快猫短视频

CATCHING light beams is the sort of trick that wizards perform in fairy
stories鈥攚hich must make Achim Wixforth and his colleagues sorcerers of the
first order. For in their lab at the University of Munich, they have found a way
of catching a beam of light, bottling it up for a while and then sending it on
its way again.

It鈥檚 a neat trick and one that could have far-reaching implications. For the
electrons that carry data around existing computers have their limitations. They
interfere with each other, they need wires to travel along and they carry
information to and fro at a snail鈥檚 pace.

The properties of a light beam, on the other hand, read like a wish list for
communication and computing. Its information-carrying capacity, or bandwidth, is
colossal. A single burst of laser light can transmit the entire
Encyclopaedia Britannica in a second. It is easily split into many
individual beams, making it ideal for parallel processing, widely seen as the
future of high-speed computing. And, of course, it鈥檚 fast: indeed, there鈥檚
nothing faster in the Universe.

Light may be a swift and capacious carrier of data, but like a runaway mail
train whose brakes have failed, if you want to get at that data, you鈥檝e got to
crash it into something. Physicists have devised some pretty fancy walls for
light to crash into in recent years, and they go by the collective name of
optoelectronics鈥攖he technology of turning data carried by light into the
electronic form used by conventional machines.

Optoelectronics allows you to send information from one place to another at
the speed of light and optoelectronic devices pop up everywhere, from
transcontinental telephone cables to your TV鈥檚 remote control. But at the end of
the line, you鈥檙e still turning the astonishing speed and carrying capacity of
light back into a sluggish stream of electrons, limited by the electrical whims
of the material that it passes through. If you could work with light instead of
electrons, it would be possible to build superfast devices 鈥攕uch as
optical computers.

To realise this dream, light must be persuaded to linger in places until it
is needed鈥攍ong enough, in fact, to act as a memory for the data
transported in the light beam.

People have been searching for ways of creating such optical memories for
years. They have tried everything from exploiting bizarre quantum effects
(鈥淎 handful of light鈥�, 快猫短视频, 27 June 1998, p 36)
to the downright kludgy, such as getting light to run round inside a coil of optical fibre for a
while.

The trouble with such devices, says Wixforth, is that they鈥檙e usually
bulky鈥攆or example, to delay light by just a millionth of a second you need
300 metres of optical fibre鈥攁nd they are hard to control too. 鈥淚deally,鈥�
he says, 鈥渁n optical memory would be a small container in which an incoming
optical signal could be stored for an arbitrarily chosen time and then released
again as light.鈥�

And that鈥檚 pretty much what his team unveiled earlier this year in the
journal Science: a practical way of storing light in a memory device
smaller than the full stop at the end of this sentence. What鈥檚 more, they did it
using semiconductors, making it much easier to manufacture such devices and
integrate them into existing electronic technology.

In theory, building an optical memory out of a chunk of semiconductor ought
to be easy. The energy of electrons in semiconductors falls into two broad
bands. Most are in the valence band, in which they are tied to specific atoms.
Give them enough energy to jump up to the conduction band, however, and they
become free to move, leaving behind holes that act as positively-charged
particles. So if you blast a semiconductor with photons of the right energy they
will be absorbed, leaving behind pairs of electrons and holes, either of which
can serve as a way to store the original light.

This is how the chips known as charge-coupled devices, or CCDs, sense and
record light. These devices, now widely used in digital cameras, store the light
as electrons. To do this, they must rip the electrons and holes apart as soon as
they form to prevent them disappearing in a puff of heat or a flash of light
within a billionth of a second.

But constructing an optical memory that can trap, hold and release light is a
far greater challenge. Just how do you pull the electrons and holes apart, yet
keep them close enough to one another so that they can recombine instantly,
releasing photons of similar wavelengths to the original light鈥攚henever
and wherever you require?

To make things even harder, the semiconductors that do the best job of
converting light energy into electron-hole pairs, such as gallium arsenide, are
pretty hopeless at keeping the pairs apart. On the other hand, semiconductors
such as silicon prevent the pairs coming together again for far longer, but are
not very good at creating them in the first place. 鈥淚t鈥檚 a real catch-22 because
for an optical memory we want both these properties in the same semiconductor,鈥�
says Wixforth.

Sound crystals

To overcome this obstacle, Wixforth and his colleagues have turned to a
rather unlikely phenomenon: sound. The team came up with their solution during
research into new ways of controlling the movement of electrons, and one very
promising method of shunting electrons around turned out to be surface acoustic
waves: tiny ripples of pressure on the surface of crystals
(鈥淭he really fast show鈥�, 快猫短视频, 16 May 1998, p 36).
鈥淭hey鈥檙e modes of elastic energy that can travel across the surface of a crystal at the speed of sound,鈥�
says Wixforth.

Creating these waves is simple鈥攋ust apply an alternating voltage to a
piezoelectric material such as lithium niobate. The changing voltage forces the
molecular lattice of the piezoelectric to stretch and flex, creating a pressure
wave that races along the material. As it travels, this wave generates an
intense electric field of its own that can be used to capture and transport
electrons.

While Wixforth and his colleagues were using these waves to move electrons
about, they realised that the waves had another use: keeping apart the
electron-hole pairs formed by light. The strong electric field created by the
waves distorts the semiconductor鈥檚 flat conduction and valence bands into
regular sine waves. When electron-hole pairs encounter these peaks and valleys,
they are pulled apart, the electrons moving to the peaks of the waves and the
holes to the troughs (see Diagram).

