Approfondimenti Cosmici

Cosmic rays: the most energetic particles in the universe
James W. Cronin
Department of Physics and Enrico Fermi Institute, The University of Chicago, Chicago Illinois 60637-1433
Cosmic rays are an ever present aspect of nature. The birth of the eld of elementary-particle physics can be traced to studies of cosmic rays. Now advances in technology and new instrumentation are changing the nature of cosmic-ray research. New forms of astronomy are being created. Ground-based instruments, spawned by cosmic-ray techniques, permit the observation of astrophysical objects emitting radiation in very-high-energy gamma rays, ( 100 GeV) , high-energy neutrinos (1 TeV), and the most energetic particles found in the cosmic radiation ( 5 1019 eV) . At these energies the galactic and intergalactic magnetic elds deect the cosmic-ray protons by only a few degrees. The interaction of these cosmic rays with the cosmic background radiation limits the possible sources to redshifts far less than unity. The origin of these highest-energy cosmic rays is not understood. The present status of knowledge of these cosmic rays and the prospects for solving the mystery concerning their origin are the subjects of this brief article. [S0034-6861(99)00602-9]
I. INTRODUCTION
Cosmic rays are a source of ionizing radiation incident on the whole earth. The intensity of this ionizing radiation varies with magnetic latitude, with altitude, and with solar activity. The attribution cosmic rays is misleading in that the radiation consists principally of fully ionized atomic nuclei incident on the earth from outer space. The eld of elementary-particle physics owes its origin to discoveries made in course of cosmic-ray research, and the study of cosmic rays has contributed to the understanding of geophysical, solar, and planetary phenomena. The existence of cosmic rays also has its practical side. An example is radio-carbon dating, rst suggested by Libby (1965). Radioactive C14 is produced by the collisions of the cosmic rays with the N14 in the atmosphere. This produces an activity of 15 disintegrations per minute per gram of natural carbon in all living matter. On death, the C14 decays with a half-life of 5600 years. Thus the specic activity of C14 provides an accurate archeological clock for the dating of objects in history and prehistory. This article presents a very personal view of the most important questions for future research. I restrict it to energies well above 1 TeV (1012 eV) where most of the observations are ground based due to low uxes. As in many elds, new technologies permit unique investigations that could only be dreamed of in the past. If we take a broad denition of cosmic rays they consist not only of electrons and nuclei, but of other particles as well, particularly gamma rays and neutrinos, which, being neutral, point back to their source. At present there are many programs under development around the world that seek to measure highenergy neutrinos in the primary cosmic radiation (Gaisser et al., 1995). These are neutrinos that come directly from astrophysical sources, as distinct from being produced by ordinary cosmic rays in the atmosphere. The detectors consist of large volumes of antarctic ice or seawater instrumented with photomultipliers. At present there are major experimental efforts under way or proposed. It is expected that high-energy neutrino detectors
Reviews of Modern Physics, Vol. 71, No. 2, Centenary 1999
will make discoveries in astronomy, cosmology, and fundamental-particle physics. In recent years astronomy has been extended to sources emitting rays with energies more than 100 GeV. Numerous galactic and extragalactic sources have been observed with ground-based instruments which de erenkov radiation emitted by the showering of tect the C the high-energy rays. At these energies the satellite detectors, the Compton Gamma-Ray Observatory, and even the new detector Gamma-ray Large Area Space Telescope (to be launched about 2005) do not have the sensitivity necessary to observe sources at energies above 100 GeV. This rapidly expanding area of astronomy has been the subject of a number of recent reviews (Weekes et al., 1998; Ong, 1998). In the remainder of this paper I shall concentrate on the cosmic rays above 1014 eV where most observations have been made with ground-based instruments.
