Issue



Emerging trends drive evolution of underfill dispensing


06/01/2000







New technologies provide solid process capabilities for a variety of interconnect strategies.

By S. J. Adamson, W. Walters,
D. Gibson and C. Ness

As new high-density electronics packaging technologies steadily push the state of the art for direct chip attachment (i.e., flip chip or CSP) to various substrates, dispensing technologies for underfill encapsulant have also undergone a dramatic evolution. With dispensing processes now evolving into third-generation capabilities, the sophistication of underfill dispensing has gone from early experimental techniques to robust and controllable processes that yield consistent repeatability and quality within high-throughput production environments.

However, the ongoing revolution in smaller semiconductor devices and more complex multi-chip design alternatives is continually presenting variations on the challenges associated with dispensing underfill encapsulant. This article briefly examines the objectives for using underfill (Figure 1), new packaging challenges and innovative dispensing techniques that are evolving to address them.


Figure 1. During temperature cycling, the larger the distance from the neutral point equates to greater stress on the corner solder balls due to differential thermal expansion mismatch.
Click here to enlarge image

The primary reason for using underfill adhesive is to reduce the impact of mismatches in global thermal expansion characteristics between the silicon die and the underlying substrate. In more traditional chip-on-board designs, these differences in coefficient of thermal expansion (CTE) are absorbed by the bond wires and the die-attach adhesive. In direct-attach designs, like flip chip, the stresses are concentrated on the solder balls.

Because the ball-array solder joints represent the weakest points in the die-attach structure, they are inherently susceptible to failures during thermal expansion, which can result in functional failures of the assembly. The use of underfill adhesive can greatly reduce the incidence of such failures by simultaneously adhering to the chip, solder balls and substrate, thereby redistributing the thermal expansion stresses over the entire chip area. By underfilling the area between the chip and the substrate, the direct stresses on the solder bump interconnects can be reduced significantly.

Secondary benefits from using underfill include increased mechanical stiffness to protect against externally induced stresses and strains, such as flexing or sharp impacts on the finished assembly. In addition, a voidless underfill encapsulation can provide added protection against moisture or other contaminants and make assemblies easier to handle.

Thus far, underfill dispensing technologies have evolved to provide solid process capabilities for a variety of interconnect strategies, such as flip-chip and chip-scale packaging (CSP). Many high-speed production environments are achieving excellent yields by using platforms that provide a combination of:

  • Highly repeatable precision positioning and motion systems
  • Third-generation, linear-positive displacement needle dispensing pumps, which give precise volumetric control over the dispensing process
  • Careful management of temperature levels throughout the dispensing and flow-out process
  • Flexible and simple programming, set-up and run-time operating environments
  • Flexible timing of dispense steps to increase productivity and underfill quality.

Building upon this base of mature dispensing capabilities, system-level manufacturers and their dispensing partners are now looking toward the resolution of additional challenges as chip-level assembly techniques continue to evolve.

Tighter Between-chip Spacing

The first objective of any underfill dispensing process must be to "do no harm" to the chips being underfilled through either direct contact or overflow contamination of other nearby components. As multi-chip CSP designs become more tightly packed, the dispensing process needs to deliver higher precision using smaller needle dimensions in order to dispense between and around the devices. Unless the needle positioning, movement and flow rate are all tightly controlled, there is a risk of the adhesive flowing under or onto adjacent components (Figure 2).


Figure 2. Underfill dispensing on substrates with closely spaced components.
Click here to enlarge image

In such an instance, instead of relieving stresses by completely and precisely encapsulating the area between each chip and the substrate, the overflow of cured adhesive can actually induce additional stresses by pulling and pushing between the chips during thermal cycling. In addition, the cross-capillary action from adjacent components can also pull the underfill material away from the intended device(s) during dispensing, thereby creating voids under the die and reducing the effectiveness of the underfill.

To counter these concerns, the module designer can focus on manufacturability by leaving adequate space for dispensing inside the package. Often, this will take the form of providing strategic access to the gap underneath the component for precise delivery of the fluid. To enable the most efficient method for underfill delivery, designers should incorporate space between components to allow insertion of the dispensing needle so as to deliver the fluid directly to the gap in which capillary force will draw it under the component. For example, a common needle used for dispensing underfill fluids has a diameter of approximately 0.8 mm. Therefore, a good first-order approximation of a design rule would be to allow 1.0 mm of space between adjacent components.

Expanding Process Windows to Deal with Larger and Smaller Die

Another arena in which underfill dispensing is being pushed to the extreme is in dealing with either larger or smaller die sizes, which are significantly extending the required dispensing process windows beyond first- and second-generation dispensing capabilities.

On one end of the spectrum, die as small as 1/16 inch square have been underfilled with third-generation systems, using precision valves and needles that can produce extremely small shot sizes down to 1 mg. While precision dispensing is paramount with such small die sizes, one up side is that if the dispensing platform can consistently deliver shot sizes in the 1- to 3-mg range, the short flow-out times can result in very high overall throughput rates.

