The KEK B-Factory Experiment

by SHIN-ICHI KUROKAWA & STEPHEN L. OLSEN

Japan’s new asymmetric B factory recorded its first events on June 1. Designed to study the B-meson system, this may give us important clues to the differences played by matter and antimatter in the evolution of the Universe.

N THE AFTERNOON OF JUNE 1, 1999, collision events were recorded for the first time by the BELLE de- tector at the KEKB collider, the new - O“B factory” at the High Energy Accelerator Research Organization in , . This facility is specifically designed to study the matter-antimatter asymmetries in decays of B mesons predicted by the theory proposed by Makoto Kobayashi and Toshihide Maskawa in 1972. In many ways, the KEKB collider is similar to the PEP-II collider described by John Seeman on page 29. The KEKB collider consists of two rings of magnets; one ring stores 8 GeV and the other 3.5 GeV . The beams are brought into collision inside the de- tector, where they produce pairs of B mesons that move along the direction of the electron beam. This configuration is designed to mea- sure charge-parity (CP) violations in decays as described earlier in this issue by Robert Cahn and in the Summer 1996 Beam Line article by Michael Riordan and Natalie Roe. Like PEP-II, the two magnet rings of KEKB occupy a tunnel that originally housed a high- er energy electron-positron collider, in this case TRISTAN, and the high energy electron ring uses many recy- cled TRISTAN components. The center-of-mass energy of 10.58 GeV coincides with the mass of the ϒ(4S) resonance, which decays into B – and B meson pairs and nothing else. The different energies of the

BEAM LINE 23 two beams cause the produced B particles with energies and trajec- mesons to travel about two tenths of tories that deviate from the ideal on- a millimeter before they decay, a dis- energy central-orbit particle. A lat- tance that is easily measured with tice of quadrupole focusing magnets modern silicon-strip acts as a series of lenses to deflect detectors. particles diverging from the beam The BELLE detec- back toward the ideal orbit, causing tor is also similar in the particles to snake back and forth many ways to PEP-II’s across the beam axis executing B AB AR detector. In “betatron oscillations.” both cases the cores Particles circulating in the mag- of the instrument are net rings radiate a few million elec- cylindrical tracking tron volts of energy in detectors; a large ar- during each turn. This lost ray of cesium-iodide energy is replaced in high voltage rf crystals situated in- accelerating cavities. The particles side a superconduct- in the storage rings cluster in bunch- ing solenoidal mag- es that are properly phased to “surf” net that provides a on the rf accelerating voltage. The Kyoto University 1972 or 1973. Standing left to right are 1.5 tesla magnetic distance between bunches depends Makoto Kobayashi, Tsuneo Uematsu, Hideo Nakajima, field. There are, how- on the frequency of this voltage. In Koichi Yamawaki, and Masako Bando. Seated left to right are ever, some impor- both KEKB and PEP-II the rf frequency Toshihide Maskawa and Keiichi Ito. (Courtesy Makoto 500 Kobayashi) tant differences in is about million hertz, and the the details between minimum bunch-to-bunch spacing the KEKB/ BELLE arrangement and is 0.6 m. Both machines try to store that of PEP-II/ BABAR. In this article as many bunches of particles as pos- we highlight those that we consid- sible in order to achieve high lumi- er to be the most significant. nosity. The magnet systems are designed KEKB so that particles with energy above the central value take a little extra The most fundamental differences time to travel around the ring and between the KEKB and PEP-II storage thus tend to lag behind their on-en- rings are the schemes used to bring ergy neighbors. Since the particle the beams into collision and the tech- bunches pass through the rf cavity niques used to provide the radio- while the voltage is decreasing, these frequency (rf) accelerating voltages. late-arriving, higher-energy particles To appreciate these differences, some receive a less than average energy awareness of the behavior of beams boost from the cavity. Likewise, par- in storage rings is necessary. ticles with lower than average energy Beam oscillations. In general, par- tend to lead their neighbors and ticles in a beam do not arrive at the rf cavities early, where all have exactly the same energy nor they get an above average energy move in exactly the same direction. boost. As a result, particles within The magnet and accelerating systems each bunch alternate between lag- are designed to accommodate beam ging behind their neighbors and

