An Ordinary LED Traps 96% of the Light It Makes. Lund Grew Branched Nanowires to Let It Out.

The dirty secret of the light-emitting diode is that most of the light never gets out. A conventional LED generates photons inside a slab of semiconductor with a high refractive index, and when those photons hit the boundary with the air at anything but a steep angle they bounce back in. Roughly 96% of the light produced is lost that way. Only about 4% escapes on its own, which is why real LEDs are covered in domes, roughening, and scattering layers whose entire job is to cheat total internal reflection.

Physicists at Lund University in Sweden have taken a different route: make the light-emitting part of the device too small for the trap to exist. Their LED is built from thin branched nanowires. A central core injects charge carriers, and light is emitted only from the side branches growing off it. Because the core itself does not emit, the researchers could confirm directly that the light was coming from the branches and not from somewhere more convenient.

The physical argument is simple to state. "If the structures are made thin enough — thinner than the wavelength of light — the light cannot be trapped inside the material in the same way," said Magnus Borgström, a professor of solid-state physics at Lund who led the work. Total internal reflection is a consequence of treating the boundary as a flat interface between two bulk materials. A wire narrower than the light it is emitting is not a bulk material in the relevant direction, and the photons leave.

The work was published on September 4 in Nano Research, with doctoral student Yue Zhao among the authors. Nanowires also come with a manufacturing argument attached: they use a small fraction of the semiconductor material a planar device needs, which in principle makes them cheaper per unit of light even before efficiency gains are counted.

The honest caveat is that these devices do not yet beat commercial LEDs. The problem is surfaces. Any nanostructure has an enormous surface-to-volume ratio, and defects at those surfaces trap charge carriers and turn them into heat rather than photons — exactly the loss channel the geometry was supposed to avoid. Getting the surfaces clean enough is the work standing between the demonstration and a product.

If it can be done, the payoff is broad. Lighting accounts for a substantial share of global electricity demand, and the same branched-wire architecture is a candidate for the light sources in optical interconnects, sensors and displays, where extraction efficiency sets the power budget for the whole system.