In most areas of technology, defects are rarely welcome. For instance, in the semiconductor material inside a computer chip, defects might trap electrons and block the flow of current. But in photonic crystals, introducing precisely engineered defects could hold the key to a range of applications, including telecommunications and quantum computing. A new principle for the design of such defects is presented in this issue, by Susumu Noda and colleagues (Nature 425, 944–947; 2003).

The periodic structure of a photonic crystal forms a 'bandgap', which forbids the propagation of light within a well-defined range of wavelengths. Making defects in the structure can then capture and confine light, creating a photonic cavity that can control light in ways that are not possible with conventional optics.

The photonic crystal investigated by Noda et al. consists of a thin slab of dielectric material, such as silicon, with a two-dimensional array of sub-micrometre-sized holes cut into it. There is a variety of ways to create defects in this structure, such as making one or more of the holes bigger or smaller, or removing them completely. Previous attempts at confining light in a photonic crystal have generally concentrated on the size or number of defects, or the manner in which light is coupled in and out of them. These authors, however, have investigated the properties not of the defect holes themselves, but of those surrounding them.

First they introduced a defect in the photonic crystal with the absence of three adjacent holes (pictured). Without further modification, this makes an abrupt interface between the defect and the surrounding array, the severity of which allows trapped light to leak out. But by shifting the position of the holes on each side of the defect, Noda et al. found that they could soften this interface and reduce light leakage. And by doing so by just the right amount, the confinement efficiency, or quality factor, of the defects improves by as much as 100-fold, compared with previous studies.

This approach, which the authors refer to as "confining light gently to confine it strongly", should improve the performance not only of similar two-dimensional systems but of photonic crystals in general.