Reliability Testing and Modeling of Anisotropic Conductive Adhesive Joints Under Temperature Cycling Test

Abstract

The use of anisotropic conductive adhesives (ACAs) in flip-chip interconnection technology has several advantages compared to solders with underfills. However, when ACAs are used with organic substrates, their reliability may cause problems. Temperature cycling testing can be used to study the effect of fluctuating temperatures on the reliability of ACA joints. However, this is fairly time consuming. If modeling can be utilized in the evaluation of reliability, significant time savings can be achieved. In this paper, the prospects of using shear stress modeling in predicting the reliability of two different anisotropic conductive adhesive films (ACFs) with an organic FR-4 substrate in temperature cycling testing are examined. Finite element models were made for the test structures. Stresses and strains during the cycling tests ae calculated. To estimate the validity of the models, experimental testing is also done. Test samples with two different ACFs are tested in a temperature cycling test where the temperatures vary from -40°C to 125 °C. Testing is carried out for 10000 cycles. In addition, failure analysis is conducted on the test samples. To model the test samples accurately, the material properties of the ACFs are tested using a dynamic mechanical analyzer and a thermomechanical analyzer. The results of the modeling are compared with those of the experimental tests by comparing typical failure locations, failure modes, and the failure order of the test samples with different ACFs. According to the modeling, the highest shear stress seems to indicate the failure location. In addition, it seems that shear stress models can be used to rank ACF flip-chip joints in the order of the anticipated failure times. On the other hand, normal stress and strain models seem to describe failure mechanisms different from that observed in the experimental tests, and they are observed to be impractical for our purposes. Delamination, which may be one manifestation of failure mechanisms related to shear stress, is observed in all samples with ACF1. However, as no delamination is found in the samples with ACF2, it cannot be positively stated whether the failure mechanism of ACF2 joints is related to shear stress or not. If the samples with different ACFs have different failure factors, simple stress or strain models seem to be impractical for reliability studies, as the results of different models cannot be easily compared.