HiRes and TA Composition measurements

In order to clarify the origin of ultra high energy cosmic rays (UHECRs), it is very important to determine the mass composition. The most effective strategy to determine the mass composition is Xmax technique. Xmax is the atmospheric depth of air shower maximum measured by fluorescence detectors (FDs). HiRes has reported Xmax measurement by FDs which indicated proton dominated mass composition. Now, Telescope Array (TA) experiment has also measured UHECRs with FDs. In this presentation, the detail of mass composition analysis and result of TA experiment will be reported and compared with HiRes experiment.


High Resolution Fly's Eye
The High Resolution Fly's Eye Experiment (HiRes), located on Dugway Proving Grounds in western Utah, is a UHECR observatory operated from 1997-2006. The experiment consisted of 2 sites, HiRes-I and HiRes-II, separated with 12.6 km. It can measure extensive air showers (EASs) of UHECRs stereoscopically by the atmospheric fluorescence technique. The fluorescence telescope in each site has spherical mirror with area of 5.2 m 2 to collect fluorescence light. At the focus of it, a cluster of photomultiplier tubes (PMTs) arranged in 16 × 16 array is mounted. Each PMT covers one degree sky and field of view (FOV) of the telescope is 14 • × 16 • . The telescopes are arranged as a ring to look around: HiRes-I has 21 mirrors in one ring with elevation angle of 3-17 • and HiRes-II has 42 mirrors in two rings with elevation angle of 3-31 • . For the data acquisition (DAQ) system, HiRes-I uses sample and hold electronics, on the other hand, HiRes-II adopts FADC system.

Telescope array
The Telescope Array (TA) is a hybrid detector consisting of a Surface Detector (SD) array and Fluorescence Detectors (FDs) to observe UHECRs. It is located in the western desert of Utah, ∼100 km away to the south from HiRes site. The SD array consists of 507 three square meter plastic scintillation counters arranged on a 1.2 km grid for a detection area of ∼700 km 2 , 7 times larger than AGASA. There are three FD stations overlooking the SD array. Two of the FD stations, located at Black Rock (BR) and Long Ridge (LR), contain 12 newly developed FD telescopes each, the other, Middle Drum, contains 14 telescopes transferred from HiRes-1. TA has been taking data with all detectors since May 2008. The optics of FD at BR and LR consists of 18 spherical mirrors with 6 m curvature radius, 3 m diameter and the mirror area is 6.8 m 2 . At the mirror focus, there is a PMT cluster which contains 256 PMTs with BG3 (Schott) filter and is covered with acrylic window (paraglas). The FOV of each telescope is 18.0 • in a e-mail: tame@icrr.utokyo.ac.jp b For the full authorlist see Appendix "Collaborations" in this volume. This is an Open Access article distributed under the terms of the Creative Commons Attribution License 2.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Article available at http://www.epj-conferences.org or http://dx.doi.org/10.1051/epjconf/20135304005 EPJ Web of Conferences azimuth and 15.6 • in elevation, each PMT looks 1.1 × 1.0 • . The total FOV of each station is 108 × 30 • . Air showers of which primary energy is 10 19 eV and shower core is 30 km apart from the FD telescope can be triggered, thus such air showers whose cores are within the SD array can be observed by 2 FD stations, stereoscopically.

Event reconstruction
For the mass composition analysis, TA FD and HiRes use stereo air shower events measured by 2 FD sites. The accuracy of stereo shower geometry is much better than that of monocular. In the case of TA FD, the determination of arrival direction is 1.9 deg for stereo mode, or 6.2 deg for monocular mode. The procedure of event reconstruction for stereo events consists of mainly tow parts, geometrical reconstruction and shower profile reconstruction as following, especially for TA FD case. For the geometrical reconstruction, at first, shower detector plane (SDP) of each FD station is determined, which includes shower axis and the detector position which is defined as the center of FD station. Next, shower axis is determined as the intersection of the SDPs determined as above.
Once the geometry of the shower axis is determined, profile of shower development is reconstructed by Inverse Monte Carlo (IMC) technique [1] using the intensity of injection photons at the detector. IMC takes contribution of direct or scattered Cherenkov light into account. Shower profile is assumed to be fit with Gaisser-Hillas function and energy is estimated as an integration of it. For proton with energy above 10 19 eV, the energy determination is 1.6 ± 7.6% and the X max accuracy is −6.2 ± 22 g/cm 2 . The atmospheric profile used is the monthly average of the radiosonde launched at Elko, Nevada, which is the closest launch site to TA. The distribution of aerosols was measured at the TA site by LIDAR [2]. The total energy deposited which is calculated by integration of the Gaisser-Hillas function along the shower axis is 93% for protons and 89% for iron.

