Design of the Central Tracker of the PANDA experiment

Scientifica Acta 4, No. 1, Ph 3-8 (21) Design of the Central Tracker of the PANDA experiment S. Costanza 1 1 Istituto Nazionale di Fisica Nucleare, Sezione di Pavia, via Bassi 6, 271 Pavia, Italy susanna.costanza@pv.infn.it Physics The PANDA experiment will be built at the FAIR facility at Darmstadt (Germany) to perform tests of the strong interaction through pp and pa annihilations, with very high accuracy. For this reason, it is fundamental that the detector is designed to achieve results with an unprecedented precision. This paper will address the design issue of the Straw Tube Tracker (STT), one of the two options proposed for the PANDA Central Tracker. Two different layouts and their influence on acceptance, reconstruction efficiency and momentum resolution of the simulated tracks will be examined here, as well as the analysis results of two simulated physics channels; in addition, the proposal to study p annihilations on light nuclei ( 4 He) will be presented. Finally, the experimental measurements with a straw tube prototype will be described, together with the spatial resolution obtained by applying a dedicated autocalibration method to the data. 1 Introduction The phenomena of the confinement of quarks, the existence of glueballs and hybrids and the origin of the mass of strongly interacting, composite systems are long standing puzzles and represent a challenge in our attempt to understand the nature of the strong interaction and of the hadronic matter. Significant progresses in fundamental issues of hadron and nuclear physics could be attained by the PANDA experiment [1], that will be installed at the international FAIR facility in the site of the GSI laboratory (Darmstadt, Germany). By taking advantage of the physics potential available using the high intensity, phase space cooled antiproton beams provided by the high energy storage ring HESR, the PANDA experiment will study the charmonium and open charm physics, gluonic excitations and the nucleon structure by means of interactions of antiprotons with nucleons and nuclei; furthermore, it will accomodate additional physics aspects, like Drell Yan and CP violating processes [2]. This rich physics program poses significant challenges on the PANDA detector. Based on the previous experience in antiproton experiments and taking advantage of ongoing detector developments performed at the laboratories for high energy experiments, PANDA will be able to combine the best ever high resolution high intensity antiproton beam with a hermetic detector for charged and neutral particles, in the energy range between 1 MeV and 1 GeV. Clearly, the design choices for the detector should represent a balance between physics needs and available resources. A detailed description of the foreseen detector layout can be found in [1]. This paper collects the major results of the work devoted to the simulation and design of the tracking system of the PANDA detector, with particular attention to the Central Tracker, together with the study of its physics performances. 2 The PANDA Central Tracker: the Straw Tube Tracker option The tracking system of the PANDA Target Spectrometer (TS), which is one of the two magnetic spectrometer the detector will be composed of, is a set of subdetectors sensitive to the passage of charged particles and used to determine their position, momentum and energy loss with high resolution. It consists in the Micro Vertex Detector (MVD), the surrounding Central Tracker and the external Gas Electron Multiplier (GEM) stations, placed downstream.

4 S. Costanza: Design of the PANDA Central Tracker Concerning the Central Tracker, two options are currently under study: a Straw Tube Tracker (STT), that will be described in detail in the following, and a Time Projection Chamber (TPC). The PANDA collaboration will evaluate the best solution as Central Tracker among these two on the basis of the performances resulting from Monte Carlo simulations and experimental tests with prototypes, of the particle identification capabilities and of the detector costs. 2.1 The STT layout The Straw Tube Tracker will consist of planar double layers of straw tubes arranged in a hexagonal layout, that has to fill up a cylindrical volume with an inner diameter of 15 mm, an outer one of 418 mm and a length varying from 12 cm to 15 cm. A straw double layer is made of two closed packed, staggered layers of tubes, each consisting of straws glued together on a reference plate with precise positioning (5 µm) and to the neighbouring tubes along their length, through glue dots. The straws are made of two layers of 12 µm mylar films glued together, and have a gold plated tungsten rhenium anode wire with 2 µm diameter stretched by a weight of 5 g at a gas overpressure in the straw tube of 1 bar. The PANDA straw tubes will be filled with a mixture of ArCO 2, with percentages still under investigation (9/1% or 8/2%); they will be operated at a high voltage of 18 V and 2 bar absolute pressure. The proposed layout will have, in radial direction, four double layers parallel to the detector axis (z axis), four skewed double layers with an angle of ± 3 with respect to the beam axis to reconstruct the z coordinate of the tracks and further two straight double layers. Additional single layers will be placed in the outer region to approach the cylindrical shape. The total number of straw tubes is 421 [3]. 