- Email: [email protected]
Particle identification for the BaBar experiment Detection of Internally Reflected Cherenkov light (DIRC)
Nuclear Instruments
and Methods in Physics Research A 379 (1996) 444-447
NUCLEAR INSTRUMENTS & METHODS IN PHYSICS RESEARCH Section A
Particle identification for the BaBar experiment Detection of Internally Reflected Cherenkov light (DIRC) R.J. Wilson *11 Departttwnt of Physics, Colorado State University, Fort Collins, CO 80523, USA
]
Representing the BaBar Collaboration [ 1
Abstract
We describe a new type of detector for particle identification which will be used in the BaBar experiment at the SLAC B factory (PEP-II). This detector, used in the barrel region of BaBar, is called the DIRC which is an acronym for Detection of Internally Reflected Cherenkov light. Similar to other Cherenkov ring imaging devices, the DIRC provides high resolution determination of the Cherenkov angle for pions and kaons but is also particularly well-matched to the momentum range and angular distribution of CP-violating final states in asymmetric B factories.
1. Introduction
A new generation of accelerators, commonly referred to as B factories, has been designed to study charge-parity (CP) violating behavior in the neutral B meson system. The high event rates and relatively high momenta of the decay products provide challenges for almost every aspect of the experiment. In this paper we will concentrate on the requirements and solution for particle identification in the BaBar experiment at the PEP-II accelerator being constructed at the Stanford Linear Accelerator Center. However, much of what we will discuss is relevant to other similar facilities. In the following sections we will present first the basic particle-ID requirements for physics at PEP-II. This will be followed by a description of the DIRC principle and the specific implementation for BaBar. Finally, we will summarize the current status and production schedule for the BaBar DIRC.
2. Particle ID requirements
at PEP-II
2.1. PEP-II The PEP-II accelerator is an asymmetric electronpositron collider with beam energies of 9 GeV (e-) on 3.1 GeV (e+) which gives rise to copious production of the Y (4s) resonance with a Lorentz boost of py = 0.56 in the laboratory frame. This boost means that the fastest decay
particle tracks occur in the direction of the high energy (electron) beam producing a significant momentum-angle correlation. Future upgrades may include an increase in the energy to run at the Y(5S) with a corresponding increase the maximum momenta. The initial luminosity of PEP-II will be 3 x 1O33cm-‘s-’ increasing to 10% cm-2s-1 after a few years operation. The beam crossing period will be 4.2 ns. This high luminosity leads to a “physics” rate of more than 100 Hz, but the BI? rate of particularly interest is a few Hz. 2.2. Primary physics goal The primary physics goal of the BaBar expkriment is to observe CP violation in the BB system. A critical objective for particle-ID in CP physics is to identify kaons for tagging the B meson flavor (b quarks decay primarily to K- and the 6 anti-quarks decay to K+) . Most of the secondary particles in B decays have a momentum below 1.0 GeVlc but efficient tagging of B mesons requires good kaon identification up to about 2.0 GeVlc. Due to the geometrical acceptance of BaBar (partially imposed by PEP-II requirements) and kaondecays in flight, the best possible efficiency for kaon tagging is approximately 22%. Measurement of dEldx by the drift chamber should give 3a separation up to about 0.7 GeV/c, but this would leave more than 45% of kaons untagged. It is important to recognize that the efictive efficiency in this type of analysis is given by
* E-mail [email protected] ’ Supported
by U.S. Department
016%9002/96/$15.?0 Copyright PlrSO168-9002(96)00600-6
of Energy Grant DE-FG03-93ER40788.
E& = E@( 1 - 2w)*,
@ 1996 Elsevier Science B.V. All rights reserved
R.J. Wson/Nucl.
