An SMD LED can pass every electrical test with a large void under its thermal pad. The junction still conducts, the die still lights, and the failure appears only as a temperature rise that shows up later as lumen depreciation or a cracked die attach. X-ray is the only routine way to see the problem, and the number it produces is meaningless until the acceptance limit is agreed in writing.
How a Void Forms
A void is trapped material, not missing solder. Flux volatiles, moisture from paste, and gas from the pad surface all expand during reflow and cannot escape through the molten alloy fast enough. The bubble freezes in place as the joint solidifies.
The usual drivers:
- Paste volume too high for the pad area, so a thick deposit traps more solvent.
- A reflow profile that ramps too fast through the preheat region, flashing off solvent instead of letting it bleed out.
- Oxidized or contaminated pads, where flux works harder and generates more gas.
- Large continuous thermal pads with no vent path, common under high-power LED packages.
- Paste left exposed too long before reflow, which absorbs ambient moisture.
What Voids Cost in Thermal Terms
A void under a thermal pad subtracts conducting area from the main heat path. If 20% of the pad area is void, the effective thermal path shrinks and junction-to-board resistance rises. For a 3W part running near its limit, that shift can be the difference between a comfortable margin and a derated drive current. Voids under signal pads matter less, because those joints carry current rather than heat.
Reading the X-Ray
X-ray reports a void percentage per pad and, often, the largest single void. Three numbers belong in the report: total void area as a percentage of pad area, the largest single void, and the location relative to the die center. A void ring at the pad perimeter is less damaging than one sitting directly under the die.
| Finding | Likely cause | Process change |
|---|---|---|
| Many small scattered voids | Solvent flashing off too fast | Extend soak, lower ramp rate |
| One large central void | Trapped air in a wide deposit | Reduce stencil aperture, vent the pad |
| Voids along one pad edge | Tilt or uneven heating | Check coplanarity and reflow balance |
| Voids plus poor wetting | Oxidized pad finish | Improve storage, review flux activity |
Acceptance limits are project-specific. Most programs write a total void limit between 15% and 25% for the thermal pad, and tighten the figure for parts running close to their thermal limit. Agree the limit with the assembly house, record it on the drawing, and state the measurement method, because hand-drawn areas on a greyscale image produce different numbers from an automated threshold.
Reducing Voids
1. Stencil aperture sized to the pad, with a paste deposit that fills the joint without overfilling.
2. A soak zone long enough to drive solvent out before the alloy melts.
3. Nitrogen reflow where the process window is tight, since it removes oxide-driven flux consumption.
4. Vacuum-assisted reflow for large thermal pads, which pulls trapped gas out while the alloy is liquid.
5. Verified pad finish and shelf-life control on both paste and boards.
FAQ
Can an assembly house pass a part with 40% void if it tests fine? It can, and a functional test will not object. Set the limit before production, not after a field return.
Does void percentage alone predict life? No, it predicts a thermal penalty. The life impact depends on how much thermal headroom the design had.
Is a reflowed part with voids reworkable? Often yes, with a hot-air or vacuum rework station, but the rework adds its own thermal cycle and should be verified with a follow-up X-ray.
Related reading