24 SUMMER/FALL 1999 The challenge to machine designers is to bring bunches from the two rings into collision inside the having higher than average energy to detector and then to particles experience non-ideal fo- leading their neighbors with lower cusing, producing effects that are than average energies, a phenomenon separate them quickly similar to chromatic aberrations in called “synchrotron oscillations.” ordinary optical systems and for Crossing scheme. At the inter- enough to get them which careful compensation must be section point, where the electron and made. These corrections are provided positron bunches pass through each back into their by sextupole magnets distributed other, the beams are focused to a very around the ring. small size. Since the electron and respective magnet Design studies for KEKB demon- positron beams have different ener- systems before they strated that standard magnet lattice gies, the two beams require magnets arrangements, such as the one used with different focusing strengths. collide with the next in PEP-II, could not provide the nec- The challenge to machine design- essary chromatic corrections while ers is to bring bunches from the two bunch. also meeting the low synchrotron rings into collision inside the exper- frequency requirement imposed by imental detector and then to sepa- the finite-angle crossing arrange- rate them quickly enough to get ment. Instead, KEKB uses a novel new them back into their respective backgrounds. The main disadvantage lattice arrangement developed by magnet systems before they collide is the introduction of a possible cou- KEK physicists Haruyo Koiso and with the next bunch—and to do so pling of each beam’s transverse be- Katsunobu Oide that satisfies all of without producing large disturbances tatron modes of oscillations with the the requirements. In the Koiso-Oide to the experimental detector. longitudinal synchrotron modes re- scheme, the chromatic corrections In PEP-II, the beams are made to sulting in the dreaded “synchro- are provided by pairs of sextupole collide head-on; they are brought to- betatron” oscillations (see box on the magnets located far apart, at posi- gether and then separated by a sys- next page). These were blamed for tions where the optical properties of tem of “separation dipoles” made beam instabilities that limited the the beams are nearly mirror images from permanent magnets that are performance of the original two-ring of each other. The mirror imaging common to both beams and located configuration of the DORIS storage provides a convenient cancellation inside the detector. This scheme lim- ring at DESY. of nonlinear effects, and the scheme its the minimum separation between Using elaborate computer simu- has the additional practical advan- bunches to 1.2 meters, twice the min- lations, Kohji Hirata and Nobu Toge tage of making it possible for the high imum bunch spacing. of KEK carefully reexamined the ef- energy electron ring to use recycled In KEKB, the beams are made to fects of beam-beam interactions in bending magnets from the recently cross at a 1.3 degree angle. Since the the case of finite-angle crossings. dismantled TRISTAN storage ring. two beams are moving in different They concluded that the deleterious Radio-frequency accelerating cav- directions, they fly apart naturally, effects of a finite crossing angle ities. In order to achieve the high and no separation dipoles are nec- would be manageable, provided that luminosity requirements of the B essary. In this case beam bunches can the frequency of synchrotron oscil- factory, the stored currents in each be as close together as the minimum lations is a small fraction (about beam must be as large as a few am- distance of 0.6 meters. 1 percent) of the beam circulation peres. To maintain the proper match The KEKB finite crossing angle frequency. between the klystron (which pro- scheme has the advantage of sim- Chromatic corrections. The fo- vides the rf power) and the cavity, the plicity. With no separation dipoles, cusing strengths of the quadrupole resonant frequency of the cavity there is no bending of off-energy magnets are set for particles that must be shifted by an amount pro- beam particles into the detector, have exactly the nominal beam en- portional to the current passing resulting in more manageable ergy. As a result, off-energy beam through the cavity. If the shift in

