Experiment: Discovery of the Top Quark
EXPERIMENT: DISCOVERY OF THE TOP QUARK
The world around us is made of two types of particles: matter particles and force particles. The former include leptons, such as the electron and the electron-neutrino, and quarks. The two lightest quarks are called up and down, and they make up the proton and the neutron. The matter particles interact by exchanging force particles. These include the photon, mediator of the electromagnetic interaction, the gluon, mediator of the strong nuclear force that holds the nucleus together, and the W and Z bosons, which mediate the weak nuclear force, responsible for nuclear beta decay. Starting in the 1960s, physicists developed the Standard Model, which describes these particles and their interactions.
The electron, electron-neutrino, up quark, and down quark form a fermion generation. This fermion generation contains all the constituents of ordinary matter. However, this is not the only generation. The muon, muon-neutrino, charm quark, and strange quark have the same properties as the particles of the first generation except they are much heavier, and most of them are not stable. They decay to the lighter particles of the first generation and therefore are not usually encountered.
In 1975, a third—yet heavier—lepton, called the tau lepton (τ lepton), was discovered. The Standard Model requires complete generations, consisting of two leptons and two quarks. This implied the existence of a third neutrino and a third generation of quarks. Experiments set out to search for a third-generation quark culminating in the discovery of the bottom or b quark in 1977. At that time, this was the heaviest known fundamental particle with a mass of about 5 GeV/c2, more than five proton masses. Now there was only one quark missing to complete the third quark generation. It was called the top quark.
The Race for the Top Quark
The race for the top quark was on. It had to be heavier than the b quark. More energetic accelerators were required to produce it. In the early 1980s, experiments in Germany and Japan, using colliding electron and positron beams, ruled out the existence of a top quark with masses below 30 GeV/c2. In 1989 to 1990 experiments at the Stanford Linear Accelerator Center (SLAC) at Stanford University and at the European Laboratory for Particle Physics (CERN) in Geneva, Switzerland, increased the limit to 46 GeV/c2. To go to even higher energies experimenters switched to colliding protons and antiprotons. At CERN, proton-antiproton collisions at energies of 630 GeV again failed to see the top quark, setting the limit at 69 GeV/c2.
In 1985, the Tevatron at Fermilab near Chicago collided proton and antiproton beams for the first time at an energy of 1.6 TeV. The newly commissioned CDF detector pushed the lower limit on the top quark mass to 77 GeV/c2. In 1992, the energy of the Tevatron was increased to 1.8 TeV, and a second detector, D0, commissioned.
Fermilab is located 30 miles west of Chicago in Batavia, Illinois. The Tevatron is an accelerator ring with a radius of 1 km. Its entire length is enclosed in about 1,000 superconducting magnets that keep the protons on their path. It is the last in a series of accelerators that accelerate protons and antiprotons to 900 GeV each. These were the particle beams with the highest energy in the world. Groups of about 1011 protons and 1010 antiprotons circle in opposite directions around the Tevatron ring and collide in the center of the two detectors every 3.5 microseconds.
Top Production and Decay
How did CDF and D0 propose to find the top quark? When the protons and antiprotons collide, their kinetic energy can be converted into mass. If enough energy is available, top quarks can be created. Most of the time, top quarks and their antiparticles are created in pairs in proton-antiproton collisions. However, the creation of top quarks is a very rare process. Only once in every 1010 proton-antiproton interaction are top quarks produced. This seems much like the proverbial search for a needle in a haystack.
Moreover, the top quark does not live long enough to be observed directly. On average it exists only for 10-24 second before decaying into a b quark
and a W boson. An antitop quark decays into an anti-b quark and an anti-W boson. The W bosons are also very short-lived and decay into electron + neutrino, muon + neutrino, or tau + neutrino with a probability of one ninth each. The remaining two thirds of their decays are into quark-antiquark pairs.
Quarks interact so strongly with each other that they cannot exist in isolation but only as constituents of other particles, such as protons. Particles that consist of quarks are called hadrons. The quarks created in the decay of the top quark immediately turn into collimated showers of hadrons, called jets.
