed in the composite due to a heterogeneous distribution of water in the matrix.
Microstructural inhomogeneity in the epoxy resin causes certain regions to have a higher affinity for water than others. The physical characteristics of the epoxy network structure and the chemical characteristics of the polymer chain that contribute to this uneven affinity for water are not clearly understood at present. However, possible causes include uneven crosslink densities due to uneven hardener concentration, trapped air bubbles, other trapped holes, or de-bonds at the filler-matrix interface. These defects promote water accumulation in the matrix surrounding the. The water-saturated regions are large enough to behave as a separate, dispersed, mocroscopic phase. The presence of this phase is observed when excess water shifts its α-transition temperature to a value lower than that of the rest of the matrix. We will refer to this phase as the soaked phase or the water-rich phase.
Microscopic-sized regions of the soaked phase of nearly identical composition are, therefore, assumed to be distributed within the neat resin. De-bonded matrix at the interface in composites with poor filler-matrix interfaces also provide similar conditions for water saturation. Therefore, water collected at the interface contributes to an increase in the total volume fraction of the water-rich resin phase.
The hypothesis will be tested by comparing the ω-transition observed in experiments to the effective volume-average dynamic moduli of epoxy resins and composites containing different volume fractions of the water-rich phase. This calculation had to be done in two stages. The first stage involves prediction of the dynamic viscoelastic moduli of the hypothesized water-rich phase. In the second stage, the dynamic moduli of the different phases in the real epoxy resins and composites are volume-averaged to obtain the effective dynamic modulus of the heterogeneous material. The accuracy of the calculated properties of the heterogeneous media will depend on the accuracy of the predicted physical properties, the volume-averaging schemes, and the accuracy of the model of the microstructure in the real material.