2.2 Design studies In preparation for the PANDA experiment, large scale simulations need to be performed in the upcoming years for the detector design, to determine analysis strategies and to be able to interpret physics results. In particular, since the Central Tracker has to fulfil strict requirements in order for PANDA to achieve high precision results in the topics of its physics program, systematic studies have been performed with the aim of determine the optimal design parameters in terms of geometrical acceptance and to test the performances of the proposed layout in terms of momentum resolution and reconstruction efficiency. The simulations are performed within PandaROOT [4], the PANDA computing framework: based on the object oriented data analysis framework ROOT and on the Virtual Monte Carlo [5], it is for both simulation and analysis. In order to perform full simulations, the physics events of interest are first generated by one of the several event generators implemented in PandaROOT, depending on the physics goal; then, the generated particles are propagated inside the detector and their interactions with the spectrometer are computed. Afterwards, the simulated data are analised with the software tools implemented in the PANDA code structure. Their aim is to reconstruct the particle path, assuming a large range of approximations at a first step (global track fitting with helix) and, in a later stage, taking into account the widest range of path distorsions (Kalman filter recursive method [6]) in order to improve the momentum resolution. To do this, a track follower is also used (GEANE [7]), which transports the track parameters and the covariance matrix from one point of the path to another one. 2.2.1 Example of resolution studies Two design options for the layout of the Central Tracker with STT have been studied: one consisting in 12 cm long straw tubes plus four GEM chambers placed upstream the STT and the other one made of 15 cm long tubes plus three GEM chambers. The tracker performances as a function of different geometric parameters have been investigated through the simulation of two different sets of single track (muon) events, generated at the interaction point (x = y = z = ): first, 1 4 tracks with fixed total momentum (.3, 1. and 5. GeV/c) and uniformly distributed in

Scientifica Acta 4, No. 1, Ph 3-8 (21) 5 φ (φ [, 36 ]) and cos θ (θ [8, 14 ]); then, again 1 4 single track events with fixed total momentum but at fixed θ values. The track fitting has been performed by applying the procedure summarised above. Concerning the first set of simulations, Table 1 summarises the momentum resolutions δp/p and reconstruction efficiencies obtained with the two geometries. The resolution is calculated as σ/µ, using the µ and σ values from the Gauss curves that fit the momentum distributions after the Kalman filter; it is then reported in percentage. The efficiency is defined by the integral below the histogram fitted region, divided by the number of generated tracks. In addition, the efficiency in peak is reported: it is the number of tracks in the fitted range (µ ± 3σ) with respect to the total number of tracks. The table shows that there are no differences between the two layouts from the momentum resolution point of view: the values are compatible within the errors, when not equal, and close to the design goal of 1-2%. Concerning the reconstruction efficiency, the STT 15 cm plus three GEMs option attains better results, apart from low momentum values. Table 1 Comparison of the Kalman fit values of momentum resolution and efficiency for the two foreseen geometries: the STT 12 cm long plus four GEMs option (a) and the STT 15 cm long plus three GEMs one (b). Momentum (GeV/c) Resolution (%) Efficiency (%) Eff. in peak (%) a b a b a b.3.72.72 45.3 43.83 32.6 31.78 1. 1.71 1.71 9.12 92.52 72.31 78.51 5. 3.64 3.66 83.95 91.13 72.1 8.1 The second set of simulations, with tracks generated at fixed values of θ, was aimed to a more detailed investigation, performed through a fine scan of the forward angular region (steps of ±.5 ), where the momentum resolution and efficiency may be mostly affected by the shortening of the tubes. The results demonstrate that the performances of the two proposed layouts are substantially equivalent in the angular region where the two options are geometrically the same (θ > 25 ). On the contrary, in the very forward region (8 < θ < 25 ) a better momentum resolution is attained in the case of the STT 12 cm plus four GEMs option: with this layout and angular range, there is a limited number of hits in the STT, hence the global resolution is dominated by that of the MVD and of the GEMs, which are very precise. The dependence of the performances on the skew angle of the tilted straw layers has been studied too. Although the choice of a higher value for the skew angle could facilitate the mechanical construction of the tracker, the performances are not too much affected: the momentum resolution does not improve significantly when going from 3 to 1, ranging from 1.63% to 1.6%, respectively. 2.2.2 Study of physics benchmark channels The performances of the Central Tracker have been tested also by studying two benchmark channels covering relevant topics of the PANDA physics program. The aim is to demonstrate that the proposed detector setup can fulfil the physics case and that the invariant mass resolution values obtained from the simulations are comparable with the well known results from other experiments. The channels that have been simulated realistically are pp η c (2979) KS K+ π and pp Ψ(377) D + D, such that only charged tracks (the only ones the Central Tracker is able to reconstruct) are present in the final states. The events have been analised by using the tracking tools mentioned above; in addition, for the Ψ(377) channel, a kinematic fit has been tested and applied, in order to further improve the results. The attention has been focused on the invariant masses: the distributions obtained for the η c decay channel are shown in Fig. 1 as an example. For this channel, the η c and KS the have been reconstructed by the PANDA Central Tracker with a resolution of 1.25% and 1.31%, respectively. In case of the Ψ(377) decay channel, the D ± mass resolutions are.73% and.78%, respectively, after the kinematic fit.

