S. Ahmad, B.E. Bonner, W.J. Llope, G. Mutchler, E. Platner, J. Roberts, H. Themann, M Wright, P. Yepes
T.W. Bonner Nuclear Laboratory
Rice University
Houston, TX 77251
1. Introduction
The physics goals of the STAR Collaboration include the experimental detection of signatures of a new form of matter called the Quark Gluon Plasma (QGP). The tracking and identification of a large fraction of the charged particles emitted in a central collision will allow the very powerful technique of the correlation of many observables on an event-by-event basis. Since it is expected that the charged particle densities will be high at mid-rapidity, (dnch/d[[eta]]~1000-1500), it will be possible to extract global observables such as the temperature, flavor composition, reaction dynamics, and energy or entropy density fluctuations from the information gathered from a single event. In view of the fact that no unequivocal experimental signature has yet been identified, we maintain that it is necessary to have this information in order to identify possible special events which may bear special scrutiny. The large acceptance and flexibility of the STAR detector will allow the study of the characteristics of the hot and dense matter that is formed. An understanding of these characteristics is important to distinguish between QGP and non-QGP models for the particle production.
In this proposal, we illustrate that the physics potential of the STAR detector will be significantly enhanced with the improved particle identification provided by a segmented Time Of Flight (TOF) system. The proposed system will cover one-third of the acceptance subtended by the STAR Time Projection Chamber (TPC) from -1<=[[eta]]<=1. Specifically, it will:
* extend the momentum limit for which unambiguous [[pi]]/K separation is possible from ~0.6 GeV/c to ~1.5 GeV/c, with <10% contamination
* extend the momentum limit for which unambiguous K/p separation is possible from ~1.2 GeV/c to ~2.4 GeV/c, with <10% contamination
* extend the range of mT for which unambiguous inclusive spectra can be obtained:
Particle From To
[[pi]] 0.2 <= mT <= 0.6 0.2 <= mT <= 2.5
K 0.5 <= mT <= 0.8 0.5 <= mT <= 2.6
p 1.0 <= mT <= 1.5 1.0 <= mT <= 3.2
This allows more accurate estimates of the characteristic [mT> for a given particle species and a sensitive inclusive search in "special" event samples for possible departures from the behavior observed at low mT
* allow high precision measurements of the mT spectra for [[pi]], K and p. This information relates to temperatures, flow, and source sizes.
* allow interferometry as a function of the transverse momentum
* allow measurements of high-pT protons and antiprotons. The QGP has been predicted to suppress such particles
* allow more sensitive measurements of the K/[[pi]] ratio versus the transverse momentum
* allow detailed studies of the [[phi]] meson to understand strangeness production
* provide an efficient means of calibrating the particle identification (PID) capabilities of the TPC and SVT in the range of momenta where these detectors overlap with the TOF system
* allow sensitive inclusive searches for events tagged as "special" using correlated event-by-event observables for signatures of quark-gluon plasma formation which are otherwise inaccessible to the STAR detector.
2. Physics Motivation
The search for the Quark Gluon Plasma (QGP) is one of the main purposes of the Relativistic Heavy Ion Collider (RHIC) being built at BNL. An intense research program has recently been carried out in ultrarelativistic heavy ion collisions, and number of QGP signatures have been proposed. Some of these signatures, like strangeness enhancement and J/[[Psi]] suppression, were observed experimentally. After the initial excitement created in the community by these observations, many models that do not include QGP formation were shown to explain the data. As a consequence, the importance of understanding the properties of the hadronic matter before making statements concerning the creation of the QGP was realized. The determination of the temperature, density, transverse and longitudinal flow and source volume reached in the collision are essential ingredients to understand the dynamics of the hadronic matter.
The temperature of a static thermalized particle source can be obtained from the [1/mT][dn/dmT] distribution of the emitted particles. This simple picture is modified in reality by the presence of resonances and transverse flow. Resonances tend to modify the mT spectrum at low momentum, especially for light particles, like pions, while transverse flow increases the transverse mass. In this context the measurement of the mT spectrum for pions, kaons and protons in a large mT range is very important in order to understand the characteristics of the hot hadronic matter produced after the collisions of heavy-ions.[1]
In STAR, such precise measurements are available in a meaningful mT range for pions, kaons, and protons only with the inclusion of the TOF system. This is illustrated in Figure 2.1, where [1/mT][dn/dmT] is plotted for the different particle species with and without the information provided by the TOF system. As apparent from this Figure, the TOF system expands the measurable range of mT considerably. It is important to note that the slopes of these distributions vary with mT. Misleading results for the temperatures extracted from the inverse slopes of these distributions would therefore be extracted in the absence of the TOF information.
