«Abstract In this document a new detector in place of the barrel time-of-ﬂight detector is proposed. This detector is based on small scintillator ...»
Proposal for a Scintillator Tile Hodoscope for PANDA
K. Goetzen, H. Orth, G. Schepers, L. Schmitt, C. Schwarz, A. Wilms
In this document a new detector in place of the barrel time-of-ﬂight detector is proposed.
This detector is based on small scintillator tiles read out by silicon photomultipliers. The
motivation in terms of physics and technical beneﬁts are summarized. Details of the detector
layout are given.
CONTENTS 2 Contents 1 Introduction 3 2 Physics Motivation 3
2.1 Hypernuclear Physics.................................. 3
2.2 Conclusions from the TAG PID............................ 4
2.3 First Physics
3 Technical Motivation 8
3.1 Event Timing...................................... 8
3.2 Relative Time of Flight................................. 12
3.3 Pattern Recognition
3.4 Conversion Detection and Charge Discrimination.................. 13 4 Detector Layout 15
4.1 Mechanics........................................ 16
4.2 Scintillator........................................ 19
4.3 Photon Detector.................................... 22
4.4 Electronics........................................ 25 5 Organization 27
5.1 Cost Estimate...................................... 27
5.2 Project Structure.................................... 27 6 Summary 29 1 INTRODUCTION 3 1 Introduction In this document a new detector in place of the barrel time-of-ﬂight detector is proposed. The main criticism of previous concepts of barrel time-of-ﬂight detectors is the material budget deteriorating the performance of the lead tungstate crystal calorimeter. In addition the available space in radial direction is extremely tight so that some detector concepts have problems to ﬁt in. The concept presented here takes the optimization of material and thickness to the extreme implementing a timing detector with less than 2% of a radiation length and less than 2 cm radial thickness including readout and mechanics. This detector is based on small scintillator tiles read out by silicon photomultipliers. A time resolution of better than 100 ps should be achieved by the system to allow for good time-of-ﬂight measurement and other timing applications. Apart from physics applications requiring time-of-ﬂight measurements in the barrel region several technical beneﬁts are important to consider, most notably the usage for software triggering and event building purposes.
In the ﬁrst sections the motivation in terms of physics and technical beneﬁts are summarized. In the following section details of the detector layout are given. Finally ﬁrst considerations concerning the project organization are given. The start of the project depends on a positive decision on the concept by the PANDA collaboration.
2 Physics Motivation
2.1 Hypernuclear Physics Hypernuclear research will be one of the main topics addressed by the PANDA experiment  at FAIR . The PANDA hypernuclear programme will reveal the strength of Λ–Λ interaction via high resolution γ spectroscopy of double Λ hypernuclei. In contrary to past hypernuclear experiments, where only a few double hypernuclei events were found, the challenge of the PANDA experiment will be to increase statistics by ﬁve orders of magnitude. Germanium detectors, despite their good energy resolution, have an eﬃciency of only a few percent.
In combination with the high luminosity of the antiproton beam at HESR, a high production rate of single and double hypernuclei under unique experimental conditions will be possible for the ﬁrst time.
In the PANDA experiment, double hypernuclei will be produced as a result of a multi–stage process. In the ﬁrst stage a Ξ− (together with its associated strange particle) is produced in a + primary target via the reaction p + p → Ξ− Ξ. The Ξ will undergo scattering or (in most cases) annihilation inside the residual nucleus. Strangeness is conserved in the strong interaction and the annihilation products contain at least two anti-kaons that can be used as a tag for the reaction.
In a second stage, the Ξ− is slowed down in a dense, solid material (e.g. a nuclear emulsion) and forms a Ξ− atom . After an atomic cascade, the hyperon is ﬁnally captured by a secondary target nucleus. If the momentum of the hyperon is too high its stopping time will exceed the lifetime and hence the Ξ− will decay prior to the atomic capture with high probability. In order to 2 PHYSICS MOTIVATION 4 reach a high capture probability it is mandatory to keep the primary momentum of the produced Ξ− as low as possible.
