Jiang Zhou

A Micromechanics-Based Vapor Pressure Model in Electronic Packages

Polymer materials have wide applications in microelectronic packaging. Some polymer materials are used in bulk form, such as encapsulant molding compound and carrier or printed circuit boards FR4 and BT . Some polymer materials are used as adhesives, such as die-attach, underfill, or other structural and thermal adhesives. Polymers are also used in thin or thick film as an isolation layer, such as a solder mask on a printed circuit board or passivation layer in the wafer level. Despite the diversities of the chemistry and compositions, the polymer materials applied in microelectronics can be either thermoset or thermoplastic materials 1 . Both types of materials have a glass transition temperature around which the material properties, such as the CTE and Young’s modulus, are very sensitive to temperature. Another common feature of polymer materials is the high porosity, which makes the material susceptible to moisture absorption. In order to understand better the capacity of the moisture absorption by a typical polymer material, let us introduce some important parameters, which will be used frequently in subsequent sections, as follows:

Advanced Numerical and Experimental Techniques for Analysis of Dynamic Responses and Solder Joint Reliability During Drop Impact

Board level solder joint reliability performance during drop test is a critical concern to semiconductor and electronic product manufacturers. A new JEDEC standard for board level drop test of handheld electronic products was just released to specify the drop test procedure and conditions. However, there is no detailed information stated on dynamic responses of printed circuit board (PCB) and solder joints which are closely related to stress and strain of solder joints that affect the solder joint reliability, nor there is any simulation technique which provides good correlation with experimental measurements of dynamic responses of PCB and the resulting solder joint reliability during the entire drop impact process. In this paper, comprehensive dynamic responses of PCB and solder joints, e.g., acceleration, strains, and resistance, are measured and analyzed with a multichannel real-time electrical monitoring system, and simulated with a novel input acceleration (Input-G) method. The solder joint failure process, i.e., crack initiation, propagation, and opening, is well understood from the behavior of dynamic resistance. It is found experimentally and numerically that the mechanical shock causes multiple PCB bending or vibration which induces the solder joint fatigue failure. It is proven that the peeling stress of the critical solder joint is the dominant failure indicator by simulation, which correlates well with the observations and assumptions by experiment. Coincidence of cyclic change among dynamic resistance of solder joints, dynamic strains of PCB, and the peeling stress of the critical solder joints indicates that the solder joint crack opens and closes when the PCB bends down and up, and the critical solder joint failure is induced by cyclic peeling stress. The failure mode and location of critical solder balls predicted by modeling correlate well with experimental observation by cross section and dye penetration tests

Multi-physics modeling in virtual prototyping of electronic packages : combined thermal, thermo-mechanical and vapor pressure modeling

The realization of virtual prototyping of electronic packages depends on the capability and reliability of multi-physics modeling. This paper focuses on the methods and solutions of combined thermal and thermo-mechanical modeling. The package-level thermal behaviors for various kinds of packages are discussed first through the thermal simulation. The impact of internal package design on thermal performance is highlighted. Then the methods and solutions of combined thermal and thermo-mechanical modeling are addressed in detail. The strong interactions of thermal and mechanical simulations, as well as the trade-off between thermal and mechanical designs are discussed through two case studies. The benefit of moisture behavior modeling for the package design is also briefed in this paper.

Investigation of nonuniform moisture distribution on determination of hygroscopic swelling coefficient and finite element modeling for a flip chip package

The coefficient of hygroscopic swelling is a material property used to measure the volumetric change with moisture absorption under given humidity/temperature conditions. Current hygroscopic swelling characterization techniques use an averaged approach based on the averaged moisture content. However, the moisture distribution is not uniform across the test specimen during measurement. This introduces analysis errors in determining the material property. In this paper, an exact analytical solution is derived to obtain the accurate coefficient of hygroscopic swelling, with the consideration of 3-dimentional moisture diffusion across the specimen. The correlation between the locally defined hygroswelling coefficient and the averaged hygroswelling coefficient is obtained analytically. The results show that the coefficient of hygroswelling obtained based on the previous method using the averaged approach overestimates the property up to 250%. The methodology and formulation developed in this paper is applied to analyze a set of existing experimental data, and results are compared to the current approach. Based on the locally accurate coefficient of hygroscopic swelling, this paper also investigates the reliability of a flip chip ball grid array package under HAST condition (120/spl deg/C/100%RH). FEA simulation results revealed the significance of contribution of hygroswelling-induced tensile stresses under bump region. The finite element results give an insight of the failure mechanism associated with moisture absorption.

