A novel concept of tuning the fracture properties of the interface through the treatment process of the coupling layer according to the cohesive critical strain energy release rate of the epoxy is proposed for optimizing the joint strength between epoxy and copper substrate. In most coupling agent application recipes, the treatment condition design has omitted the influence of the fracture properties of the corresponding adhesive. Conceivably, excessive strengthening of the adhesive-substrate interface may not lead to optimal interfacial strength. Synergistic toughening of the interface takes place when there is simultaneous interfacial debonding and failure of adhesive under a comparable critical stress state. Under critical applied load, energy is concurrently dissipated through the fracture of the interface, the fracture in the adhesive, and possible non-reversible failure processes such as shear yielding or micro-cracking of the adhesive. These combined energy dissipation processes result in extensive energy absorption around the crack tip. The adhesive joint, therefore, becomes more crack resistant. In this study, the interfacial adhesion promotion concept with synergistic toughening was demonstrated using three different epoxy systems bonded to copper substrates modified by a thiol-based coupling layer. The coupling layer was formed by treating the copper substrate with a thiol-based coupling agent. Critical strain energy release rate of the treated tapered double cantilever beam samples in different treatment conditions was measured for each of the epoxy systems. From the failure path analysis, mixed interfacial and cohesive failure was observed. This observation indicated that extensive energy dissipation occurs around the crack tip that results in synergistic toughening of the interface. This work shows the significance of matching the fracture property of the coupling layer with the adhesive. Up to 2.3 times improvement in the critical strain energy release rate was achieved with optimized thiol treatment compared to non-optimized treatment. (C) Koninklijke Brill NV, Leiden, 2011