MARTA: A high-energy cosmic-ray detector concept with high-accuracy muon measurement
P. Abreu, S. Andringa, P. Assis, A. Blanco, V. Barbosa Martins, P. Brogueira, N. Carolino, L. Cazon, M. Cerda, G. Cernicchiaro, R. Colalillo, R. Conceição, O. Cunha, R. M. de Almeida, V. de Souza, F. Diogo, C. Dobrigkeit, J. Espadanal, C. Espirito-Santo, M. Ferreira, P. Ferreira, P. Fonte, U. Giaccari, P. Gonçalves, F. Guarino, O. C. Lippmann, L. Lopes, R. Luz, D. Maurizio, F. Marujo, P. Mazur, L. Mendes, A. Pereira, M. Pimenta, R. R. Prado, J. Rídký, R. Sarmento, C. Scarso, R. Shellard, J. Souza, B. Tomé, P. Trávnícek, J. Vícha, H. Wolters, E. Zas
MMARTA: A high-energy cosmic-ray detector concept forhigh-accuracy muon measurement
P. Abreu , , S.Andringa , P. Assis , , A. Blanco , V. Barbosa Martins , P Brogueira , N. Carolino , L. Cazon , M.Cerda , G. Cernicchiaro , R. Colalillo , R. Concei¸c˜ao , , O Cunha , R. M. de Almeida , V. de Souza , F. Diogo , C.Dobrigkeit , J. Espadanal , C. Espirito-Santo , M. Ferreira , P. Ferreira , P. Fonte , U. Giaccari , P. Gon¸calves , ,F. Guarino , O. C. Lippmann , L. Lopes , R. Luz , D. Maurizio , F. Marujo , P. Mazur , L. Mendes , A. Pereira ,M. Pimenta , , R. R. Prado , J. ˘R´ıdk´y , R. Sarmento , C. Scarso , R. Shellard , J. Souza , B. Tom´e , , P.Tr´avn´ı˘cek , J. V´ıcha , H. Wolters , and E. Zas LIP - Laborat´orio de Instrumenta¸c˜ao e F´ısica Experimental de Part´ıculas, Braga, Coimbra and Lisbon, Portugal IST - Instituto Superior T´ecnico, Lisbon, Portugal Observat´orio Pierre Auger, Malarg¨ue, Argentina CBPF - Centro Brasileiro de Pesquisas F´ısicas, Rio de Janeiro, Brazil INFN - Istituto Nazionale di Fisica Nucleare, Sezione di Napoli and Dipartimento di Fisica “E. Pancini”, Universit`a di Napoli“Federico II”, Naples, Italy Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil Universidade de S˜ao Paulo, Instituto de F´ısica de S˜ao Carlos, - IFSC/USP, S˜ao Paulo, Brazil Universidade Estadual de Campinas, IFGW, Campinas, SP, Brazil Fermilab, Chicago, USA Universidade Federal Fluminense, Rio de Janeiro, Brazil Institute of Physics of the Czech Academy of Sciences, Prague, Czech Republic Universidad de Santiago de Compostela, Santiago de Compostela, SpainReceived: date / Revised version: date
Abstract.
A new concept for the direct measurement of muons in air showers is presented. The concept isbased on resistive plate chambers (RPCs), which can directly measure muons with very good space andtime resolution. The muon detector is shielded by placing it under another detector able to absorb andmeasure the electromagnetic component of the showers such as a water-Cherenkov detector, commonlyused in air shower arrays. The combination of the two detectors in a single, compact detector unit providesa unique measurement that opens rich possibilities in the study of air showers.
PACS.
XX.XX.XX No PACS code given
Over the last decade, the results from the Pierre AugerObservatory [1] and Telescope Array [2] have greatly ad-vanced our understanding of the highest-energy cosmicrays. Despite this fact, several fundamental questions re-main to be tackled by the upgrades of current experimentsor by the next generation of experiments. The determina-tion of the nature of these highest-energy particles is oneof the biggest challenges and an essential ingredient forthe astrophysical interpretation of the data [3].At the highest energies, cosmic rays are scarce andmeasured indirectly through the detection of extensiveair showers (EAS). Large air-shower detectors based onlyon one technique have limited capabilities for separating
Send offprint requests to : a Corresponding author the electromagnetic and muonic shower components. Themuon component of air showers is rich in valuable infor-mation although poorly known. Muons carry informationfrom the first few, high-energy, interactions in the shower.Together with the depth at which the shower reaches itsmaximum ( X max ), the number of muons in the shower( N µ ) can play a crucial role in the determination of the na-ture of the primary. To obtain as much information as pos-sible from the showers, disentangling the electromagneticand muonic components, is paramount for compositionstudies [4]. In particular, a direct and accurate measure-ment of the muonic component would be a breakthrough.For the bulk of air showers, information on the muoniccomponent is currently obtained using indirect methods,which lack precision and direct validation. The first mea-surement of the mean number of muons in inclined ultra-high-energy air showers is presented in [5]. The sensitivity a r X i v : . [ phy s i c s . i n s - d e t ] A p r P. Abreu et al.: MARTA: A high-energy cosmic-ray detector concept for high-accuracy muon measurement of the number of muons to the cosmic-ray mass composi-tion is demonstrated, and a muon deficit in model predic-tions is observed. To fully explore the constraining powerof muon measurements in mass composition, the appar-ent muon deficit in air shower simulations needs to be un-derstood and the uncertainty on the muon measurementfurther reduced. Improvements in the description of themuonic component will also reduce the systematic uncer-tainty in the simulation of the other shower components.Measurements presented in [6] further support that theobserved muonic signal is significantly larger than pre-dicted by models. It should be noted that the showermuon content is not being measured directly but instead,these measurements explore shower characteristics to es-timate it. It is clear though that current hadronic interac-tion models cannot provide a consistent description of allshower quantities, signalling deficiencies in their abilityto describe the interactions that rule the shower devel-opment. The discrepancy between models and data canonly be elucidated by extending this type of event-by-event analysis to include observables with complementarysensitivity to hadronic physics and composition, such asmuonic variables, and by improving the ability of detectorsto separately measure the muon and electromagnetic com-ponents from the signal recorded at