Drive a white LED hard and two things happen at once. The chip gets hotter, and its output falls faster than the current rise should explain. Part of that loss is series resistance inside the semiconductor. The larger and more surprising part is phosphor thermal quenching: the phosphor that converts blue light into white stops converting efficiently as its own temperature climbs. It is a reversible effect, which is why it shows up in any thermal transient measurement, and it is one of the main reasons a datasheet quote taken at 25°C does not hold at 85°C.
What Quenching Actually Is
A phosphor such as a cerium-doped yttrium aluminium garnet absorbs blue photons and re-emits at longer wavelengths. That conversion is not perfectly efficient, and the leftover energy stays in the crystal as heat. At room temperature most of that heat leaves through the package. Above roughly 120°C, non-radiative recombination paths inside the phosphor begin to win: the excitation energy is lost as lattice vibration instead of emitted photon. The conversion efficiency drops, less blue is down-converted, and the package output falls. The emitted spectrum also shifts, usually toward the blue side, which changes colour temperature as well as brightness.
Where the Temperature Sits
Quenching is a property of the phosphor temperature, not the junction temperature directly. The two differ by the thermal resistance of the die attach, the reflector, the encapsulant and the board. A phosphor layer sitting on the die runs 10°C to 30°C above the junction reading taken by an infrared camera on the package surface. That gap is why two packages with the same thermal resistance can still behave differently, depending on how far the phosphor sits from the heat source and how much encapsulant volume sits above it.
| Condition | Phosphor temperature | Typical conversion loss |
|---|---|---|
| Cold start, low current | below 60°C | under 2 percent |
| Steady state, rated current | 80 to 110°C | 2 to 6 percent |
| High current, thin board | 120 to 150°C | 6 to 15 percent |
| Overdriven with poor heat sink | above 150°C | 15 percent and rising |
How to Measure It
The cleanest measurement is a pulsed test. Drive the device with a short current pulse short enough that the junction barely warms, then compare with a steady-state reading at the same current. The difference is the thermal contribution, and separating it into chip loss, phosphor loss and encapsulant loss requires a spectroradiometer at both states. A simpler field check uses the forward voltage: Vf drops predictably as the junction warms, so a constant-current supply with poor thermal design produces a Vf that keeps sliding downward, which is the thermal runaway signature seen on overdriven boards.
How Design Keeps It Out
Three choices move the number. First, keep the phosphor close to a good heat path and keep encapsulant volume above the conversion layer small but not so small that blue escapes sideways. Second, derate the current. Running at 60 percent to 70 percent of rated current typically keeps the phosphor under 110°C and keeps the conversion loss inside a few percent, which is a larger saving in lifetime than any driver efficiency gain. Third, match the phosphor to the board: a high-power phosphor with a large Stokes shift carries more self-heating per watt, so it needs more headroom than a small-shift formulation used at the same luminous target.
Key Takeaways
Is quenching permanent? No. When the phosphor cools, conversion recovers. Permanent loss comes from encapsulant yellowing and phosphor settlement, not from quenching.
Why does colour shift with temperature? Less blue is down-converted, so the mix moves toward blue and correlated colour temperature climbs.
Does a higher efficacy bin help? A phosphor with a smaller Stokes shift produces less self-heating per emitted lumen, so it resists quenching better at the same output.
Related reading
- junction temperature limits for UV LEDs in medical devices
- chip-level reliability questions in LED Diode Q&A
