Processability and Reliability of Nonconductive Adhesives (NCAs) in Fine-Pitch Chip-on-Flex Applications

Journal of Electronic Materials, Mar 2006 by Chiang, W K, Chan, Y C, Ralph, Brian, Holland, Andrew

The use of NCAs to form direct contact interconnections between chip bumps and substrate pads have become a viable option in interconnection technology for fine-pitch applications. However, the primary concerns with NCAs are their long-term reliability, stability, and consistent electrical performance in particulate interconnections. Results of assembly process studies and environmental testing using NCAs on flexible substrates are analyzed and discussed herein. An extensive design experiment was performed to determine which process parameters were critical in obtaining good electrical connections. A reliability evaluation of NCAs for flexible substrate applications was carried out to gain more insight into the failure mechanisms of this type of interconnect. Pressure cooker test results showed that failures occurring in NCA joints are primarily due to moisture absorption, which could lead to interfacial delamination at the substrate/adhesive interface, accompanied by hygroscopic swelling. NCAs with lower coefficients of thermal expansion also exhibited better contact resistance stability during high-temperature storage tests.

Key words: Fine pitch, nonconductive adhesive, NCA, reliability, contact resistance

INTRODUCTION

Flip chips using adhesive technology have been used extensively in microelectronics packaging. This technology has offered numerous advantages over conventional solder joining, including lighter weight, smaller size, lower processing temperature, and greater environmental friendliness (lead free).1,2

To realize fine-pitch flip-chip interconnections, three promising candidates have emerged in adhesive joining techniques: isotropic conductive adhesives (ICA), anisotropic conductive adhesives (ACA), and nonconductive adhesives (NCA). ICA and ACA consist of a thermoset or thermoplastic polymer matrix with high filler and conductive metallic particle loadings, respectively. ICA conducts electricity equally in all directions, while ACA has only unidirectional conductivity in the z direction. However, neither of these adhesives can meet increasing demands for high-density, miniaturized packaging applications. ACA and ICA both have drawbacks, particularly in fine-pitch interconnections. For instance, in ICA joints, decreasing the pad pitch increases the risk of short circuits, while ICA bond strength decreases with more filler and reduced bond area.3,4 For ACA interconnections, decreasing the pad pitch could lead to a reduction in current-carrying capacity of the interconnects. Moreover, bridging or forming particle chains due to high particle loading inside the matrix could cause serious problems.5

On the other hand, the use of NCA on flexible substrates has gained widespread interest. Unlike ACA and ICA, the "bump-to-pad" contact using NCA does not require conductive particles or silver flakes to ensure a good electrical connection. Without the fillers, the electrical conduction of NCA is characterized purely by mechanical contacts between the chip bumps and substrate pads.6,7 Also, this type of interconnect is normally formed by applying pressure and temperature simultaneously to bind the flip chip and substrate together. Other bonding methods include ultraviolet (UV) and low-temperature curing; however, these are still at the development stage.8 The contacts are retained by the adhesive strength of the cured polymer adhesive. Nevertheless, flipchip joining using NCA is a relatively new technique in microelectronics packaging. Owing to the lack of documentation of materials and process requirements for enhanced reliability performance, the failure mechanisms of NCA interconnections are not yet well understood.

The objective of this study is to enhance the understanding of process and various bonding parameters on the package performance using NCAs. To deal with numerous variables in the process assembly and reliability, a systematic design experiment was carried out to reduce the complexity and increase the effectiveness of the tests. Optimum bonding parameters for NCAs were determined by means of a design matrix. Reliability data were obtained by measuring contact resistance while subjecting the joints to a pressure cooker test (PCT, 121°C, 100% RH, 2 atm) and high-temperature storage (HTS, 150°C). Failure analysis was conducted on bonds after reliability testing.

EXPERIMENTAL PROCEDURES

Materials, Chip, and Substrate

Two commercial NCAs were used in this study. The general properties of both epoxy-based NCAs are summarized in Table I. The test chips used in this study were 3 × 11 mm^sup 2^, with 300 rectangular bumps (50 × 70 µm^sup 2^, pitch 60 µm) around the periphery. The bumps were gold 17 µm thick. Twelve sets of "daisy-chained" bumps were present within the chip. Five bumps were arranged as a group, with two adjacent bumps used in the insulation test, and the other five bumps were designed for measuring the contact resistance. The flexible substrate used for this study had the advantages of extreme thinness with a very fine pitch and trace routability. The flexible substrate was approximately 40 µm thick, and the electrodes were gold/electroless nickel coated with copper (Au/Ni/Cu). A 7-µm thick copper pad was electrodeposited onto a 25-µm thick polyimide layer, followed by 3-µm thick layer of electroless nickel; this was then sputter coated with a 0.4-µm layer of gold.