ALMOST since the moment of its birth, our Universe has been divided into four
kingdoms, ruled by four different forces. Across astronomical distances, gravity
reigns. The electromagnetic force holds swarms of electrons around nuggets of
protons and neutrons to form atoms. Deep within the atomic nucleus, the weak
force sparks some forms of radioactivity. And the strong force acts on quarks,
clasping them together in trios to form protons and neutrons. But the strong
force must also hold together a strange type of particle that contains no
matter. For more than twenty years, I have been one of several physicists trying
to hunt it down.
Each of the fundamental forces has a particle associated with it. The most
familiar is the photon, that carries the electromagnetic force. Fortunately for
us, photons travel in straight lines, and have no attraction for each other. If
they did attract, the photons reflected from this page might clump together into
“atoms” of light before reaching your eyes. Electrons might be permanently
trapped within atoms, and chemistry as we know it would not exist. In fact, life
only exists because photons can roam free in space.
The force carriers that rule in the atomic nucleus, however, are a different
kettle of fish. These are gluons, the carriers of the strong force between the
quarks that make up protons and neutrons. So strong is the force between them
that they should stick together into “glueballs”—the particles that I and
many others have been seeking for so long. Within the past two years, the first
possible signs of glueballs have turned up in particle experiments. And late
last year, my colleagues and I discovered a magical new way of sifting out
glueballs from more familiar particles. A whole new family of glueballs looks
set to emerge from the mist.
Advertisement
The theory of the strong force, known as quantum chromodynamics (QCD),
suggests that the strong force does more than bind quarks together to form
protons and neutrons. It should also act on the gluon carriers themselves, and
bind them to each other. Shortly after the big bang, so the theory goes, gluons
must have congealed into billions of glueballs similar in size and mass to a
proton. These would have decayed within a split second. But if glueballs formed
then, it should also be possible for them to be produced today in the debris of
collisions between protons in particle accelerators here on Earth. Accelerators
are already powerful enough to do this, so with the right detectors, glueballs
should show up in their thousands.
Lost in the labyrinth
This idea has been around for over 25 years, but evidence that it actually
happens has been tantalisingly hard to find. The problem for experimental
glueball hunters is that QCD does not tell us the exact mass of particle we are
looking for. The theory gives a near-perfect mathematical description of how
quarks and gluons interact at the highest energies, when they behave almost
independently of each other. But glueballs are likely to appear only at low
energies, when quarks and gluons become much more “sticky”. At these energies,
the interactions between quarks and gluons become so labyrinthine that it is not
possible to solve the QCD equations by normal means.
Over the past few years, however, a computer technique known as lattice QCD
has started to give us a coherent picture of how quarks and gluons behave at
these low energies, and the sort of glueball that should form. The trick is to
pretend that space and time form a discrete lattice of points rather than being
continuous. Each point of the lattice has a number attached, representing the
chance a quark or a gluon is there. The computer then simulates the effect of
the strong force on the quarks and gluons, according to QCD theory, and reveals
how they respond, move and cluster into massive particles.
The result can only be approximate. For a perfectly accurate picture of how
these particles behave, we would have to reduce the spacing between the lattice
points to zero, to represent continuous space-time. This would give us an
infinite number of points that would take an infinite amount of time to compute,
even on the fastest computers. However, the discovery of mathematical short cuts
in the past three years has made it possible to compute with finer lattices than
before, and with powerful modern computers this can get us fairly close to
reality. The technique can now calculate the mass of proton, for instance, to
within 20 per cent of its known, measured mass.
To work out the properties of glueballs using lattice QCD, we remove the
quarks and put gluons alone onto the lattice. It turns out that the simplest
stable glueball occupies a connected set of lattice points that forms a closed
loop. Projected into the real, continuous world of three dimensions, this
becomes a sphere of glue. It has no intrinsic angular momentum, or “spin”, and
is known as a “scalar” glueball.
Two leading groups working on lattice QCD have calculated the mass of this
glueball. The UKQCD collaboration, which involves several British universities,
calculated in 1993 that it is about 1.5 times the mass of the proton. A team led
by Don Weingarten of IBM in Yorktown Heights, New York, calculated in 1994 that
it is roughly 1.7 times the proton mass. And last year, the IBM team also
estimated how long this unstable glueball lives. It turns out to be a fraction
longer than 10-23 seconds.