鈥淎nd once they鈥檝e been separated, the electrons and holes would be too far
apart from each other to recombine,鈥� says Wixforth. 鈥淭hey would remain trapped
in the surface acoustic waves like surfers at Sunset Beach.鈥�

The idea of electrons and holes riding waves of potential difference like
subatomic beach-bums is certainly picturesque, but is it anything more than a
sophisticated party trick? In 1997, a team led by Carsten Rocke, one of
Wixforth鈥檚 students at Munich, announced that they鈥檇 constructed a minute
sandwich made from layers of indium and gallium-based semiconductors on a
piezoelectric base and created a surface acoustic wave using a high-frequency
electric field.

A burst of infrared laser light spawned electron and hole pairs, which were
promptly pushed apart by the electric field. Held about a micrometre apart and
unable to recombine, the electron-hole pairs had no choice but to preserve the
energy of the photon that gave them birth.

Rocke and his colleagues managed to store the energy for several
microseconds鈥攖housands of times longer than the natural lifetime of the
electrons and holes. And it could have been longer, says Rocke: it鈥檚 just that
the crystal was only a few millimetres long, so the sound wave, which travels at
around 3000 metres per second in the crystal, simply ran out of space.

Reporting their achievement two years ago in Physical Review Letters
(vol 78, p 4099), the researchers pointed out that they could do a lot more than
simply store the original energy of the photon. They could also dictate when and
where on the device it re-emerged again鈥� an impressive level of
control.

To get the electrons and holes to recombine, the researchers simply smooth
out the hills and valleys of the electric field. They found they could do this
in two ways, by depositing a thin metal electrode on top of the crystal or
running one sound wave into another with an electric field of opposite phase.
When the electrons and holes recombine, they produce a burst of photons with
essentially the same energy as the ones originally injected. In short, Rocke and
his colleagues had bottled up a light beam, stored it for a while and then
released it again.

But there was a catch: all their experiments had been performed at liquid
helium temperatures, just 4 degrees above absolute zero鈥攏ot exactly
convenient for everyday applications. Now Wixforth and his colleagues have shown
that by using a tiny semiconductor chip made from layers of gallium arsenide and
aluminium arsenide, with a transparent electrode deposited on top to provide the
electric field, the same trick can be done at liquid nitrogen temperatures (
Science, vol 283, p 1292).

They managed to store light for up to 35 microseconds. What鈥檚 more, by
separating the electrons and holes using a stationary electric field, they were
able to make the chip a fraction of the size of the crystal used by Rocke. By
tweaking the design still further, they see no reason why a room-temperature
device can鈥檛 be up and running soon.

Flexible memories

Like so much in life, says Wixforth, building an optical memory is pretty
easy鈥攐nce you know how. 鈥淔rankly, what we鈥檝e done was technically possible
some years ago. It was thinking up the idea that took so long.鈥�

With the basic physics now well in hand, Wixforth and his colleagues are
increasingly looking at possible applications for their 鈥渁cousto-optic鈥� devices.
The flexibility of these memories, they believe, opens the way to a whole range
of devices, capable not only of storing light but also of performing tricks such
as multiplexing and demultiplexing鈥攖he combining of many incoming optical
signals into one and vice versa. Wixforth has discovered that he can even change
the wavelength of the re-emitted light simply by squeezing the semiconductor.
Eventually, the researchers could exploit this to encode extra information, he
says.

But their attention is focusing on dynamic random access memories, or DRAMS,
which are crucial in conventional communications and computing. An acousto-optic
version of these devices, made from an array of semiconductor cells or 鈥減ixels鈥�
(see diagram) and able to deal with photons, could handle tasks
that bring electron-based devices to their knees, says Wixforth. 鈥淎n optical
DRAM would be potentially very attractive for tasks like optical pattern
recognition and image processing.鈥�

In addition to using light, he envisages loading and reading each of the
memory cells using electron-hole pairs carried by surface acoustic waves. Stored
information could even be shifted from one cell to the next for processing.FIG-mg21894902.JPG

In the longer term, acousto-optic components might also play a role in the
development of that quintessentially futuristic bit of kit, the optical
computer. Using laser light instead of wires and exploiting the innate
parallelism of light beams, such a computer is potentially the ultimate number
cruncher鈥攓uantum computer excepted. But, like its sluggish electronic
counterpart, an optical computer will need to store information.

Is Wixforth鈥檚 acousto-optic device the answer? Maybe, though the fact that
the photons coming out of it are not quite the same as those that went in could
be a problem. Unlike laser light, the photons created when the electrons and
holes recombine are incoherent, that is, out of phase with each other. 鈥淭here鈥檚
some controversy at the moment about whether incoherent light is sufficiently
good for transmitting data around an optical computer,鈥� says Andrew Walker, a
leading optical computing expert at Heriot-Watt University in Edinburgh.

Even so, the ingenuity of the Munich team in producing so tiny a photon store
has impressed many in the field. 鈥淭hey鈥檙e a very good group and this is a highly
innovative approach to the problem,鈥� says Clivia Sotomayor Torres, an expert on
electronic materials research at the University of Wuppertal in Germany. 鈥淲hat
you鈥檝e got to remember is that with each factor of ten by which you can make
devices smaller, the volume shrinks by a factor of 1000. In the late 1980s,
people were working with optical devices around 10 centimetres square, and now
Wixforth鈥檚 group is working on the millimetre scale. That鈥檚 a big difference,
and it is what makes their work important.鈥