快猫短视频

LIGHT is just about the most elusive stuff in nature. It never stays in one
spot and nothing can catch it, much less travel faster. But what if light had
finally met its match? What if you could stop light in its tracks, bend its dead
straight beams into closed loops, and keep it trapped inside a labyrinth of
winding pathways?

You might then be able to build far more efficient lasers, and medical
imaging devices that could see clearly through tissues using ordinary light
instead of X-rays. And maybe even realise the big prize鈥攐ptical computers
that would process data at the speed of light. An optical computer would work
much like an electronic computer, but would substitute pure light for
electricity. But to make one, you would need a way to control bits of light with
precision, and to store them for more than a mere instant.

For years, physicists have been trying to pull off these tricks with
鈥減hotonic band gap鈥� materials, crystals that can mould the flow of light in much
the same way as semiconductors mould the flow of electrons. These have been
demonstrated in experiments, but their promise as light-powered switches and
memory devices is still more than rivalled by the monumental complexities of
creating them.

Quantum pinball

To make a photonic band gap material that would work for visible light, you
need huge crystals in which the distances between atoms are some thousand times
larger than in ordinary semiconductor crystals. 鈥淭hey are quite unnatural,鈥� says
Ad Lagendijk, professor of physics at the University of Amsterdam. 鈥淵ou need
clever tricks to make even an impure photonic crystal.鈥� Lagendijk should know.
For he is one of a team of physicists who have now shown that trapping
light not in a crystal but in a disordered material鈥攐ne that contains an
element of randomness in its structure鈥攊s an entirely different
proposition.

What exactly is their breakthough? They鈥檝e ground up a crystalline
semiconductor in an ordinary mortar and pestle to create a powdered material,
honeycombed with microscopic channels and holes. This is their disordered
material鈥攖heir trap for light. Bombard it with infrared light, and the
rays, you might think, would travel through those winding channels, bouncing
along haphazard paths. And then inevitably the photons would escape like balls
bouncing their way through a pinball machine. Indeed, this conclusion would be
unavoidable, but for one thing鈥攍ight has an irreducibly quantum nature.
And quantum pinballs don鈥檛 always act like classical ones.

In 1956, Philip Anderson, a young physicist working at Bell Laboratories,
pointed out that there is a peculiar conspiracy between disorder and the laws of
the quantum world that can keep quantum pinballs pinned to the spot. Anderson
had electrons in mind, not light: he was trying to explain why electrons had a
very hard time moving through a semiconductor doped with impurities, whereas
they slipped quite easily through the pure crystal of undoped semiconductor. The
impurities, Anderson argued, acted as obstacles, scattering the travelling
electrons and knocking them about randomly as they moved through the crystal鈥檚
structure. But that was only part of the story.

If electrons really did move in this way, then they should have been able to
escape the forest of impurities, albeit slowly. Experiments, however, showed
that they didn鈥檛. Anderson鈥檚 achievement was to prove mathematically that if
enough impurities were littered through the material, then out of all the
conceivable paths an electron might follow, quantum effects should single out
some to be especially important. In particular, 鈥渃losed loops鈥� come to dominate
an electron鈥檚 behaviour (see Diagram).
For any path an electron might
take through the obstacle course and back to its original position, it can also
follow that path in reverse. And because electrons act like waves, the electron
waves moving along the closed loops interfere with each other constructively,
increasing the likelihood that an electron will stay.

This effect became known as 鈥淎nderson localisation鈥�. In 1984, Harvard
doctoral student Sajeev John 鈥渂orrowed鈥� Anderson鈥檚 idea to build a similar
theory for light. But most physicists working in photonics鈥攖he young
science of using light to manipulate information鈥攊gnored John鈥檚 notions
until Anderson himself took up the concept in an influential paper published in
1985. By then, photonics researchers had learnt enough about light鈥檚 behaviour,
and about structuring materials at a molecular level, to see an opportunity: if
it was possible to localise electrons, surely it should be possible to localise
photons in a practical way.