6 S. Costanza: Design of the PANDA Central Tracker Entries 22 2 18 16 14 12 1 8 6 4 2 M K S Entries 25769 Mean.4957 RMS.2774 Constant 1911 Mean.4972 Sigma.6452.4.45.5.55.6 M 2 (GeV/c ) KS Entries 35 3 25 2 15 1 5 M ηc Entries 5228 Mean 2.898 RMS.2924 Constant 312.3 Mean 2.976 Sigma.3917 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 2 (GeV/c ) Fig. 1 KS (left) and η c (right) invariant mass distributions. The parameters reported in the statistics boxes are referred to the Gauss functions that fit the distributions in the mass windows indicated by the red dashed lines in figure (m K ± 15 MeV/c 2 and m ηc ± 15 MeV/c 2, respectively). S M ηc Some tests have been also performed to evaluate the dependence of the invariant mass resolution on the spatial resolution of the single straw tubes. It has been shown that a factor two in the spatial resolution of the single tube turns out into a loss in invariant mass resolution. In particular, this loss amounts to 1% for the D ± mesons in the case of the pp Ψ(377) D + D channel and of 9% and 17% for the K S and the η c, respectively, in the case of the other decay channel. Finally, starting from recent results on p 4 He annihilations, it has been investigated the possibility to study antiproton annihilations on light nuclei, since they could allow strangeness and charm production studies in exclusive annihilation channels and could be a powerful tool to discover possible exotic signals, like the quark gluon plasma (QGP) formation [8]. The p 4 He simulated annihilations have been analised with the tracking algorithms already presented and through the study of the (p u, E) plots, which show how the reconstructed events are distributed as a function of the total measured energy E and unmeasured momentum p u : p u p u = E 2 u M 2 u = (E m s E) 2 M 2 u, (1) where E is the initial energy of the system p 4 He (E = E p + m4 He), m s is the mass of the spectator nucleons ( hm p, h = 1, 2, 3) and M u is the mass of the unmeasured particles [9]. The results demonstrate that PANDA would be able to disentangle an exotic signal from the background with a statistical significance of 5σ if the signal amount is down to 4-5% of the background. Moreover, the (p u, E) plots would allow to distinguish between annihilation on a single nucleon (SNA) and SNA combined with the final state interaction (FSI). 3 Test measurements with a STT prototype Concerning the experimental tests, an R&D program which foresees the construction of a complete full scale prototype of the STT is ongoing: at the Institut für Kernphysik of the Jülich Research Center (IKP FZJ) about 2 straw tubes have been assembled and glued together to double layers (Fig. 2, left), but such a prototype is not yet ready for experimental tests. Nevertheless, a smaller laboratory setup is available at IKP FZJ and it has been used to develop calibration techniques for the PANDA STT, to understand signal formation, to test the electronics, to investigate the potentiality of the tracker in particle identification based on energy loss measurements.

