Contact Us
  • Home
  • BLOG
  • Which BGA Packages Actually Need Underfill?

Which BGA Packages Actually Need Underfill?

Underfill earns its cost on a large body package that sees a wide thermal swing or repeated mechanical shock, and it is wasted money on a small package inside a sealed enclosure that never moves. What decides it is the distance from the neutral point to the corner ball, the standoff under the package, the board thickness and the environment, and all four can be checked before the build starts.

A BGA joint fails in shear fatigue, and the shear comes from the package and the board expanding by different amounts. Take a 25 mm body on 1.6 mm FR-4. The corner ball sits about 17.7 mm from the neutral point, the package and the board differ by roughly 6 ppm per degree, and a 165 K swing moves that ball 17.5 um sideways. Across a 0.3 mm standoff that is 5.8 percent shear strain, close to 58,000 microstrain, on every cycle. The same package with a 0.45 mm standoff sees 3.9 percent. Gap is a lever as large as body size.

Underfill gives that displacement a second path. The layer under the package and the fillet around it join the same two bodies across the full footprint, while the balls join them across their own cross sections. The ratio of those two areas, weighted by modulus, is what brings 5.8 percent down to roughly one percent at the joint. That is the whole mechanism, and it explains why the material helps most where the arithmetic is worst: large body, fine pitch, tall standoff, thin board.

Where the numbers point

Package and boardEnvironmentWhat usually pays
Body under 15 mm, 0.8 mm pitch, 1.6 mm boardSealed indoor, 40 K swingNothing. Fix the void and the profile first
15 to 25 mm body, 0.5 to 0.8 mm pitchOutdoor, 100 to 125 K swingCorner bond, package stays removable
Over 25 mm body, 0.5 mm pitch or finer165 K swing, under hood or base stationFull underfill, Tg above the field peak
Board under 1.0 mm with heavy partsDrop, vibration, transport shockFull underfill or corner bond plus stiffener

The table is a starting point, not a qualification. A corner bond, three beads at the package corners, is the cheap middle option: it cuts the corner ball displacement and keeps the package removable, which matters on any board that still expects rework. A full underfill is a one way decision, and it should be taken with the material data in hand rather than with a supplier promise.

Three numbers that choose the material

Filler size against the gap. The rule of thumb is that the largest particle should be about a third of the gap or smaller. A 10 um filler under a 0.3 mm standoff is 30 to 1 and flows. A 25 um filler under a 0.2 mm standoff is 8 to 1, and the particles filter out at the ball field, so the material reaches the die shadow late and with the wrong ratio of resin to filler. The visible result is a void line under the middle of the package, which the X-ray shows as a dark band and the cycle test finds as the first failure.

Tg against the hottest hour. Below its glass transition the material has a CTE near 25 to 35 ppm per degree and a useful modulus. Above it the CTE climbs to 90 or 110 ppm and the modulus falls by an order of magnitude, so the ball carries the load again exactly when the joint is hottest. An underfill with Tg at 80 C on a board whose field peak is 105 C is not protection, it is a soft layer. Match Tg above the highest field temperature with margin, and remember that some reworkable grades lose their cycling benefit once the peak crosses their Tg.

Flow and cure against the footprint. A 25 mm flow path at 90 C preheat takes minutes, not seconds, and the dispense pattern decides whether the front arrives as a straight line or as two streams that trap air in the middle. An L or U pattern from one edge, a fillet of 0.5 to 1 mm on the wet side, and a cure that follows the data sheet. A partially cured layer looks and feels right and contributes nothing.

What goes wrong

Underfilling everything as insurance. It adds cost per board, it removes rework, since removal at 200 to 250 C damages the site, and it moves the failure to the pad and to the package substrate where it is harder to see. If the void control is weak, the underfill does not fix the void, it seals it in and makes the joint impossible to reflow a second time.

Underfilling a wet or dirty board. Moisture and flux residue trapped under a package become vapour at reflow temperature and delaminate the die shadow; residue under the fillet becomes the leak path that shows up as an intermittent fault months later. Bake to J-STD-033, clean or plasma when the process needs it, and keep the dispensed board inside its floor life.

Choosing by brand instead of by data. Ask for the CTE below and above Tg, the modulus at 25 C and at 125 C, the filler size, the pot life and the cured Tg. Two products from the same maker can differ by a factor of three in modulus, and the stiffer one is not automatically better: it protects the ball and loads the pad.

What SUPERB Assembly controls on the build

The material approval and the cycle test belong to your reliability laboratory or to a partner laboratory, and we do not quote a cycle count we did not run. What we control is the build that decides whether the material performs: 7 SMT lines, placement from 01005 to 55 mm connectors, 3D SPI on every print, 3D AOI plus DIP AOI, X-ray in 2D and CT with void held under 5 percent to IPC-7095, a 12 zone reflow with nitrogen under 500 ppm, and bake and dry pack records to J-STD-033 that travel with the lot.

For the dispense step the board is preheated, the pattern is set per package, cure follows the material data sheet, and the fillet is inspected on the first article and at intervals after that. Final inspection is to IPC-A-610 Class 2 or Class 3 as the product requires, with first article comparison against CAD data and ICT or functional test on a dedicated fixture. The assembly scope is on turnkey PCBA assembly, the board side including Tg and copper balance is on high reliability PCB, and the same question under hood is on automotive PCBA.

What to send with the question

Body size, pitch and ball count per package, standoff or collapsed ball height, board thickness and layer count, the largest copper plane under the package, the field swing with its dwell, any drop or vibration requirement, your rework and repair policy, and the underfill grade if one is already approved. With those seven items the answer is a decision with a number behind it, instead of a habit.

Need a quote or a review of your build against these limits? pcba@superb-tech.com · WhatsApp +86 133 9608 1443