Instr. and Meth. in Phys. Rex A 379 (1996) 444-447
where w is the fraction of incorrect tags. This means that the purity of the sample is also very important. With dE/dx alone the tagging efficiency would be only 9%. Even more challenging than tagging is the requirement for exclusive B decay reconstruction for rare modes such as B0 --+ ~+rr- (branching ratio ~1.2 x 10v5), used to determine the CKh4 matrix parameter sin( 2cu). The asymmetric boost of the two-body final state means that r-K separation up to 4.0 GeVlc is required. In summary, physics requirements imply that we will need a PID system with high efficiency and purity over a large solid angle and large momentum range. Accommodating the additional rich variety of rare B decay, charm and tau physics also will have an impact on the overall detector optimization.
3. The DIRC principle In Fig. 1 we show the basic principle of the DIRC. As the name implies, the device measures Cherenkov radiation produced by a charged particle traversing a solid radiator which then acts as a light guide for the internally reflected light. The portion of the Cherenkov cone trapped in the radiator depends on the entry angle of the incident particle and the ratio of the refractive indices at the interface (ni and ptg in the diagram). The internally reflected light will propagate down the bar until it reaches either end. A mirror placed at one end will reflect that portion of the light back to the opposite end. If the sides of the bar are parallel the photon production angle will be preserved during the multiple reflections. As indicated in the diagram, a position sensitive photodetector surface placed at some distance from the bar end will determine the photon exit angle (after correction for refraction at the bar/stand-off region interface). The “zerodegree” mirror in the stand-off region reduces the size of the imaging array, at the cost of an overlap of a portion of the image. The Cherenkov angle may be determined from these conic section images combined with the particle incidence angle provided by a tracking chamber. The basic requirements of the DIRC can be summarized as: - n3 << nl for efficient internal reflection; - nr x n2 for good light transfer at the end of the bar; - very low absorption in the near UV for long bars (tens of meters) ; - high surface reflectivities; - bars with parallel faces, perpendicular sides and sharp
445
comers; - low absorption for near-UV in the stand-off medium. 3.1. DIRC advantages
Among the many potential advantages of the DIRC compared to other ring-imaging or threshold Cherenkov devices are: compact radiator (a stable solid yielding many photons/cm); relatively simple mechanical structure; no active components in the fiducial volume; no complex fluids system; conventional photodetector technology can be used (photomultiplier tubes) ; sensitivity well matched to an asymmetric collider (light production and capture efficiency improve with dip angle); acceptable mass in close proximity to the calorimeter. In a realistic device, of course, some of these advantages may be obscured by the usual constraints imposed by integration of the system into the rest of the apparatus. Nevertheless, the final implementation of a DIRC in a colliding beams environment will share many of the features of a traditional, and robust, time-of-flight system.
4. BaBar DIRC
In this section we describe a schematic implementation of a DIRC for the BaBar experiment. In Fig. 2 we show the mechanical assembly of the DIRC. The system consists of two major components: the Radiator structure and the Stand-off Box. 4.1. Radiator The radiator barrel consists of 144 synthetic quartz bars arranged in a 1Zsided polygon. Synthetic quartz was chosen over natural quartz due to the superior transmission of
/c-2 m-y Fig. I. DIRC principle.
Fig. 2. DIRC mechanical a.&mbly.
IV. PARTICLE IDEtNTIFICATION
446
R.J. Wilson/Nucl.