Scientifica Acta 4, No. 1, Ph 3-8 (21) 7 Although it is not a complete full scale prototype of the PANDA STT and it is equipped with non dedicated electronics, it consists of four double layers of straw tubes similar to the PANDA ones: the tubes have the same geometrical properties (radius, length, wall thickness), are filled with the same gas mixture and are operated at the same high voltage and pressure (Fig. 2, right). So tests with this laboratory setup can provide useful hints for the design and construction of the PANDA STT, as well as for the data analysis. Fig. 2 Left: assembly of a full scale semi barrel prototype; right: straw tube prototype used for experimental tests at the Institut für Kernphysik at Forschungszentrum Jülich (IKP FZJ). Design and construction by IKP FZJ. 3.1 Experimental results The experimental data collected with the prototype (cosmic ray events) have been analysed with software tools implemented ad hoc. In particular, a dedicated method has been developed to obtain an accurate knowledge of the r(t) relation, i.e. the relation between the measured drift time t and the distance of closest approach of the particle trajectory to the wire (drift distance r), which is necessary to perform a good track fitting. This procedure is called autocalibration, since it makes use just of the information from the tubes under investigation. The method works as follows: at each step of the procedure, the r(t) relation derived in the previous iteration is used to convert the measured drift times into drift radii, that will be used in the track fitting; at the first step, the r(t) relation obtained directly from the integration of the drift time spectra is used. Once a track candidate is identified by the pattern recognition, the track is reconstructed through a dedicated track reconstruction algorithm. For each tube of the pattern associated with a track, the residuals ri = i rfi it rraw are then computed: if the r(t) relation was exact, the average residuals would be zero at all radii; deviations from zero mean miscalibrations in the r(t) relation, which is then directly corrected by using the average deviations themselves. The procedure is iterated until the corrections become less and less relevant; at this point, the r(t) relation has converged to a stable and optimal solution [1]. In the analysis of the experimental data, the attention has been mainly focused to the spatial resolution of a single straw tube: this is usually intended as the standard deviation σ of the distribution of the residuals as a function of the drift distance r. The obtained resolution curve is shown in Fig. 3, left. Its mean value is in agreement with the deviation σ of about 177 µm (parameter p5 of the fit) of Fig. 3, right: it shows the distribution of the mean residuals of 162385 straw hits after the last iteration of the autocalibration procedure, whatever the value of the corresponding track to wire distance is, fitted with two Gauss functions to better describe both the peak c 21 Università degli Studi di Pavia

8 S. Costanza: Design of the PANDA Central Tracker and the tails. The result obtained is very close to the design goal of 15 µm, which could be certainly reached by using a dedicated electronics with a better time resolution. Spatial resolution (mm).35.3.25.2 Entries 16 14 12 1 8 6 Iteration 6 Mean.5152 RMS.3847 p 1192 p1 -.3596 p2.6853 p3 1.341e+4 p4.1385 p5.1773.15 4 2.1 1 1.5 2 2.5 3 3.5 4 4.5 Drift distance (mm) -2.5-2 -1.5-1 -.5.5 1 1.5 2 2.5 Residuals (mm) Fig. 3 Left: single tube spatial resolution σ as a function of the drift distance, fitted with a third order polynomial (red line); right: distribution of the residuals after the last iteration of the autocalibration procedure; the statistics box reports the parameters of the Gauss fit (red curve). 4 Conclusions The results presented above show that the performances of the PANDA Straw Tube Tracker are close to the design goals, both concerning the momentum resolution δp/p and the spatial resolution σ r (r). The work is still ongoing from the experimental point of view, as well as from the software side: consequently, further improvements will certainly be attained. The present scenario, however, already demonstrates the feasibility of the STT and makes a decisive step towards its realisation. Acknowledgements A deep acknowledment goes to Prof. A. Rotondi for supervising this work, for enlightening discussions and scientific advices. I am grateful to all the colleagues of Gruppo III, in particular Dr. L. Lavezzi for the helpful suggestions. Prof. J. Ritman and Dr. P. Wintz from the Institut für Kernphysik of the Forschungszentrum Jülich (Germany) are gratefully acknlowledged for the hospitality in their research group and for the fruitful collaboration. References [1] M. Kotulla et al., Technical Progress Report for PANDA Strong Interaction Studies with Antiprotons (Feb 25). [2] W. Erni et al., Physics Performances Report for PANDA (Mar 29), arxiv:93.395v1. [3] S. Costanza et al., Proceedings of the 11th Pisa Meeting on Advanced Detectors, La Bioloda, Isola d Elba, Italy, May 24 3, 29; Nuclear Instruments and Methods in Physics Research A 617, 148 (21). [4] PANDA Computing Group, A data analysis and simulation framework for the PANDA collaboration, Scientific Report GSI (Darmstadt, 26). [5] I. Hrivnacova, D. Adamova, V. Berejnoi, R. Brun, F. Carminati, A. Fasso, E. Futo, A. Gheata, I. Gonzalez Caballero, A. Morsch (ALICE Coll.), The Virtual Monte Carlo, ArXiv Computer Science e-prints, cs/365. [6] R.E. Kalman, Transaction of the ASME Journal of Basic Engineering 82 (Series D), 35 (196); R. Frühwirth, Nuclear Instruments and Methods in Physics Research A 262, 444 (1987). [7] V. Innocente and E. Nagy, Nuclear Instruments and Methods in Physics Research A 324, 297 (1993). [8] J. Rafelski, Physics Letters 91 B, 281 (198). [9] P. Montagna et al., Nuclear Physics and Methods in Physics Research A 7, 159 (22). [1] G. Avolio et al. Nuclear Instruments and Methods in Physics Research A 523, 39 (24); M. Bellomo et al., Nuclear Instruments and Methods in Physics Research A 573, 34 (27).