The origin of extra solar cosmic rays (CR) is one of the important unresolved astrophysical questions. Galactic shell type supernova remnants (SNR) have been proposed as plausible acceleration sites for cosmic rays up to energies of several PeV (Völk 1997) and - for very massive SN progenitors - to even higher energies (Biermann et al. 1995). Recently direct experimental evidence for electron acceleration in these objects has been found in the X-ray (Koyama et al. 1995; Koyama et al. 1997; Allen et al. 1997) and TeV -ray range (Tanimori et al. 1998). Somewhat surprisingly, similar searches for evidence of hadron acceleration have only yielded upper limits on the expected -ray emission from the interaction of the hadrons with interstellar matter up to now (Prosch et al. 1996; Hess 1997; Buckley et al. 1998).
An indirect approach to distinguish between different theoretical models aiming to describe the acceleration of charged cosmic rays (CR) is to measure the energy spectrum and composition of CR and compare the results with model predictions. Here the energy regime around the so called "knee" between 1 and 10 PeV is especially interesting(Watson 1997). In this energy range the all-particle CR energy spectral slope - that is constant within measurement errors for lower energies - suddenly increases. The riddle of the origin of the knee and of the cosmic radiation with energies exceeding it, is not yet finally resolved. The following general solutions have been discussed:
1. The change in index is due to some propagation effect in an "original" cosmic-ray population that displays an unbroken power law from low energies up to energies above the knee. The most popular idea is that the energy dependence of the diffusion constant of cosmic rays in the Galaxy could change in the knee region (Peters 1961; Ptuskin et al. 1993). Because of the dependence of the diffusion constant on the nuclear charge Z a modest decrease in the fraction of "light" elements (hydrogen and helium) would be expected. In the simplified chemical model we use below (heavy elements modelled by 65 oxygen and 35 iron, light elements by 40 hydrogen and 60 helium) the fraction of light elements would be expected to decrease from an assumed value of 60% below the knee to 43% above the knee. In such a model (barring a special cancellation of effects) the knee is expected to be a relatively smooth feature, extending over about a decade in energy. A principal problem with this approach is that no plausible Galactic source of cosmic rays has been identified which is quantitatively capable of producing the "original" cosmic-ray population.
2. The knee signals in some way the maximum energy for the sources responsible for low energy cosmic rays. The cosmic rays at higher energies could be "re-accelerated" low energy cosmic-rays, e.g. at the shock front of a Galactic wind (Jokipii & Morfill 1987) or an ensemble of shock fronts in clusters of massive stars (Bykov & Toptygin 1997). In this case a phenomenology similar to the diffusion model in the previous paragraph would be expected. Alternatively, above the knee a completely new population of cosmic rays dominates. In this case typically dramatic changes in chemical composition are expected, e.g. to pure hydrogen in the extragalactical model of Protheroe & Szabo (1992) and nearly pure heavy elements (fraction of light elements 0.3 far above the knee) in a model with special SNRs by Stanev et al. (1993). The special properties of these new sources could in principle allow to understand a knee relatively "sharp" in energy.
To definitely discriminate between a composition changing as expected in models with an energy dependent diffusion constant (discussed above under 1.) and an unchanging composition, it is necessary to achieve an error of smaller than 10 in the experimental determination of the fraction of light elements in the total cosmic radiation above the knee.
While the cosmic-ray composition and energy spectrum are well known from direct balloon and space-borne observations up to energies of about 100 TeV, no general agreement has been reached at higher energies (Watson 1997). The results obtained for CR around the "knee" suffer seriously from the fact that due to the low flux of CR above 1 PeV, only large ground based installations observing the extensive air showers (EAS) induced by cosmic rays in the atmosphere can provide experimental data. However the sensitivity of EAS observables to the mass of the primary CR is weak. The analyses are rendered even more difficult due to theoretical uncertainties concerning the high energy interactions in the atmosphere (Knapp et al. 1996; Gaisser 1997).
Here we present an analysis of EAS between 300 TeV and 10 PeV which restricts to observables related to the electromagnetic shower component. In the following sections the experimental setup (Sect. 2), the Monte-Carlo simulations (Sect. 3), the event reconstruction (Sect. 4) and analysis methods (Sect. 5) are described. Sect. 6 presents the results concerning the CR energy spectrum and the coarse mass composition. A more detailed study of the systematic errors and a discussion of methods to analyse the composition without relying on the absolute penetration depth are discussed in Sect. 7. The paper ends with conclusions in Sect. 8.
© European Southern Observatory (ESO) 2000
Online publication: July 7, 2000