Electrons and muons are electrically charged and live long enough to be observed directly in the detector. τ leptons decay in a variety of ways and are very difficult to identify.
Neutrinos are electrically neutral and have a fair chance of making it all the way through the earth without interacting. Therefore the chance of catching them in a detector is nil. However, since the initial protons and antiprotons carry only momentum along the beam direction, the total transverse momentum (with respect to the beam direction) of all particles produced in a proton-antiproton collision must be zero. If a neutrino carries away a lot of momentum, that momentum will be missing from the visible particles, and their transverse momenta will not add up to zero. Thus large missing transverse momentum implies that one or more neutrinos were produced that carried away the missing momentum.
In collisions in which a top quark and an anti-top quark are created and decay, a b quark and an anti-b quark are always produced. In addition, other particles appear, depending on the decay channel of the W bosons:
- dilepton channel: two electrons or muons, and two neutrinos (5 percent of all top-antitop decays)
- lepton + jets channel: one electron or muon, one neutrino, and two additional quarks (30 percent of all top-antitop decays)
- all-jets channel: four quarks (44 percent of all top-antitop decays)
- decays that include tau leptons (21 percent of all top-antitop decays).
When protons and antiprotons collide, they typically break up into quarks and gluons, which turn into jets. At the Tevatron, this happens several thousand times more often than top quark production. Of the top-antitop signatures above, the ones that contain electrons or muons are easiest to pick out of this background. In the dilepton channels one expects two electrons or muons, missing transverse momentum, and two jets. In the lepton + jets channel one expects one electron or muon, missing transverse momentum, and four jets. Although the other channels make up about two thirds of all top-antitop decays, the experiments did not even try to look for them in the beginning because they are so hard to separate from the background.
Although the D0 and CDF experiments look quite different, they have the same basic detector components.
As the particles emerge from the collision, the experimenters first measure their direction. Thus innermost in both experiments are tracking detectors, which are built very light and with as little material as possible, since they should not disturb the particles very much. They are mostly gas chambers with thin wires strung across. Particles that carry electrical charge ionize the gas, which generates an electrical pulse on the wire closest to the particle path. Many such pulses can be used to reconstruct the trajectory of the particle to about 100-micron precision. CDF also had a detector made of thin slices of silicon. This allows even more precise measurements of the particle path near the collision point. The CDF tracking chambers are placed in a magnetic field such that the particle trajectories curve. The curvature determines the momentum of the particle.
Next the experimenters measure the energy of the particles by stopping them and measuring the energy transferred to the material of the detector, which equals the kinetic energy of the particles. The detector that does this is called a calorimeter, and, in contrast to the tracking detectors, it is built of heavy materials. D0 has a calorimeter made of uranium and liquid argon. CDF's calorimeter consists of lead and scintillator wedges in the center and lead and gas chambers at the ends of the detector. The calorimeter also allows the distinction of electrons and hadrons by their different energy loss patterns. Hadrons are much more penetrating than electrons.
However, not all particles can be stopped. Electrons and hadrons interact strongly with matter and are easily stopped. To contain muons requires tens of meters of steel, not very practical. The experimenters turn this to their advantage. Since muons are electrically charged, they can be detected in tracking chambers, and both experiments have large tracking chambers outside their calorimeters. With high probability, any charged particles that penetrate the calorimeter and are detected in these chambers are muons. Therefore these chambers are called muon detectors. D0 also has a magnet that deflects the muons so that their momenta can be measured.
How do we know when we have discovered the top quark? CDF and D0 were looking for dilepton and lepton + jets events. Seeing such events, however, is not equivalent to discovering the top quark. Other processes can also produce these signatures. For example, in proton-antiproton collisions a W boson and four jets can be produced. Then the W boson decays to an electron and a neutrino, which is exactly the signature of a lepton + jets event. This background is so overwhelming that seeing the actual top signal requires more sophisticated analysis techniques. The top-antitop decays differ in two ways from this W + jets background:
- the top-antitop decays have two jets from b or anti-b quarks
- the jets in top-antitop decays have much higher momentum transverse to the beam.
