Three years ago in a Colorado laboratory, scientists realised a long-standing dream, bringing the quantum world closer to the one of everyday experience
by Eric A. Cornell and Carl E. Wieman
In June 1995 our research group at the Joint Institute for
Laboratory Astrophysics (now called JILA) in Boulder, Colo., succeeded in
creating a minuscule but marvellous droplet. By cooling 2,000 rubidium atoms
to a temperature less than 100 billionths of a degree above absolute zero
(100 billionths of a degree kelvin), we caused the atoms to lose for a full
10 seconds their individual identities and behave as though they were a single
"superatom." The atoms' physical properties, such as their motions, became
identical to one another.This
Bose-Einstein condensate (BEC), the
first observed in a gas, can be thought of as the matter counterpart of the
laser-except that in the condensate it is atoms, rather than photons, that
dance in perfect unison.
Our short-lived, gelid sample was the experimental realisation
of a theoretical construct that has intrigued scientists ever since it was
predicted some 73 years ago by the work of physicists
Albert Einstein and Satyendra
Nath Bose. At ordinary temperatures, the atoms of a gas are scattered throughout
the container holding them. Some have high energies (high speeds); others
have low ones. Expanding on Bose's work, Einstein showed that if a sample
of atoms were cooled sufficiently, a large fraction of them would settle
into the single lowest possible energy state in the container. In mathematical
terms, their individual wave equations-which describe such physical
characteristics of an atom as its position and velocity-would in effect merge,
and each atom would become indistinguishable from any other.
Progress in creating Bose-Einstein condensates has sparked
great interest in the physics community and has even generated coverage in
the mainstream press. At first, some of the attention derived from the drama
inherent in the decades- long quest to prove Einstein's theory. But most
of the fascination now stems from the fact that the condensate offers a
macroscopic window into the strange world of quantum mechanics, the theory
of matter based on the observation that elementary particles, such as electrons,
have wave properties. Quantum mechanics, which encompasses the
uncertainty principle, uses these wavelike properties to describe the
structure and interactions of matter.
We can rarely observe the effects of quantum mechanics in the
behaviour of a macroscopic amount of material. In ordinary, so-called bulk
matter, the incoherent contributions of the uncountably large number of
constituent particles obscure the wave nature of quantum mechanics, and we
can only infer its effects. But in Bose condensation, the wave nature of
each atom is precisely in phase with that of every other. Quantum-mechanical
waves extend across the sample of condensate and
can be observed with the naked eye. The sub- microscopic thus becomes
New Light on Old Paradoxes
Nothing in our experience, based as it is on familiarity with
matter at normal temperatures, helps us comprehend this paradox. That is
because at normal temperatures and at the size scales we are all familiar
with, it is possible to describe the position and motion of each and every
object in a collection of objects. The numbered Ping-Pong balls bouncing
in a rotating drum used to select lottery numbers exemplify the motions
describable by classical
At extremely low temperatures or at small size scales, on the
other hand, the usefulness of classical mechanics begins to wane. The crisp
analogy of atoms as Ping-Pong balls begins to blur. We cannot know the exact
position of each atom, which is better thought of as a blurry spot. This
spot-known as a wave packet-is the region of space in which we can expect
to find the atom. As a collection of atoms becomes colder, the size of each
wave packet grows. As long as each wave packet is spatially separated from
the others, it is possible, at least in principle, to tell atoms apart. When
the temperature becomes sufficiently low, however, each atom's wave packet
begins to overlap with those of neighbouring atoms. When this happens, the
atoms "Bose - condense" into the lowest possible energy state, and the wave
packets coalesce into a single, macroscopic packet. The atoms undergo a quantum
identity crisis: we can no longer distinguish one atom from another.
The current excitement over these condensates contrasts sharply
with the reaction to Einstein's discovery in 1925 that they could exist.
Perhaps because of the impossibility then of reaching the required
temperatures-less than a millionth of a degree kelvin-the hypothesised gaseous
condensate was considered a curiosity of questionable validity and little
physical significance. For perspective, even the coldest depths of intergalactic
space are millions of times too hot for Bose condensation.
In the intervening decades, however, Bose, condensation came back into fashion. Physicists realised that the concept could explain superfluidity in liquid helium, which occurs at much higher temperatures than gaseous Bose condensation. Below 2.2 kelvins, the viscosity of liquid helium completely disappears - putting the "super" in superfluidity.