Figure 2.1.1 Invariant cross section versus the transverse mass for pions, kaons and protons from the full simulation of the STAR detector for central AuAu collisions using the Hijing model for particle production. The shaded points show the region that can be measured using only dE/dx from the TPC for PID. The open points are obtained with inclusion of the TOF system.
Particle correlation measurements have been widely used to extract the source sizes at freeze-out. Most commonly pions are used to form the interferometric correltation functions. However, the presence of abundant resonances tend to distort the source radius extracted from pion correlations, especially for low momentum pairs. The situation improves when analyzing kaon correlations and/or high momentum pairs.
Kaon interferometry has also being used to deduce the size of the kaon emitting source. Kaons are less sensitive to resonances and the lifetime of the source can be extracted from it.[2] In addition, the study of interferometry as a function of transverse momentum provides useful information about particle flow. Furthermore, the presence of QGP droplets in hadronic matter has been predicted to produce characteristic features in the kaon correlation function.[3] The presence of the TOF system will significantly extend the pT range where kaon and (identified) pion interferometry can be studied.
2.2. Strangeness enhancement
The K/[[pi]] ratio does not by itself provide a signature for QGP formation. However, it has been proposed that the variation of this ratio with temperature and chemical potential in the presence of a QGP could be unique.[4] A very precise study of the ratio as a function of transverse momentum, which can only be achieved with the information from the TOF system, will shed light on the origin of possible enhancements to this ratio.
The production cross section of [[phi]] meson can be measured inclusively from the decay [[phi]]->K++K-. The measurement of the yield of the [[phi]], which is an s+s-bar particle, places more stringent constraints on the origin of the observed flavor composition than the K/[[pi]] ratio. The production rate is also expected to be extremely sensitive to changes in the quark masses,[5] due to the chiral phase transition at high energy densities which is predicted by lattice QCD calculations. The [[phi]] pT distribution with/without the proposed TOF system is shown in Figure 2.2.1. As can be observed, the pT dependence can only be studied with the TOF system.
Figure 2.2.1 The [[phi]] meson pT spectrum using only dE/dx from the TPC (solid points), and including a one-sixth patch of the TOF system (open points). The distributions are not corrected for the detection efficiencies or the geometrical acceptance.
2.3. Jet quenching
When the QGP is formed in a heavy-ion collision, most of the particles produced in this new state of the matter must undergo freeze-out and hadronization processes before reaching the experimental apparatus. Unfortunately, most of the particles that are detected will have lost all memory of the primordial QGP. Only jets and leptons are predicted to retain any memory of the transient QGP. The study of jets is particularly interesting since RHIC will be the first heavy ion machine with a high enough energy to produce them. However the experimental challenges that jet detection presents in AuAu collisions are substantial.
The TOF system can provide a unique tool to study jet quenching by measuring the ratio of high pT proton/antiproton spectra. The energy loss produced by QGP will affect protons and antiprotons differently. Therefore, modified pT spectra are expected at mid-rapidity where the production mechanism is the same for both particles.[6]
Other mechanisms due to the presence of QGP have been also proposed for the suppression of high pT nucleons.[7] Annihilation processes have been predicted to suppress intermediate-pT antiproton production in heavy ion collisions. This could be used as a probe of the space-time evolution of the reaction.[8] This effect is opposite to that predicted for jet quenching, and at a lower momentum. The TOF system will allow the study of the such antiproton pT spectra, which are potentially very interesting.
3. Particle Identification
In the STAR detector, particle identification will be accomplished by exploiting the fact that the [pT> of pions, kaons, and protons in the STAR acceptance is such that they exhibit different energy losses while traversing the Silicon Vertex Tracker (SVT) and the Time Projection Chamber (TPC). Figure 3.1a shows the expected energy loss curves as measured in the TPC in the momentum range 0 - 3 GeV/c for pions, kaons, and protons. These data were obtained by running the STAR GEANT Monte Carlo program to simulate detector response and energy loss processes in the detector. From this figure, for example, one can see that it is possible to distinguish pions and kaons below 600 MeV/c by determining their momentum and energy loss.
Particles also deposit energy in the SVT, and the measured energy loss in this detector can be used for particle identification as well. At the time of this proposal, the software needed to evaluate the performance of the combined SVT and TPC dE/dx PID capabilities was not available. However, Figure 3.1b shows the TPC dE/dx distributions with the resolution improved by [[radical]]2, which would be the effect on the combinaed PID capability if the SVT dE/dx information was equivalent to that from the TPC. From the curves shown in Figure 3.1b, it is apparent that the additional information provided by the SVT has the effect of raising the upper momentum limit by ~100 MeV/c for the separation of pions and kaons.
