杏吧原创

MANIPULATING beams of light is old hat. People have been making mirrors since
2000 years before Cleopatra, and the ancient Greeks knew lenses as burning
glasses. In the much more recent past, scientists have managed to create devices
that can focus, reflect and split different types of beams. Electrons, for
example, are easily guided using electric and magnetic fields. But it has taken
until the past few years to create a set of tools that can manipulate beams of
atoms, which carry no net charge. The trick sounds counterintuitive, but
physicists have found that they must fashion their mirrors and lenses not from
matter but from light.

It is hard to say what the practical value of these devices will be because
they are still confined mostly to the realm of pure research. But people are
hopeful. J眉rgen Mlynek of the Centre for Modern Optics at the University of
Konstanz in southern Germany, for example, foresees laser-like devices that
produce intense beams of atoms, and a new generation of atom microscopes. In one
area, exciting advances have already been made: physicists have learnt to use
atom beams to build staggeringly small structures. The prize for mastering this
branch of 鈥渁tom optics鈥� is a wholly new way to build electronic circuits beyond
the reach of today鈥檚 chip makers.

鈥淚t鈥檚 generally difficult to exert a force on neutral atoms,鈥� says Jabez
McClelland of the National Institute of Standards and Technology (NIST) in
Gaithersburg near Washington DC. 鈥淭here鈥檚 no very easy way to push them around
and make them go where you want them to go.鈥� The ability to guide beams of atoms
has 鈥渟parked the whole field and got things going鈥�, he says.

Atom optics exploits the interactions between atoms and laser light. When an
atom enters an electric field, its cloud of negatively charged electrons is
distorted relative to its positive nucleus. It becomes an induced electric
dipole, rather like a spring with a positive charge at one end and a negative
charge at the other. If the electric field changes from place to place, the
dipole is attracted to the region where the field is strongest.

The elements designed to manipulate beams of atoms take advantage of the fact
that light is made up partly of an oscillating electric field. By reflecting
laser light back on itself, physicists can generate a standing wave similar to
the standing sound wave in an organ pipe. The regular intensity peaks of this
wave stay static in space, and each one can be used as a lens. A fine beam of
atoms passed through just one of these intensity peaks will experience a force
acting towards the centre of the peak. Mlynek and his colleagues were the first
to demonstrate this effect in 1992 when they focused a beam of helium atoms with
a red laser.

One of the intellectual leaps made in the 1920s with the formulation of
quantum theory was that subatomic particles also behave as waves. Physicists
have now shown that atoms also have this split personality. For example, in 1991
Mlynek and Oliver Carnal, also then at Konstanz, performed the first 鈥渄ouble
slit鈥� experiment with atoms. In the early 19th century, the English scientist
Thomas Young divided a light beam by passing it through two parallel slits. On
the far side, the light waves from the slits recombined to produce a series of
light and dark bands, or interference fringes. Mlynek and Carnal found a similar
interference pattern when they replaced the beam of light with one of atoms.

The 鈥渨avelength鈥� of a minute particle decreases as its momentum increases,
and momentum depends on the particle鈥檚 mass and velocity. Because atoms are much
heavier than electrons, their wavelengths tend to be shorter. A beam of
slow-moving helium atoms, for example, has a wavelength of about 0.1 nanometres:
electrons with the same energy have a wavelength of about 10 nanometres. In
turn, this is much lower than the wavelength of light, which is measured in
hundreds of nanometres.

The electron鈥檚 short wavelength is already used to great effect in electron
microscopes, which can resolve much finer structures on the surface of a sample
than can a light microscope. Mlynek sees a parallel here for atom beams,
suitably focused with light-force lenses. 鈥淵ou can use this beam to study
surfaces, so you can try to study the topography of the surface and special
properties of the surface,鈥� he says. Electrons could be produced with the same
wavelength as slow-moving atoms, but they would have to travel at such high
speeds that they would obliterate the sample鈥檚 surface features.

