Multiscale perspectives of interface delamination in microelectronic packaging applications

Abstract

With the increasing complexity and ongoing miniaturisation of microelectronic systems, reliability issues and their associated structural dimensions cross over from the microscale to the nanoscale. From this perspective, fracture of materials and material interfaces for microelectronic components is essentially a multiscale process. In this paper, interface delamination at the individual scales (atomistic, meso and micro) is considered, and specific analysis methods are discussed in order to compile understanding of contributions from each scale towards the macroscale response of an epoxy moulding compound. As will be addressed, the contributions from each scale can be applied to the next scale, and so the multiscale impact is derived sequentially rather than simultaneously in a single model. First, results on each scale are presented, considering the multilevel impact

Using Coarse-Grained Molecular Models (Molecular-Mesoscale) of a Copper Oxide-Epoxy Interface to Obtain Stress–Strain Failure Predictions Which Include Interfacial Roughness, Water and Filler Effects

“…From these examples, it appears that a bond breaking criterion is needed for simulations which qualitatively match expected reality. Previous studies [36][37][38] used perfectly flat copper oxide-epoxy interfaces, however in reality materials, do not exhibit perfect flatness [1,42], so one question that arises is whether the effect of roughness can be captured using the same coarsegraining method. In order to approach the interface roughness question, in relationship to how the polymer may be coupled into the adhesive interface [43][44][45][46][47][48], the polymer structure was fixed at the highest crosslink case studied previously [36][37][38] (all epoxies reacted with a phenolic unit from bisphenol A) and only the copper oxide-polymer interface was considered.…”

“…The derivation of the basic bead bond, bead non-bond, bond stretch and non-bond parameters were previously described [36][37][38][39][40] and are shown in Table 1 in which both the cohesive and adhesive bead bond and non-bond parameters were derived from atomistic dynamics models of interacting repeat units. The bond stretch was estimated from the end-end length of the repeat unit and the non-bond distance was approximated from molecular models as average closest distances between two units optimized together.…”

“…The entire novolac epoxy plus its' bisphenol A curative represented one bead, as described previously [36]. As before with the perfectly smooth interfaces [36][37][38], the polymer was incrementally compacted onto the Cu 2 O bead surface in order to attain a global minimum before the interfaces were separated by incrementally moving the Cu 2 O layer into the vacuum space.…”

“…In all of these studies, the previously described polymer has been used [36][37][38][39][40] comprising a novolac epoxy (using its' repeat unit as the fundamental coarsegrained bead), and a bead layer of copper oxide coarse-grained using the copper (I) oxide unit cell. Building of the coarse-grained layers starts from the molecular repeat unit (Fig.…”

“…For this reason, coarse-grained mesoscale (or what we have called molecular-mesoscale or mesoscopic) methods were investigated [36][37][38][39].…”