A representative molecular model of the carbon-reinforced PEEK interphase with stochastic vacancy defects

Carbon fibre-reinforced polymer composites rely on efficient stress transfer across the fibre-matrix interphase. Yet, real carbon fibre surfaces exhibit nanoscale roughness and vacancy disorder rarely captured in molecular dynamics (MD) models, which typically assume a pristine graphitic surface. This is particularly limiting for thermoplastic matrices such as poly(ether ether ketone) (PEEK), which lack functional groups capable of covalent crosslinking and rely primarily on π - π interactions with the fibre surface, making surface morphology a key determinant of interfacial performance. Here, a stochastic, Perlin-noise-based defect-generation framework is developed to produce graphene-based reinforcements with spatially correlated vacancy defects, motivated by the nanoscale roughness observed via SEM and high-speed AFM characterisation of IM7-GP carbon fibre. The resulting defective reinforcements show progressive degradation in elastic properties, with Young’s and shear moduli falling over 80% from pristine to the highest defect threshold, consistent with loss of lattice continuity. Nine stacked configurations were combined with an experimentally validated PEEK matrix to isolate the effect of surface-vacancy morphology on interfacial shear stress (ISS). The defect-free interphase yields composites with an ISS of 63 ± 4 MPa, while low-to-moderate vacancy fractions produce no significant change. In contrast, large, spatially connected vacancy clusters increase ISS by 37 − 44 %, a trend confirmed across two pull-out rates, indicating ISS is controlled by mechanical anchoring of PEEK chains within the vacancy network during pull-out. These results show that the framework generates physically meaningful defect morphologies without an annealing step, and demonstrate that vacancy architecture influences interfacial performance, guiding carbon fibre surface design in thermoplastic composites.