Search

Abstract: Isotropic conductive adhesives (ICAs) are promising candidates for low temperature joining technologies in microelectronics, enabling ultra-fine pitch sizes. Especially in solar and automotive applications, long-term reliability is a prerequisite in new generation electronics. It is essential that reliability predictions take processing history into account in order to correctly address premature failures as well as to make sound long-term predictions. In this paper, residual stresses that develop in a nano-Ag ICA interconnect during the assembly of a flip-chip pin grid array are investigated. A multiscale modeling framework is adopted to link the nano-sized particles to the interconnect level. This is achieved by the numerical analysis of the mechanical response during the curing process through the computational homogenization approach, in which two boundary value problems, one at each scale are formulated and solved simultaneously, in a fully nested manner. The mechanical response of the interconnect is analyzed with respect to the particle volume fraction and distribution properties. It is shown that, although the overall residual stresses at the interconnect scale decrease with increasing the amount of conductive particles, at the particle scale local stress concentrations increase, indicating the possibility of damage and decohesion that might compromise mechanical integrity and interrupt the conductive path. Hence, the multiscale scheme proved crucial for the sound analysis of the nano-particle ICA interconnect problem, where consideration of only the interconnect level would lead to misleading conclusions. [Copyright &y& Elsevier]

This paper presents a novel approach for establishing microstructure statistics-property relations for a silver particle-based thermal interface material (TIM). Several sintered silver TIMs have been prepared under different processing conditions, generating samples with distinct microstructures. The 3D microstructure is revealed and visualized using the combination of Focussed Ion Beam (FIB) milling and Scanning Electron Microscopy (SEM) imaging. Representative synthetic model microstructures have been generated based on Gaussian random field models, having well defined analytical descriptions. The statistical characteristics of the samples and the synthetic models are shown to have a good correspondence, indicating that the linear effective properties of these complex materials can be predicted based on analytical estimates available for the synthetic models. This is verified by computing the effective elastic and thermal material properties using the computational homogenization approach based on the finite element models of the real samples. The computational homogenization, providing the reference solution, and the higher-order statistical estimates for the synthetic models are in very good agreement. These results can be used in the development of new silver particle-based materials, whereby the expensive and time consuming effective material property characterization can be replaced by efficient estimation based on the synthetic random field models. [ABSTRACT FROM AUTHOR]

Locally resonant acoustic metamaterial (LRAM) panels are excellent candidates for the low-frequency flexural wave attenuation in thin structures. To enable the efficient analysis and design of LRAM beam/shell structural elements for practical applications, a computational homogenization method for modelling wave propagation phenomena in LRAM panels is presented in this work. The approach is based on the notion of a relaxed separation of scales, which tailors the methodology to the phenomena governed by local resonators embedded in a host medium. The macroscopic LRAM panel is modelled as a thin continuum beam/shell described by proper structural kinematics and momentum balance relations. At the microscale, a LRAM unit cell is considered with lamina-wise in-plane boundary conditions and zero out-of-plane tractions, representing the free top and bottom surfaces of the macroscopic LRAM panel. Under the assumption of linear elasticity, in the relaxed separation of scales regime, the microscale unit cell problem can be represented by the superposition of a static and a dynamic problem, hereby enabling a significant model reduction. As a result, the macroscale effective material properties can be computed once and for all off-line for a given unit cell. In addition, a new macroscale evolution equation emerges describing the effect of the microscale internal dynamics on the macroscopic fields through the introduction of new macroscale enrichment variables, which reflect the modal amplitudes of localized microscale modes. The proposed homogenization methodology reveals a high level of accuracy and numerical efficiency compared to the reference direct numerical simulation (DNS) for all relevant analyses: computation of real and complex dispersion spectra for an infinite LRAM panel, as well as steady-state frequency response and transient behaviour for a finite LRAM panel. • Computational homogenization methodology for LRAM panels (beams/shells) is presented. • Model reduction leads to closed-form frequency-independent effective properties. • A new macroscale evolution equation emerges describing microscale internal dynamics. • Approach is applied for computing dispersion properties of an infinite LRAM beam. • Approach is applied for transient analysis of a finite LRAM beam in time domain. [ABSTRACT FROM AUTHOR]

This paper addresses the two-scale problem underlying the enriched continuum for transient diffusion problems, which was previously developed and tested at the single scale level only (Waseem et al., Comp.Mech, 65, 2020). For a linear material model exhibiting a relaxed separation of scales, a model reduction was proposed at the micro-scale that replaces the micro-scale problem with a set of uncoupled ordinary differential equations (ODEs). At the macro-scale, the balance law, the ODEs and the macroscopic constitutive equations collectively represent an enriched continuum description. Examining different discretization techniques, distinct solution methods are presented for the macro-scale enriched continuum. Proof-of-principle examples are solved for a mass diffusion system in which species diffuse slower in the inclusion than in the matrix. The results from the enriched continuum formulation are compared with the computational transient homogenization (CTH) and direct numerical simulations (DNS). Without compromising the solution accuracy, significant computational gains are obtained through the enriched continuum approach. • Enriched continuum, emerging via model reduction, describes diffusion phenomenon. • Solution is obtained by multi-field and internal-variable spatial discretizations. • Accurate compared to computational homogenization and direct numerical simulation. • Approach provides computational gains up to four orders of magnitude. [ABSTRACT FROM AUTHOR]