Not until the late 1970s did refrigeration technology advance
to the point that physicists could entertain the notion of creating something
like Einstein's original concept of a BEC in a gas. Laboratory workers at
M.I.T., the University of Amsterdam, the University of British Columbia and
Cornell University had to confront a fundamental difficulty. To achieve such
a BEC, they had to cool the gas to far below the temperature at which the
atoms would normally freeze into a solid. In other words, they had to create
a supersaturated gas. Their expectation was that hydrogen would supersaturate,
because the gas was known to resist the atom-by-atom clumping that precedes
Although these investigators have not yet succeeded in creating
a Bose-Einstein condensate with hydrogen, they did develop a much better
understanding of the difficulties and found clever approaches for attacking
them, which benefited us. In 1989, inspired by the hydrogen work and encouraged
by our own research on the use of lasers to trap and cool alkali atoms, we
began to suspect that these atoms, which include caesium, rubidium and sodium,
would make much better candidates than hydrogen for producing a Bose condensate.
Although the clumping properties of caesium, rubidium and sodium are not
superior to those of hydrogen, the rate at which those atoms transform themselves
into condensate is much faster than the rate for hydrogen atoms. These much
larger atoms bounce off one another at much higher rates, sharing energy
among themselves more quickly, which allows the condensate to form before
clumping can occur.
Also, it looked as if it might be relatively easy and inexpensive
to get these atoms very cold by combining ingenious techniques developed
for laser cooling and trapping of alkali atoms with the techniques for magnetic
trapping and evaporative cooling developed by the researchers working with
hydrogen. These ideas were developed in a series of discussions with our
friend and former teacher, Daniel Kleppner, the co-leader of a group at M.I.T.
that is attempting to create a condensate with hydrogen.
Our hypothesis about alkali atoms was ultimately fruitful.
Just a few months after we succeeded with rubidium, Wolfgang Ketterle's group
at M.I.T. produced a Bose condensate with sodium atoms; since that time,
Ketterle's team has succeeded in creating a condensate with 10 million atoms.
At the time of this writing, there are at least seven teams producing
condensates. Besides our own group, others working with rubidium are Daniel
J. Heinzen of the University of Texas at Austin, Gerhard Rempe of the University
of Konstanz in Germany and Mark Kasevich of Yale University. In sodium, besides
Ketterle's at M.I.T., there is a group led by Lene Vestergaard Hau of the
Rowland Institute for Science in Cambridge, Mass. At Rice University Randall
G. Hulet has succeeded in creating a condensate with lithium. All these teams
are using the same basic apparatus.
As with any kind of refrigeration, the chilling of atoms requires
a method of removing heat and also of insulating the chilled sample from
its surroundings. Both functions are accomplished in each of two, steps.
In the first, the force of laser light on the atoms both cools and insulates
them. In the second, we use magnetic fields to insulate, and we cool by
Laser Cooling and Trapping
We adjust the frequency of the laser radiation so that the atoms
absorb it and then reradiate photons. An atom can absorb and reradiate many
millions of photons each second, and with each one, the atom receives a minuscule
kick in the direction the absorbed photon is moving. These kicks are called
radiation pressure. The trick to laser cooling is to get the atom to absorb
mainly photons that are travelling in the direction opposite that of the
atom's motion, thereby slowing the atom down (cooling it, in other words).
We accomplish this feat by carefully adjusting the frequency of the laser
light relative to the frequency of the light absorbed by the atoms [see
In this setup, we use laser light not only to cool the atoms
but also to
keeping them away from the room-temperature walls of the cell. In fact, the
two laser applications are similar. With trapping, we use the radiation pressure
to oppose the tendency of the atoms to drift away from the centre of the
cell. A weak magnetic field tunes the resonance of the atom to absorb
preferentially from the laser beam that is pointing toward the centre of
the cell (recall that six laser beams intersect at the centre of the cell).
The net effect is that all the atoms are pushed toward one spot and are held
there just by the force of the laser light.
These techniques fill our laser trap in one minute with 10 million
atoms captured from the room-temperature rubidium vapour in the cell. These
trapped atoms are at a temperature of about 40 millionths of a degree above
absolute zero-an extraordinarily low temperature by most standards but still
100 times too hot to form a BEC. In the presence of the laser light, the
unavoidable random jostling the atoms receive from the impact of individual
light photons keeps the atoms from getting any colder or denser.
To get around the limitations imposed by those random photon
impacts, we turn off the lasers at this point and activate the second stage
of the cooling process. This stage is based on the magnetic-trapping and
evaporative-cooling technology developed in the quest to achieve a condensate
with hydrogen atoms. A magnetic trap exploits the fact that each atom acts
like a tiny bar magnet and thus is subjected to a force when placed in a
magnetic field . By carefully controlling the shape of the magnetic field
and making it relatively strong, we can use the field to hold the atoms,
which move around inside the field much like balls rolling about inside a
deep bowl. In evaporative cooling, the most energetic atoms escape from this
magnetic bowl. When they do, they carry away more than their share of the
energy, leaving the remaining atoms colder.