II. A BRIEF HISTORY
The history of research in cosmic rays is a fascinating one, lled with serendipity, personal conict, and experiments on a global scale. The discovery of cosmic rays, attributed to Victor Hess (1912), had its origin in the obsession of some scientists to understand why a heavily shielded ion chamber still recorded radiation. It was assumed that this was some residual radiation from the earths surface and by placing the ion chamber at some distance above the earths surface the detected radiation would be reduced. When Victor Hess took an ion chamber several thousand meters above the earth in a balloon, it was found that the radiation level actually rose, leading to the conclusion that the radiation was arriving from outer space. It took more than 30 years to discover the true nature of the cosmic radiation, principally positively charged atomic nuclei arriving at the top of the atmosphere (Sekido and Elliot, 1985; Simpson, 1995). Many hypotheses were offered for the nature of these cosmic rays. One of the most interesting ideas was that of Robert A. Millikan (Millikan and Cameron, 1928). Millikan noted Astons discovery of nuclear binding energies. He sug1999 The American Physical Society
0034-6861/99/71(2)/165(8)/$16.60
S165
S166
James W. Cronin: Cosmic rays
gested that the cosmic rays were the result of the formation of complex nuclei from primary protons and electrons. In the 1920s electrons and ionized hydrogen were the only known elementary particles to serve as building blocks for atomic nuclei. The formation of atomic nuclei was assumed to be taking place throughout the universe, with the release of the binding energy in the form of gamma radiation, which was the cosmic radiation. A consequence of this hypothesis was that the cosmic radiation was neutral and would not be inuenced by the earths magnetic eld. A worldwide survey led by Arthur Compton demonstrated conclusively that the intensity of the cosmic radiation depended on the magnetic latitude (Compton 1933). The cosmic radiation was predominately charged particles. This result was the subject of an acrimonious debate between Compton and Millikan at an AAAS meeting that made the front page of the New York Times on December 31, 1932. In 1938, Pierre Auger and Roland Maze, in their Paris laboratory, showed that cosmic-ray particles separated by distances as large as 20 meters arrived in time coincidence (Auger and Maze, 1938), indicating that the observed particles were secondary particles from a common source. Subsequent experiments in the Alps showed that the coincidences continued to be observed even at a distance of 200 meters. This led Pierre Auger, in his 1939 article in Reviews of Modern Physics, to conclude One of the consequences of the extension of the energy spectrum of cosmic rays up to 1015 eV is that it is actually impossible to imagine a single process able to give to a particle such an energy. It seems much more likely that the charged particles which constitute the primary cosmic radiation acquire their energy along electric elds of a very great extension. (Auger et al., 1939). Auger and his colleagues discovered that there existed in nature particles with an energy of 1015 eV at a time when the largest energies from natural radioactivity or articial acceleration were just a few MeV. Augers amazement at Natures ability to produce particles of enormous energies remains with us today, as there is no clear understanding of the mechanism of production, nor is there sufcient data available at present to hope to draw any conclusions. In 1962 John Linsley observed a cosmic ray whose energy was 1020 eV (Linsley, 1962). This event was observed by an array of scintillation counters spread over 8 km2 in the desert near Albuquerque, New Mexico. The energetic primary was detected by sampling some of the 5 1010 particles produced by its cascade in the atmosphere. Linsleys ground array was the rst of a number of large cosmic-ray detectors that have measured the cosmic-ray spectrum at the highest energies.
III. COSMIC-RAY SPECTRUM
FIG. 1. Spectrum of cosmic rays greater than 100 MeV. This gure was produced by S. Swordy, University of Chicago.