Figure 3.1c shows the same 1/[[beta]] versus momentum spectra for pions, kaons, and protons, assuming the inclusion of the STAR-TOF system. The clean separation of these species at still higher momentum is evident. The range of momenta over which inclusive spectra of [[pi]]'s, k's and p's can be studied unambiguously is extended significantly.
Figure 3.1 Particle identification distributions for a) dE/dx in the TPC alone, b) dE/dx for TPC+SVT (see text for explanation), and c) Time-of-flight information. Plotted is the mean value of the measured quantity as a function of particle momentum. The error bars are one standard deviation of the distributions at the measured momenta.
Figure 3.2 shows the efficiency and contamination as a function of pT for kaons and protons with only the TPC dE/dx information, and including the information from the TOF system. The efficiency is defined as the number of correctly identified particles divided by the number of those reconstructed by the TPC. The contamination is defined as the number of incorrectly identified particles divided by the total number of TPC tracks.
Figure 2.2 Efficiency and contamination levels for particle identification plotted as a function of transverse momentum for kaons and protons. Figures (a) and (b) are for TPC dE/dx particle identification only, and (c) and (d) are obtained with combined dE/dx and time-of-flight information.
We estimate that for pions and kaons with pT>150 MeV/c approximately 30% and 40%, respectively, of the spectra of these particles cannot be measured without the information of the TOF system. Most importantly, without the TOF system only a very reduced pT range can be studied. With the inclusion of the TOF system, the measurable pT range is increased by a factor ~4(~1.5) for kaons(protons).
In conclusion, a very significant fraction of the pT spectra for pions, kaons and protons cannot be identified without the particle identification provided by the TOF system. The additional information is particularly important for the identification of kaons and protons. At least a partial implementation of the TOF detector system is essential to study the pion, kaon, and proton pT (or mT) spectra. As pointed out in Section 2, this is essential to understand the hadronic matter produced in the heavy-ion collision.
4. System Design
4.1. Occupancy
Given the large number of particles expected at mid-rapidity for central AuAu collisions at [[radical]]snn = 200 GeV, an obvious concern is the occupancy of the proposed TOF system, and the importance of multiple hits on singler TOF counters. The segmentation of the TOF system has been chosen to insure an acceptable single hit occupancy. Table 4.1 shows the number of hits per counter, which were obtained from a full GEANT simulation with all processes turned on.
Table 4.1 Number of hits per TOF counter
Hits Probability(%)
0 72.5
1 20.5
2 5.5
3 1.2
>3 0.3
Although the average occupancy is relatively high (~23%), the number of counters with more than one hit is less than 10%.. This sets a lower limit on the efficiency for correctly identifying particles which hit the TOF counters. The actual efficiency for identifying a given particle species has been shown to be somewhat better than that presented above[9], since the spread in the particles' momenta may be sufficient to unravel the correct TOF even in the presence of multiple hits.
4.1. Acceptance
The acceptance of the proposed TOF system is driven partially by cost, and partially by the desire to optimize the acceptance for measuring the [[phi]] meson. Due to the fact that the mass of the [[phi]] is very close to the mass of 2 kaons, the [[phi]] mass and width are particularly sensitive to possible modifications of the kaon dispersion relation in the nuclear medium, and therefore, for example, to the possible restoration of chiral symmetry. For decay kaons above 0.5 GeV/c in momentum, the efficiency for detection without the TOF detector is reduced substantially due to the decreasing efficiency for unambiguous PID in the TPC and SVT. This corresponds to a [[phi]] meson having a momentum of ~1.0 GeV/c. While the decay kaons from [[phi]] mesons of this momenta would all fall within an envelope of approximately +/-15deg. in the absence of the STAR magnetic field, GEANT simulations of [[phi]] decay in the STAR detector indicate the envelope necessary to insure effective detection of the decay kaons increases to +/-30deg. when the magnetic field is properly taken into account. Therefore, [[phi]] detection is possible if the TOF patch covers at least 60deg. in azimuth. The proposed system will cover 120deg. in azimuth over a rapidity range of -1<=[[eta]]<=1.