Package structural integrity analysis considering moisture

There are three types of failures when an electronic package is exposed to a humidity environment: the popcorn failure at soldering reflow, the delamination and cracking at HAST without electrical bias, and the corrosion under biased HAST. Despite the difference in failure mechanisms, the common and most contributory factor to moisture induced failures is the degradation of adhesion strength in the presence of moisture. Therefore, understanding interface behavior with moisture becomes a key for package integrity and reliability analysis. In this paper, first, several methods to characterize the interfacial fracture toughness or adhesion strength at elevated temperature with moisture effect are presented. The interfaces between polyimide on silicon chip and underfill (PI/UF) are used as a carrier to investigate the influence of moisture. Both interface fracture mechanics based fracture toughness measurement techniques and the quick-turn method such as die shear test are applied to investigate the interface behaviors with moisture. Details of several sample preparation methods, by which the fracture can be made to stay along the desired interface, are illustrated. Key results on the influence of the moisture on fracture toughness are presented. Next, the hygroscopic swelling characterization techniques are reviewed. Due to the fact that the moisture diffusion is a slow process, specimens used in hygroscopic swelling measurement are often subjected to a non-uniform moisture distribution. This becomes a potential hidden error in obtaining the coefficient of hygroscopic swelling. Analytical solutions have been devised to predict the errors caused by the non-uniform moisture distribution. A simple procedure in obtaining the accurate swelling characteristics is proposed. Both TGA-TMA method and Moire interferometry method are applied to measure the hygroscopic swelling behaviors of several underfills. A very good agreement between these two methods is achieved. Subsequent to hygroscopic swelling characterization, the paper presents a novel method to allow a time-dependent nonlinear finite element analysis for package deformations and stresses induced by hygroscopic as well as thermal mismatches. This has been a challenging problem since commonly used commercial finite element software such as ABAQUS and ANSYS do not explicitly allow the fully coupled nonlinear thermal and hygroscopic stress analysis. The existing linear superposition method, which couples hygroscopic stress with thermal stress analysis, can not apply to the problem with nonlinear materials such as polymers and solder materials. A fully integrated finite element stress modeling methodology is demonstrated through an example of flip chip package subjected to a multi-step humidity/temperature loading profile. The results of the effect of hygroscopic swelling on the inter-layer dielectric (ILD) and under bump metallurgy (UBM) structures reveals that the overall ILD stresses under HAST can be twice as high as those considered without the moisture effect. The contribution of these hygroswelling-induced tensile stress is very significant to lead to the interfacial damage and electrical failures during HAST.

Effect of Non-Uniform Moisture Distribution on the Hygroscopic Swelling Coefficient

The coefficient of hygroscopic swelling is a material property used to measure the volumetric change with moisture absorption under given humidity and temperature conditions. Current hygroscopic swelling characterization techniques use an averaged approach based on the averaged moisture content. However, the moisture distribution is not uniform across the test specimen during measurement. This introduces analysis errors in determining the material property. In this paper, an exact analytical solution was derived to obtain the accurate coefficient of hygroscopic swelling, with the consideration of 3-D moisture diffusion across the specimen. The correlation between the locally defined and the averaged coefficient of hygroscopic swelling was obtained analytically. The results showed that the coefficient of hygroscopic swelling obtained based on the previous method using the averaged approach may overestimate the property up to 250%. The methodology and formulation developed in this paper was then applied to analyze a set of existing experimental data, and results were compared to the current approach. This paper also investigated the reliability of a flip chip ball grid array package under high accelerated stress test condition (120degC/100%RH). Finite element analysis simulation results revealed the significance of contribution of hygroscopic swelling induced tensile stresses under bump region. The finite element results gave an insight of the failure mechanism associated with moisture absorption.

Board-level solder joint reliability analysis of thermally enhanced BGAs and LGAs

Thermally enhanced ball grid arrays (BGAs) are designed to have reduced thermal resistance through features such as heat slug, heat spreader, and thermal solder joints. This paper studies the design comparison of five types of thermally enhanced BGAs, i.e., conduction cooled BGA (C/sup 2/BGA), metal-core BGA, exposed-die land grid array (LGA), slug LGA, and spreader LGA. The solder joint reliability performance of thermally enhanced BGAs is benchmarked with conventional thin-profile fine-pitch ball grid array (TFBGA). Both global and local three-dimensional finite-element analysis (FEA) models are established to predict the fatigue life of solder joints during thermal cycle testing. Detailed pad design with realistic geometry of solder balls and nonlinear material properties are considered in the model. The fatigue model is based on a modified Darveauxs approach with nonlinear viscoplastic analysis of solder joints. For the test vehicles studied, the critical solder joints are located near the package corner. Design variations investigated include the effects of key package dimensions and material properties. Design variations are mainly reported using C/sup 2/BGA package as the trend for the other four thermally enhanced BGAs was similar. The choice of mold compound (MC) material is critical, and a material with higher coefficient of thermal expansion (CTE/sub 1/) and lower modulus is preferred. Die size, die attach, and slug-attach material have little effect on solder joint reliability. It is observed that there is good correlation of fatigue life between modeling prediction and thermal cycle testing for C/sup 2/BGA. Reliability of C/sup 2/BGA thermal solder joints is proven to be excellent, and heat can be effectively conducted away from the die to the PCB. This is crucial to the design of C/sup 2/BGA. In addition, solder joint fatigue life is found to be related to package warpage induced during thermal cycling test. A design with less package warpage usually has a longer fatigue life.