Instr. and Meth. in Phys. Res. A 379 (1996) 444-447
near-UV light (above 290 nm) after irradiation of the bar. The quartz has a refractive index of 1.474. Each bar has dimensions 1.75 cm x 3.5 cm x 480 cm with a mirror on one end and is constructed by gluing together four shorter bars of length 120 cm. Twelve long bars are mounted together in a thin-walled box which forms the sides of the polygon. A very thin air gap is maintained between each bar in the box to prevent optical cross-talk. Including the support structure and the bar-box material, the central cylinder has a radial extent of less than 10 cm and presents about 15% of a radiation length at normal incidence. This arrangement defines a fiducial volume in polar angle of 25.5” < 0 < 147’ (or 0.9 > cos 0 > -0.84) and covers approximately 95% of the full azimuth. 4.2. Stand-off Box/photodetector array The Stand-off Box is the mechanical structure which supports the photodetector array and contains high purity water which is the interface medium between the end of the radiator bars and the photodetectors. The vessel contains approximately 6 tonnes of lo-18 Ma water which will be recirculated and purified to maintain good transmission of near-UV light. The photodetector array consists of 11000 conventional 1.125 in. bialkali photomultiplier tubes (EMI 9125) submersed in water for optimal light coupling. The PMTs are arranged in a close-packed configuration on a toroidal surface subtending an angular range of O”-54’ in 12 identical azimuthal sectors corresponding to the symmetry of the radiator structure. The faces of the PMTs are 117 cm from the end of the radiator bars. Reflective hexagonal “light catchers” will be mounted on the circular PMT faces to recover the light which would be lost in the interstices. Since the experiment may run for many years, the PMTs will be operated at a relatively low gain of 106. This provides an average signal of -20 mV (into 50 a). The frontend electronics currently under test will allow a discriminator threshold of around -2 mV and provide 1 ns timing resolution. In Fig. 3 we show a simulation of the PMT hit distribution in the array for an event containing a B” -+ .~+‘rrdecay. This simulation includes background hits from both correlated and uncorrelated interactions in the detector.
Fig. 3. Photodetector array hits simulation. Images from B + rr+rrare shown as solid markers, images from two other charged particles are open markers.
direction is particularly important at an asymmetric collider due to the angle-momentum correlation; the higher particle momenta lead to a corresponding decrease in separation of Cherenkov angles for pions and kaons. We estimate that the 0, resolution per pion will be 2-3 mrad for high momentum tracks (2.0-4.0 GeVlc) for all dip angles, increasing to 4-5 mrad for lower momenta. These assumptions are supported by results from beam tests of a large prototype discussed by Besson at this conference [ 21. In Fig. 4 we show the r-K separation (units of sigma) as a function of cos B for several momenta. Care must be taken in the direct application of this figure to PEP-II physics due to the momentum-angle correlation and the kinematic cutoff in B decays. A detailed simulation of the process B” -+ rr+r- indicates a r-K separation of more than 4a over the entire kinematically allowed range of momenta and angles. If this level of performance is maintained in the final system, we project an efficiency for K-tagging of 19% compared to the maximum possible of 22% (and only 9% with just
5. Projected performance The transverse size of the radiator bar, the photomultiplier tube diameter and the stand-off distance combine to give an intrinsic resolution for single photoelectrons, S&, of =lO mrad. Detailed simulation of the device indicates that we can expect 30-40 detected photoelectrons for particles in a momentum range 0.8-4.0 GeV/c and dip angles in the range f30”, increasing to 60 photoelectrons in the forward direction (cos ~9< 0.85)‘. This enhancement in the forward
Fig. 4. Simulated a/K
separation.
R.J. Wilson/Nucl.
Instr. and Meth. in Phys. Res. A 379 (1996) 444-447
dE/dx). In addition to the primary task of P-K ID the DIRC will also provide muon/proton separation, complementary to BaBar Instrumented Flux Return, of better than 2u below 0.5 GeV/c. 6. Status and schedule The conceptual design for the BaBar DIRC was approved in January 1996 and detailed engineering is under way leading to the final design review in September of this year. The current schedule indicates completion of the quartz assemblies by the summer of 1997 and delivery of the Standoff Box to SLAC in May 1998. The primary goal is for the BaBar detector to be ready for commissioning with cosmic rays by December 1998 followed by colliding beam physics in June 1999.
441
Acknowledgements
I would like to thank my U.S. and French collaborators in the DIRC group, especially members of Group B at the Stanford Linear Accelerator Center for their continuing hospitality. References
[ 11 BaBar Technical Design Report, BaBar Collaboration, SLAC-R-950457 (March 1995). [21 P. Besson, presented at this Conference (6th Int. Conf. on Instrumentation for Experiments at e+eColliders, Novosibirsk, Russia, 1995).