BEAM LINE 25 Crossing Angles, Synchro-Betatron Oscillations and Crab Cavities

HE DISADVANTAGE of a finite angle crossing of the beam bunch- es is the introduction of couplings between the transverse betatron Toscillation modes of beam particles with their longitudinal synchro- tron oscillation modes. This establishes an additional set of possible beam- destroying resonances that must be avoided. frequency exceeds the circulation fre- quency of the beam in the ring (100 kilohertz), beam-destroying insta- bilities can occur. Only large rings with large circulating currents, such Electron Positron as KEKB, suffer from this type of in- stability. In order to reduce the effects of Positron Electron the beam currents on the cavity res- onant frequency, KEKB uses cavity systems where the energy stored in the resonant mode is very large. The large stored energy provides suffi- cient inertia to reduce the effects of the beam current. Superconducting The coupling mechanism illustrated above shows electron and cavities naturally have lots of stored positron beam bunches when they start and finish passing through each energy and these are used in the KEKB other at an exaggerated crossing angle. The positively charged high energy ring. This is not the case positrons in the front end of the e+ bunch pull the negatively charged − for normal-conducting cavities, electrons in the back end of the e bunch sideways in one direction, where heating in the cavity walls while the positrons in the back end tug the electrons in the front end in limits the amount of energy that can the opposite direction. Changes in the front-to-back particle positions, be stored in the cavity itself. How- caused by synchrotron oscillations, result in different excitations of the ever, it was not possible to supply all transverse betatron oscillations, giving rise to coupled “synchro-betatron” of KEKB’s radio frequency require- oscillations. ments with superconducting rf cav- This coupling can be avoided using the “crab-cavity” trick invented by ities. Normal-conducting rf cavities Robert Palmer of Brookhaven National Laboratory for use in high energy are also needed. Tsumoru Shintake linear electron-positron . In this scheme (shown below) rf of KEK proposed a two-cell system where the accelerating cavity is RF Deflector closely connected to a second, very Electrons large energy-storage cavity, thereby Positrons increasing the total stored energy by about a factor of ten. This idea was Crossing Angle improved to include a third coupling Head-on cavity. Collisions Kick In KEKB, all of the rf power for the Kick low energy ring and about half of it for the high energy ring is provided cavities on either side of the interaction point provide transverse electric by normal-conducting systems based fields that rotate the beam bunches by pushing the front of the beam on this three-cell scheme. bunch in one direction and the rear in the opposite direction. In this way, even though the beam bunches pass by each other at an angle, THE BELLE DETECTOR the bunches go through each other head-on, and the transverse- longitudinal coupling mechanism shown above is eliminated. The technical challenges presented by the B-factory physics program to the experimental detector are not

26 SUMMER/FALL 1999 as severe as those faced by the ma- chine builders. This is because there are existing detectors, primarily the CLEO detector at the Cornell Elec- tron Storage Ring, that provide use- ful guidance. The relative maturity of detector technologies is reflected in the fact that differences between BABAR and BELLE are not nearly as pronounced as the differences in the storage rings. Both detectors sur- round the beam intersection region with a 1.5 tesla magnetic field, pro- vided by large superconducting so- lenoids that encompass most of the about 99 percent of the speed of light; Eiichi Nakano from Osaka City University detector elements. Immediately out- the more massive K mesons are only checks the cabling of the endcap side of the electron-positron collision a bit slower—their velocities are typ- chambers in the completed BELLE detector while it sits in its “rolled-out” point are high resolution track de- ically about 90 percent of the speed position in the Tsukuba experimental hall tectors made of silicon that pin down of light. Particle identification sys- at KEKB. On May 1, the detector was the decay position of the B mesons. tems have to exploit these small dif- moved into its final location at the Surrounding the silicon detectors are ferences in particle velocity. electron-positron interaction point, tracking chambers that measure the One way to do this is to make a which is inside the temporary tunnel trajectories of charged particles from direct measurement of it. This is of shielding blocks seen at the right in the B meson decays as they travel done in BELLE with an array of large the photograph. (Courtesy KEK) though the magnetic field. The cur- plastic scintillation counters ar- vature of the trajectories is used to ranged in a barrel that surrounds the determine the particle momenta. tracking system and measures the Outside of the tracking chambers but time-of-flight of particles as they still inside the coil are large arrays of cross the detector volume. Using cesium-iodide crystals for detecting state-of-the-art techniques, the BELLE gamma rays. The major differences time-of-flight system measures the between CLEO-II, BABAR, and BELLE particle transit times with a preci- are the techniques used to distin- sion of about 100 trillionths of a sec- guish between different species of ond. This is good enough to distin- charged particles. BELLE uses arrays guish π and K mesons up to particle of plastic scintillation time-of-flight energies of about 1 GeV. counters and aerogel Cerenkov At energies higher than 1 GeV, π counters. and K mesons can not be reliably dis- Particle Identification. For the CP tinguished by the time-of-flight sys- violation measurements, it is essen- tem; for these particles Cerenkov tial to distinguish charged K mesons techniques are used. These tech- from other particles, especially from niques rely on the fact that when a the more copiously produced π charged particle passes through a mesons. The π and K mesons from transparent material with a speed B decays are quite relativistic: typi- that exceeds the speed of light in that cal π mesons have velocities that are material, it emits measurable