ground. Furthermore,studies show that the energy evolution of the moments of X max and N µ can be used to assess the validity of a masscomposition scenario, surpassing current uncertainties inthe shower description [7].In short, current muon measurements are not well de-scribed by models and, while this raises interesting ques-tions concerning hadron interaction at the highest ener-gies, large uncertainties come along in the inference ofcomposition. A direct and accurate measurement of themuon component with 100% duty cycle would allow us tohave composition sensitivity on a shower-by-shower basis,to study hadronic interactions at the highest energies, toimprove the sensitivity to photon primaries and to bet-ter understand and reduce the systematic uncertainties ofmany different measurements.In this paper, we introduce MARTA, an innovativeconcept combining high-accuracy muon measurements pro-vided by resistive plate chambers (RPCs) with the calori-metric measurement of standard air shower detectors. WhileMARTA is a generic detector concept, a detailed imple-mentation for the Pierre Auger Observatory has been de-veloped and is used in this work to provide concrete andrealistic performance expectations.This paper is organized as follows. In section 2, theMARTA concept is presented. The generic detector unitlayout and the choice of RPCs are explained. In section3, details on the built prototypes and performed tests aregiven. Performances are discussed in section 4. Finally, insection 5, prospects and conclusions are drawn. MARTA (Muon Array with RPCs for Tagging Air show-ers) is a hybrid detector concept that combines the in-formation from RPCs with data from another detectorable to perform a calorimetric measurement of air show-ers. The RPCs are placed under the calorimeter, using itas an active shield to most of the electromagnetic showercomponent, and thus allowing the RPCs to assess directlythe muons in the shower. The calorimeter is sensitive bothto electromagnetic and muonic components of the shower.As such, it can be used to trigger more efficiently.The absorption of the shower electrons and gammas,and thus the purity of the muonic signal, increases withthe quantity of matter over the RPCs. It should be noted,however, that also the energy threshold for muon detectionincreases with the mass overburden. Nevertheless, giventhe typical muon energies, the effect is expected to besmall, as discussed below. Also, a mass overburden of theorder of few hundreds of g cm − is sufficient to grant theaccurate reconstruction of the muon component in a widerange of energies and zenith angles. The high segmenta-tion of the RPCs allows for the definition of fiducial areas.These fiducial areas, defined selecting RPC pads, are cho-sen to take into account the effective mass overburden overeach individual pad. This quantity depends on the arrivaldirection of particles, which can be estimated from thereconstructed shower direction, as discussed in the nextsection.The hybrid concept of MARTA not only allows for theseparate measurement of the shower components, givingadditional insight on the shower development mechanismsbut also enables better control of systematic uncertain-ties inherent to each detector through, for instance, cross-calibrations. RPCs are widely used in nuclear and particle physics ex-periments. These gaseous detectors have shown to be ro-bust while having a high particle detection efficiency. More-over, they are relatively low-cost detectors that can easilyoffer excellent spatial and time resolutions [8]. In the re-cent years, much R&D on the ability to run RPCs in out-door environments and with low maintenance has beenperformed [9,10]. The results demonstrate that RPCs area good candidate to be used in cosmic-ray experiments.RPCs are composed of millimetre-thick gaseous vol-umes that are contained by highly resistive parallel plates.High-voltage (HV) electrodes are applied to these plates,creating an intense and uniform electric field. The passageof ionising particles through the detector creates avalanchesof electrons which induce signals in the readout electrodes.The high resistivity of the plates prevents electrical dis-charges, which would affect the whole detector. . Abreu et al.: MARTA: A high-energy cosmic-ray detector concept for high-accuracy muon measurement 3
RPCs have significant advantages with respect to moreconventional detectors, as for example scintillators, partic-ularly concerning cost and feasibility. Moreover, the seg-mentation level is very flexible and constrained essentiallyby the readout. The signal pick-up electrodes are physi-cally separated from the sensitive volume. This approachallows us to achieve high-voltage insulation and gas tight-ness, reducing the number of breakthroughs considerably.An aluminium case is used to host the RPC, the DataAcquisition system (DAQ), the high-voltage and monitor-ing systems. Details on the design of the assembled andtested prototypes are presented in section 3. The MARTAdesign is based on a multi-gap gaseous volume. The us-age of thin gas gaps guarantees fast detector response toavalanche development, yielding very good time resolu-tions. Moreover, the multi-gap approach enhances detec-tion efficiency. The chambers require low gas flux and usetetrafluorethane (R-134a), a common refrigerator gas, andthe main component of the mixture used in most modernRPC installations.The MARTA concept also has advantages when com-pared to muon detectors buried underground, below an airshower detector. Firstly, the energy threshold for muonsremains essentially the same in the MARTA sub-detector(WCD + RPCs), while it would differ considerably be-tween underground detectors and surface ones. The im-plementation of MARTA would also consume much lesstime and resources. Finally, as mentioned before, the de-tection of the same particles by both detectors providesan invaluable tool both to understand the detector perfor-mance and to further exploit the shower physics. Other in-teresting possibilities for combined measurements includeprimary photon identification and air shower physics nearthe core.