So what’s the best way to look for this fleeting glueball? I suggested in
1983 that glueballs might appear in experiments at CERN, the European particle
physics laboratory near Geneva, using the Low Energy Antiproton Ring or LEAR.
This accelerator shoots protons and antiprotons at each other, and the quarks
and antiquarks from inside the colliding particles might occasionally annihilate
each other, leaving lots of gluons behind. The gluons could then clump into a
variety of ephemeral particles, glueballs among them. As each of these
short-lived species decayed, it would leave a characteristic collection of more
familiar particles.
The principle is simple: find that pattern of particles that is
characteristic of decaying glueballs (see
Diagram). But the practice is a
great deal trickier because the proton-antiproton collisions themselves also
spawn showers of particles containing quarks, which in turn can decay to the
same products as are yielded by glueballs. In the end it all has to come down to
statistics. Each particle produces a given collection of decay products with a
given probability. So by looking at thousands of collisions and counting each
decay product every time, it is possible to work out which particles have
appeared.
An international team of scientists has applied this technique to results
from a detector at LEAR called the Crystal Barrel, which has gathered data since
1989. The Crystal Barrel is especially sensitive to neutral particles, which
should be common in the decay products of a glueball. Most of the particles
revealed by analysing these data have turned out to be old friends made of
quarks. But among them were some strangers. It now seems likely that in future,
one of them will be recognised as the first clear sighting of a glueball.
It was in the spring of 1994 when the excitement began for me. I knew that
Chris Batty, a colleague at the Rutherford Appleton Laboratory, was due to
report the latest results from the Crystal Barrel experiment at a conference in
Glasgow, so I went up to hear what he had to say. In his talk, he mentioned that
one of the unidentified particles had a mass of about 1.5 times the proton mass.
The UKQCD collaboration had just announced their estimate of the mass of the
lightest glueball from lattice QCD calculations and come up with an almost
identical figure. The similarity was striking, and I began to wonder how to
prove the connection.
Three months later, at the International Conference on High Energy Physics,
also in Glasgow, it emerged that the Crystal Barrel particle was cropping up
elsewhere. A team from another experiment at LEAR, known as Obelix, reported
seeing it, as did a team at Fermilab near Chicago. However, the results from all
the teams were a tantalising mix of what we expected from a glueball, together
with other features that did not seem to fit. According to theory, glueballs
should decay to produce a range of particles that includes an equal mix of three
different types of mesons (quark-antiquark pairs) called kaons, pions and
etas. But the mystery particle never produced any kaons when it decayed.
But while the mystery particle did not behave as we thought a glueball
should, neither was it a meson, the most likely conventional candidate, as
Claude Amsler of the Crystal Barrel team and I quickly showed. Like a glueball,
a meson that decays to produce etas must also produce the kaon from time to
time. This result was an important first step for us, but not enough. To draw a
legal analogy, we wanted to prove this particle guilty of being a glueball. And
while a jury would agree that it had not been totally cleared of the charge,
neither was it guilty beyond reasonable doubt. There had to be another piece to
the jigsaw.
In January 1995, while I was waiting at Gatwick Airport for the plane that
would whisk me from wintry England to give a series of lectures in Brazil,
inspiration struck. For a second, the realisation was exhilarating; a moment
later I felt a fool for not having seen the answer much sooner. A scalar
glueball at this mass would sit in the middle of two other scalar particles made
of quarks and antiquarks, with masses of roughly 1.4 and 1.7 times the proton
mass. A well-known property of quantum mechanics would then effectively mix the
three particles together, disturbing the way they would be expected to decay.
The absence of kaons, completely wrong for a pure glueball, was a natural
consequence of this mixing. Now the evidence was starting to point much more
strongly to the guilty verdict.
The extra piece in the jigsaw also added a new twist: the lightest glueball
should implicate three suspects, not just one. The arch criminal must be more or
less a 50:50 mix of glueball and quark, and there will be two
accomplices—mainly quark, but with a little glueball added. Theory says
that all three should be scalar particles with masses that are within about 25
per cent of one another.
