A study on reliability of anisotropic conductive joints under thermally induced warpage for advanced packaging

Abstract

Anisotropic Conductive Film (ACF) offers miniaturization of package size, reduction in interconnection distance and high performance, cost-competitive packaging and improved environmental impact. ACF, even though it has merits, suffers from a major drawback in the issue of reliability. The factor of thermal warpage may lead to a highly unreliable electrical connection in the assembly. The major limitation of ACF is its instability caused by thermal warpage. The warpage issue and its consequent effect is a critical issue when dealing with ACF in modern electronic manufacturing. The following study of the ACF joints and the warpage has been carried out in three stages. Initially, a chip containing a bump (Au/Ni) was studied for its thermal warpage and effect of this warpage on the ability to conduct current. Then an ACF interconnection on a bumpless chip was studied under the same operating conditions. Finally, the results from both studies were correlated. Flip chips of dimensions of 11x3 mm were used. The flex substrate used was made of polyimide film, with Au/Ni/Cu electrodes and a daisy-chained circuit matched to the die bump pattern. The ACF used was based on epoxy resin in which nickel and gold-coated polymer particles were dispersed. A comparative study was carried out on the results obtained. ACF assemblies were subjected to thermal cycling with different temperature profiles at high temperatures below and above the glass transition temperature (Tg) of ACF. The results showed that the glass transition temperature (Tg) of the ACF material played an important role in the high temperature contact resistance. Above Tg,the ACF matrix became more viscous, which reduced its adhesive strength and assisted bumps on the chip to slide away from the pads on the substrate. Even though a flex substrate was used in this study, the effect of chip bump sliding over the substrate pad was severe at the corner joints of the chip, where cumulative forces were generated due to the thermal mismatch. For every thermal cycling profile, there was an incubation period encountered from this work that would have a significant impact on the application of ACF. After the incubation period the online contact resistance increased rapidly and assemblies were therefore no longer reliable. The project focuses on the online contact resistance behavior of the ACF joint during thermal shock and compares the results of two different types of dies (Au/Ni bump & Bumpless). The ACF used was epoxy resin in which nickel and gold-coated polymers balls were dispersed. Tests for three different thermal cycling profiles (125oC to -55oC, 140oC to -40oC and 150oC to -65oC) were carried out. The samples bonded at temperature 180oC; and pressure 80N was used. The initial contact resistances of Au/Ni bump and bumpless samples were 0.25Ω and 0.4Ω respectively. Initially the Au/Ni bumped chip was studied along with its failure mechanism and resistance to thermal warpage. In another phase, a similar study was carried out on a bumpless chip. A comparative study was carried out using the results obtained. The results showed that for the Flip Chip On Flex (FCOF) packages having an Au/Ni bump, the increase in online contact resistance was higher than in the FCOF packages which had bumpless chips. For example in the thermal cycling profile 140oC to -40oC, the online contact resistance for an Au/Ni bump rose to a higher value after 180 cycles, whereas it was comparatively less for bumpless samples. The bump height and bump materials were found to be the main factor for such variation. Results show that above glass transition temperature (Tg), the ACF matrix became more viscous which reduced its adhesive strength and lets the height of the bump on the chip increase resulting in a higher standoff of the package, and so it is easier for sliding to take place. The responses of the assemblies to hot and cold conditions were examined and in-chamber behavior of the assembly was studied and explained. The results obtained from this work show that joints at the corner of the assembly are more likely to fail, than joints in the middle. An assembly with higher standoff height has poor reliability; and temperature cycles above the glass transition temperature of the ACF material have a higher failure rate. Hence we suggest using chips with low bump height and ACF with a lower coefficient of thermal expansion (CTE). It is suggested that an ACF, which has higher glass transition temperature is used; and it is also vital to select a substrate material which has a CTE more or less similar to that of the chip material.

Date of Award	15 Feb 2007
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Yan Cheong CHAN (Supervisor)