A possible design of MARTA has been elaborated in de-tail for the Pierre Auger Observatory. Several prototypeshave been built and tested both in laboratory and at theobservatory site.At the Observatory, MARTA units could be installedat the entire surface array under the water-Cherenkov de-tectors (WCD). The WCD would remain unchanged, act-ing as shielding for the electromagnetic shower compo-nent, and sitting on top of a concrete structure hostingthe RPC modules. The water (1.2 m depth) and the con-crete (20 cm thickness) correspond to a mass overburdenof 170 g/cm . A schematic view of the MARTA imple-mentation for the Pierre Auger Observatory is displayed ifigure 1. RPC unit
The baseline configuration foresees four RPCs per tank.The structure of each chamber is as follows: – An area of 1.2 x 1.5 m , for a total of over 7 m ofRPC per WCD; Fig. 1.
The MARTA implementation for the Pierre AugerObservatory: the RPCs (in brown) are placed under the water-Cherenkov detector (in green) which provides active shieldingand trigger. The concrete support structure is shown in black. – A total of three resistive plates made of soda-limeglass, each 2 mm thick, mounted on top of each other; – The resistive plates are separated by 1 mm gaps forthe gas, making it a double gap chamber, filled withR-134a; – The detector is glued to an acrylic box of 3 mm thick-ness; – The readout plane is external and segmented in 8 x 8pick-up electrodes (pads), each with dimensions 14 ×
18 cm and separated by a 1 cm guard ring; – Coaxial cables transmit the signal induced in each padto the DAQ.In figure 2, a photograph and a scheme of the RPCunit are shown. The high-voltage electrode and the activedetector layers are enclosed inside the acrylic box, guaran-teeing high-voltage insulation and gas tightness. Only twobreakthroughs for the gas and two for the high-voltage arerequired. The asymmetric design, with the readout elec-trode at only one side of the gaps, has the advantage ofeasing the cabling by having it only at one side of thedetector.
Data acquisition system
A new front-end acquisition system [11] was developedfor MARTA RPCs. It is a hybrid system capable of count-ing active RPC pads and measuring the charge inducedin the detector. To comply with the strict demands ofMARTA operations in the field, the system was designedto be low-power consuming (a few Watts per RPC), com-pact (due to space limitations inside the aluminium case),stable and reliable, for low maintenance operation. Theserequirements made a system based on an application spe-cific integrated circuit (ASIC) the most appealing option.The Multi-Anode Readout Chip, MAROC 3 [12], is a lowpower (3.5 mW per channel) and compact (16 mm ) ASICthat fulfils all the stated criteria with 64 input channels,64 discriminated outputs and that can measure chargesup to 15 pC.The ASIC counts particles by applying a simple thresh-old to the signal after a pre-amplifier and a fast shaper. To P. Abreu et al.: MARTA: A high-energy cosmic-ray detector concept for high-accuracy muon measurement . .
14 cm18 cm
Fig. 2.
Photo and scheme of the RPC detector. Top: RPCwith visible cabled pad plane showing the detector spacial reso-lution. Bottom: scheme of the RPC box (1), readout plane (2),I C sensors layer (3), aluminium case base, cover and junction(4a, 4b, 5), and breakthroughs for gas and high-voltage (6). measure charge, a slow shaper is applied to the signal afterthe pre-amplifier. The slow shaper peak is then convertedto digital using an analog to digital converter (ADC) andtaken as the charge induced in the RPC. Both measure-ments have been tested [11]. The results show that theDAQ can measure the fast RPC signals without introduc-ing any unwanted inefficiencies in the setup.A low power field-programmable gate array (FPGA)processes and stores the ASICs digital outputs. It is alsoresponsible for all the communications using low voltagedifferential signalling (LVDS) lines. These lines, connectedto a concentrator central unit, are also used as trigger in-terface. Alternative communication with a PC via USBis also available and used mostly for debugging. Addi-tional features are available to increase the system flex-ibility, namely an analog acquisition of the sum of theRPC signals, environmental and power monitoring as wellas multi-purpose input and output ports. In figure 3, ascheme of the MARTA DAQ is shown.
Several prototypes of MARTA RPCs were built in theLIP-Coimbra workshops according to the aforementionedspecifications, following the requirements for counting in-dividual muons in extensive air showers. These have beentested indoors and outdoors, including at the Pierre AugerObservatory site, with the objective to study the long-term behaviour of the detector, evaluating its resilience tothe environmental conditions and monitoring the opera-tional parameters.
