Underfill Dispensing for BGA and CSP

A ball grid array or chip scale package carries its thermal and mechanical load through hundreds of small joints that cannot flex. Underfill spreads a filled polymer into the gap between package and board, surrounds each joint and transfers stress away from the solder. It is a reliability measure rather than a manufacturing one, and it adds a process step that has to be controlled or it creates more problems than it solves.

What Underfill Does

The material fills the space between the package body and the board, bonding to both and to every joint in between. When the assembly heats and the package expands at a different rate from the board, the polymer carries part of the resulting strain, so the solder joints see a smaller displacement for the same temperature change.

The benefit shows up as a longer fatigue life under thermal cycling, and under mechanical shock where the joint would otherwise be the weakest link. It is routinely specified for large packages, for thin boards and for products that see repeated temperature swings.

Capillary Flow Mechanics

The material is dispensed along one or two edges of the package and drawn under it by capillary action. The driving force is the surface energy of the gap and the wetting of the surfaces; the resistance is the viscosity of the polymer and the obstacle course formed by the solder joints themselves.

Flow therefore depends on gap height, joint pitch, board finish and the temperature of the assembly. A large package with a fine pitch is the hardest case, because the material has to travel a long distance through a narrow, obstructed channel before it cures. The flow front is measured in millimetres per minute, not centimetres, which sets the practical limit on package size.

Dispensing Patterns and Volume

The dispensing pattern decides where the flow starts and how evenly it advances. A single line along one edge is the simplest, while an L shape along two adjacent edges shortens the longest path and suits larger packages. Some processes use a dot pattern around most of the perimeter to reduce the travel distance further.

Volume is set slightly above the theoretical cavity volume so that a small fillet forms around the package. Too little leaves the corners dry, while too much produces a large fillet that can wick onto neighbouring components or reach a via and create a path for further flow.

Filler Content and Rheology

Underfill is a polymer loaded with a mineral filler, and the filler content controls both the thermal expansion of the cured material and its flow behaviour. A high filler content lowers expansion, which is good for reliability, but it raises viscosity and can filter at narrow gaps, leaving a dry region downstream.

The particle size distribution matters as much as the total loading. A filler whose largest particles approach the gap height will not pass the tightest part of the channel, and the flow stalls. Matching the filler size to the smallest gap in the assembly is a design decision, not a purchasing one.

Cure and Process Window

The cure schedule has to complete the flow before the material gels. A slow cure lets the polymer travel further, but it also raises the chance that the filler settles or that the flow front stops in a narrow region. A fast cure risks incomplete fill and a weak bond at the trailing edge.

The correct combination is established by dispensing trials on the real package, with the board held at a defined temperature. Preheating the assembly to around 60 to 90 degrees Celsius is common, because warm underfill flows faster and wets better while remaining stable long enough to complete the fill.

Voiding and Incomplete Fill

Two defects dominate: voids and incomplete fill. Voids come from air trapped as the flow front advances, from moisture driven out of the board, and from uneven wetting of the joints. They concentrate at the corners furthest from the dispense edge, where the flow front is slowest and most likely to close around a pocket.

Incomplete fill shows as a dry region or as a fillet that stops short of the package perimeter. It is caused by insufficient volume, a dispense pattern that leaves a long path, a gap that is too narrow for the filler, or a cure that started before the flow finished. The four causes are distinguished by examining where the fill stopped.

Inspection Methods

Visual inspection sees the fillet and the perimeter, which confirms that material arrived but says nothing about what happened underneath. Scanning acoustic microscopy maps the interface and reveals voids and delamination that are invisible from outside, and it is the standard method for a critical assembly.

X-ray is less useful for underfill because the polymer is radiolucent, though it remains the tool for checking the joints themselves. The strengths and limits of each technique for package inspection are set out in our comparison of X-ray and AOI.

When Underfill Is Not Needed

Underfill adds cost, process time and a rework obstacle, since a filled package cannot be replaced without removing the polymer. Where the package is small, the board is thick, the thermal excursion is modest and the product is not safety critical, the added reliability may not be worth the added risk.

The decision should follow an analysis of the actual stress. Our notes on thermal cycling test design and on microvia reliability describe how the relevant loads are characterised, and the same data supports the decision about underfill as it does about any other reliability measure.

Process Control and Verification

Where the requirement is not written down, the shop supplies its own default, and the default is chosen for the process rather than for the design.

FAQ

Can underfilled packages be reworked? Only with difficulty. The polymer has to be removed mechanically and thermally, which risks damaging the pads. Designing the assembly so that underfilled parts do not need replacement is the practical approach.

Does underfill need a primer? Usually not on a clean, properly finished board. Contamination or an incompatible solder mask can prevent bonding, and the answer is a cleaning step rather than an extra material.

How is complete fill confirmed? By scanning acoustic microscopy on a sample, and by sectioning a coupon. Visual inspection of the fillet alone cannot prove that the material reached the far corners of the package.