By 1991, researchers had managed to localise microwaves鈥攍ight waves of
typically a few centimetres in wavelength, far longer than visible
light鈥攊n two-dimensional networks of metallic rods. It was a step forward.
鈥淏ut when you use microwaves, it鈥檚 difficult to keep the material from absorbing
the waves instead of just trapping them and holding them intact,鈥� Lagendijk
explains. 鈥淵ou must be able to distinguish true localisation from mere
补产蝉辞谤辫迟颈辞苍.鈥�

In order to localise light at shorter wavelengths, Lagendijk and his
colleagues knew that they had to find a material that would reflect and scatter
light as much as possible, flinging photons along the tortuous paths that might
keep them trapped. And they also had to devise a way to distinguish localised
light from light that was simply being absorbed. Because something that is white
scatters light of all colours, Lagendijk鈥檚 group first set out to find the
whitest stuff possible. In technical terms, they were hunting for something with
a high refraction index.

A material鈥檚 index of refraction is a measure of how much speed light loses
when it hits the material from empty space. Water, for example, has an index of
refraction of 1.3, indicating that light travels about 30 per cent more slowly
through water than through space. The index also tells you how much of the light
attempting to jump the boundary from space into the material will be reflected
backward, due to the sudden change in speed at the boundary. Materials with a
high refraction index reflect photons better. This is why earlier researchers
had been enticed by titanium dioxide, dazzlingly white with an index of
refraction of 2.7, the highest known within the spectrum of light visible to the
human eye. Using it, they had tried all kinds of schemes to trap light, from
microscopic spheres of the material suspended in air, to Swiss-cheese-like
arrangements of air bubbles in titanium dioxide. But nothing worked. The trouble
was that the compound couldn鈥檛 scatter light quite strongly enough.

So in 1995, Lagendijk and Diederik Wiersma, a former student of his by then
based at the European Laboratory for Nonlinear Spectroscopy in Florence, decided
to take a different tack. Instead of using visible light, they went for infrared
light and used gallium arsenide, a semiconductor. Photonics researchers knew
that gallium arsenide has a refractive index of 3.48 for infrared light. But in
an odd sort of scientific oversight, no one had ever tried to use it to achieve
localisation. 鈥淭he idea of using gallium arsenide to do the experiment wasn鈥檛
new,鈥� Wiersma says. 鈥淲hat was new was to just go ahead and actually do the
experiment using this different material.鈥�

Gallium arsenide also offered another benefit: it has just the properties to
avoid absorbing infrared light. The reason has to do with the arrangement of its
atoms, and how these atoms affect the movement of electrons in the material. In
a semiconductor, some electrons aren鈥檛 tied to specific atoms, but spread out
through the material as waves. Different electrons have different energies, but,
due to the structural 鈥減ersonalities鈥� of semiconductors, not all possible
energies can be occupied by electrons. The permitted energies form 鈥渂ands鈥�,
separated by the forbidden energy levels which form 鈥渂and gaps鈥�
(see Diagram).
When a photon enters the material and has enough energy to send an
electron from the 鈥渧alence鈥� band to the 鈥渃onduction鈥� band, across the band gap,
that photon can be absorbed. On the other hand, a photon with energy less than
the band gap can鈥檛 persuade the electron to jump across the gap. So this photon
will pass through the material unimpeded rather like light going through a
window.

Setting the trap

As it turns out, gallium arsenide鈥檚 band gap is just right to prevent the
absorption of infrared light. So by using this semiconductor, the teams solved
two problems at once: they found a material that would not only reflect back
most of the light trying to get in from the air, but would not absorb any of the
light that manages to slip inside its boundaries.