tion (Zatsepin et al., 1966; Berezinskii et al., 1990; Watson, 1991; Cronin, 1992; Sokolsky et al., 1992; Swordy, 1994; Nagano, 1996; Yoshida et al., 1998). In Fig. 1 the spectrum of cosmic rays is plotted for energies above 108 eV. The cosmic rays are predominately atomic nuclei ranging in species from protons to iron nuclei, with traces of heavier elements. When ionization potential is taken into account, as well as spallation in the residual gas of space, the relative abundances are similar to the abundances of elements found in the sun. The energies range from less than 1 MeV to more than 1020 eV. The differential ux is described by a power law: dN / dE E , (3.1) where the spectral index is roughly 3, implying that the intensity of cosmic rays above a given energy decreases by a factor of 100 for each decade in energy. The ux of cosmic rays is about 1/cm2/sec at 100 MeV and only of order 1/km2/century at 1020 eV. The bulk of the cosmic rays are believed to have a galactic origin. The acceleration mechanism for these cosmic rays is thought to be shock waves from supernova explosions. This basic idea was rst proposed by Enrico Fermi (1949), who discussed the acceleration of cosmic rays as a process of the scattering of the charged cosmic-ray particles off moving magnetic clouds. Subsequent work has shown that multiple bounces off the turbulent magnetic elds associated with supernova shock waves is a more efcient acceleration process
After 85 years of research, a great deal has been learned about the nature and sources of cosmic radiaRev. Mod. Phys., Vol. 71, No. 2, Centenary 1999
S167
(Drury, 1983). At present there is no direct proof of this hypothesis. The argument for it is based on the fact that a fraction of the energy released by supernova explosions is sufcient to account for the energy being pumped into cosmic rays. A second point in favor of the hypothesis is that the index of the spectrum, 2.7 below 5 1015 eV, is consistent with shock acceleration when combined with the fact that the lifetime of the cosmic rays in our galaxy is about 107 years due to leakage of the rays out of the bottle provided by the magnetic eld of our galaxy. Shock acceleration would provide an index of 2.0. The leakage out of the galaxy accounts for the steeper spectrum given by 2.7. The spectrum steepens (the knee) to an index of 3.0 at about 5 1015 eV. In the most recent experiments, this bend in the spectrum is gradual. The conventional explanation of the knee is that the leakage of the cosmic rays from the galaxy depends on the magnetic rigidity E / Z . The knee results from the fact that, successively, the lighter components of cosmic rays are no longer contained in the galaxy as the energy increases. This hypothesis requires that the mean atomic number of the cosmic rays becomes progressively heavier as the energy rises. At the present time this prediction has not been convincingly demonstrated.
IV. TECHNIQUES OF MEASUREMENT
of uorescence light. The limitation of this technique is that it can only function on dark moonless nights, which amounts to only 10% of the time. The positive aspect of the technique is that it rather directly measures the energy of the shower dissipated in the atmosphere, which in most cases is a large fraction of the primary energy. Absolute knowledge of the uorescence efciency of the nitrogen, the absorption of the atmosphere, and the quantum efciency and gain of the photomultipliers is required. Neither technique is particularly effective in identifying the nature of the primary (nucleon, nucleus, or photon). The mean fraction of energy contained in the muonic component of the shower particles increases as the primary becomes heavier. The mean depth in the atmosphere where the cascade is at its maximum moves higher as the primary becomes heavier. Because of uctuations in these quantities, neither technique offers hope of identifying the nature of the primary on an event by event basis.
V. PROPERTIES OF COSMIC RAYS ABOVE 1017 eV
At energies below 1014 eV the ux of primary cosmic rays is sufcient to be measured directly with instruments on balloons and satellites. Above 1014 eV the ux is about 10/m2/day. At this energy very-large-area detectors are required to measure the cosmic rays directly. But fortunately at this energy the cascades in the atmosphere produce a sufcient number of particles on the earths surface so that the primary cosmic ray can be observed indirectly by sampling the cascade particles on the ground. This technique is just an application of Augers experiment with modern technology. Observations made with a surface array of particle detectors can adequately measure the total energy and the direction of the primary cosmic ray. It should be noted that the atmosphere is an essential part of a surface detector. The technique has been extended to instruments that cover as much as 100 km2 with individual detector spacings of 1 km. Much larger arrays will eventually be built. At energies above 1018 eV the density of particles at a xed distance (500 1000 m) from the shower axis is proportional to the primary energy. The constant of proportionality is calculated by shower simulation. A second technique has been used to measure the spectrum above 1017 eV. Optical photons in the range 300 nm to 400 nm are produced by the passage of the charged particles through the nitrogen of the atmosphere (Baltrusaitus et al., 1985; Kakimoto, 1996). About four uorescence photons are produced per meter for each charged shower particle. With an array of photomultipliers, each focused on a part of the sky, the longitudinal development of a shower can be directly measured and the energy inferred from the total amount