The energy release of about 28 MeV during the conversion of the Ξ− into two Λ hyperons may give rise to the emission of particles from the nucleus (double Λ compound nucleus), where the conversion took place. As a consequence, a variety of double, single or twin hypernuclei as well as ordinary nuclei may be produced in excited states.
The hypernuclei study will make use of the modular structure of the PANDA detector. Removing the backward end-cap calorimeter will allow to add a dedicated nuclear target station and the required additional detectors for γ spectroscopy close to the entrance of PANDA.
The major diﬃculty to accomplish this project resides in the complexity of the hypernuclei production mechanism and in the identiﬁcation procedure. Furthemore, the pp → Ξ− Ξ cross section of 2µb is about a factor 25000 smaller than the total pp cross section of 50 mb at 3 GeV/c.Therefore an eﬃcient background suppression is mandatory. As it was remarked before, the associated antihyperon annihilates with large probability ( 85% 1 ) within the primary target nucleus releasing at least two positive kaons which can be used to tag the hypernuclei production. Kaons produced in this way are emitted in the forward direction (beam direction) and with a momentum distribution around 500 MeV/c. Here the diﬃculty resides in ﬁnding a proper detector system to identify eﬃciently positive kaons.
In the current PANDA design, particle identiﬁcation for slow particles (below 700 MeV/c) may be provided at large polar angles by a Time Of Flight (SciTil) detector in combination with the central tracker. This issue is the topic of a diﬀerent report  and can be summarized as follows: The associated Ξ Ξ production allows to trigger on the Ξ to suppress the background by many orders of magnitude. For the planned 12 C primary target 15% of the Ξ leave the target and decay. The other 85% annihilate within the target and produce due to strangeness conservation two kaons.
The number of positive charged kaons produced in this process amounts to approximately 40%.
Here, the reconstruction probability of the kaons is about 28%. These results are the predictions from a URQMD+SMM model  containing the production of Ξ Ξ and the partial annihilation of the Ξ. The background contribution to the kaon trigger amounts to 7.7% and is modeled by the URQMD model  not containing the above mentioned processes. Triggering on the positive kaons increases the total trigger rate by a factor of 7. The barrel DIRC can measure 55% of these kaons while 45% are below its momentum threshold. The beneﬁt of a positive kaon identiﬁcation below the momentum threshold of the barrel DIRC is a factor of 2 higher trigger rate.
2.2 Conclusions from the TAG PID In the report of the Technical Assessment Group on Particle Identiﬁcation (TAG PID)  to the ¯ PANDA collaboration all informations to the PID sub-detectors planned at that time are collected as well as a method is introduced to deﬁne and evaluate the performance of detector parts and a complete detection system for a positive particle identiﬁcation.
As one of the PID detectors a Barrel-Time-of-Flight detector was evaluated in combination with the two options of the central tracker for the target spectrometer being a STT or a TPC. No 1 Probability is given by the UrQMD model 2 PHYSICS MOTIVATION 5 Endcap-ToF was considered in this report. The Barrel-ToF detectors proposed were two types of RPCs and a ToF scintillator barrel covering the polar angles Θ from 22◦ to 140◦. Their position was allocated inside the Barrel-DIRC radius. As simulated by the working groups behind these diﬀerent types all solutions for the Barrel-ToF provide a valuable pion-Kaon separation below the Cherenkov-threshold and a Kaon-proton separation up to 1.5 GeV/c.
Since at the time of the preparation of the report no full simulation was available a fast simulation was introduced  using the parameterization of the diﬀerent PID processes. All known speciﬁc detector eﬀects were input for this simulation package.
With this tool the separation power, deﬁned in the report, could be determined for each combination of two diﬀerent particle species. The separation power was calculated for a ﬁne binning of the solid angle θ covered (by each sub-detector) and the momentum p of the produced particles.