FGPA INTEL CYCLONE IV
EP4CE30F23I8LN
Flash ADC
USB FT2232H 100 MHz OSC GPIO LVDS RJ45 User PC Boundary Scan Active Serial Eprom I C Temperature Humidity SMA Trigger In
SMA Trigger Out RPC 64 Channels
ASIC MAROC 3
Power Whatchdog JTAG
Trigger In Trigger Out Clock Data
Fig. 3.
MARTA DAQ schematic representation with its maincomponents.
The initial laboratory tests were conducted with pureR-134a gas at a very low flow rate, of the order of 0.4cm /min or 1 kg/year, with the purpose of evaluating theimperviousness to humidity [9]. The background current,measured directly from the HV power supply, was cho-sen as the parameter to monitor the detector conditions,since it depends not only on the gas ionization rate andaverage charge per ionization, but also on additional con-tributions from leakage currents. There was no observedincrease in the background current with the detector im-mersed in water for two weeks. Moreover, detector opera-tion was achieved with a fraction of streamers below 10%and high detection efficiency for cosmic rays. The back-ground current was also monitored during seven monthsin a chamber placed outdoors, showing no strong depen-dence on the detector gas pressure or relative humidity.The indoor tests were prolonged for nine months, usinggas flow rates of 1, 4 and 12 cm /min. A detection effi-ciency of the order of 90% was observed in all cases [13].This value coincides with the fraction of the detector sen-sitive area that is covered by the readout pads, meaningan almost 100% intrinsic detection efficiency for cosmicmuons. Moreover, the efficiency dependence on the detec-tor reduced electric field was measured, a curve which isto be used for keeping track of the efficiency under the fi-nal measurement conditions. Also, for the first time, someprototypes were installed outdoors at the Pierre AugerObservatory site, where large daily variations in tempera-ture and pressure occur. Nevertheless, after four months ofoperation, a temperature variation in the detector (6 ◦ C)much lower than the ambient thermal amplitude (28 ◦ C)was observed, which is explained by the thermal inertia ofthe tank with its concrete support structure, as predictedby a thermal simulation.Later software developments allow to dynamically ad-just the applied high voltage in function of the averagepressure and temperature, to keep a constant value of thereduced electric field. The main results of these develop-ments, reported in [10] and [14], were the confirmation ofsmall daily thermal amplitudes in the detector and a re- . Abreu et al.: MARTA: A high-energy cosmic-ray detector concept for high-accuracy muon measurement 5 markable stability of the efficiency, at the level of 85%,measured during nearly one year of operation in the field.Constant and uniform efficiency across all the detectionarea, independent of the temperature or pressure gradi-ents, was also observed. After almost two years of fieldmeasurements at the Pierre Auger Observatory, it hasbeen shown that these RPCs can be operated in a harshoutdoor environment, and perform suitably for a cosmic-ray experiment.During the test phase an application of this detec-tor concept was started: the study of the response of anAuger water-Cherenkov detector (used for tests) to at-mospheric muons, by using a hodoscope built with RPCprototypes and custom-made electronics [15], with verysuccessful results [16]. Currently, a first shower-dedicatedmeasurement is in progress.
MARTA is a generic detector concept designed to fulfilthe requirements of large high-energy cosmic-ray exper-iments. A detailed implementation for the Pierre AugerObservatory has been developed and is used in this workto provide concrete and realistic performance expectation.A detailed simulation of this implementation of MARTAhas been performed using the GEANT4 toolkit [17], ac-cording to the baseline design described in section 3.1.EAS simulations for several primary species, zenith anglesand energies were undertaken using CORSIKA [18]. TheQGSJet-II.04 [19] and EPOS-LHC [20] have been used ashadronic interaction models.
A measurement of the number of muons can be obtainedfrom each individual MARTA detector unit. The first crudeestimator of the number of muons is the number of hits inthe pads within a fiducial area defined as the set of padsin a given shower that have a mass overburden greaterthan a chosen value. In the case of the MARTA configu-ration, the definition of the fiducial area required a slantmass greater than 170 g cm − , corresponding to the verti-cal mass overburden from the water tank and the concretetank support, defining a minimum criteria - with 100% offiducial volume for vertical events. The number of padswithin the fiducial area is then a function of the showergeometry. An example of a slant mass map, computed forincident particles at 40 ◦ zenith angle, is shown in figure 4.In this case 2/3 of the pads are contained in the fiducialarea. For a vertical shower, all the pads located below theshielding detector are contained in the fiducial area.The dependence of the energy threshold for muon de-tection with the mass overburden was studied using sim-ulated CORSIKA showers. The muon energy spectrum at1400 m above sea level peaks above 1 GeV and about 15%of these muons are absorbed after crossing the additional170 g cm − . Fig. 4.