Meanwhile, in the light of the new ideas, my colleague David Bugg of Queen
Mary and Westfield College in London had become a sort of glueball
palaeontologist, searching through data from old experiments for glueballs that
no one had spotted before. One experiment he looked at ran at Stanford,
California, in the 1970s. The experiment had recorded the decays of the
short-lived psi particle, which is composed of a charmed quark and its
antiquark. The psi dies when the charmed quark and antiquark destroy one
another, and it is in just this sort of circumstance that you might expect
glueballs to show up.
Bugg noticed a familiar pattern in the data. With Glennys Farrar of Rutgers
University in New Jersey and Zhenping Li of the University of Beijing, we
analysed the data over the best part of a year. And once again, we found a
particle with mass 1.5 times that of the proton. This must be roughly a 50:50
mixture of glueball and quark, the same as the particle that turned up in the
Crystal Barrel experiment.
Although the evidence has become compelling, the case is not closed. But at
the end of last year, I stumbled on an efficient new way of making glueballs
that should help us prove their existence beyond doubt. It has been known for a
long time that glueballs could appear when protons collide, because the
collision can leave behind gluons which can then form a glueball as the protons
leave the scene. Andrew Kirk of Birmingham University has been leading a team
doing these experiments at CERN since 1990.
Proton problem
A problem is that proton-proton collisions can also produce conventional
particles made of quarks, and no one knew how to encourage the glueballs at the
expense of the conventional particles. But in October 1996, Kirk and I chanced
upon the answer. We were discussing an unusual result that Kirk and his team had
just published. They had found that the particles produced in the middle region
are not always the same, and that the collection of particles seems to depend on
the relative orientation of the outgoing pair of protons.
A possible reason for this dawned on me there and then. I felt that when the
protons bounced apart in more or less the same direction—”upwards”,
say—gluons would be more likely to coagulate on the other
side—”downwards” in this case—where they would free from
interference by the departing protons. However, QCD theory suggests that if the
protons bounced apart more violently, so that one went upwards and the other
recoiled downwards, contamination by unwanted quarks and antiquarks would be
more likely, and they would swamp the glueballs (see
Diagram). I asked
Kirk what happened if you compared the two situations. He told me that his team
had never looked at that, but he could let me know the answer within a week
I didn’t have to wait that long. The next morning I got an excited phone call
from Kirk in which he told me that the signals changed dramatically going from
one configuration to the other. By the following week he had shown that when the
outgoing protons move apart nearly in opposite directions, the particles that
appear are mainly well-known ones made from quarks and antiquarks. But in
glancing collisions, where the protons emerge more or less in parallel, the
quark companions disappear to leave other particles including the glueball
culprits. In particular, the scalar particle that had appeared in the Crystal
Barrel in 1995 stands out clearly in Kirk’s data when he selects the magic
glueball-friendly configuration.FIG-mg20694602.GIF
This was terrific news. But it was also ironic, because in the days between
suggesting to Kirk that he should try out my idea and his discovery that it
worked, I had found that it was actually flawed. However, given that all too
often the best ideas fail to produce the desired result, it seems only fair that
just for once we got the right answers for the wrong reasons.
Anyway, through a lucky break we have stumbled on a way to filter out
glueballs. We do not yet understand why it works so well, but more data should
eventually answer that question. In the meantime we hope to use this new method
to establish beyond any doubt the existence of the lightest glueball, and
possibly to identify a whole family of heavier glueballs. These would reveal
lots of new information about the strong force, which could be used to help
fine-tune lattice calculations.
For the moment, however, the proof of the pudding is in the eating. Having
spent 20 years chasing glueballs from a theorist’s perspective, I have been
seduced into taking part in experiments this year. The prospect of mining this
new seam is just too exciting to miss.
- Further reading: Also by Frank Close: “Glueballs and Hybrids: new
states of matter”, Contemporary Physics, vol 38, p 1, January 1997; “A glueball
filter in central hadron production”, Physics Letters B, in press; The Cosmic
Onion, Heinemann, 1983, ISBN 0 88318 4915.