CC BY
29
A study of UHECR composition using multi-component analysis of the Yakutsk array data
UDK 537.591.15
A. A. ívanou, S, P. Knurenko, A. A. Lagutin, R.I. Raikin and I. E. Sleptsov
A study of UHECR composition using multi-component analysis of the Yakutsk array data
Longitudinal and radial development parameters of extensive air showers measured with charged particle and air Gherenkov light detectors of the Yakutsk array are considered. To estimate the average mass composition of the primary particles, the two-dimensional spread of shower parameters is analyzed in comparison with cascade simulations using CORSIKA/QGSjet code. As a result, three fractions of the primary nuclei groups are estimated with the reconstruction error below 30% in the energy range En 6 Í10 .1019) eV.
1. Introduction
Cosmic ray (CR) mass composition in ultra-high energy region (UHE) remains unknown because of still insufficient event number of extensive air showers (EAS) detected. Several attempts have been made to determine the mean CR mass using measurements of the depth of shower maximum, X{E), or the muon content at the observation level, as a function of energy [1]. At present, there is some consensus for a composition becoming as if proton dominated around 1019 eV.
In order to improve the informativeness of data, other methods were proposed, mostly combining two or more shower parameters for multi-component analysis of the dataset [2—6]- For example, in papers [5, 6] a method was offered to evaluate the average mass composition using a combination of EAS parameters - Xmax and the density of electrons and muons at the distance 600 m from a shower core, peoo-
This method allows to reliably distinguish showers initiated by primary protons and iron nuclei analyzing a distribution of the event rate on the plane (Xmax,/96oo) supposed to be similar to that given by CORSIKA/QGSjet simulations.
Our aim here is to apply this method to the data of the Yakutsk array, namely Xmax and pmo distribution in the energy range Eo £ (1017.10m) eV, in order to estimate the fractions of nuclei divided into three groups in the primary beam.
2. Method of analysis
In this work two-dimensional probability distributions of two experimental observables (J£„u,x and peoo) are used. According to the procedure suggested in [6], the normalized variables r and p
This paper is an extended version of contribution to 20(h European Cosmic Ray Symposium in Lisbon, Portugal, September 5th-8th 2006.
have been introduced instead of Xmax and /jgoo-
_ __ --max / ' max \ olX max
Ig ¿600 _ / lg ¿>600 \
cr(lgP60o) VOgPGOn)/ '
where a denotes standard deviations and brackets mean the averaging.
Experimental data of the Yakutsk array are compared to the distributions obtained using CORSIKA code v. 6.0 employing QGSjet hadronic interaction model. We have generated 1500 artificial showers of fixed primary particle energies (1037,10JS, 1019 eV) initiated by five type of nuclei: p, He, C, Si, Fc\ For each of considered energy and kind of the primary nucleus two dimensional distributions of showers /(r, p) were analysed
As it was shown, it is possible to distinguish zones on the (r, p) plane, which allows to clearly separate showers initiated by light (p + He), medium (C) and heavy (Si + Fe) nuclei [6]. This method was subjected to validation by simulation of experimental procedure of primary mass reconstruction assuming different models of composition. The error in reconstruction of light, medium and heavy nuclei fractions does not exceed -30% at £„ € (T017,1019) eV.
3. Data analysis and discussion
Experimental data of the Yakutsk array are used to derive A"ma;< and psoo- Charged particle density at 600 m is an immediate output of our scintillation detectors while JVmax is reconstructed basing on air Cherenkov light measurements |7j.
In order to minimize systematic uncertainties the following data selection criteria were applied: i) shower axes are within the array area; at Eo < 1018 eV an additional cut was applied for showers with
A. A. Ivanov, S. P. Knurenko, A. A. Lagutin et al.
axes close to the array periphery; ii) shower detection efficiency is greater than 0.9; iii) zenith angle doesn't exceed the Cherenkov light detector aperture; iv) light extinction in atmosphere is less than 0.41 at A = 420 nm. As a result, 587 showers with primary energy above 1017 eV observed during the period of 1993-2005 were selected.
In Figure 1 the two-dimensional distribution of showers in the plane (Armax,p6oo) is shown. Here !,„„ characterizes a maximum depth of individual showers and p6oo is the density of particles at the observation level. One can see from Figure that there is some obvious correlation between the shower maximum and the charged particle density.
