Solid information

Computers and the Internet have led to an explosion of easily available information. Somewhere, somehow, all of this information must be stored and processed, quickly and economically. Indeed, the decreasing cost of storing data has been a crucial impetus for the information revolution. But magnetic and conventional optical data storage, in which individual bits are stored on the surface of a recording medium, are approaching physical limits beyond which individual bits may be too small to easily write, store or read. Storing information throughout the volume of a medium, rather than just on its surface, is an intriguing high-capacity alternative. One such volumetric approach, holographic data storage, was conceived decades ago. But it tends to be fragile — reading out information tends to destroy what is written. On page 665of this issue1, Buse et al. describe a way to make holographic storage much more durable.

In holographic data storage2, an entire page of information — a million or so data bits — is stored at once as an interference pattern within a thick, photosensitive optical material. This is done by crossing a so-called object wave (typically a laser beam, on which the information has been imprinted by a suitable light modulator, such as a small liquid-crystal TV screen) with a reference beam carrying no added information. A large number of these interference gratings or patterns can be stored in the same piece of material, as long as they are distinguishable by their direction or spacing. That can be accomplished by changing the angle between the object and reference wave or by changing the laser wavelength. A page is read out by illuminating the material with the reference wave that was used to store that page: the wave is scattered in such a fashion that the object beam is reconstructed. The theoretical limit for the storage density of this technique is around tens of terabits per cubic centimetre, although the densities that have been demonstrated so far are three orders of magnitude smaller, comparable to a standard disk drive.

In addition to high storage density, holography promises fast access times, because the laser beams can be moved rapidly without inertia, unlike the actuators in disk drives; and also high data rates, because of the inherent parallelism of its page-wise storage and retrieval. But despite all of these obvious advantages — and the availability of cheap components such as liquid-crystal displays for spatial light modulators and CCD camera chips from video recorders for detector arrays — no commercial holographic storage products are available today. Why? The chief hurdle is the lack of a suitable recording medium. Despite the efforts of academic and industrial laboratories around the globe, there remains a critical problem. Recording or reading out a page tends to partially erase any holograms recorded earlier.

Non-destructive read-out requires some process that renders the stored information permanent, similar to fixing a photograph during film development. Depending on the specific mechanism used to record the holograms, fixing could involve applying an electrical field or annealing with heat. But the required high voltages, ovens or high-power lasers would be expensive, and possibly troublesome in a consumer device. Such fixing techniques would also make it much more difficult to verify information immediately after it had been written, as is typically done with magnetic and optical recording today.

One might avoid fixing by using a material whose optical properties are wavelength or optical-power dependent. The simplest method is to record a hologram at a wavelength for which the material is photosensitive (green, for example) and reads it out at another wavelength at which the material is not sensitive to light (infrared, for example). This is like a photographer using a red light in the darkroom to observe pictures being printed using white light. Using a different read-out wavelength doesn't work well, however, because it changes where the individual pixels appear on the detector array, introducing errors that are difficult to avoid. Another approach uses a high-power laser — data are written at high power, taking advantage of nonlinear effects, and read back with low-power light pulses well below the writing threshold3. Unfortunately, high-power, pulsed lasers are complex and will be expensive for the foreseeable future.

Figure 1: Light writing: object and reference laser beams intersect in a doped LiNbO3crystal to record a hologram containing 1 Mb of information.

(Courtesy of IBM Almaden Research Center.)

Still another approach is a one-two punch using two different wavelengths1,4,5,6. In the device described by Buse et al.1, an ultraviolet light source excites electrons in the recording medium from a low-energy state (on a manganese ion ‘trap’) to an intermediate energy state (on an iron ion) via the conduction band — a process called gating or biasing the material. Red light then has enough energy to write the hologram in the material by exciting the electrons from the intermediate state up into the conduction band; these electrons quickly diffuse the relatively short distance into the deeper manganese traps (see Fig. 1 on page 665). The electric fields between these trapped electrons and the ions left behind establish localized changes in the refractive index of the host material, forming the hologram's interference grating.

Such a hologram is stable because further illumination at the writing wavelength (red) cannot excite the trapped electrons back into the conduction band. For read-out, light at the writing wavelength is scattered from these localized inhomogeneities, and the stored wavefront is reconstructed. As the first light source need not be coherent, or even very narrowly peaked in wavelength, light-emitting diodes or lamps with colour filters can be used, instead of an expensive laser emitting visible light. And as a by-product, the holographic medium can be bulk erased by an exposure to the gating light.

As mentioned at the beginning, holograms are easily erased by those recorded later in the same volume. This can be alleviated partially by anticipating the partial erasure: the first holograms can be recorded for much longer times, thus achieving in the end a relatively equal intensity for all recorded holograms. Another approach would be to refresh the information periodically, as is done in dynamic ‘DRAM’ semiconductor memory. For both solutions, the crucial property is the ‘ M number’ — the ratio of the speed at which holograms are recorded and erased. In the new work1, M would reach about 7 for a 1-cm-thick sample. That means that erasure is seven times slower than the recording speed — a very desirable ratio.

A practical material must meet many other criteria, too. High sensitivity at the writing wavelength would eliminate the need for a large, expensive, high-power laser. In the present system, the high M was accomplished without sacrificing the sensitivity to the point that the results would be of only academic interest.

This study is a step towards a practical holographic storage device. But it is by no means the last one. A truly useful holographic recording medium must also have the right optical and mechanical properties, and be insensitive to temperature and cheap to make. Many more advances are necessary before we can buy a device at a price that is competitive with established data storage technologies. Moreover, holography fans should not overlook the fact that its well-entrenched competitor technologies are improving rapidly — by about 60% a year.


  1. 1

    Buse, K., Adibi, A. & Psaltis, D. Nature 393, 665–668 (1998).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Psaltis, D. & Mok, F. Sci. Am. 273, 70–76 (1995).

    Google Scholar 

  3. 3

    von der Linde, D., Glas, A. M. & Rodgers, K. F. Appl. Phys. Lett. 25, 155 (1974).

    Google Scholar 

  4. 4

    Bai, Y. S. & Kachru, R. Phys. Rev. Lett. 78, 2944–2947 (1997).

    Google Scholar 

  5. 5

    Guenther, H., Wittmann, G., Macfarlane, R. M. & Neurgaonkar, R. R. Opt. Lett. 22, 1305–1307 (1997).

    Google Scholar 

  6. 6

    Lande, D., Orlov, S. S., Akella, A., Hesselink, L. & Neurgaonkar, R. R. Opt. Lett. 22, 1722–1724 (1997).

    Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Coufal, H. Solid information. Nature 393, 628–629 (1998). https://doi.org/10.1038/31353

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.