Open a failed SMD LED under a microscope and the die is usually intact. That is the point of a package: the semiconductor itself is the most robust part, and everything around it does the work of carrying heat, moving current and shaping light. Field returns cluster into a short list of package-level mechanisms. Knowing which mechanism you are looking at changes the fix, because each one points back to a different design or process decision upstream.
Bond Wire Lift
The bond wire carries current from the die pad to the lead frame. Thermal cycling makes the wire, the die adhesive and the mould compound expand at different rates, and the fatigue accumulates at the heel where the wire bends over the die pad. When the bond lifts, the path opens and the device goes dark or intermittent. Slightly wetter conditions make it worse: moisture that enters the package becomes vapour during a power cycle, which peels the bond mechanically. This is the mechanism behind most sudden dead-on-arrival and early-life dark failures in high-current parts.
| Failure mode | Field symptom | Usual root cause | Confirming test |
|---|---|---|---|
| Bond wire lift | sudden dark or flicker | thermal cycling, moisture ingress | decap and visual, forward curve open |
| Phosphor settlement | beam narrows, colour shifts | phosphor applied too heavily or uncured | spectral measurement over time |
| Encapsulant yellowing | warm shift, output drop | heat plus UV or sulfur exposure | spectrophotometer, chromaticity shift |
| Solder joint crack | flicker under vibration | reflow profile or board flex | x-ray, cross-section, cycling |
| Sulfur corrosion | high Vf, no light | silver reflector exposed to H2S | EDAX element scan |
| Die attach void | hot spot, early dimming | poor solder wetting | scanning acoustic microscopy |
Phosphor Settlement
Phosphor is distributed inside the encapsulant. If its mass loading is too high, or if the encapsulant was assembled before it finished curing, the particles drift downward under vibration and thermal soak. The physical result is a denser layer near the die and a clearer layer above it. Beam angle narrows, and correlated colour temperature moves. Unlike bond lift, this is gradual and visible on a spectrum over hundreds of hours, which makes it easy to miss in a simple on/off burn-in.
Encapsulant Yellowing
The silicone or epoxy above the die is the part that ages in public. Heat turns it amber; ultraviolet and airborne sulfur accelerate that. Yellowing removes blue as well as white, because the layer sits in front of everything, and it does the damage without any electrical change. In industrial and marine environments, sulfur compounds in the air reach the reflector through any gap in the moulding, and silver-plated reflectors tarnish, which raises forward voltage before luminous output visibly falls.
Solder Joint Fatigue
The LED sits on a board, and the board moves. A large heat sink, a long lead and an aluminum substrate all flex differently during a thermal cycle. The solder joint at the pad takes the strain, and microcracks grow until the connection opens intermittently. This failure is temperature-range-specific: products rated for -40°C to 85°C need a reflow profile and a board stack that survive many more cycles than an indoor fixture at 25°C to 60°C sees.
How to Tell Them Apart in a Return
Start electrically before you open anything. An open circuit means a broken bond or a cracked joint. A forward voltage that has climbed while light output stays low points at corrosion or reflection loss. A device that still draws normal current and produces a shifted spectrum points at phosphor or encapsulant. From there, decap the package for visual inspection, or run scanning acoustic microscopy for voids under the die. Each step narrows the list, and none of them need a new part number to fix.
Key Takeaways
Do LEDs fail on the die? Rarely. Package-level mechanisms account for most returned units.
Which one causes gradual dimming? Encapsulant yellowing and phosphor settlement, both driven by heat and assembly quality.
How do you catch it before shipment? Burn-in at the top of the rated temperature with spectral monitoring, not just an on/off test.
Related reading
- how AEC-Q102 stress tests qualify LEDs for automotive use
- more package and failure mode notes in LED Diode Q&A
