Quantum efficiency measures how effectively an LED turns electrical carriers into photons. It never appears as a line item on a datasheet, yet it is the physics behind every efficacy figure in lumens per watt, every heat-sink decision and much of the price spread between chips. Two related numbers matter: internal quantum efficiency and external quantum efficiency.
Two Levels: IQE and EQE
Internal quantum efficiency (IQE) is the fraction of electron-hole recombinations inside the crystal that emit a photon rather than heat. Recombination is a race: radiative channels produce light, while non-radiative channels — crystal defects, Auger processes at high carrier density — release the same energy as lattice vibration.
External quantum efficiency (EQE) counts only the photons that actually leave the chip. The two connect through light extraction efficiency (LEE):
EQE = IQE × LEE
A photon generated inside a high-index semiconductor strikes the chip surface; beyond the critical angle it undergoes total internal reflection, bounces back and is eventually absorbed as heat. Early chips lost most photons this way, so EQE sat far below IQE. Chip shaping, surface roughening and patterned extraction layers raise LEE by giving trapped photons repeated chances to escape.
Why Modern Chips Extract So Much More
Decades of improvement in epitaxial growth cut defect densities in the active layer, pushing blue LED IQE above 90 percent at low current density. Extraction engineering then lifted blue EQE from a few percent in the 1990s toward 80 percent in modern devices. Multiply the two advances and the practical result follows: a modern white LED produces several times the light of a first-generation product at the same electrical input.
| Metric | Counts | Typical modern value (blue) |
|---|---|---|
| IQE | Photons generated per recombination inside the crystal | > 90% at low current density |
| LEE | Fraction of generated photons that escape the chip | 80–90% with patterned extraction |
| EQE | Photons leaving the chip per injected carrier | 70–80% |
Droop: Efficiency Depends on Drive Current
Efficiency is not a constant. As current density rises, non-radiative Auger recombination and carrier overflow grow faster than light output — the phenomenon called efficiency droop. The same chip can show its best EQE at a few A/cm² and lose 20–30% of it when driven hard. Comparing chips only at peak efficiency misleads; reputable suppliers publish efficacy at multiple current levels instead of one headline number.
What This Explains Between Suppliers
For anyone specifying LEDs, quantum efficiency explains observable differences without ever appearing in a spec table. A chip with higher EQE needs less electrical power for the same lumens, so it runs cooler. A cooler junction relaxes thermal design, extends lifetime and stabilizes color. In dense applications — automotive ceramic packages, UV arrays where many chips share one substrate — the advantage compounds across every watt and every operating hour.
The practical discipline follows: compare products under identical drive current, junction temperature and measurement conditions (IES LM-79 for photometry, LM-80 for lumen maintenance). When quoted efficacy differs between suppliers at the same drive condition, the gap usually traces back to IQE, extraction or droop — physics, not marketing.
FAQ
Why is EQE always lower than IQE?
Because EQE includes extraction losses. Photons emitted inside the semiconductor hit the surface beyond the critical angle and undergo total internal reflection; since EQE = IQE × LEE, trapped light lowers EQE even when internal generation is near-perfect.
Can two chips with the same datasheet efficacy behave differently?
Yes. Efficacy is quoted at one drive condition. If chip A droops faster than chip B, it loses its advantage at high current. Compare efficacy and lifetime at the current your driver actually uses.
How does higher EQE reduce system cost?
Less electrical input for the same lumens means less heat, so the heat-sink budget, fan requirements and temperature derating all shrink — savings that compound in dense arrays over the product lifetime.