IV. PARTICLE IDENTIFICATION
and Methods in Physics Research A 379 (1996) 444-447
NUCLEAR INSTRUMENTS & METHODS IN PHYSICS RESEARCH Section A
Particle identification for the BaBar experiment Detection of Internally Reflected Cherenkov light (DIRC) R.J. Wilson *11 Departttwnt of Physics, Colorado State University, Fort Collins, CO 80523, USA
]
Representing the BaBar Collaboration [ 1
Abstract
We describe a new type of detector for particle identification which will be used in the BaBar experiment at the SLAC B factory (PEP-II). This detector, used in the barrel region of BaBar, is called the DIRC which is an acronym for Detection of Internally Reflected Cherenkov light. Similar to other Cherenkov ring imaging devices, the DIRC provides high resolution determination of the Cherenkov angle for pions and kaons but is also particularly well-matched to the momentum range and angular distribution of CP-violating final states in asymmetric B factories.
1. Introduction
A new generation of accelerators, commonly referred to as B factories, has been designed to study charge-parity (CP) violating behavior in the neutral B meson system. The high event rates and relatively high momenta of the decay products provide challenges for almost every aspect of the experiment. In this paper we will concentrate on the requirements and solution for particle identification in the BaBar experiment at the PEP-II accelerator being constructed at the Stanford Linear Accelerator Center. However, much of what we will discuss is relevant to other similar facilities. In the following sections we will present first the basic particle-ID requirements for physics at PEP-II. This will be followed by a description of the DIRC principle and the specific implementation for BaBar. Finally, we will summarize the current status and production schedule for the BaBar DIRC.
2. Particle ID requirements
at PEP-II
2.1. PEP-II The PEP-II accelerator is an asymmetric electronpositron collider with beam energies of 9 GeV (e-) on 3.1 GeV (e+) which gives rise to copious production of the Y (4s) resonance with a Lorentz boost of py = 0.56 in the laboratory frame. This boost means that the fastest decay
particle tracks occur in the direction of the high energy (electron) beam producing a significant momentum-angle correlation. Future upgrades may include an increase in the energy to run at the Y(5S) with a corresponding increase the maximum momenta. The initial luminosity of PEP-II will be 3 x 1O33cm-‘s-’ increasing to 10% cm-2s-1 after a few years operation. The beam crossing period will be 4.2 ns. This high luminosity leads to a “physics” rate of more than 100 Hz, but the BI? rate of particularly interest is a few Hz. 2.2. Primary physics goal The primary physics goal of the BaBar expkriment is to observe CP violation in the BB system. A critical objective for particle-ID in CP physics is to identify kaons for tagging the B meson flavor (b quarks decay primarily to K- and the 6 anti-quarks decay to K+) . Most of the secondary particles in B decays have a momentum below 1.0 GeVlc but efficient tagging of B mesons requires good kaon identification up to about 2.0 GeVlc. Due to the geometrical acceptance of BaBar (partially imposed by PEP-II requirements) and kaondecays in flight, the best possible efficiency for kaon tagging is approximately 22%. Measurement of dEldx by the drift chamber should give 3a separation up to about 0.7 GeV/c, but this would leave more than 45% of kaons untagged. It is important to recognize that the efictive efficiency in this type of analysis is given by
* E-mail [email protected] ’ Supported
by U.S. Department
016%9002/96/$15.?0 Copyright PlrSO168-9002(96)00600-6
of Energy Grant DE-FG03-93ER40788.
E& = E@( 1 - 2w)*,
@ 1996 Elsevier Science B.V. All rights reserved
R.J. Wson/Nucl.