Rev. Mod. Phys., Vol. 71, No. 2, Centenary 1999
Above 1017 eV the cosmic-ray spectrum shows additional structure. This structure is displayed in Fig. 2, where the differential spectrum has been multiplied by E 3 to better expose the observed structures. These data are the combined results of four experiments that have operated over the past 20 years. They are from the Haverah Park surface array in England (Lawrence et al., 1991), the Yakutsk surface array in Siberia (Afanasiev et al., 1995), the Flys Eye uorescence detector in Utah (Bird et al., 1994), and the AGASA surface array in Japan (Yoshida et al., 1995). Before plotting, the energy scale of each experiment was adjusted by amounts 20% to show most clearly the common features. The method of energy determination in each of these experiments is quite different, and the fact that they agree within 20% is remarkable. Above 5 10 17 eV the spectrum softens from an index of 3.0 to an index of 3.3. Above 5 10 18 eV the spectrum hardens, changing to an index of 2.7. Beyond 5 10 19 eV the data are too sparse to be certain of the spectral index. There is no clear explanation of this structure. Above 10 18 eV, the galactic magnetic elds are not strong enough to act as a magnetic bottle even for iron nuclei. If the cosmic rays continue to be produced in the galaxy, they should show an anisotropy that correlates with the galactic plane. No such anisotropy has been observed. The hardening of the spectrum to an index of 2.7 above 5 10 18 eV may then be a sign of an extragalactic component emerging as the galactic component dies away.
VI. THE DIFFICULTY OF ACCELERATION
Above 1019 eV the precision of the spectrum measurement suffers from lack of statistics. There have been about 60 events recorded with energy greater than 5 1019 eV. Yet it is above this energy that the scientic mystery is the greatest. There is little understanding of how known astrophysical objects could produce par-
S168
FIG. 2. (Color) Upper end of the cosmic-ray spectrum. Haverah Park points (red; Lawrence et al., 1991) serve as a reference. Yakutsk points (black; Afanasiev et al., 1995) have been reduced in energy by 20%. Flys Eye points (green; Bird et al., 1995) have been raised in energy by 10%. AGASA points (Yoshida et al., 1995) have been reduced by 10%.
ticles of such energy. At the most primitive level, a necessary condition for the acceleration of a proton to an energy E in units of 1020 eV is that the product of the magnetic eld B and the size of the region R be much larger than 3 1017 G-cm. This value is appropriate for a perfect accelerator such as might be scaled up from the Tevatron at Fermilab. The Tevatron has a product BR 3 109 G-cm and accelerates protons to 1012 eV. Analogous acceleration of cosmic rays to energies above 1019 eV seems difcult, and the literature is lled with speculations. Two reviews that discuss the basic requirements are those of Greisen (1965) and Hillas (1984). While these were written some time ago, they are excellent in outlining the basic problem of cosmic-ray acceleration. Biermann (1997) has recently reviewed all the ideas offered for achieving these high energies. Hillas in his outstanding review of 1984 presented a plot that graphically shows the difculty of cosmic-ray acceleration to 1020 eV. Figure 3 is an adaptation of his gure. Plotted are the size and strength of possible acceleration sites. The upper limit on the energy is given by E 18 0.5 ZB GL kpc . (6.1) Here the E 18 is the maximum energy measured in units of 1018 eV. L kpc is the size of the accelerating region in units of kiloparsecs, and B G is the magnetic eld in G. The factor was introduced by Greisen to account for the fact that the effective magnetic eld in the accelerator analogy is much less than the ambient eld. The factor in Hillass discussion is the velocity of the shock wave (relative to c ), which provides the acceleration.
The plotted lines correspond to a 1020 eV proton with 1 and 1/300. A line is also plotted for iron nuclei ( 1). With Z 26, iron is in principle easier to accelerate. Realistic accelerators should lie well above the dashed line. The gure is also relevant for one-shot acceleration, as it represents the electromotive force (emf) induced in a conductor of length L moving with a velocity through a uniform magnetic eld B . Synchrotron energy loss is also important. For protons the synchrotron loss rate at 1020 eV requires that the magnetic eld be less than 0.1 G for slow acceleration (the accelerator analogy; Greisen 1965). From Fig. 3 it can be seen that the acceleration of cosmic rays to 1020 eV is not a simple matter. Because of this, some authors have seriously postulated that cosmic rays are not accelerated but are directly produced by top down processes. For example, defects in the fabric of spacetime could have huge energy content and could release this energy in the form of high-energy cosmic rays (Bhattacharjee, Hill, and Schramm, 1992).
