Frozen light

Electromagnetism is the fundamental mediator of all interactions in atomic physics and condensed-matter physics — in other words, the force that governs the structure of ordinary matter. It is rare to see an entirely new electromagnetic phenomenon, but Diederik Wiersma and colleagues report one on page 671 of this issue1. In carefully prepared semiconductor powders, Wiersma et al. have shown that light can be forced to stand still through the process of multiple scattering and wave interference. This experiment will lead to applications in optical data processing, spectroscopy and laser physics.

Our understanding and manipulation of electromagnetic waves has a long and rich history. The propagation of electromagnetic waves was predicted by James Clerk Maxwell in the nineteenth century, and Maxwell's equations are one of the pillars of modern physics. But the localization of light waves was predicted2,3,4 only in the 1980s, and confirmatory experiments (first in the microwave spectrum5,6, and now in the near-visible spectrum1) are milestones in our endeavour to create materials that mould the flow of light.

We all experience the effects of multiple scattering when it becomes dark on a cloudy day. Light from the Sun scatters many times from water droplets, following a tortuous diffusion path before reaching the ground. The distance that light travels within the cloud before being scattered into a random direction is called the mean free path. The effect of multiple light scattering is that the total amount of light transmitted through the cloud is reduced by a factor of the ratio of the cloud thickness to the transport mean free path.The rest comes back out of the other side, which is why clouds (and powders) look white.

Multiple scattering of light also takes place in human tissue. Here, the transport mean free path for infrared light of one micrometre wavelength is about a millimetre. Understanding the diffusive propagation of light in tissue is already paving the way to infrared imaging, a cheap and safe alternative to X-rays and nuclear magnetic resonance in diagnostic imaging of tumours7.

But neither clouds nor human tissue can scatter light strongly enough to cause localization. To achieve that, Wiersma et al. slowly ground samples of the semiconductor gallium arsenide to create a fine powder with a high refractive index, which scatters light roughly a thousand times more strongly than tissue. In their experiment, the transport mean free path is about the size of the wavelength of light — a regime of scattering never before attained. In this regime, the photons no longer travel like billiard balls bouncing randomly through a maze. Instead, there are strong interference effects, due to the wave-like nature of the photon, which severely impede diffusion (Fig. 1). This becomes evident in the transmission properties of the medium: for grain sizes near the onset of localization, the transmission coefficient falls off as the square of the optical depth (the ratio of mean free path to sample thickness); and when localization occurs, transmission falls off exponentially.

Figure 1: A phase transition in light waves, from classical diffusion to localization.

a, In classical diffusion through a scattering medium (such as white paint), the intensity distribution shows many fluctuations due to interference. Diffusion is possible because the peaks overlap. b, When the scattering becomes strong enough, the light is localized in separate peaks. The lack of spatial overlap between the peaks prohibits light transport.

Unlike electrons in a semiconductor, which are always conserved in number, photons are readily absorbed by matter. This complicates the interpretation of experiments, because absorption alone can lead to exponential decay of the transmitted light intensity. In order to separate the effects of absorption from true localization, Wiersma et al.1 varied the temperature of their semiconductor powders over a range of about 500 K. This leads to characteristic variations in the wavelength of the material's optical absorption edge, allowing localization and absorption to be distinguished. To further confirm that wave interference causes the localization transition, the authors compared their results with coherent back-scattering experiments8,9 on the same samples, which directly measure the interference between time-reversed optical paths.

The history of light localization has unfolded in a manner that is almost the reverse of electron (‘Anderson’) localization. In the case of electrons, the invention of the semiconductor preceded considerations of localization. In the case of photons, considerations of localization are a driving force in the development of photonic band-gap materials10,11, the photonic analogue of the semiconductor. These materials consist of a periodic array of scatterers with a lattice constant comparable to the wavelength of light. Photonic band-gap materials have important technological applications — in the development of micro-lasers and optical transistors, for example — because they can coherently localize light when disorder is introduced (either by structural defects or by doping the material with resonant atoms or molecules).

Powders are much easier to make than photonic band-gap materials, and the localization length in a powder is much longer: the photon may propagate hundreds of wavelengths before realizing that it is completely trapped. That may be very useful for cooperative effects involving photons and atoms, such as laser action.

Further progress in the field of light localization will depend on the optical purity of the materials. Optical fibres with an extinction length of many kilometres are now quite routine. But unlike optical fibres, which have a refractive index of about 1.5, materials that localize light must have a refractive index of at least 3.0, and at the same time be optically pure. (Unfortunately, a high index usually means operating close to a resonance, and absorption also becomes large near a resonance.) Candidates include silicon and germanium, either powdered or in the ordered form of a photonic band-gap material. These developments bring closer the new age of photonics, in which zero-threshold microlasers, sub-picosecond optical switches and all-optical transistors will take over from conventional electronics.


  1. 1

    Wiersma, D., Bartolini, P., Lagendijk, A. & Righini, R. Nature 390, 671–673 ( 1997).

  2. 2

    John, S. Phys. Rev. Lett. 53, 2169–2173 (1984).

  3. 3

    Anderson, P. W. Phil. Mag. B 52, 505–509 (1985).

  4. 4

    John, S. Phys. Today 44, 32–40 ( 1991).

  5. 5

    Genack, A. Z. & Garcia, N. Phys. Rev. Lett. 66, 2064–2068 (1991).

  6. 6

    Garcia, N. & Genack, A. Z. Phys. Rev. Lett. 66 , 1850–1854 (1991).

  7. 7

    Yodh, A. & Chance, B. Phys. Today 48, 34–40 (1995).

  8. 8

    van Albada, M. P. & Lagendijk, A. Phys. Rev. Lett. 55, 2692–2696 ( 1985).

  9. 9

    Wolf, P. E. & Maret, G. Phys. Rev. Lett. 55, 2696–2700 (1985).

  10. 10

    Yablonovitch, E. Phys. Rev. Lett. 59, 2059–2063 (1987).

  11. 11

    John, S. Phys. Rev. Lett. 59, 2486–2490 (1987).

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

John, S. Frozen light. Nature 390, 661–662 (1997).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.