4.2. Phototubes and Bases
To achieve the physics goals of STAR, the TOF system must measure the time of flight of as many particles as possible in the central region with a timing resolution of at least 100 picoseconds for a flight path of 2 m. This requirement can be met by a highly segmented cylindrical counter of 2 m radius and 4.3 m length, capable of operating in a 5 kG magnetic field. The proposed design is a "shingle" design, consisting of tiling a portion of the cylinder with ~ 2000 single ended scintillators arranged in 40 trays of 50 scintillators each. Each tray is 2.4 meters long and 21 cm wide. Individual trays can be inserted from one end of the magnet. They are mounted and supported on the TPC outer field cage.
Each shingle counter consists of a 1" diameter 15 stage R5500 Hamamatsu proximity mesh dynode phototube optically coupled to a scintillator with a plexiglas light guide. The scintillator to be used will be BC404 with dimensions 1.5 x 4 x 20 cm. Plexiglas light guides of fish tail design are used to couple the 1.6 cm diameter active area of the phototube to the 4 x 1.5 cm edge of the scintillator. These wedges are machined from 1.5 cm thick acrylic sheets. The counters are placed 5 abreast (in [[phi]]) and 10 deep (in [[eta]], over the range [[Delta]][[eta]]=1) in the module trays.
We have shown that the basic concept of using a mesh dynode photomultiplier in a 5 kG magnetic field to achieve better than 100 ps timing is viable. The experimental results leading to this conclusion are reviewed in Section 5.1 below.
4.3. Support Structure
The mechanical support system consists of the 60 five meter long aluminum rails already bolted to the TPC outer field cage for the support of the Central Trigger Barrel (CTB). The 40 module trays to be used for the TOF counters are each 2.4 meters long and 21 cm wide and are inserted into the slots formed by the support rails. Twenty trays are inserted from each end of the STAR detector. Each tray holds 50 TOF shingle counters, 5 across (in [[phi]]) and 10 deep (in [[eta]]). A cross sectional view of a tray is shown in Fig. 4.3.1. Also shown is a fraction of the lengthwise cross section of a tray.
Figure 4.3.1 Mechanical design of the TOF tray showing the mounting bracket that is attached to the outer field cage of the TPC. The five abreast by ten deep groups of counters are shown in the bottom two diagrams.
The module trays are fabricated out of molded plastic. A shingle counter is supported by a molded foam block, and secured by plastic restraint struts. Each tray will have water cooling for the electronics. This is provided by an aluminum cooling coil running around the top of the tray. An aluminum heat sink joins the electronics to the coil. The power and signal cables, as well as the cooling water lines, are run inside the module trays.
4.4 Readout Electronics
An overall block diagram of the TOF electronics is given in Figure 4.4.1. The TOF system must provide mid-rapidity multiplicity information as well as timing to better than 100 ps. The PMT signals are processed in electronics adjacent to the phototube. Each tube feeds a fast discriminator and amplifier. The number of discriminators firing (multiplicity), amplitude, and time signals are digitized within the 110 ns crossing time. The multiplicity is transmitted to the trigger matrix via optical data links for each crossing. Upon acceptance by Level 1 Trigger, the time and amplitude information is transmitted to Level 2 Trigger by other optical data links.
Figure 4.4.1 Block Diagram of TOF electronics.
To avoid excessive heat load and cable runs, the high voltage for each phototube is provided by a 15 stage Cockroft-Walton high voltage supply capable of operating in the 5 kG magnetic field of the STAR solenoidal magnet. The signals will be connected directly to the fast discriminators and amplifiers mounted on the TDC and ADC boards in the tray, which are adjacent to the phototubes. As shown in Figure 4.4.1, each board contains 5 constant fraction discriminators, time to pulse height converters (TPH), 12 bit ADC's and associated logic. The present plans are to use the circuit designed at Los Alamos, which will also be used in the PHENIX detector. In addition to feeding the TDC, the discriminator will be used to provide a multiplicity trigger. Each ADC board also serves 5 adjacent tubes. The board contains 5 amplifiers, 10 bit ADCs and associated logic. The ADC information can be used either for a multiplicity trigger or for TOF walk corrections.
The ADC and TDC data are collected in a tray mounted driver board and sent over high-speed fiber-optic links to a Receiver Module located near the Level-1 and Level-2 Trigger system. Upon Level-1 acceptance, all information of interest is sent to Level-2 to undergo further processing. These links are capable of transmitting 1 Gigabit/sec. The link consists of a G-link module, a FOXMIT module and a fiber optic cable. The data is transmitted synchronously with the 110 ns crossing time. An additional link per tray is needed for the RHIC clock.
Each tube will have a custom designed chip to provide slow control functions such as setting high voltage, setting discriminator thresholds, reading back high voltage and threshold settings, et cetera. The chip required for this is being developed by BNL and LeCroy under a CRADA agreement. The signals for the CRADA units will be transmitted on a slow duplex optical link for each tray.