VII. NATURES DIAGNOSTIC TOOLS
There are some natural diagnostic tools that make the analysis of the cosmic rays above 5 1019 eV easier than at lower energies. The rst of these is the 2.7-K cosmic background radiation (CBR). Greisen (1966) and Zatsepin and Kuzmin (1966) pointed out that protons, photons, and nuclei all interact strongly with this radiation, a phenomenon that has become known as the GZK effect.
S169
FIG. 4. Proton energy as a function of propagation distance through the 2.7-K cosmic background radiation for the indicated initial energies.
FIG. 3. Modied Hillas plot (Hillas, 1984). Size and magnetic eld of possible sites of acceleration. Objects below the dashed line cannot accelerate protons to 1020 eV.
As an example, a collision of a proton of 1020 eV with a CBR photon of 10 3 eV produces several hundred MeV in the center-of-mass system. The cross section for pion production is quite large so that collisions are quite likely, resulting in a loss of energy for the primary proton. In Fig. 4 we plot the results of the propagation of protons through the CBR. Regardless of the initial energy of the proton, it will be found with less than 1020 eV after propagating through a distance of 100 Mpc (3 108 light years). Thus the observation of a cosmic-ray proton with energy greater than 1020 eV implies that its distance of travel is less than 100 Mpc. This distance corresponds to a redshift of 0.025 and is small compared to the size of the universe. Similar arguments can be made for nuclei or photons in the energy range considered. There are a limited number of possible sources that t the Hillas criteria (Fig. 3) within a volume of radius 100 Mpc about the earth. The fact that the cosmic rays, if protons, will be little deected by galactic and extragalactic magnetic elds serves as the second diagnostic tool. The deection of protons of energy 5 1019 eV by the galactic magnetic eld ( 2 G) and the intergalactic magnetic elds ( 10 9 G) is only a few degrees (Kronberg, 1994a, 1994b), so that above 5 1019 eV it is possible that the cosmic rays will point to their sources. We approach an astronomy, even for charged cosmic rays, in which the distance to the possible sources is limited.
VIII. COSMIC-RAY ASTRONOMY
The energy 5 1019 eV represents a lower limit for which the notion of an astronomy of charged particles
from local sources can be applied. The GZK effect enhances the number of events from sources within a distance of 100 Mpc. Of these events, two particularly stand out with energies reported to be 2 1020 eV by the AGASA experiment (Hayashida et al., 1994) and 3 1020 eV by the Flys Eye experiment (Bird et al., 1995; Elbert and Sommers, 1995). More recently a total of six events with energy 1020 eV have been reported by the AGASA experiment (Takeda et al., 1998). For all these events the probable distance to the source is less than 50 Mpc. The events above 5 1019 eV are too few to derive a spectral index. It is not clear that a single spectrum is even the proper way to characterize these events. Since they must come from nearby, the actual number of sources may not form an effective continuum in space, so the spectrum observed may vary with direction. The matter within 100 Mpc is not uniformly distributed over the sky. It is probably more fruitful to take an astronomical approach and plot the arrival directions of these events on the sky in galactic coordinates. Arrival-direction data are available for the Haverah Park experiment (Watson, 1997), the AGASA experiment (Hayashida et al., 1996), and for the most energetic event recorded by the Flys Eye experiment (Bird et al., 1995; Elbert and Sommers, 1995). In Fig. 5 we plot the arrival directions of 20 AGASA events and 16 Haverah Park events. The size of the symbols corresponds to the angular resolution. In addition, the error box for the most energetic event recorded by the Flys Eye experiment is plotted. What is remarkable in this gure is the number of coincidences of cosmic rays coming from the same direction in the sky. Of the 20 events reported by AGASA, there are two pairs. The probability of a chance coincidence for this is about 2%. The addition of the Haverah Park events shows a coincidence with one of the AGASA pairs. However, the Flys Eye event co-
S170
FIG. 5. (Color) Plot of arrival directions of cosmic rays with energy 5 1019: red points, Haverah Park (Lawrence et al., 1991); blue points, AGASA (Yoshida et al., 1995); green point, Flys Eye event with energy 3 1020 eV. The size of the symbols represents the resolution of each experiment. The empty region marked by the blue line is the part of the sky not seen by the northern hemisphere location of the observations.
