Elsevier

Microelectronics Reliability

Volume 49, Issues 9–11, September–November 2009, Pages 1315-1318
Microelectronics Reliability

Numerical prediction of failure paths at a roughened metal/polymer interface

https://doi.org/10.1016/j.microrel.2009.06.019Get rights and content

Abstract

Debonding of polymer–metal interfaces often involves both interfacial and cohesive failure. This paper extends the investigation of Yao and Qu presented in [Yao Q, Qu J. Interfacial versus cohesive failure on polymer–metal interfaces in electronic packaging – effects of interface roughness. J Electr Packag 2002;124;127–34] towards a numerical fracture mechanics model that is used to quantitatively predict the relation between cohesive and adhesive failure on a metal–polymer interface. As example, an epoxy–aluminum interface is investigated. The competition between adhesive and cohesive failure depending on surface roughness parameters will be studied. Understanding of these phenomena could enable the optimization of interface properties for different applications.

Introduction

Delamination of interfaces is one of the key failure mechanisms in semiconductor packaging (see [2]). Failure of solder/underfill, leadframe/moulding compound, leadframe/adhesive or adhesive/silicon interfaces are examples that lead to severe reliability and durability problems. Most of the interfaces in microelectronics are metal(-oxide)/polymer interfaces. Therefore, improvement of the adhesion properties of metal(-oxide)/polymer interfaces to increase thermo-mechanical reliability receives great attention inside packaging communities. In this paper, delamination is defined as the work of separation that contributes to the separation of two materials including all dissipative effects. Hence, not only adhesive failure of the interface, but also cohesive failure inside the bulk materials occurring near the interface contributes to interface separation. This means that interface separation may occur along three different paths (as shown in Fig. 1):

  • 1.

    At the interface (adhesive failure).

  • 2.

    In the material near to the interface (cohesive failure).

  • 3.

    A combination of 1 and 2.

To be able to predict which of these failure paths are most likely to occur, the competition between adhesive and cohesive failure phenomena needs to be investigated. In general the following interactions (see [3], [4]) play a role in adhesion or cohesion of materials and interfaces:

  • 1.

    Chemical interactions: Chemical interactions inside materials and between material surfaces refer to primary bonds, i.e. covalent, ionic or metallic bonds. The interaction scale of the chemical interaction is approximately 0.2–1.0 nm.

  • 2.

    Physical interactions: Physical interactions refer to secondary bonds inside materials and between interfaces, like the Coulomb force or Van der Waals force. Although in general the magnitude of the physical interaction is weaker than the one of the primary bond, it is formed at most interfaces while chemical interaction requires certain chemical conditions to be met. The interaction scale of the physical interaction is approximately 5.0–10.0 nm.

  • 3.

    Mechanical interlocking: Mechanical interlocking is the interaction between material surfaces or inside materials due to geometric effects at macroscopic scale. This interaction is believed to play a dominant role in interface bonding. The scale of surface roughness of metal–polymer interfaces, which is a typical interlocking feature, is in the order of micrometers.

Inside the European NanoInterface project, these three interactions are investigated and a sequential multi-scale framework will be constructed to be able to predict the mechanical behavior of metal–oxide–polymer interfaces. At the molecular scale the chemical and physical interactions will be analyzed by molecular dynamics simulations. The results of this scale will be used in a FE micro-scale analysis to analyze the mechanical interlocking features. Finally, insight in the interaction mechanisms will be used in a macro-scale analysis investigating the reliability of semiconductor packages. The current work is part of the micro-scale investigation and focuses on the effect of mechanical interlocking at the interface due to surface roughness. In this paper, an energy criterion is used to study the conditions and at which position on the interface the crack kinks from the interface into the polymer. For this purpose, a numerical model of an aluminum–epoxy moulding compound (EMC) interface is developed to calculate energy release rate (ERR) values at the different locations. Next, a parameter study shows the dependence of the cohesive/adhesive failure ratio on roughness parameters. This paper ends with conclusions and recommendations for further investigations.