Splitting beams

So far nobody has built an atom microscope. It would take great efforts to
improve on the resolution of devices such as scanning tunnelling microscopes,
says Tilman Pfau, a colleague of Mlynek鈥檚 at Konstanz. But such a device would
have some unusual advantages, he stresses. 鈥淎toms have magnetic
properties鈥攁 magnetic spin,鈥� says Pfau. An atom microscope should be able
to sense these magnetic properties.

Mlynek鈥檚 original lens consisted of a single intensity peak from a standing
light wave. It is actually easier to fire atoms at a string of these peaks to
produce not a lens, but a 鈥渄iffraction grating鈥� (see
Diagram). In conventional
optics, a diffraction grating can be made from a glass slide marked with opaque,
parallel lines every 2000 nanometres or so. Light waves emerging from between
these lines 鈥渞einforce鈥� each other only if they are in phase, which happens
along the central axis and at certain angles relative to the axis. At other
angles, they tend to cancel each other out. In effect, an incident light beam is
split into several secondary beams that appear as bright bars on a screen beyond
the grating.FIG-20264001.jpg

The ability to turn the tables and diffract atom beams with light was first
demonstrated in the mid-1980s by a pioneering group led by David Pritchard at
the Massachusetts Institute of Technology. For the first time, physicists could
split one atom beam into two or more. Several groups, including Pritchard鈥檚,
have since used a series of three parallel standing wave gratings to split one
atom beam into two, and then recombine them.

One such system has been built by Siu-Au Lee and her team at Colorado State
University in Fort Collins. 鈥淭he first standing light wave acts as a 50:50
beam-splitter,鈥� she says. The two atom beams head off in different directions
until they hit the second grating. This sends the two beams back towards each
other鈥攁s if they had glanced off a pair of mirrors placed perpendicular to
the standing wave鈥攁nd the third grating recombines them. The overlapping
atom waves from the two beams create a pattern of interference fringes that can
be seen with an atom detector.

Taking a lead once more from traditional optics, such devices have opened up
an entirely new field of research鈥攁tom interferometry. An optical
interferometer splits a beam of light and then recombines the two halves. If
something changes the path of one beam but not the other, the phase relationship
between the two is disturbed and the interference pattern shifts. This effect is
used to measure the refractive index of a material, for example, or very small
changes in length.

The advantage of atoms over light is that their wavelengths are much shorter
so more precise measurements of length can be made. Pritchard鈥檚 group has also
hung its interferometer from the lab ceiling and twisted it very slightly. The
tiny jolt lengthened the path of one beam, and shortened that of the other,
shifting the interference pattern. Pritchard believes that this effect could be
used to create an ultraprecise gyroscope.

Back in Colorado, Lee points out that each diffraction grating can be tuned
to give the best results. 鈥淭he very interesting thing is that by changing the
laser intensity or the laser frequency, or both, we can vary the reflectivity of
a beam-splitter,鈥� says Lee. 鈥淚ndeed, it is possible to have the reflectivity
high enough that the light wave acts like a mirror. In principle, it should be
possible to get close to 100 per cent reflectivity.鈥�

This flexibility of laser light is a real bonus, according to Mlynek. But
lasers also bring drawbacks. At present lenses, mirrors and beam-splitters for
atoms work only when the laser light has a frequency that is close to鈥攂ut
not the same as鈥攁 natural resonant frequency of the atoms. This limits the
scope of atom optics.

The resonant frequencies of atoms are dictated by the frequencies at which
their electrons absorb light energy. If atom opticians use a laser with exactly
the same frequency as the atoms鈥� resonance, the mirror, lens or beam splitter
loses its properties. This happens because the atoms鈥� electrons absorb the light
and jump to a higher energy state, before dropping down again and re-radiating
the absorbed energy in any old direction. The atoms then recoil and appear to
scatter at random.