X max technique
The longitudinal development of a UHECR EAS depends strongly on its primary energy and particle type. The depth in the atmosphere at which the number of particles in the shower reaches a maximum, X max , is a good indicator of primary particle type. Since FDs observe longitudinal development of air showers, this technique has the advantage over SDs of measuring the energy calorimetrically and determining primary particle type. However, primary particle type of EAS cannot be determined shower by shower due to fluctuations in development of individual showers. Thus, the mass composition should be determined on a statistical basis by comparing the X max distribution of the data and expected from a Monte Carlo (MC) simulation. Here, it should be noted that the uncertainty of the MC depends strongly on hadron interaction models that have been extrapolated from measured cross sections at much lower energies.
An expected distribution of X max is estimated with a MC shower simulation using CORSIKA [3]. QGSJET-01 [4], QGSJET-II [5] and SIBYLL [6] are used for the hadronic interaction models. Primary particles are assumed to be either protons or iron nuclei. The left side of Fig. 1 shows the averaged X max of each energy based on only the shower MC simulation. As energy increases, the X max of air showers increase. At a given energy, the X max of a light primary particle will be deeper than that of a heavy primary particle. Since the FDs only can see showers in certain geometric regions, X max may be either above the FOV or below it, or it may be inside the field of view but the FD cannot reconstruct the shower (for instance, the shower may be coming nearly directly toward the FD). Moreover, reconstruction bias may affect X max distribution, systematically. This means that the distribution of observed X max can be different from the expected distribution estimated only by shower simulation such as CORSIKA. This means that actual FD configuration and reconstruction method should be take into account.  The right side of Fig. 1 shows the averaged X max in which the actual TA FD configuration are taken into account using detector simulation and the same reconstruction procedure as data is applied. Averaged X max rails are shifted down systematically 20 g/cm 2 due to detector effect and reconstruction bias in TA FD case. In the case of HiRes, we can see the similar systematic shift of 16 g/cm 2 and it comes from detector effect called as acceptance bias. Reconstruction bias can be negligible in HiRes case. As above, X max technique depend on the detector simulation, which should be understood well. Agreement between data and MC can be one of good indicator to evaluate our simulation performance. Figure 2 shows the comparison of data and MC above the energy of 10 18 eV in HiRes case for distribution of several parameters of zenith angle, height of X max , first viewed depth and last viewed depth for two primary particle type of proton and iron nuclei. Hadronic interaction model shown here is QGSJET-II. We can see good agreement between data and MC, especially for proton primary model. In case of iron, differences can be found. The comparison of zenith angle distribution of data and MC with TA SD shows similar tendency with much higher statistics. Figure 3 shows the TA FD stereo case for several parameters of zenith angle, azimuth angle, core location, impact parameter (Rp) and track length. As above, we check carefully the detector simulation by comparison of data and MC to evaluate our analysis. Figure 4 shows the X max distributions compared with MC based on QGSJET-II above energy of 10 18 eV. Left of Fig. 4 are HiRes case compared with proton (upper) and iron (lower) model. Right of Fig. 4 is TA FD stereo, proton (red) and iron (blue). We can see good agreement between data and MC with proton primary model for each experiment. We applied Kolmogorov-Smirnov (KS) test to estimate the degree of agreement of the X max distribution in each energy region. Figure 4 are the results of KS test. Each of them, in the whole energy regions, the X max can be compatible with proton primary model. At the low energy, iron model can be excluded but at high energy, iron model can be also compatible with data due to low statistics.

X max distribution
Comparison of averaged X max plot is obvious way to determine the mass composition of UHECRs. Figure 6 show the results of averaged X max of HiRes (left) and TA FD stereo (right). In each figure, X max rails are derived from shower simulation with detector simulation; the same as the left of Fig. 1 for TA FD case. The results of both experiments show that data is consistent with proton primary model, especially QGSJET model in whole energy region.

SUMMARY
In this presentation, the HiRes and TA results of UHECR mass composition is presented. Mass composition analysis derived from X max technique is affected by acceptance or reconstruction bias which is well understood by comparison of data and MC. X max distribution is consistent with proton primary model with QGSJET model above 10 18 eV. X max distributions for each energy are tested by KS test and P values show that proton model is compatible with proton model for whole energy region. On the other hand, iron model can be excluded, but below 10 19.4 eV for TA case. Averaged X max shows that data is consistent with proton model, especially QGSJET model. Both TA and HiRes results of the UHECR mass composition are consistent with proton model from not only the distribution of X max but also averaged X max above ∼10 18.2 eV.     Figure 6. Averaged X max of MC and Data. MC prediction rails of proton or iron primary include biases estimated by detector simulation: HiRes (left), TA FD stereo (right). Hadronic interaction models are QGSJET-01, QGSJET-II and SIBYLL.