Underfill Options Become More Creative

For as long as flip-chip technology has been around, underfills have been necessary. Continually decreasing die thicknesses and standoff heights have worsened the challenges associated with using underfills to the point where new processes must be investigated. Several creative approaches for using pre-applied underfills were presented at this year's Electronic Components and Technology Conference (ECTC).

The biggest problem with dispensed underfill processes is low throughput. Dispensing the material at the edges of a die and waiting for it to wick underneath the die can be time-consuming. People still perform it because the processes are established, and there are some problems associated with pre-applied and no-flow processes. In some stacked-die applications, underfill is dispensed between the die after the stack passes some electrical tests, so the stack can be reworked more easily if necessary. Even with recent advances in dispensing and jetting technologies, pre-applied technologies still have a throughput advantage.

In no-flow processes, where the underfill is applied to the board prior to attach, common problems include die float and filler trapping. Usually, the die is simply placed on the site, then heated. Sometimes the die can move. Also, filler particles can get trapped between the ball and its corresponding pad. Except for applications with a very small die, filler particles are required to modulate the properties of the underfill material.

Researchers at STATS ChipPAC (Fremont, Calif.) presented work on a "thermode-assisted" process, in which the die is heated before it is placed on the board. The heated die and solder bumps are thought to help sweep filler particles away from the contact sites, and form a temporary metallurgical bond between the ball and pad. This bond is strong enough to hold ball-to-pad contact until full solder attach is performed.

One of the problems with pre-applied and no-flow processes is that moisture and volatiles absorbed by the board or substrate cannot outgas and escape during solder attach. Though steps can be taken to reduce absorbed moisture, it is useful to take every opportunity to eliminate it.

1. An expanding underfill allows volatiles to escape in a process that enables higher throughput.

2. Coating solder balls with underfill resin can boost reliability while reducing hassle.

Researchers at Emerson & Cumming (Billerica, Mass.) have developed a pre-applied underfill that expands during reflow (Fig. 1 ). Applied at a partial thickness on the die, it expands into a foam when heated to reflow temperatures. With appropriate temperature ramping, the volatiles absorbed in the board escape before the foam fills the space between the die and board. Drop tests indicate that the foam acts as a cushion. Life tests are still in progress.

There are often marginal cases in which reliability is good without an underfill, but not quite good enough. An underfill helps to spread the stress from the whole die and its entire footprint area, but there are times where reinforcing the solder balls will provide the necessary reliability. This is the idea behind the polymer collars used in some wafer-level packaging technologies. ¹

A researcher at Dell (Round Rock, Texas) presented work on a solder ball coating process that provides support for the whole solder ball, including both ball-pad interfaces (Fig. 2 ). After reflow and before attach, the die is dipped in a pool of underfill material, coating the solder balls. During solder attach, the material cures, providing the necessary support while leaving most of the space between the die and board free. This process promises to be fast and inexpensive. Since there is little, if any, contact between the underfill and the board, many of the delamination problems caused by moisture in the board are simply avoided.

There is another challenge: Getting the optimal fillet shape is more difficult and more important as the standoff height and die thickness decrease. A convex fillet can cause as many problems as the underfill solves. The fillet must come up the edge of the die, but not over the edge, and it must be concave. Obtaining this shape with less material and thinner die can be very difficult. The last two examples, foaming underfill and coated solder balls, both sidestep the fillet challenge.

Sometimes, when the limits of a given packaging approach are reached, simpler and more elegant approaches take hold that address the key concerns and obviate others. This seems to be happening with underfill processes, and it will happen with other processes down the road. Simple innovations like these are indicators of the growing importance of packaging and interconnection. They are also consistent with the idea that sometimes technologies must get simpler to make systems that are more complex.