For the dipole force to have maximum effect, the atoms must be excited using
laser light of a frequency close to the atoms鈥� resonant frequency. Light is an
oscillating electric field, so the dipole鈥攖he atom鈥攎ust also
oscillate. If the chosen laser oscillates at slightly less than the atom鈥檚
resonant frequency鈥攌nown as red detuning鈥攖hen the atom will
oscillate in step with the field, and the atom will be driven towards the region
of highest field. If the laser is detuned farther from the atom鈥檚 resonant
frequency, the induced dipole鈥攁nd hence the light force鈥攂ecomes
weaker.

Near-resonance laser techniques opened the door to another way to make a
mirror for neutral atoms, dreamt up in 1982 by Richard Cook and Richard Hill at
the US Air Force Institute of Technology in Ohio. When light travelling in a
block of glass hits the glass-air boundary at a glancing angle, it is 鈥渢otally
internally reflected鈥� back into the glass. It is this effect that keeps light
inside optical fibres as they twist and turn.

The light鈥檚 electric field, however, leaks out of the glass and into the air,
falling off over a distance of a couple of thousand nanometres. This 鈥渆vanescent
wave鈥�, which moves along the surface in the direction of the original beam, is
strong enough to exert a dipole force on neutral atoms. Cook and Hill calculated
that atoms travelling at about 1 metre per second would bounce off an evanescent
wave mirror without ever touching the glass.

Since then, research on evanescent wave mirrors has moved on by leaps and
bounds. Jean Dalibard, Claude Cohen-Tannoudji and their colleagues at the Ecole
Normale Sup茅rieure in Paris, have dropped slow-moving atoms on their
mirror and watched them bounce. 鈥淲e used a `trampoline鈥� geometry, composed of a
single, horizontal, concave mirror,鈥� says Dalibard. Using gravity and a mirror
is one way to create a cavity for trapping atoms鈥攁nother area that has
attracted huge attention in recent years (鈥淎toms caught in a web of light鈥�,
New 杏吧原创, 29 January 1994, p 32).

Another French group, led by Alain Aspect at the Institute of Optics in
Orsay, near Paris, is experimenting with coatings on the glass that can
strengthen the evanescent wave by as much as a thousand times. These mirrors can
reflect much faster atoms or may be driven by laser light that is further
detuned from resonance.

Atom lasers

With atom mirrors has come the prospect of the 鈥渁tom laser鈥濃�攚hich would
be to an atom beam what a laser is to light from a bulb. Not only should the
beam from an atom laser be highly directional, says Pfau, all the atoms would
have the same velocity. Light lasers need two things: a 鈥減ump鈥� to raise
electrons into a higher energy state, and a resonant cavity so that light given
off when the electrons drop back to their lower state can bounce back and forth
through the lasing medium and reinforce the beam. The resonant cavity of a light
laser is usually a pair of mirrors, precisely positioned at either end of the
lasing medium. Pfau is trying to build an atom laser using a single evanescent
wave mirror and an envelope of laser light in place of a second mirror. He
directs low energy atoms through the envelope into the cavity, where they are
鈥減umped鈥� using another laser.

鈥淭he output is some new state of matter,鈥� says Pfau. This is because the
atoms in an atom laser beam will all have sunk, or condensed, into a single
quantum state, just as light photons do in a laser, and the atoms then become
indistinguishable. This state, known as a Bose-Einstein condensate was first
seen fleetingly last year by Eric Cornell, Carl Wieman and colleagues at the
Joint Institute for Laboratory Astrophysics in Boulder, Colorado (Science,
New 杏吧原创, 22 July 1995, p 16). 鈥淵ou can look at these experiments as a
pulsed laser-like source of atoms. They generate a condensate and release it,
and that鈥檚 the pulse,鈥� says Pfau. 鈥淥ur goal is to make a continuous source.鈥�

One area of atom optics that is already producing results is atom
lithography. The aim here has largely been to build tiny structures with the
atoms in a beam. At NIST, McClelland has been piling up chromium atoms on
silicon. 鈥淲e put a standing wave across the surface of a substrate, and we
deposit the chromium atoms through it, and as they go through they get
concentrated by the lenses into an array of little lines,鈥� he says. In 1993,
McClelland and his colleagues, including Robert Celotta and Rajeev Gupta,
deposited lines of chromium just 65 nanometres wide and separated by 213
nanometres. 鈥淭he major progress that we have made is to generate a
two-dimensional array, that is, an array of dots instead of an array of lines,鈥�
says McClelland. 鈥淭his we do by superimposing two standing waves on each other
at 90 degrees.鈥� Each dot is about 80 nanometres across.