Hadrons that contain b quarks often live much longer (10-12 second) before they decay than other hadrons (10-20 second). In this time, they can move several millimeters away from the point at which the proton and antiproton collided. If the detector can measure the particles from their decay very precisely, this displaced secondary decay point can be reconstructed. If such a displaced decay point is found in a jet, it is likely that the jet originated from a b quark. This is called vertex tagging. CDF was able to tag jets that originated from b quarks in this way using the
silicon tracking detector. Both experiments could tag b quark jets using muons. Approximately one in ten hadrons that contains b quarks decays into a muon and other particles. Other hadrons are not likely to decay in this way. The muon can be detected, and a jet in which a muon is found likely originates from a b quark. This is called lepton tagging. It was used by D0 and CDF.
When jets are produced in proton-antiproton collisions, their direction is often close to that of the original proton or antiproton. Such a jet may have high momentum, but the component of its momentum transverse to the beam direction is not large. When a heavy object such as the top quark is produced and decays to jets, however, the jets are emitted in all directions with about the same probability. Thus, in this case, the transverse components of the jet momenta could be quite large. D0 found that the sum of the transverse momentum components of all jets, called HT, and the angular distribution of the jets provide good discrimination between the top quark signal and the background.
Using all these tricks, the experimenters were able to reduce the background in their data samples so that signal and background were comparable. Since the background originates from known processes, the number of events expected from these processes can be calculated. If the top quark is produced, the observed number of events should be larger than that calculated under the background-only hypothesis. However, even if there is no top quark, the experimenters could observe more events than calculated because of statistical fluctuations. For example, if 3.8 events are expected, most probably 3 or 4 will be observed, but there is a 33 percent probability to observe more than 4, and a 2 percent probability to observe more than 10. The standard for a discovery is much higher. For a discovery, the probability of the observed number assuming there is only background must be around one in a million or less!
D0 observed three dilepton events and fourteen lepton + jets events for a total of seventeen events when a background of 3.8 ± 0.6 was expected. The probability to see seventeen or more, assuming the background-only hypothesis, was 2 × 10-6. CDF observed six dilepton events with 1.3 ± 0.3 expected. In the lepton+jets channel, CDF did not count events but b-tagged jets. They saw twenty-seven vertex tags (6.7 ± 2.1 expected) and twenty-three muon tags (15.4 ± 2.0 expected). The combined probability to see this many events and b tags without a top signal was 10-6.
The papers published by the D0 and CDF collaborations announcing the discovery of the top quark have 414 authors for the D0 paper and 539 for the CDF paper. This large number indicates the complexity of the effort behind large high-energy physics experiments.
The time scale of such a project from conception to completion is typically a decade or more. For example, the D0 experiment received preliminary approval in 1983. Construction was completed in1990. The experiment took its first data in 1991. In 1995 the discovery of the top quark was published. To date, the D0 collaboration has published over one hundred research papers.
The D0 detector is about three stories high and weighs about 5,000 tons. It provides about 120,000 electronic signals for every proton-antiproton collision. There are 5,500 km of cables to distribute these signals, to supply power, and to control operation of the detector. All detector components and most of the electronics were developed and built by physicists. This required thousands of person-years of effort. This task was therefore distributed over research groups from many collaborating institutions. For example, parts of the tracking detectors were built at Saclay in France, Lawrence Berkeley Lab in California, State University of New York in Stony Brook, and Northwestern University in Evanston, Illinois. The components were brought to Fermilab and assembled. Every channel had to be tested and problems fixed before data could be taken. This was often the task of Ph.D. candidates and postdoctoral students, who resided at Fermilab. After everything was working and the accelerator started to provide colliding beams, data taking began. The accelerator ran seven days a week, twenty-four hours a day for almost two years, and a crew of six physicists manned the control room of the detector at all times. All collaborators contributed to this, taking eight-hour shifts to acquire the data. When they were not taking shifts, they spent their time analyzing the data to determine hundreds of thousands of calibration constants that were stored in databases, or they developed the millions of lines of computer code necessary to operate the detector and acquire, store, and analyze the data. This comprised a huge body of work that was required before the search for the top quark in the data could even begin. The search for the top quark was of course not the only goal of the experiment. Many other important results were also based on this work.