Instr. and Meth. in Phys. Rex A 379 (1996) 444-447
where w is the fraction of incorrect tags. This means that the purity of the sample is also very important. With dE/dx alone the tagging efficiency would be only 9%. Even more challenging than tagging is the requirement for exclusive B decay reconstruction for rare modes such as B0 --+ ~+rr- (branching ratio ~1.2 x 10v5), used to determine the CKh4 matrix parameter sin( 2cu). The asymmetric boost of the two-body final state means that r-K separation up to 4.0 GeVlc is required. In summary, physics requirements imply that we will need a PID system with high efficiency and purity over a large solid angle and large momentum range. Accommodating the additional rich variety of rare B decay, charm and tau physics also will have an impact on the overall detector optimization.
3. The DIRC principle In Fig. 1 we show the basic principle of the DIRC. As the name implies, the device measures Cherenkov radiation produced by a charged particle traversing a solid radiator which then acts as a light guide for the internally reflected light. The portion of the Cherenkov cone trapped in the radiator depends on the entry angle of the incident particle and the ratio of the refractive indices at the interface (ni and ptg in the diagram). The internally reflected light will propagate down the bar until it reaches either end. A mirror placed at one end will reflect that portion of the light back to the opposite end. If the sides of the bar are parallel the photon production angle will be preserved during the multiple reflections. As indicated in the diagram, a position sensitive photodetector surface placed at some distance from the bar end will determine the photon exit angle (after correction for refraction at the bar/stand-off region interface). The “zerodegree” mirror in the stand-off region reduces the size of the imaging array, at the cost of an overlap of a portion of the image. The Cherenkov angle may be determined from these conic section images combined with the particle incidence angle provided by a tracking chamber. The basic requirements of the DIRC can be summarized as: - n3 << nl for efficient internal reflection; - nr x n2 for good light transfer at the end of the bar; - very low absorption in the near UV for long bars (tens of meters) ; - high surface reflectivities; - bars with parallel faces, perpendicular sides and sharp
445
comers; - low absorption for near-UV in the stand-off medium. 3.1. DIRC advantages
Among the many potential advantages of the DIRC compared to other ring-imaging or threshold Cherenkov devices are: compact radiator (a stable solid yielding many photons/cm); relatively simple mechanical structure; no active components in the fiducial volume; no complex fluids system; conventional photodetector technology can be used (photomultiplier tubes) ; sensitivity well matched to an asymmetric collider (light production and capture efficiency improve with dip angle); acceptable mass in close proximity to the calorimeter. In a realistic device, of course, some of these advantages may be obscured by the usual constraints imposed by integration of the system into the rest of the apparatus. Nevertheless, the final implementation of a DIRC in a colliding beams environment will share many of the features of a traditional, and robust, time-of-flight system.
4. BaBar DIRC
In this section we describe a schematic implementation of a DIRC for the BaBar experiment. In Fig. 2 we show the mechanical assembly of the DIRC. The system consists of two major components: the Radiator structure and the Stand-off Box. 4.1. Radiator The radiator barrel consists of 144 synthetic quartz bars arranged in a 1Zsided polygon. Synthetic quartz was chosen over natural quartz due to the superior transmission of
/c-2 m-y Fig. I. DIRC principle.
Fig. 2. DIRC mechanical a.&mbly.
IV. PARTICLE IDEtNTIFICATION
446
R.J. Wilson/Nucl.