Structures more interesting than simple dots might be possible by shuffling
the substrate about as the atoms are deposited, or even by creating exotic
patterns of light intensity capable of focusing atoms into the desired shapes.
鈥淭hat鈥檚 a very difficult problem and we haven鈥檛 tackled it yet, but we鈥檙e
working on it,鈥� says McClelland. Nevertheless, even the dot pattern is a step
closer to the team鈥檚 ultimate goal: to make the smallest ever microcircuits.

Radical departure

Building individual structures with atom beams is a radical departure from
the way chips are made at present. Today, silicon is painted with a substance
called photoresist, the desired pattern is masked off and the chip exposed to
light, which chemically alters the photoresist. The unchanged areas of
photoresist can then be etched away. Chips inside modern computers have a
feature size of around 350 nanometres and the next generation of chips will have
features half this size. Nevertheless, chip makers know that the relatively long
wavelength of light will prevent them going below about 100 nanometres.

However, beams of atoms鈥攚ith their shorter wavelengths 鈥攑romise
structures with much shorter dimensions. 鈥淚 see the strength as being able to
produce a large number, on the order of a million, of very small features in an
extremely precise fashion, and in a relatively short time,鈥� says Lee. Her group
has used standing light waves to lay down aluminium lines on a silicon surface.
鈥淭he weakness is that the method is atom-specific, that is to say, different
atoms will require different lasers for deposition, and most atoms cannot be
deposited this way because there are no lasers suitable for the job,鈥� she
says.

A different approach, more akin to today鈥檚 chip-making methods, is being
tried out by another group of researchers of which McClelland is also a part.
The team includes another NIST group, led by Bill Phillips, chemists from
Harvard University led by George Whitesides, and Mara Prentiss, a physicist from
Harvard. Their technique takes advantage of the fact that electrons within atoms
can stay for some time in energetic states called metastable states. Last year,
the team reported the results of directing a beam of metastable argon atoms onto
a substrate covered by a film of alkanethiolate just 1 molecule thick.

鈥淲hen these little metastable atoms hit the surface, they give off their
energy鈥攊t鈥檚 like dropping a little bomb on the surface鈥攁nd it
damages the molecular layer,鈥� says McClelland. The damaged areas, and the
metallic layers below, can later be etched away. McClelland and his colleagues
found that with a physical mask the technique could carve out patterns with
dimensions as small as 20 nanometres.

Light mask

Going a step further, the team has now made a mask out of light. This
technique works because although metastable argon atoms have enough energy to
damage the alkanethiolate film, the same atoms in their low-energy 鈥済round
state鈥� do not. 鈥淲e can transfer atoms between internal states using light and
thus control whether an atom will damage the resist,鈥� says Prentiss. As
metastable atoms pass through the light mask, their electrons are excited into a
still higher energy state and then almost immediately drop back to their ground
state鈥攇iving off the surplus energy as photons. The atoms in their ground
state bounce harmlessly off the alkanethiolate film (see
Diagram).FIG-20264002.jpg

The past six years have largely been devoted to creating the elements, such
as lenses and mirrors, that underpin atom optics, says Mlynek. Only now are
people beginning to explore the potential of these elements. McClelland points
out that atom lithography is still experimental, and Pfau reckons he will need a
year to build his atom laser. But both men are excited by the notion of
combining the two techniques. The resolution of today鈥檚 lithographic techniques
is limited by the wide spread of velocities within atoms beams. But the atoms
streaming from an atom laser would all travel at the same speed. The combination
could prove something very special indeed. Watch this space.