Over four hundred physicists and uncounted engineers and technicians from forty-two institutions in the United States, Brazil, Colombia, France, India, Korea, Mexico, and Russia contributed to the construction and operation of the experiment. Professor Paul Grannis from the State University of New York at Stony Brook led this operation, from conception to completion.
Once the detector was calibrated and the computer code debugged that found electrons, muons, and jets in the data from the detector, smaller groups of physicists began to analyze the data for signals of different processes of interest. The group that performed the search for the top quark was one of five physics groups. The top physics group consisted of about eighty members, of which twenty were Ph.D. students, and was led by Dr. Boaz Klima from Fermilab and Professor Nick Hadley from the University of Maryland. Within this group, smaller working groups were formed, each concentrating on one channel. These groups reported back to the top physics group in weekly meetings. For example, Meenakshi Narain of Fermilab worked on the dielectron channel with students V. Balamurali and Bob Kehoe from the University of Notre Dame. Kehoe was one of the students whose theses composed the top quark discovery. He worked tirelessly many nights trying to find a candidate event in "his" channel. To his great disappointment the dielectron channel ended up being the only channel in which no candidates were observed.
At the time of the annual winter conference in Aspen in January 1995, no excess close to meeting the standards for a discovery had been seen. The mass limit had increased to 131 GeV/c2, and it had become clear that the top quark had to be much heavier than originally expected. If it were that heavy, only very few top quarks would be produced, and they would be swamped by background. A concentrated effort was started to develop a strategy to isolate the signal from a very heavy top quark. Studying a computer simulation of the top quark signal, the HT cut described above was developed, which would have cut out the signal of a lighter top quark but accepted much of the signal from a heavy top quark. The new analysis strategy soon showed hints of a signal. Soon also rumors circulated that CDF was planning to come out with a discovery publication.
The experimenter had agreed with Dr. John Peoples, the director of Fermilab, to give each other notice before they published a discovery. The other experimental group would then have one week to prepare its own paper. This mechanism was designed to prevent one of the groups from jumping the gun with a premature publication that the other experiment would contradict. In February 1995, the CDF collaboration announced that it saw a significant excess and was preparing a publication.
Once the top physics group had concluded that they saw a signal, the next step was to convince the rest of the collaboration. The usual process for publishing results was to set up an editorial board. This was comprised of several physicists from the collaboration who were not involved in the analysis as well as one or two proponents of the analysis. This board would meet and examine the analysis in detail to check for any weak points. It also would review the proposed paper. Once the board was satisfied, the collaboration had a few weeks to comment, and if no objections were voiced, the paper was submitted for publication. In case of the top discovery, there was very tight time pressure, because nobody wanted to let CDF publish first. The editorial board worked tightly with the analyzers, and the collaboration review was accomplished in days. On Friday, February 24, 1995, at 11 a.m. Central Standard Time, Herb Greenlee hit the computer key submitting the paper to Physical Review Letters at exactly the same time as his counterpart, Mel Shochet, at CDF. The announcement for the usual weekly top meeting that week read: "Since the top quark has been discovered, there will not be a top group meeting this week."
CDF Collaboration. "The CDF Detector: An Overview. " Nuclear Instruments and Methods A271 , 387–403 (1988).
CDF Collaboration. "Observation of Top Quark Production in Anti-p p Collisions." Physical Review Letters74 , 2626–2631 (1995).
D0 Collaboration. "The D0 Detector." Nuclear Instruments and Methods in Physics Research A338 , 185–253 (1994).
D0 Collaboration. "Observation of the Top Quark." Physical Review Letters74 , 2632–2637 (1995).
Liss, T. M., and Tipton, P. L. "The Discovery of the Top Quark." Scientific American277 , 54–59 (1997).