incides with one of the AGASA events. It is not possible to estimate properly the probability of chance overlaps, but the possibility that these overlaps may be real should not be ignored. The triple coincidence contains the AGASA event of 2 1020 eV, the Haverah Park event of about 1 1020 eV, and the AGASA event of 5 1019 eV. The Flys Eye event of 3 1020 eV is in coincidence with the AGASA event of 6 1019 eV. The third pair contains AGASA events of 6 1019 eV and 8 1019 eV, respectively. The triple coincidence is particularly interesting if it is not the result of pure chance. It contains cosmic rays separated by a factor of 4 in energy that have not been separated in space by more than a few degrees. This is an encouraging prospect for future experiments in which, with many more events, one may observe point sources, clusters, and larger-scale anisotropies in the sky. The crucial questions will be: Does the distribution of cosmic rays in the sky follow the distribution of matter within our galaxy or the distribution of nearby extragalactic matter, or is there no relation to the distribution of matter? Are there point sources or very tight clusters? What is the energy distribution of events from these clusters? Are these clusters associated with specic astrophysical objects? If there is no spatial modulation or no correlation with observed matter, what is the spectrum? This situation would imply an entirely different class of sources, which are visible only in the light of cosmic rays with energy 5 1019 eV. Of course there may be a combination of these possibilities. If even crude data on primary composition are available, they
can be divided into categories of light and heavy components, which may have different distributions. Crucial to these considerations is uniform exposure over the whole sky. And a nal and fundamental question is: Is there an end to the cosmic-ray spectrum?
IX. NEW EXPERIMENTS
The ux of cosmic rays with energy 5 1019 eV is about 0.03/km2/sr/yr. It required ve years for the AGASA array, with an acceptance of 125 km2-sr, to collect 20 events above this energy. In 1999 an improved version of the Flys Eye experiment (HiRes) will begin operation (Abu-Zayyad, 1997). It will have an acceptance of about 7000 km2-sr above 5 1019 eV. With a 10% duty cycle it should collect about 20 events per year. The experiment will be located in northern Utah. Only half of the sky will be observed. Experiments with far greater statistical power are required to make real progress. It is very likely that a combination of types of sources and phenomena are responsible for the highest-energy cosmic rays. Thus the experiment must be constructed so as not to have a bias towards a particular or single explanation for the cosmic rays. An ideal experiment should have uniform coverage of the entire sky. It should also be fully efcient at energies beginning at 1019 eV, as the present data available above that energy are very sparse. It should have the best possible means to identify the primary particle, although no experiment can make a unique identication on an event by event basis.
S171
A number of experiments have been proposed or will be proposed in the next few years. These are all described in the Proceedings of the 25th International Cosmic-Ray Conference held in Durban in 1997. One of these seeks to satisfy all the general requirements outlined above. The experiment of the Pierre Auger Observatories (Boratav, 1997) consists of two detectors with acceptance 7000 km2-sr. They will be located at midlatitude in the southern and northern hemispheres, which will provide nearly uniform sky coverage. An important feature of the Auger experiment is its use of a hybrid detector that combines both a surface array and a uorescence detector. Such an experiment will collect 450 events 5 1019 eV each year. Some 20% of the events may originate from point sources or tight clusters if the AGASA results (Hayashida et al., 1996) are used as a guide. Also being proposed is an all-uorescence detector called the Telescope Array (Telescope Array Collaboration, 1997) to be located in the northern hemisphere. It would have an aperture of 70 000 km2-sr (7000 km2-sr with the 10% duty cycle). It would also co-locate two of its uorescence units with the northern Auger detector. A visionary idea has been offered in which the uorescence light produced by a cosmic ray in the atmosphere would be viewed from a satellite (Linsley, 1997; Krizmanic, Ormes, and Streitmatter, 1998). There are many technical difculties in such a project. It would, however, represent a next step in the investigations if the projects above are realized and no end to the cosmic-ray spectrum is observed. The estimated sensitivity of such a satellite detector for cosmic rays with energy 1020 eV would be 10 100 times that of the Pierre Auger Observatories.
