IBM's flip-chip technology, which is commonly known as Controlled-Collapse Chip Connection (C4), is very reliable, yet expensive. During the past decade, there have been many worldwide R&D projects aimed at low-cost flip-chip on organic boards. And like many new technologies, high-performance, low-cost flip-chip-on-board (FCOB) is facing some technical issues. Issues A key issue involves low-cost epoxy underfills (Figures 1 and 2), in particular the reworkable underfills that facilitate chip removal for rework or repair-and how to employ these underfills without damaging the solder mask and/or copper pads. Other issues include low-cost wafer-level solder bumping, under-bump metallurgy, the flip-chip infrastructure, known-good die, fluxing issues, pick-and-place, the high cost of wafer bumping, etc. To put it bluntly, most of the available flip-chip expertise and infrastructure are still lacking for most newcomers, particularly when they are smaller system integrators. The exceptions are the few fully integrated companies, like Delco, IBM, Lucent and Motorola. Reworkable Underfills We'll examine a few of these flip-chip processing issues in forthcoming columns. In this issue, however, we will discuss reworkable underfills for low-cost flip-chip applications. The advantage of reworkable underfills is the cost savings for recovering expensive devices from multichip module packages. By employing modern rework technology, BGAs, CSPs and non-underfilled flip-chip devices can all be reworked.
Flip-Chip Limitation However, due to the intractability and infusibility of the epoxy underfill after curing, the underfill in a flip-chip package makes rework much more difficult or even impossible-which is a severe limitation to flip-chip technology. This difficulty arises because the cross-linked underfill does not allow easy chip removal or underfill residue removal. Therefore, flip-chip devices can only be reworked before underfilling.
Research Much research has been done attempting to make flip-chip devices reworkable. This includes technologies not based on underfills and attempts to develop reworkable underfills. Tsukada, et al., conducted mechanical chip removal using a chip-grinding approach. The test was performed on ceramic pin grid array modules. The grinding depth was maintained at a level until the solder joints were sliced halfway. Following the grinding, the parts were cleaned for debris and contamination. Reliability Data The eutectic solder (60/40 Sn/Pb) was then directly applied on the half-cut solder pads (95 Pb/5 Sn). New chips were assembled on the rework site by reflowing the eutectic solder, and the reworked chip was underfilled using the non-reworkable underfill. Reliability data showed that the reworked parts were quite similar in reliability to the non-reworked parts. Parylene Coating Another IBM option, a "mold release layer," utilizes a non-stick release coating (Parylene) on all surfaces intermediate of the chip and the substrate to remove the chip without damaging either the device or the bump. Limited Improvement Parylene by itself offered limited fatigue life improvement for the solder joints, so the epoxy underfill was applied directly to the Parylene-coated parts. The Parylene offers no resistance to the underfill flow under the chip. A combination of Parylene and underfill offers good reliability. Buchwalter, et al.,1-3 at IBM pioneered research on reworkable epoxy underfills by developing epoxy compositions that are soluble in an organic acid after curing, which fits into the chemically reworkable category. The IBM researchers achieved this by introducing acetal/ketal groups into the structure of diepoxide. They discovered that acetal/ketal group inside the diepoxide did not affect epoxy curing, and that cured epoxies containing acetal/ketal could slowly be dissolved by acid. Unfortunately, it is very time-consuming for the acid to penetrate through the chip-substrate gap to dissolve the underfill, and usage of solvents makes localized rework difficult. Diepoxide Synthesis Ober, et al.,1-2 at Cornell University synthesized a series of diepoxides containing a secondary, or tertiary, ester group. The Cornell researchers compared the thermal stability of these diepoxides to the epoxy from ERL-4221, a commercial diepoxide. This diepoxide contains a primary ester group and has been widely used in underfill formulations, due to its low viscosity and good electrical properties. Replacing the primary ester group with a secondary ester group on the diepoxide reduced the degradation temperature of the epoxy. A tertiary ester group reduced the degradation temperature (< 250°C) even further than the secondary ester group. From several diepoxides containing the tertiary ester group, the one with the lowest degradation temperature (~220°C) was selected and applied to a rework test. This material was found to meet the desired application and reliability demands.
Special Additives Wong, et al., at Georgia Tech took a two-sided approach towards the thermally reworkable epoxy underfills. The first approach was to identify some special additives for epoxies. These additives act as the blowing agent. When incorporated into the epoxy formulation, they are stable during epoxy curing and thermal cycling. When the temperature reaches the eutectic solder reflow temperature, however, they start to decompose and emit a large amount of gasses. This additive causes a mini-explosion within the epoxy matrix, which makes the chip that initially bonded strongly to the bond pad of the substrate much easier to remove. The second approach employed at Georgia Tech was the development of thermally degradable epoxies. The concept is similar to the approach by Ober, but the thermally labile groups used are different. These two approaches are eventually combined, with the reworkable underfill formulation consisting of thermally degradable epoxy and the special additive to provide both localized chip removal and underfill residue removal capability. Non-Epoxy-Based Approach Very recently, a non-epoxy-based approach by Iyer, et al., at Shell Chemical developed a thermally reworkable cross-linked resin that utilizes the reversibility of a Diels-Alder reaction. This cross-linked material is a solid at room temperature, and is heated to liquid before it fills the chip-substrate gap. It changes to an uncrosslinked liquid at 175°C. References 1. S. L. Buchwalter and L. L. Kosbar, "Cleavable Epoxy Resins: Design for Disassembly of Thermo-set," J. Polym. Sci., Part A: Polym Chem., Vol. 34, P. 249 (1996). 2. A. Afzali-Ardakani, S. L. Buchwalter, et al., "Cleavable Diepoxide for Removable Epoxy Comp-ositions," U.S. Patent 5,560,934, October 1996. 3. S. L. Buchwalter, A. J. Call, et al., "Reworkable Epoxy Underfill for Flip-Chip Packaging," First International Symposium on Advanced Packaging Materials, Process, Properties, and Interfaces, ISHM, February P. 7 (1995). 4. S. Yang, J. Chen, et al., "Reworkable Epoxies: Thermosets with Thermally Cleavable Groups for Controlled Network Breakdown," Chemistry of Materials, Vol. 10, P. 1475 (1998). 5. L. Crane, A. Torres-Filho, et al., "Development of Reworkable Underfills, Materials, Reliability and Processing," Proc. 3rd International Conference on Adhesive Joining and Coating Technology in Electronics Manufacturing, P. 262 (1998).
|