The power supplies for the tray electronics will be placed in racks on the cart. Each tray will have + 5 volt and - 5.2 volt supplies connected by 4 #1 wires. This will be a total of 240 5 volt supplies. Five trays will share the +24 and -24 volt supplies. This will require 48 of the 24 volt supplies. Each tray will have 4 #10 wires to the power racks for the 24 volt power.
Two N2 UV lasers (one at each end of the STAR detector) will be used to provide calibration pulses to each TOF shingle. The UV light will be piped to each of the TOF trays via an optical fiber. These fibers will be subdivided inside each tray, supplying calibration inputs to each TOF shingle.
5. Report on TOF R&D
5.1. Results from Tests
The TOF system should measure the flight time of as many of the charged particles in the central region of the STAR detector as possible, with a timing resolution of at least 100 picoseconds. It is now a technical reality to place phototubes directly in 5 kG magnetic fields and achieve the timing resolution that are required for the present physics goals. This was made possible by the advent of a new phototube design, the mesh dynode tube.[10],11 Further advances in this tube design, the proximity mesh dynode structure with the dynodes positioned near the photo cathode in order to reduce the transit time spread (TTS) have resulted in a magnetic field insensitive tube with a single TTS of 100 to 150 ps. Four such tubes (Hamamatsu R3432-01) were evaluated. These are 15 stage, one inch diameter tubes, with an active area of 2 cm2 . Tube bases with a linear dynode resistor string supplied by Hamamatsu were used. No attempt was made to optimize the performance of the bases.
Tests of the shingle geometry using the R3432-01 phototube at zero field were done in the M-11 beam line at TRIUMF with a minimum ionizing beam of 300 MeV/c [[pi]]- and in a 5 kG field in the MPS facility at the AGS using an 8 GeV/c [[pi]]- beam. The results have been reported in two STAR Notes[12], and a NIM paper[13]. We summarize the results here. From our measurements of the time resolution obtained in a variety of geometries for various scintillator materials, we derived empirical fits that could be used to predict the resolution to be expected over a wide range of scintillator dimensions. The time resolution ([[sigma]]) attainable for a 2.6 cm x 1.5 cm x 23 cm shingle counter made from BC420 varies from better than 60 ps near the phototube to 100 ps at the far end of the counter in a 5 kG field. This shows that the required resolution can be obtained.
5.2 Future Plans
An R&D effort is under way to develop proximity mesh dynode PMTs under MEPhI and PPL-Dubna direction supported by State Department funding. The goal is to produce a few prototype phototubes with Russian industry. According to the schedule now being finalized with our collaborators at MEPhI and PPL-Dubna, these tubes should be available for testing by the end of FY95. A follow-on request for State Department funding in FY95 is planned. This effort would be directed to develop large quantity production capability with the expectation that that the TOF phototubes can be produced for less than $250 each.
Also presently under investigation is the feasibility of using Avalanche PhotoDiodes (APDs), instead of the proximity mesh dynode PMTs, to provide a significant savings in the overall cost of the TOF system. It has been known for some time that small APDs (i.e. with active areas on the order of ~1 mm2) that are run in Geiger mode are capable of providing a timing resolution near 60 ps.[14] Important advances in the electronic circuitry for these devices have been made,[15] which allow these devices to be run at high rates via the active quenching of the avalanches that signal particle hits. These devices are insensitive to 5 kG magnetic fields, and require generally simpler (i.e. less expensive) bias supply circuitry as compared to the proximity mesh dynode PMTs. Larger area APDs (i.e. ~16-25 mm2, such as would be necessary for the present application) have generally suffered from higher costs and larger terminal capacitances, which significantly degrade the best timing resolution that can be obtained. However, Hamamatsu and other manufacturers have recently fabricated large area, low capacitance, "blue-enhanced" APDs for special orders of large quantities. The extent to which such APDs provide both a significant cost savings and the performance needed for the present physics goals is currently being studied.
6. Cost Estimate
Table 6.1 gives our best estimate of the cost of a TOF system for STAR covering one-third of the full acceptance within -1 <= [[eta]] <= 1.
Table 6.1 The projected cost of the proposed TOF system for STAR.
7. Schedule
The schedule projections for the TOF system are shown in Figure 7.1. It is noted that if funding becomes available starting in FY95, the TOF detector system as proposed would be ready by RHIC start-up.
Figure 7.1 The projected schedule for the STAR-TOF detector.