The analogy here is to cooling coffee. The most energetic water
molecules leap out of the cup into the room (as steam), thereby reducing
the average energy of the liquid that is left in the cup. Meanwhile countless
collisions among the remaining molecules in the cup apportion out the remaining
energy among all those molecules. Our cloud of magnetically trapped atoms
is at a much lower density than water molecules in a cup. So the primary
experimental challenge we faced for five years was how to get the atoms to
collide with one another enough times to share the energy before they were
knocked out of the trap by a collision with one of the untrapped,
room-temperature atoms remaining in our glass cell.
Many small improvements, rather than a single breakthrough,
solved this problem. For instance, before assembling the cell and its connected
vacuum pump, we took extreme care in cleaning each part, because any remaining
residues from our hands on an inside surface would emit vapours that would
degrade the vacuum. Also, we made sure that the tiny amount of rubidium vapour
remaining in the cell was as small as it could be while providing a sufficient
number of atoms to fill the optical trap.
Incremental steps such as these helped but still left us well
shy of the density needed to get the evaporative cooling under way. The basic
problem was the effectiveness of the magnetic trap. Although the magnetic
fields that make up the confining magnetic "bowl" can be quite strong, the
little "bar magnet" inside each individual atom is weak.This characteristic
makes it difficult to push the atom around with a magnetic field, even if
the atom is moving quite slowly (as are our laser-cooled atoms).
In l994 we finally confronted the need to build a magnetic trap
with a narrower deeper bowl. Our quickly built, narrow-and-deep magnetic
trap proved to be the final piece needed to cool evaporatively the rubidium
atoms into a condensate. As it turns out, our particular trap design was
hardly a unique solution. Currently there are almost as many different magnetic
trap configurations as there are groups studying these condensates.
Shadow Snapshot of a "Superatom"
In the plot of the velocity distribution, the condensate appears
as a dorsal-fin- shaped peak. The condensate atoms have the smallest possible
velocity and thus remain in a dense cluster in the centre of the cloud after
it has expanded. This photograph of a condensate is further proof that there
is something wrong with classical mechanics. The condensate forms with the
lowest possible energy. In classical mechanics, "lowest energy" means that
the atoms should be at the centre of the trap and motionless, which would
appear as an infinitely narrow and tall peak in our image. The peak differs
from this classical conception because of quantum effects that can be summed
up in three words: Heisenberg's uncertainty principle.
The uncertainty principle puts limits on what is knowable about
anything, including atoms. The more precisely you know an atom's location,
the less well you can know its velocity, and vice versa. That is why the
condensate peak is not infinitely narrow. If it were, we would know that
the atoms were in the exact centre of the trap and had exactly zero energy.
According to the uncertainty principle, we cannot know both these things
Einstein's theory requires that the atoms in a condensate have
energy that is as low as possible, whereas Heisenberg's uncertainty principle
forbids them from being at the very bottom of the trap. Quantum mechanics
resolves this conflict by postulating that the energy of an atom in any
container, including our trap, can only be one of a set of discrete, allowed
values-and the lowest of these values is not quite zero. This lowest allowed
energy is called the zero - point energy, because even atoms whose temperature
is exactly zero have this minimum energy. Atoms with this energy move around
slowly near-but not quite at-the centre of the trap. The uncertainty principle
and the other laws of quantum mechanics are normally seen only in the behaviour
of submicroscopic objects such as a single atom or smaller. The Bose-Einstein
condensate therefore is a rare example of the uncertainty principle in action
in the macroscopic world.
Bose-Einstein condensation of atoms is too new, and too different,
for us to say if its usefulness will eventually extend beyond lecture
demonstrations for quantum mechanics. Any discussion of practical applications
for condensates must necessarily be speculative. Nevertheless, our musings
can be guided by a striking physical analogy: the atoms that make up a Bose
condensate are in many ways the analogue to the photons that make up a laser
Ultimate in Precise Control?
Meanwhile many groups have begun a variety of measurements on
the condensates. In a lovely experiment, Ketterle's group has already shown
that when two separate clouds of Bose condensate overlap, the result is a
fringe pattern of alternating constructive and destructive interference,
just as occurs with intersecting laser radiation. In the atom cloud, these
regions appear respectively as stripes of high density and low density. Our
group has looked at how the interactions between the atoms distort the shape
of the atom cloud and the manner in which it quivers after we have "poked"
it gently with magnetic fields. A number of other teams are now devising
their own experiments to join in this work.