According to the analysis of probability distributions f(r,p) reconstructed using the simulated showers database [6], we roughly divide the final spot by two lines mi and m2 into three zones corresponding to primary particle type at fixed primary energy. In the first and third zones light (p + He) and heavy (Si + Fe) nuclei initiated showers respectively are well separated from each other with the probability ~ 90%. In the medium zone showers initiated by different nuclei are strongly intermixed. Our calculations based on different assumptions about primary composition show that in medium zone the fraction of carbon-initiated showers is ~ 50% [6].
Figure 2 presents the results of multi-component analysis of (r,p) distribution derived from the Yakutsk array data. The analysis was carried out in three energy bins with average energies 2.4 x 1017 eV, 9.8 x 1017 eV and 4.8 x 1018 eV. Lines represent borders of the above mentioned zones.
p(600) (1 / m")
Figure 1. A distribution of the depth of shower maximum, Xmhx, and the charged particle density at 600 m from the shower core, P60O1 observed at the Yakutsk array.
i A/ VV "V #
-2
/«•.7*?* rv
-■ m, (p + He) ■ m (Si + Fe)
- E =4.8 * 10" eV
-2
Figure 2. Normalized experimental data (r,p) in different energy bins, mi is a borderline between light and medium nuclei initiated shower groups (p + He and C), while m? is a borderline between medium and heavy nuclei groups (C and Si + Fe). Average energies are indicated for each energy bin.
Namely, line m\ is dividing light and medium nuclei initiated showers, and m^ is between medium and heavy nuclei zones. As it is seen from Figure 2, the points are spread over zones non-uniformly. The main part is located in the first and the second zones, in percentage, the zone-2 sample is twice larger than that of zone-!.
We assume that carbon nuclei are responsible for a half of all events in the second zone. Significant contribution to this zone is made by nuclei of the first group (up to 30%) and by nuclei from the third group 15%). The remaining part of the set (16-27%) falls into the zone-3 where the heavy nuclei (silicon and iron) induced showers are concentrated.
In the Table 1 the shower samples are given in the energy bins and three zones, in the last column the average logarithm of the primary nuclei mass is given. The fraction of light nuclei increases from 50% to 53%, and a fraction of medium nuclei is increasing with energy from 23% to 31%. At the same time, the fraction of heavy nuclei decreases from 27% to 16% in the primary CR flux. These changes result in the average composition, (ln/l), changing from a heavy to lighter mix as the energy
increases, as it is seen from the last column.
All these conclusions are inferred under the necessary condition of similarity of observed and simu lated distributions in the (r,p) plane. The recognition error of the nuclei groups in the whole energy range E0 e (1017,1019) eV does not exceed 30% in this case.
The results obtained here using multi-component analysis of Jima.x and p6oo distribution are in qualitative agreement with our previous conclusion (derived by other methods) concerning the fraction of protons and helium nuclei in the primary beam increasing with energy [4], [8].
The further improvement of this technique could be implemented as suggested in [9] involving other observables sensitive to the primary mass, for example, mean square radius of lateral distribution of electrons/charged particles or some other parameter. This will be the subject of our forthcoming publication.
Acknowledgment
This work is partially supported by RFBR (grant #06-02-16973), MSE (NSh-#75i4.2002J| and INTAS (#03-51-5112).
Table 1
EAS event numbers
{E0), eV n Light Medium Heavy (In A)
2.4 • 1017 177 89/0.50 40/0.23 48/0.27 2.05 ± 0-61
9.8 • 1017 266 133/0.50 70/0.26 63/0.24 1.95 ±0.59
4.8-1018 144 77/0.53 44/0.31 23/0.16 1.68 ±0.50
References
1. Nagano M. and Watson A. A.//Rev. Mod. Phys. - 2005. - 72. - P.689.
2. Nikolsky S. I., Stamenov I. N. and Ushev S. Z.//JETP. - 1984. - 60,- P.10.
3. Wibig T. and Wolfendale A. W'./fJ. Phys. G: Nucl. Part. Phys. - 2000. - 26. - P.825.
4. Knurenko S. P. et a\.//Nucl. Phys. B (Proc. Suppl.) - 2006. •- 151,- P. 92.
5. Abu-Zayyad T et al. Preprint astro-ph/9911144. - 1999,
6. Lagutin A. A. and Stanovkina N. V.//Izv. AGU.
- 2005. - 5. - P.76 (in Russian).
7. Dyakonov M. N. et al .//Proc. 16th JCRC (Kyoto). - 1979. - Í6. - P. 174.
8. Dyakonov M. N. et al.//JETP Lett. - 1989. -50, - P.442.
9. Lagutin A. A., Raikin R. I., Inoue N. and Mis-aki A.//J. Phys. G: Nucl. Part. Phys. - 2002.
- 28. - P. 1259.