Slant mass crossed before reaching the MARTA RPCs,under a 170 g/cm vertical mass overburden, for incident par-ticles at 40 ◦ zenith angle. The circle indicates the area coveredby the water tank. In figure 5, an example of a trace in the MARTARPCs is shown and compared with the traces in the water-Cherenkov detector. The muonic signal separation as afunction of the pad overburden is also shown.The RPC segmentation and the chosen readout elec-tronics allow for the digital counting of muons with a timeresolution of 5 ns and a position resolution limited by thepad dimensions. For the baseline design described in sec-tion 3, this corresponds to a maximum particle densityof 35.6 per m (assuming that all particles arrive at thesame time). This density is equivalent to that of muonsat about 500 m from the shower axis for a proton showerwith E = 10 . eV and θ = 40 ◦ . Due to the spread of themuon arrival times and the small dead time of the readout,pile-up effects become relevant only at smaller distancesand the number of muons can be successfully recovered,in the case exemplified, down to about 300 m by applyingdedicated algorithms [21]. For the purpose of measuringthe signal very near the shower core at the highest ener-gies, the analog channel is expected to provide countingcapabilities up to 20000 particles per RPC.The bias and the resolution of the reconstructed muonsignal have been estimated using the digital mode onlyand no pileup correction. A bias (due to electromagneticsignal contamination) of around 20% is, as expected, seendown to a distance to the core of 500 m. Below 500 m, thepileup effect starts to be visible and must be corrected for.At E = 10 eV, the resolution of the reconstruction ofthe muonic signal is below 20% up to 1000 m. For largedistances to the shower core, the muon signal resolutionis dominated by the low number of secondary particles.The atmospheric muon flux can be used to monitor andcalibrate the efficiency of the pads. Estimations of back-ground muons per pad were done using dedicated simu-lations. The atmospheric particle flux at ground (1400 m P. Abreu et al.: MARTA: A high-energy cosmic-ray detector concept for high-accuracy muon measurement Muon hits in RPC Tank traces h i t s A DC c oun t s Red – muons Blue – e+/e- Violet - photons t (ns) t (ns) ] pad overburden [g/cm ] - a v e r age den s i t y pe r pad [ m a v e r age h i t s pe r pad T h r . f o r f i du c i a l a r ea eV Iron QGSJetII.04 =38 deg θ =41 deg φ Station at r = 448 m μ Fig. 5.
Top: Example of a MARTA simulated trace – thesimulated WCD and RPC traces are shown; Bottom: Separa-tion of the electromagnetic and muonic shower components ina simulated MARTA unit. a.s.l.) was obtained through CORSIKA simulations. Thesesimulations were obtained injecting the particles accord-ing to the known primary cosmic ray energy spectrum,with energies ranging from E = 10 eV to E = 10 eV.The output of these simulations was afterwards injectedinto a MARTA Geant4 dedicated simulation, leading to anestimate of the number of muons able to reach the RPCpads. The rate is obtained using the expected number ofmuons in the water-Cherenkov detector, taken from [22].From this study, one can conclude that the hits due toatmospheric particles are of the order of 5-7 Hz.The results, presented in figure 6, show that the num-ber of atmospheric particles in each pad per minute is x [m]-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 y [ m ] -2-1.5-1-0.500.511.52 ] - pa r t i c l e s [ m i n Fig. 6.
Number of atmospheric particles in each RPC pad perminute. Result obtained from a dedicated simulation at 1400 ma.s.l. (see text for details). r [m] - -
10 1 R P C H i t s D en s i t y f o r one e v en t Fig. 7.
LDF of one QGSJET-II.04 proton shower with E =10 . eV and θ =38 ◦ . RPC hits densitiy are given in squaremeters. higher than 300, which means that a statistical precisionof 1% is reached every half an hour. Many possibilities arise from the knowledge of the arrivalposition and time of individual muons in MARTA stations.Some of the main aspects are outlined below.
From muon density to composition
In figure 7 is shown the lateral distribution function (LDF)obtained by the MARTA RPCs for a simulated protonshower with E = 10 . eV and θ = 38 ◦ simulated withCORSIKA and QGSJET-II.04. The LDF, which repre-sents the density of particle at a given distance from theshower core, was, in a first approximation, parametrizedby [1]: ρ LDF ( r, β ) = ρ (cid:16) r (cid:17) β · (cid:18)
700 + r (cid:19) β (1) . Abreu et al.: MARTA: A high-energy cosmic-ray detector concept for high-accuracy muon measurement 7 Fig. 8.