Instr. and Meth. in Phys. Res. A 379 (1996) 444-447
near-UV light (above 290 nm) after irradiation of the bar. The quartz has a refractive index of 1.474. Each bar has dimensions 1.75 cm x 3.5 cm x 480 cm with a mirror on one end and is constructed by gluing together four shorter bars of length 120 cm. Twelve long bars are mounted together in a thin-walled box which forms the sides of the polygon. A very thin air gap is maintained between each bar in the box to prevent optical cross-talk. Including the support structure and the bar-box material, the central cylinder has a radial extent of less than 10 cm and presents about 15% of a radiation length at normal incidence. This arrangement defines a fiducial volume in polar angle of 25.5” < 0 < 147’ (or 0.9 > cos 0 > -0.84) and covers approximately 95% of the full azimuth. 4.2. Stand-off Box/photodetector array The Stand-off Box is the mechanical structure which supports the photodetector array and contains high purity water which is the interface medium between the end of the radiator bars and the photodetectors. The vessel contains approximately 6 tonnes of lo-18 Ma water which will be recirculated and purified to maintain good transmission of near-UV light. The photodetector array consists of 11000 conventional 1.125 in. bialkali photomultiplier tubes (EMI 9125) submersed in water for optimal light coupling. The PMTs are arranged in a close-packed configuration on a toroidal surface subtending an angular range of O”-54’ in 12 identical azimuthal sectors corresponding to the symmetry of the radiator structure. The faces of the PMTs are 117 cm from the end of the radiator bars. Reflective hexagonal “light catchers” will be mounted on the circular PMT faces to recover the light which would be lost in the interstices. Since the experiment may run for many years, the PMTs will be operated at a relatively low gain of 106. This provides an average signal of -20 mV (into 50 a). The frontend electronics currently under test will allow a discriminator threshold of around -2 mV and provide 1 ns timing resolution. In Fig. 3 we show a simulation of the PMT hit distribution in the array for an event containing a B” -+ .~+‘rrdecay. This simulation includes background hits from both correlated and uncorrelated interactions in the detector.
Fig. 3. Photodetector array hits simulation. Images from B + rr+rrare shown as solid markers, images from two other charged particles are open markers.
direction is particularly important at an asymmetric collider due to the angle-momentum correlation; the higher particle momenta lead to a corresponding decrease in separation of Cherenkov angles for pions and kaons. We estimate that the 0, resolution per pion will be 2-3 mrad for high momentum tracks (2.0-4.0 GeVlc) for all dip angles, increasing to 4-5 mrad for lower momenta. These assumptions are supported by results from beam tests of a large prototype discussed by Besson at this conference [ 21. In Fig. 4 we show the r-K separation (units of sigma) as a function of cos B for several momenta. Care must be taken in the direct application of this figure to PEP-II physics due to the momentum-angle correlation and the kinematic cutoff in B decays. A detailed simulation of the process B” -+ rr+r- indicates a r-K separation of more than 4a over the entire kinematically allowed range of momenta and angles. If this level of performance is maintained in the final system, we project an efficiency for K-tagging of 19% compared to the maximum possible of 22% (and only 9% with just
5. Projected performance The transverse size of the radiator bar, the photomultiplier tube diameter and the stand-off distance combine to give an intrinsic resolution for single photoelectrons, S&, of =lO mrad. Detailed simulation of the device indicates that we can expect 30-40 detected photoelectrons for particles in a momentum range 0.8-4.0 GeV/c and dip angles in the range f30”, increasing to 60 photoelectrons in the forward direction (cos ~9< 0.85)‘. This enhancement in the forward
Fig. 4. Simulated a/K
separation.
R.J. Wilson/Nucl.
Instr. and Meth. in Phys. Res. A 379 (1996) 444-447
dE/dx). In addition to the primary task of P-K ID the DIRC will also provide muon/proton separation, complementary to BaBar Instrumented Flux Return, of better than 2u below 0.5 GeV/c. 6. Status and schedule The conceptual design for the BaBar DIRC was approved in January 1996 and detailed engineering is under way leading to the final design review in September of this year. The current schedule indicates completion of the quartz assemblies by the summer of 1997 and delivery of the Standoff Box to SLAC in May 1998. The primary goal is for the BaBar detector to be ready for commissioning with cosmic rays by December 1998 followed by colliding beam physics in June 1999.
441
Acknowledgements
I would like to thank my U.S. and French collaborators in the DIRC group, especially members of Group B at the Stanford Linear Accelerator Center for their continuing hospitality. References
[ 11 BaBar Technical Design Report, BaBar Collaboration, SLAC-R-950457 (March 1995). [21 P. Besson, presented at this Conference (6th Int. Conf. on Instrumentation for Experiments at e+eColliders, Novosibirsk, Russia, 1995).
IV. PARTICLE IDENTIFICATION