BEAM LINE 27 amounts of light in the form of of the BELLE detector was “the mid- look forward to a rich program of “Cerenkov” radiation. In BELLE, a dle of Japanese Fiscal Year 1998,” measurements that will elucidate the cylindrical mosaic of nearly a thou- which started in April 1998. Despite nature of CP violations and confirm sand blocks of transparent silica aero- major changes from KEKB’s original or reject the Kobayashi-Maskawa gel with indices of refraction that plans for the injection scheme, the theory. range from n = 1.01 to 1.03 occupies interaction region configuration, the the radial space between the track- magnet lattice and the rf cavities, ing region and the barrel of time-of- both rings were completed in flight counters. Each aerogel block is November 1998 and beam was first viewed by very sensitive phototubes. stored in the high energy ring on Charged pions with energy above December 12. Also, in spite of a num- about 1 GeV have velocities that are ber of design changes, the assembly above the Cerenkov threshold and of the entire BELLE detector was fin- produce light as they traverse the ished on December 18. aerogel material. In contrast, charged In a four-month beam commis- kaons, which are more massive, do sioning run that ended in April, high not have velocities above the current electron and positron beams Cerenkov threshold until they have (about 0.5 amperes each) were energy in excess of 3.5 GeV, which is achieved with tolerable levels of near the highest energy possible for background radiation at the ultimate particles from B meson decays. Kaons location of the BELLE detector. These below this energy do not produce any beam currents are sufficient for op- light. Thus, the response of the aero- eration at a luminosity of about one- gel counters can be used to distin- fifth of the ultimate design goal. The guish high momentum pions from magnet lattice, chromatic correc- kaons in BELLE. tions , and the rf systems all behaved The aerogel system can be nicely as expected. During beam-beam col- tailored to the variation in particle’s lision studies, no evidence was seen momentum range with polar angle for deleterious syncho-betatron that results from the asymmetric na- oscillations. ture of the electron-positron col- During the KEKB commissioning lisions in KEKB, but it has the dis- run, the BELLE detector remained in advantage of consuming a sizable the “rolled out” position, where it volume inside the detector and inter- accumulated large samples of posing a considerable amount of ma- cosmic-ray events both with and terial in front of the cesium iodide without the magnetic field excited. calorimeter in a complicated non- These events were used to align and uniform pattern calibrate the detector components. In May, the detector was moved KEKB AND BELLE—CURRENT into location at the KEKB interaction STATUS point and operations resumed. The first collision events were recorded In 1994, when BELLE and the KEK B- by BELLE about a week later. This factory project were started, the pro- was an important proof of principle: posed date for the start of KEKB beam all systems in BELLE and KEKB op- commissioning and the completion erated very nearly as expected. We

28 SUMMER/FALL 1999