IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 3, NO. 11, NOVEMBER 2013 1859 Experiments and Three-Dimensional Modeling of Delamination in an Encapsulated Microelectronic Package Under Thermal Loading Siow Ling Ho, Shailendra P. Joshi, and Andrew A. O. Tay, Member, IEEE Abstract— Interfacial delamination in encapsulated silicon devices has been a great reliability concern in IC packaging. Experimental testing of a transparent quad flat no leads package is carried out with the goal of studying the delamination characteristics and investigating the viability of cohesive zone modeling in simulating delamination patterns and trends. The pattern of initiation and propagation of delamination under thermal loading is the focus of this paper. A microscope is focused on the interface between the pad and the encapsulant to capture the progressive delamination in a package that was molded without a die. When the temperature reaches a critical value, delaminations are observed to initiate and propagate in a certain pattern. The experimental setup is then modeled within the finite element framework, with the failure of the interface described through a cohesive-zone surface interaction approach. With a slight modification to the experimental procedure and through a separate finite element model, the fracture energy of the interface is estimated. It is found that the 3-D numerical model is able to capture the experimentally observed delamination pattern satisfactorily. Index Terms— Coefficient of thermal expansion mismatch, cohesive zone model (CZM), delamination, delamination initiation, delamination propagation, electronics packaging, encapsulant, failure analysis, failure pattern, finite element methods (FEMs), fracture mechanics, interface fracture, interfa- cial fracture energy, quad flat no leads (QFN) package, reliability, stress concentration thermal stress, thermomechanical stress. I. I NTRODUCTION O BSERVATIONS of failures in microelectronicspackages reveal that delamination is a major concern. Among the interfaces, between the copper leadframe pad and the epoxy molding compound, is highly susceptible to delamina- tion. Copper is widely used as a leadframe material because of its superior electrical and thermal conductivity but its adhesion to epoxy is relatively weak. At elevated temper- atures, residual stresses that arise from thermal mismatch between the lead frame and the epoxy in combination with the degradation of adhesion strength with temperature can Manuscript received May 21, 2012; revised January 21, 2013; accepted May 7, 2013. Date of publication July 29, 2013; date of current version October 28, 2013. Recommended for publication by Associate Editor G. Q. Zhang upon evaluation of reviewers’ comments. S. L. Ho is with the Institute of Microelectronics, Agency for Science, Tech- nology and Research, 117685 Singapore (e-mail: hosl@ime.a-star.edu.sg). S. P. Joshi and A. A. O. Tay are with the Department of Mechanical Engineering, National University of Singapore, 117576 Singapore (e-mail: mpejsp@nus.edu.sg; mpetayao@nus.edu.sg). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TCPMT.2013.2266406 lead to delamination during solder reflow or qualification tests. Experimentally, details such as the location of the initiation of delamination, shape of the crack front, and delamination propagation characteristics with increased thermal loading are inadequately addressed. On the numerical analysis front, linear elastic fracture mechanics have been applied to predict delamination in IC packages [1]–[5]. Past research primarily focused on the capabilities of numerical methods to predict delamination by comparing the inferred conclusions from numerical study with the experimental results. For example, by comparing the energy release rate (ERR) obtained from the virtual crack closure method (VCCM) for two designs, Hu et al. [6] predicted that the design of the stacked die ball grid array (BGA) was more prone to delamination which corroborated well with experimental results. Although such an approach is useful, it suffers from the disadvantage that an a priori assumption of the location, size, and shape of an initial crack is required. In contrast, no knowledge of the initial crack is required in cohesive zone modeling (CZM), which has become a popular numerical method for modeling delamination initiation and propagation in microelectronics packages. Experimental studies of delamination in electron- ics packaging using C-mode scanning acoustic microscopy (C-SAM) have been reported [7], but this method could only give delamination patterns at a particular instant but not capture the initiation and progressive delamination in the package, especially during catastrophic delamination stages that are inherently unstable. The demand for realistic simulations arises because experi- ments and physical testing are expensive and time consuming. As a step toward realistic virtual prototyping, there is a need for a numerical approach that is able to explicitly simulate the delamination process, including initiation and propagation. CZM is a promising approach in this regard and is currently an active area of research [8]–[14]. Among its numerous applica- tions, CZM can be employed in parametric studies of materials and geometric variations. Tambat et al. [11] studied the impact of material parameters on the failure of interlayer dielectric structures, while Xiaopei et al. [9] addressed the impact of CZM parameters, material properties, and geometrical varia- tions on delamination in an embedded die package. Although significant insights can be derived from numerical studies, it is of equal importance for a numerical model to be able to capture physical observations. A number of works had cor- related numerical results with experimental observations. Lall 2156-3950 © 2013 IEEE