Top: Average MARTA LDF (over 300 events) fordifferent primaries with QGSJET-II.04 at E = 10 eV and θ = 38 ◦ ; Bottom: ρ mean and β mean . where the parameters ρ and β represent, respectively,the normalization and the shape. In the fit, β was fixed to-2.1.In figure 8(top) the average LDF for proton, helium,nitrogen and iron QGSJET-II.04 showers at E = 10 eV and θ =38 ◦ is shown. This figure was obtained using300 showers for each primary. A net separation is visi-ble. Considering an energy bin of log( E/ eV) = 0 . E = 10 eV, and taking the ultra-high-energy cosmic rayflux [23], an experiment such as the Pierre Auger Obser-vatory would be able to reach this event statistics in lessthan half a year. It is then possible to fit such averagedistributions taking as free parameters both ρ and β ( ρ mean and β mean ). In figure 8(bottom), ρ mean and β mean for proton and iron primaries at E = 10 eV and E = 10 . eV are shown. A clear separation between pro-ton and iron is observed, showing that β mean may be apowerful variable to assess the beam composition. The im-pact on the choice of the high-energy hadronic interactionmodel was evaluated repeating the analysis using EPOS-LHC instead of QGSJet-II.04. Although the results for β and ρ obtained using EPOS-LHC are slightly higherthan from QGSJet-II.04 (about 15% for ρ and 2% for β ), the discrimination power regarding primary mass com-position is not altered significantly.Fitting the individual muon LDF distributions, andfixing β as a function of the zenith angle, one obtains for / eV ) MC log( E
19 19.5 20 - ) M C r / M A R T A r ( s protonheliumnitrogeniron / eV ) MC log( E
19 19.5 20 æ - M C r / M A R T A r Æ protonheliumnitrogeniron Fig. 9. ρ MARTA1000 resolution (top) and bias (bottom) for dif-ferent primaries at θ = 38 ◦ , for E = 10 eV and E = 10 . eV obtained with QGSJET-II.04. Similar results were obtainedwith the EPOS-LHC interaction model. each shower ρ , the measured muon density at 1000 mfrom the core ( ρ MARTA1000 ). The β parameter was obtainedfrom a mixed composition simulation (50% proton and50% iron). The bias and resolution with respect to the truemuon density are shown in figure 9, for different primarytypes at θ = 38 ◦ , for E = 10 eV and E = 10 . eV. Thisfigure shows that the bias in the muon density estimator ρ MARTA1000 is nearly energy-independent. It has also beenobserved that the bias decreases with the increasing zenithangle and has no significative dependence on the hadronicmodel considered. In figure 10 we show the distributionsof ρ MARTA1000 for p and Fe at different angles and for twoenergy values.
From muon production depth to composition
In the muon production depth (MPD) technique, the showergeometry is combined with the arrival times of muons toreconstruct the longitudinal profile of muon production[24,25]. While the reconstruction of the MPD is the sameas for the Auger WCDs, the direct detection of muons
P. Abreu et al.: MARTA: A high-energy cosmic-ray detector concept for high-accuracy muon measurement
MARTA1000 r M A R T A r d N / d (cid:215) / N p Fe (cid:176) = 21 q log(E/eV) = 19.0, MARTA1000 r M A R T A r d N / d (cid:215) / N p Fe (cid:176) = 38 q log(E/eV) = 19.0, MARTA1000 r M A R T A r d N / d (cid:215) / N p Fe (cid:176) = 52 q log(E/eV) = 19.0, MARTA1000 r M A R T A r d N / d (cid:215) / N p Fe (cid:176) = 21 q log(E/eV) = 19.8, MARTA1000 r M A R T A r d N / d (cid:215) / N p Fe (cid:176) = 38 q log(E/eV) = 19.8, MARTA1000 r M A R T A r d N / d (cid:215) / N p Fe (cid:176) = 52 q log(E/eV) = 19.8, Fig. 10. ρ MARTA1000 distributions for proton and iron primaries at different zenith angles (from left to right, θ = 21 ◦ , ◦ , ◦ with E = 10 eV (top plots) and E = 10 . eV (bottom plots) obtained with QGSJET-II.04. Similar results were obtainedwith the EPOS-LHC interaction model. opens great possibilities to extend the use of this tech-nique.The maximum of this profile is known to be composition-sensitive variable [26] which could add relevant informa-tion about the hadronic interaction processes that rulethe shower development. As an example, the depth ofthe maximum of the MPD for proton and iron showersis shown in figure 11, for θ = 38 ◦ and two different pri-mary energies. The event resolution stays the same as itis dominated by the number of muons entering the recon-struction, which will remain practically the same in eachstation. The largest impact comes from the reduction ofsystematics which can be lowered by ∼ / cm , since theuncertainties associated to the time response of the WCDare reduced by the RPCs to negligible values (comparedto the remaining sources of systematics). In addition, theseparation of the electromagnetic and muonic shower com-ponents at ground allows the MPD technique to be appliedto stations which are closer to the core and also in morevertical showers, which permits the enlargement of the en-ergy window and angular window of applicability. Finally,it is worth noting that the angular dependence of MPDcan give information about the EAS muon energy spec-trum [27], providing an additional handle to evaluate andconstrain hadronic interaction models. The redundant use of the two sub-detectors, detectingthe same portion of the shower at consecutive depths,
Fig. 11. X µmax distributions for simulated showers initiatedby protons and iron at θ = 38 ◦ with E = 10 eV (left) and E = 10 . eV (right) obtained with QGSJET-II.04. opens several possibilities for systematic studies and cross-calibration, reducing global uncertainties associated withthe measurements. Additionally, there is the possibility toexplore new variables and increase the amount of recon-structed information by cross-analyses between the twosub-detectors, benefiting for example the precision of theenergy spectrum measurement or the sensitivity to pho-ton primaries, which have a much-reduced muon contentcompared to hadron primaries [28].While a detailed exploitation of such possibilities is outof the scope of this concept paper, we give as an examplethe impact on the energy reconstruction of correlating theinformation of the two detectors to disentangle the muonicand electromagnetic shower components. The electromag-netic and muonic contents of the shower can be estimated . Abreu et al.: MARTA: A high-energy cosmic-ray detector concept for high-accuracy muon measurement 9 R[m] S D W C T s i gna l [ VE M ] R[m] A ll R P C [ c oun t s ] R[m] R P C - F i du c i a l S i gna l [ c oun t s ] Fig. 12.