Section snippets

Adhesive and cohesive failure at an interface

In semiconductor processing metal(-oxide) surfaces are microscopically roughened to increase adhesion to polymers. When a polymer material is applied to a rough metal surface, it conforms to the rough surface and tends to fill up the irregularities of the metal surface. During curing of the polymer mechanical interlocks form. In this way roughening of the metal could prevent the interface to decohere completely along the interface path only. Instead, if the metal is assumed to be much tougher

Numerical model to predict the point of crack kinking

Eq. (1) is used to find the point of crack kinking and subsequently to calculate the cohesive/adhesive failure ratio. In [1] an approximate expression is derived to calculate the Gi and Gp terms for the proposed interface profile and to derive the cohesive/adhesive failure ratio once the material and interface properties are given. Because the proposed engineering approach needs to empirically fit two parameters of interest (for more details see [1]), it is not capable to predict the

Results

The ERR was calculated at different positions of the interface. The crack position at the incline is indicated by parameter ε ranging from 0 (crack at bottom) to 1 (crack at top of incline) as indicated in Fig. 3b. In Fig. 4, the Von Mises stress contours for the deformed geometry of the adhesive and the cohesive crack at the middle of the incline (ε = 0.5) are shown. The global deformation of both results is identical, however, at the crack tip the adhesive failure model opens along the

Conclusions

Interface delamination is a critical failure mechanism in semiconductor packaging. Interface separation is possible by decohesion at the interface, decohesion in the bulk material near the interface or a combination of both. In this work delamination of mechanically interlocked polymer/metal interfaces is investigated. To investigate the influence of interface roughness parameters on the fracture path, a parametric FE model is constructed. The example shows that the method enables to predict

Acknowledgement

We thank the European Commission for partial funding of this work under project NanoInterface (NMP-2008-214371).

References (8)

There are more references available in the full text version of this article.

Cited by (14)

  • Modeling and simulation of interface failure in metal-composite hybrids

    2021, Composites Science and Technology
    Citation Excerpt :

    In contrast, there are models within a fracture mechanics concept that explicitly resolve the local interface. This makes it possible to investigate the amount of adhesive and cohesive failure depending on the interface roughness [8,16]. A combination of cohesive zone models with a sinusoidal interface for similar [17,18] and dissimilar materials [19,20] was also presented for studies of pure adhesive failure.

  • Microscale simulation of adhesive and cohesive failure in rough interfaces

    2017, Engineering Fracture Mechanics
    Citation Excerpt :

    Yao and Qu [3] developed a fracture mechanics model to predict the amount of adhesive and cohesive failure in rough polymer-metal interfaces. These investigations were extended to a numerical fracture mechanics model by Noijen et al. [9] and applied to an aluminum-epoxy interface. Pure adhesive mode I failure of sinusoidal structured interfaces were investigated numerically by Zavattieri et al. for identical elastic [10] and elastic-plastic [12] materials.

  • Multi-scale simulations for predicting material properties of a cross-linked polymer

    2015, Computational Materials Science
    Citation Excerpt :

    For a non-stoichiometric mixture (∼90:10 wt. ratio [43]), Poisson’s ratio is reported as 0.345 which is in the range of typical values for various epoxy systems 0.3–0.4 [6,44,45]. The elastic modulus values found in literature show a variation depending on the epoxy-amine composition.

  • An improved atomistic simulation based mixed-mode cohesive zone law considering non-planar crack growth

    2013, International Journal of Solids and Structures
    Citation Excerpt :

    As the crack propagates beneath the interface, it creates a spall that incorporates the film and a portion of the underlying substrate (see e.g. Drory et al. (1988) and Okazaki et al. (2005). Another example is – in polymer/metal interface, often found in electronic packaging, not only adhesive failure (i.e. failure due to crack growth along the interface) but cohesive failure due to crack kinking in the polymer phase and propagating near the interface are often believed to be the cause of interfacial debonding (Noijen et al., 2009). It is therefore apparent that realistic scaled-down atomistic model of these systems to characterize interfacial traction–separation law, could very well exhibit such mechanisms of interfacial debonding at the atomic-to-nano scale particularly under mixed mode loading conditions.

  • Effect of a hard coat layer on buckle delamination of thin ITO layers on a compliant elasto-plastic substrate: An experimental-numerical approach

    2011, Engineering Fracture Mechanics
    Citation Excerpt :

    Consequently, the cohesive zone parameters correspond to the intrinsic interface properties. Clearly, this approach requires a fully coupled multi-scale approach which is a topic of ongoing research, see e.g. [28,29]. In this paper, approach (a) is applied.

View all citing articles on Scopus
View full text