At the other end of the spectrum, specialized die as large as 0.75 to 1.0 inch square with relatively small under-chip spacing are challenging the limits of capillary underfill systems to achieve full, voidless flow-out while maintaining acceptable throughput rates. These designs push the envelope of conventional underfill chemistry in a high-speed assembly environment. However, some interesting work has been done that looked at filler particle properties, such as particle shape and concentration, which have an important effect on successful underfilling of large die.1 Designers can assist by reducing barriers to flow on the substrate side, such as raised copper traces, raised solder mask features or inter-layer trenches.

The challenge of creating an optimal wetting agent can be complicated by the fact that the under-chip environment consists of several different materials, such as silicon, lead, FR-4 or soldermask. One particular area that can lead to flow problems is the presence of flux residue, especially tacky flux. However, flux residue problems are becoming less of an issue through the use of advanced selective flux application techniques, such as precision flux jetting.

Dispensing Through Holes in RF Shielding

With the trends toward ever-smaller wireless devices, system manufacturers have not only turned to increased usage of direct die to substrate methods, they have also had to tighten the close integration of radio frequency (RF) shielding around miniaturized multi-chip assemblies. This often means that the underfill dispensing process must be carried out through access holes in the RF shielding. Design of the RF shield can either enable or hinder the underfill dispensing process depending upon the number and position of holes in the shield. Since the presence of too many holes or holes that are too large can also defeat the overall effectiveness of the RF shielding, the product and process designers have to play a balancing act between designing for manufacturability and retaining sufficient shielding effectiveness.


Figure 3. Stacked-die design with mixed die types.
Click here to enlarge image

In addition, the extremely tight spacing between chips required in many miniaturized wireless designs also complicates the challenge by requiring ultra-precise needle placement to avoid cross-contamination of the underfill between multiple chips. Besides worrying about chip-to-chip contamination, process and product designers also have to be concerned about the close proximity between the die to be underfilled and the RF shield itself. In some instances, the nearby RF shield can induce a wicking action that causes the underfill to flow up the inside of the shield rather than under the chip. The use of higher flow rates or positioning the needle too high can contribute to this risk.

However, slowing down the dispensing rate may have the negative impact of reducing overall production throughput. Some process designers have dealt with this dilemma by partially dispensing a location and then moving on to dispense through different holes to other devices before returning to the first hole. Although multiple moves can potentially impact throughput, process designers have found that highly programmable systems, which precisely control both quantity and timing of dispensing, are capable of maintaining acceptable flow out and overall production rates while shuttling between different dispensing locations.

Avoiding Contamination of Specialized Components

In addition to multiple die that must be underfilled, many designs also require dealing with the presence of other components on the substrate that either interfere with or can be contaminated by the underfill. For example, nearby surface mount connector leads can provide a natural capillary wicking action that pulls the underfill up into the connector. Upon curing, the underfill material will then degrade or completely defeat the mechanical spring action of the connector, rendering it unusable. Similarly, the inadvertent flow of underfill beneath an RF crystal can effect its oscillation frequency.

It is sometimes possible to deal with these problems by using a multi-head dispensing system to pre-dispense a dam around the adjacent component using a higher viscosity material that won't flow beneath it. Then, during the subsequent underfill dispensing pass, the dam effectively prevents any unintended capillary flow from going under the adjacent device.

The Challenges of Encapsulating Stacked Die

Stacking multiple die on top of each other is becoming more prevalent for some ultra-compact applications, such as cardiac pacemakers. In some instances, stacked-die designs also mix die types, such as using wire-bonded die on the bottom and flip chip on the top (Figure 3).

Here again, with stacked-die applications the use of highly programmable, precision-dispensing platforms becomes critical in order to support design of multi-pass dispensing processes. Tight control over needle positioning, dispensing patterns, flow rates and volumetric mass are key factors for achieving consistent results and acceptable production throughput rates. In addition, stacked-die applications are also driving new requirements for very small needle sizes, higher dispensing pressures and the need for stiffer dispensing chambers, which are becoming characteristic of third-generation dispensing pumps.

The Bottom Line

Although third-generation underfill dispensing processes have matured to the point of delivering very robust, controllable process characteristics for most mainstream die-attach underfill requirements, the range of extreme challenges is continually expanding. In order to maintain adequate design latitude for creating increasingly smaller and more complex devices, process designers need to ensure that their dispensing capabilities can accommodate the widest possible range of dispensing requirements, while also providing the extensibility to handle tomorrow's as-yet unconsidered possibilities.
AP

Reference

  1. Michael Todd, Kishor Desai and Lan Hoang, "Evaluation of Key Underfill Formulation Parameters on the Performance of Flip Chip Devices," presented at Pan Pacific Microelectronics Conference, January 2000.

STEVEN J. ADAMSON, semiconductor product manager, WES WALTERS, applications manager, DAVID L. GIBSON, industry specialist, and CHRISTIAN Q. NESS, technical staff member, can be contacted at Asymtek, 2762 Loker Avenue W., Carlsbad, CA 92008; 760-431-1919; Fax: 760-431-2678; E-mail: [email protected], [email protected], [email protected] or [email protected].