Combined fit for one event (QGSJET-II.04 proton with θ = 38 ◦ and E = 10 eV) in the tank and in the RPC’s totaland fiducial area. The large black dots are the measurements used in the fit. The black line is the sum of the two dashed linescorresponding to the muonic (blue) and electromagnetic (red) LDFs. The true (simulated) information about the muonic andelectromagnetic signals is shown as smaller dots. performing combined fits to the LDFs measured both inthe RPC fiducial and non-fiducial areas as well as in thewater-Cherenkov detector. The fitted parameters S em and S µ should be proportional, respectively, to the total elec-tromagnetic energy and the total number of muons in theshower. In figure 12 an example of such a fit is shown.Taking advantage of the excellent timing of RPCs, thetrajectory of single EAS muons can be reconstructed us-ing an MPD-like algorithm. The RPC segmentation allowsone to determine the path of the muon through the SDtank, and the track length can be reconstructed with aresolution of around 6%. With this information, one cancompare the expected muon signal in the WCD with thecollected one, constraining in this way either the WCD properties or the muon energy spectrum. The usual WCDcalibration with atmospheric muons can be improved re-quiring a coincidence with MARTA RPCs. These studies,combined with the data collected by an RPC hodoscopeinstalled in a test WCD at the Observatory site can helpus understand and improve the monitoring tools for eachwater-Cherenkov detector in the field equipped with aMARTA unit [16]. The MARTA concept for the direct measurement of muonsin air showers has been presented. An RPC-based detector measures muons with very good space and time resolutionand is shielded by a detector able to absorb and measurethe electromagnetic component of showers. The combina-tion of the two detectors in a single, compact detector unitprovides a unique measurement that opens rich possibili-ties in the study of air showers.MARTA was designed to fulfil the requirements of largehigh-energy cosmic-ray experiments, proposing low costunits able to operate stably and at high efficiency in out-door environments with minimal maintenance and lowpower consumption. To provide a concrete and realisticperformance expectations a detailed implementation forthe Pierre Auger Observatory has been developed.Extensive R&D on RPCs for MARTA was carried out.Long-term performance tests confirm stable operation with ∼
90% efficiency and low gas flow (0 . / min). Abouttwenty MARTA RPC units are installed and taking datain the laboratory and at the field at the Pierre Auger Ob-servatory. A MARTA mini-array, equipping eight water-Cherenkov detectors with four RPC modules each, willbe installed at the Pierre Auger Observatory site. EachMARTA station will count muons with a time resolutionof 5 ns and a position resolution limited by the pad dimen-sions. In each individual station, the number of muons willbe reconstructed with a bias and resolution below 20% ina wide range of distances to the shower core.In a MARTA array, many possibilities arise from theknowledge of the arrival position and arrival time of muons.In particular, composition sensitivity is greatly enhanced.Simulation studies show that the lateral density of muonsat 1000 m from the core is measured with a resolution of10-20% and a bias below 10%, and that differences in thelateral distribution of muons provide a clear separationbetween proton and iron. In addition, the MPD techniqueis applicable to a wider range of zenith angles, and theassociated systematics greatly benefit from the excellenttime resolution. The use of two systems detecting the sameportion of the shower at consecutive depths opens severalpossibilities for systematic studies, cross-calibrations andcross-analyses, benefiting, for example, the sensitivity tophoton primaries.The MARTA concept for the direct and precise mea-surement of muons in air showers may find applicationin future cosmic-ray experiments, but also in dedicatedarrays for high-energy gamma rays. In this case, RPCsprovide very good space and time resolution, while thecalorimetric detector ensures trigger efficiency and back-ground rejection. Combined, they can bring the sensitiv-ity of air shower gamma-ray arrays to lower energies [29].While the concept applies, developments concerning oper-ation at very high altitudes will have to be addressed. Acknowledgements
The authors greatly acknowledge all the fruitful discus-sions held with Pierre Auger colleagues. We are also verythankful for all the support at the site and the use ofPierre Auger Observatory infrastructure, in particular tothe Observatory staff for the assistance and dedication. The authors also thankfully acknowledge the support ofFAPESP/FCT (2014/19946-4). S. Andringa, L. Cazon,R. Concei¸c˜ao, R. Luz, R. Sarmento want to thank fund-ing by Funda¸c˜ao para a Ciˆencia e Tecnologia (PD/ BD/113488 /2015, SFRH/ BPD/ 84304/ 2012). J. ˘R´ıdk´y, P.Tr´avn´ı˘cek, J. V´ıcha acknowledge the financial support un-der Grant No. MSMT CR LTT18004, LM2015038 andCZ.02.1.01/ 0.0/ 0.0/ 16 013/ 0001402.
References
1. A. Aab, et al., Nucl. Instrum. Meth.
A798 , 172 (2015).DOI 10.1016/j.nima.2015.06.0582. T. Abu-Zayyad, et al., Nucl. Instrum. Meth.