The combination of all detectors results in a map of separation power over θ and p. In regions (bins) covered by more than one detector the global separation power was calculated as the quadratic sum of the separation power of the contributing detectors. Due to its higher sensitivity to small diﬀerences the separation power has been translated into the so called mis-identiﬁcation level for the process of evaluation and comparison.
The connection to the envisaged physics is achieved by comparing phase space plots over θ and p with the map of separation power, whereas the only interesting regions are those where the signal ¯ overlaps with the background particles. For all PANDA relevant physics channels and its relevant background channels phase space plots were produced and the overlap regions were determined.
Only for these regions the average fraction of mis-identiﬁcation was determined.
For all four scenarios of the detector setup, i.e. with the STT, the STT plus a Barrel-ToF, the TPC and the TPC plus a Barrel-ToF, a factor relative to the above value has been computed.
These numbers allow to evaluate the performance of the four options.
In the fast simulation a start detector was assumed having with δt=100 ps the same uncertainty in the time measurement as the Barrel-ToF detector. This lead to an over-all time resolution for the ToF of 141 ps. (Running without a start detector improves the time resolution but this requires to run the ToF with a relative timing.) As emphasized with three examples identiﬁed (and shown in dedicated plots) in the report  the Barrel Time of Flight detector can add in some cases beneﬁcial positive PID information to the
case where otherwise only dE/dx from the central tracker (the STT or the TPC) is available:
of the STT tracker option a Barrel TOF signiﬁcantly improves the identiﬁcation power.
Obviously in this particular situation the dE/dx information of the TPC alone already provides better identiﬁcation potential than the combined information of STT and TOF.
These cases can be seen as example for all charmed baryons.
Table 1: Inﬂuence of a Barrel-ToF detector on the mis-identiﬁcation value of two particle species.
The commented examples are marked with a point.
Since no Endcap-Time-of-Flight detector as it is now proposed with the Forward-Endcap SciTil was evaluated by the PID-TAG an addition to the conclusions of the report may be allowed and
an example might be given where the inﬂuence of this proposed detector becomes clear:
It has to be noted again that the report and its conclusions are based on a fast simulation. This means that for the results no microscopic simulation was done. The PID processes were parameterized and some estimations were done for simpliﬁcation or since no better knowledge was available.
Nevertheless the positive inﬂuence of a Time-of-Flight measurement in the Target-Spectrometer to ¯ the charged particle identiﬁcation in PANDA -ﬁrst of all below the relevant Cherenkov threshold is obvious. The SciTil solution proposed here corresponds to a reduced material amount, a larger ﬂight length from the interaction point, a closer distance to the Electromagnetic Calorimeter and a larger solid angle coverage compared to the solutions evaluated in the report. All these advantages lead to a further improvement in the performance of the Time-of-Flight system.
2.3 First Physics PANDA has no hardware trigger but continuously digitizes all detector signals after autonomous hit-detection. Once the data of one time frame of approx. 500 µs (super burst) is assembled the processing for the event selection can start. The complete processing of all events at the full interaction rate is not possible since this would require computing resources exceeding the cost of the entire experiment. This holds even for the beginning when data taking will still run at a factor of 10-100 lower rate. Therefore simple signatures have to be used for a software trigger, i.e. a very fast ﬁrst selection level which takes only few microseconds. Depending on the available computing power a reduction of 100-1000 has to be achieved in the ﬁrst fast selection.
The ﬁrst step in the event selection is called software trigger. This trigger is based on a simple algorithm based on one or few detectors. It does not require long computational operations but is based on correlations of digital signals. The goal is to achieve high rejection factors from simple signatures. More reﬁned processing takes place at higher levels. Typical examples for software triggers are very much in analogy to their familiar hardware ancestors and cut on particle multiplicity, particle coincidences in time, or simple track pointing.