Analytical and numerical bound analysis of hygroscopic swelling characterization

In this paper, a comprehensive study is developed on the effects of non-uniform moisture distribution and hygroscopic stress in hygroscopic swelling characterization. Two different averaged methods are introduced to analyze the experimental data obtained using TMA/TGA techniques with respect to the different reference states chosen, i.e., absolute dry and saturation conditions, respectively. It turns out that these two methods give the upper and lower bound estimates of the coefficient of hygroscopic swelling of materials. Three-dimensional moisture diffusion solution is employed to consider the non-uniform moisture distribution. The analytical expressions of both methods are obtained for the averaged coefficient of hygroscopic swelling, which are functions of time, diffusivity and specimen dimensions. The effects of aspect ratio, and time-dependency on measurement accuracy are discussed. Mathematical proof and numerical results are given for the lower and upper bound analysis. The analytical predication shows a remarkable agreement with the experiment results. Because the analytical solutions are developed without considering the effect of the hygroscopic stress induced deformation, a sequentially coupled moisture diffusion and hygroscopic stress analysis is utilized to investigate the impact of hygroscopic stresses in the hygroscopic swelling characterization. It is interesting to notice that even through the elastic strain caused by hygroscopic stress accounts for about one third of the total strain, the displacement induced by the elastic deformation at probe measurement point is negligible compared to the displacement of hygroscopic swelling. We conclude that our analytical solutions give the accurate analysis of hygroscopic swelling characterization

Transient analysis on hygroscopic swelling characterization using sequentially coupled moisture diffusion and hygroscopic stress modeling method.

The characterization of hygroscopic swelling properties for polymeric materials in electronic packaging presents unique challenges that have not been adequately addressed in existing literature. One of those challenges is the accurate determination of the coefficient of hygroscopic swelling. Due to the fact that the moisture diffusion is a slow process, specimens used in hygroscopic swelling measurements are often subjected to a non-uniform moisture distribution condition during measurement period. This introduces potential sources of hidden errors in obtaining the coefficient of hygroscopic swelling. Previously analytical solutions have been devised to predict the errors caused by the non-uniform moisture distribution and the use of averaged approach. The elastic deformation by non-uniform hygroscopic swelling was not considered. However, when the non-uniform moisture distribution is present, it is inevitable that the measured deformation includes both hygroscopic and elastic deformation. It is vital to separate the elastic deformation from the total deformation so that only hygroscopic swelling strain is documented. The transient analysis using sequentially coupled moisture diffusion and hygroscopic stress finite element modelling has been applied in this paper to simulate the TMA hygroscopic swelling characterization process. Results showed that the elastic strain caused by the hygroscopic stress accounts for about one-third of the total strain during the measurement period. It is also found that the elastic strain along the thickness direction changes from tensile state to compressive state, thus the overall contribution of elastic deformation to the total deformation along the thickness direction is negligible. The simulation results confirmed the excellent agreement with the previous analytical study without considering elastic deformation. This implies that the previously developed analytical solutions are still valid and accurate in analyzing the characterization of swelling property.

Novel process warpage modeling of matrix stacked-die BGA

A comprehensive warpage analysis is performed on the matrix stacked-die ball grid array BGA (SDBGA) by means of finite-element modeling and experimental warpage measurements. By comparing the block warpage results from conventional linear small deformation simulation and the experimental measurement results, it is found that the linear method is not able to capture the warpage behavior of the SDBGA matrix, because the change of the centroidal moment of inertia of cross section after deformation cannot be considered due to small deformation assumption. The nonlinear large deformation analysis must be taken instead. Based on the nonlinear analysis, an advanced warpage prediction methodology for matrix SDBGA is established. This methodology is then used to characterize the warpage behavior of matrix SDBGA and to study the different effects on the warpage. Warpage of matrix SDBGA during the whole assembly processes is also predicted. For the SDBGA matrix investigated, the crossbow dominant warpage and buckling phenomena are observed for the matrix after bottom die bonding and after interposer bonding, which are new findings in the warpage study for electronic packages. It is also found that not only the total die length, but also the dice distribution will affect the warpage pattern of the matrix. For the matrix after top die bonding and after molding, normal warpage patterns are observed, i.e., both crossbow and coilset warpage are comparable. Bending interaction and the warpage competition mechanisms are proposed to explain the warpage characteristic for matrix SDBGA.