A689 , 87(2013). DOI 10.1016/j.nima.2012.05.0793. A. Aab, et al., Phys. Rev.
D90 (12), 122006 (2014). DOI10.1103/PhysRevD.90.1220064. A. Letessier-Selvon, P. Billoir, M. Blanco, I.C. Mari¸s,M. Settimo, Nucl. Instrum. Meth.
A767 , 41 (2014). DOI10.1016/j.nima.2014.08.0295. A. Aab, et al., Phys. Rev.
D91 (3), 032003 (2015). DOI 10.1103/PhysRevD.91.059901,10.1103/PhysRevD.91.032003.[Erratum: Phys. Rev.D91,no.5,059901(2015)]6. A. Aab, et al., Phys. Rev. Lett. (19), 192001 (2016).DOI 10.1103/PhysRevLett.117.1920017. R.R. Prado, R. Concei¸c˜ao, M. Pimenta, V. de Souza, As-tropart. Phys. , 40 (2016). DOI 10.1016/j.astropartphys.2016.07.0038. P. Fonte, in Proceedings, 11th Workshop on Resistive PlateChambers and Related Detectors (RPC2012): Frascati,Rome, Italy, February 5-10,2012 , vol. RPC2012 (2012),vol. RPC2012, p. 0339. L. Lopes, P. Fonte, M. Pimenta, in
Proceedings, 11th Work-shop on Resistive Plate Chambers and Related Detectors:Frascati, Rome, Italy, 5-10 Feb 2012 , vol. 8 (2013), vol. 8,p. T03004. DOI 10.1088/1748-0221/8/03/T0300410. L. Lopes, et al., in
Proceedings, 13th Workshop on ResistivePlate Chambers and Related Detectors (RPC2016): Ghent,Belgium, February 22-26, 2016 , vol. 11 (2016), vol. 11, p.C09011. DOI 10.1088/1748-0221/11/09/C0901111. P. Assis, et al., To be published
TBD , TBD (2017)12. S. Blin, P. Barrillon, C. de La Taille, in
Proceedings, 2010IEEE Nuclear Science Symposium and Medical ImagingConference (NSS/MIC 2010): Knoxville, Tennessee, Oc-tober 30-November 6, 2010 , vol. 2010 (2010), vol. 2010,pp. 1690–1693. DOI 10.1109/NSSMIC.2010.587406213. L. Lopes, et al., JINST (10), C10023 (2014). DOI 10.1088/1748-0221/9/10/C1002314. R. Concei¸c˜ao, et al., PoS ICRC2017 , 379 (2017)15. P. Assis, P. Brogueira, M. Ferreira, R. Luz, L. Mendes,JINST (08), T08004 (2016). DOI 10.1088/1748-0221/11/08/T0800416. P. Assis, PoS ICRC2015 , 620 (2016)17. S. Agostinelli, et al., Nucl. Instrum. Meth.
A506 , 250(2003). DOI 10.1016/S0168-9002(03)01368-818. D. Heck, G. Schatz, T. Thouw, J. Knapp, J.N. Capdevielle,
CORSIKA: A Monte Carlo code to simulate extensive airshowers (1998)19. S. Ostapchenko, Phys. Rev. D , 014018 (2011). DOI 10.1103/PhysRevD.83.014018. URL http://link.aps.org/doi/10.1103/PhysRevD.83.014018 . Abreu et al.: MARTA: A high-energy cosmic-ray detector concept for high-accuracy muon measurement 1120. T. Pierog, I. Karpenko, J.M. Katzy, E. Yatsenko,K. Werner, Phys. Rev. C , 034906 (2015). DOI10.1103/PhysRevC.92.034906. URL https://link.aps.org/doi/10.1103/PhysRevC.92.034906
21. A.D. Supanitsky, A. Etchegoyen, G. Medina-Tanco,I. Allekotte, M.G. Berisso, M.C. Medina, Astropart. Phys. , 461 (2008). DOI 10.1016/j.astropartphys.2008.05.00322. X. Bertou, et al., Nucl. Instrum. Meth. A568 , 839 (2006).DOI 10.1016/j.nima.2006.07.06623. J. Abraham, et al., Phys. Lett.
B685 , 239 (2010). DOI10.1016/j.physletb.2010.02.01324. L. Cazon, R.A. Vazquez, E. Zas, Astropart. Phys. , 393(2005). DOI 10.1016/j.astropartphys.2005.01.00925. A. Aab, et al., Phys. Rev. D90 (1), 012012 (2014). DOI10.1103/PhysRevD.92.019903,10.1103/PhysRevD.90.012012,10.1103/PhysRevD.90.039904. [Erratum: Phys.Rev.D92,no.1,019903(2015)]26. S. Andringa, L. Cazon, R. Concei¸c˜ao, M. Pimenta,Astropart. Phys. , 821 (2012). DOI 10.1016/j.astropartphys.2012.03.01027. J. Espadanal, L. Cazon, R. Conceio, Astropart. Phys. ,32 (2017). DOI 10.1016/j.astropartphys.2016.11.00328. R. Engel, D. Heck, T. Pierog, Ann. Rev. Nucl. Part. Sci. , 467 (2011). DOI 10.1146/annurev.nucl.012809.10454429. P. Assis, et al., Astropart. Phys.99