To prepare their traps, each group ground gallium arsenide crystals into
powder, and suspended the granules in methanol. The liquid not only trapped
toxic dust, but also sorted the particles: the largest sank to the bottom, the
smallest floated at the top, and those of in-between sizes arrayed themselves in
between. By skimming off the particles at specific depths in the liquid, the
researchers found they could make samples containing mostly large granules, and
others of mostly medium or small grains. Within each sample the particles were
approximately the same size. 鈥淭hat was close enough,鈥� says Wiersma. 鈥淯niformity
in the particles鈥� size isn鈥檛 important. It doesn鈥檛 matter if they鈥檙e identical
spheres or a combination of eggs and potatoes as long as they scatter very
蝉迟谤辞苍驳濒测.鈥�

All of these powders shared one key property: the microscopic tunnels and
chambers formed out of the gaps between particles were about a millionth of a
metre in length and diameter鈥攔oughly equal in size to one wavelength of
infrared light. That dimension is crucial. For theory shows that localisation
should be strongest in convoluted passages where light can鈥檛 stretch out for
more than a single wavelength before it has to turn and go in a different
direction. The light trapped in these nooks and crannies should create the kind
of constructive interference that Anderson had envisioned for electrons, and
should keep the light firmly in its place.

Caged light

When the teams directed an infrared laser on the samples, they saw
tantalising signs of caged light. The first clue came when they tried to shine
light through samples of slightly different thicknesses. If the light passing
through the samples was being merely scattered and not localised, the intensity
of light emerging from the other side of the samples would have obeyed Ohm鈥檚
law, declining proportionately with the thickness of the sample. Which is
exactly what they found in the samples made out of the largest particles. But in
samples of moderate-size grains the intensity declined in proportion to the
square of the sample鈥檚 thickness. And in samples made up of the finest grains,
the intensity of light passing through the sample declined exponentially in
proportion to the thickness鈥攕uggesting that the light was trapped and just
couldn鈥檛 make it through the fine networks of spaces in these powers.

While these results suggested localisation, they couldn鈥檛 absolutely rule out
absorption. Which was it? To find out, the team analysed the 鈥渂ackscattering
cone鈥�. The idea is to measure the intensity of light scattered from an object in
all directions. Plot that intensity on a graph, and you find that the scattered
light is most intense at the centre of the graph. This is the light that bounces
back in a straight line to its source. The shape of the data near but not at the
graph鈥檚 centre is a unique signature of a particular behaviour of light. When
the sample scatters light strongly, the graphed points sweep sharply upward at
the graph鈥檚 centre to form a peak or cusp, meaning the light intensity is high
and that the sample retains little of it. By contrast, says Wiersma, 鈥淲hen light
is being absorbed, the cusp at the centre disappears.鈥� For all of the group鈥檚
samples, the graph鈥檚 cusp remained 鈥攊ndicating that the light wasn鈥檛 being
absorbed, but was being trapped inside.

So it seems that light鈥攊nfrared light, at least鈥攈as been trapped,
brought to a screeching halt in some unassuming piles of powder. But not all
scientists are convinced just yet. Physicist Georg Maret and his colleagues at
Konstanz University believe that Lagendijk鈥檚 team may have overinterpreted its
results. 鈥淲e find that all reported data can be accounted for by classical
diffusion of light combined with reasonable amounts of absorption,鈥� Maret writes
in a recent letter submitted to Nature.

But University of Pennsylvania physicist Arjun Yodh, an expert on the
behaviour of light, voices the view of most scientists who have reviewed the
data. 鈥淚t鈥檚 fairly convincing to me,鈥� he says, 鈥渕ore convincing than anything
else I鈥檝e seen.鈥� Adds John, now at the University of Toronto: 鈥淭he group has
given a fairly convincing presentation. The practical implications have yet to
be worked out, but at the time the semiconductor was invented no one envisioned
how many applications it would have. This is probably a similar case.鈥�

John himself envisions several, including more efficient lasers. In a typical
laser, he notes, you have to pump in a lot of energy into a crystal before it鈥檚
stimulated enough to produce a laser beam. Most of the energy is lost. But if
you can localise a small amount of light in one place, that light could
stimulate a material during the entire period it鈥檚 contained. The great
advantage, he stresses, is that you鈥檇 need to put less energy in to get laser
radiation out.

Binding light up in the practical ways needed to make the parts of a
functioning computer may still be far off. But now that light鈥檚 ever-elusive
essence has been enslaved, the next goal is to put it to work.