X. CONCLUSION
It is now widely recognized that the investigation of the upper end of the cosmic-ray spectrum will produce new discoveries in astrophysics or fundamental physics. There are a number of complementary proposals for new experiments that will provide the needed observations within the next ten years.
ACKNOWLEDGMENTS
Over the years I have learned much from discussions with many individuals concerning the highest-energy cosmic rays. Prominent among these are V. Berezinskii, P. Biermann, M. Boratav, T. Gaisser, A. M. Hillas, P. P. Kronberg, M. Nagano, R. Ong, and A. A. Watson. Support from the U.S. National Science Foundation is gratefully acknowledged.
REFERENCES Abu-Zayyad, T., 1997, in Proceedings of the 25th International Cosmic Ray Conference, Durban, edited by M. S. Potgieter, B. C. Raubenheimer, and D. J. van der Walt (World ScienRev. Mod. Phys., Vol. 71, No. 2, Centenary 1999
tic, Singapore), Vol. 5, p. 321; this paper and the eleven that immediately follow describe various aspects of the HiRes detector. Afanasiev, B. N., et al., 1995, in Proceedings of the 24th International Cosmic Ray Conference, Rome, edited by N. Lucci and E. Lamanna (University of Rome, Rome), Vol. 2, p. 756. Auger, P., and R. Maze, 1938, C. R. Acad. Sci. Ser. B 207, 228. Auger, P., P. Ehrenfest, R. Maze, J. Daudin, Robley, and A. on, 1939, Rev. Mod. Phys. 11, 288. Fre Baltrusaitis, R. M., et al., 1985, Nucl. Instrum. Methods Phys. Res. A 240, 410. Berezinskii, V. S., S. V. Bulanov, V. A. Dogiel, V. L. Ginzburg, and V. S. Ptuskin, 1990, in Astrophysics of Cosmic Rays, edited by V. L. Ginzburg (Elsevier Science, New York/ Amsterdam). Bhattacharjee, P., C. T. Hill, and D. N. Schramm, 1992, Phys. Rev. Lett. 69, 567. Biermann, P., 1997, J. Phys. G 23, 1. Bird, D. J., et al., 1994, Astrophys. J. 424, 491. Bird, D. J., et al., 1995, Astrophys. J. 441, 144. Boratav, M., 1997, in Proceedings of the 25th International Cosmic Ray Conference, Durban, edited by M. S. Potgieter, B. C. Raubenheimer, and D. J. van der Walt (World Scientic, Singapore) Vol. 5, p. 205; this paper and the ve that immediately follow describe various aspects of the Pierre Auger Observatories. Compton, A. H., 1933, Phys. Rev. 43, 387. Cronin, J. W., 1992, Nucl. Phys. B (Proc. Suppl.) 28, 213. Drury, L. OC., 1983, Rep. Prog. Phys. 46, 973. Elbert, J. W., and P. Sommers, 1995, Astrophys. J. 441, 151. Fermi, E., 1949, Phys. Rev. 75, 1169. Gaisser, T. K., F. Halzen, and T. Stanev, 1995, Phys. Rep. 258, 173. Greisen, K., 1965, in Proceedings of the 9th International Cosmic Ray Conference, London (The Institute of Physics and The Physical Society, London), Vol. 2, p. 609. Greisen, K., 1966, Phys. Rev. Lett. 16, 748. Hayashida, N., et al., 1994, Phys. Rev. Lett. 73, 3491. Hayashida, N., et al., 1996, Phys. Rev. Lett. 77, 1000. Hillas, A. M., 1984, Astron. Astrophys. 22, 425. Hess, V. F., 1912, Z. Phys. 13, 1084. Kakimoto, F., E. C. Loh, M. Nagano, H. Okuno, M. Teshima, and S. Ueno, 1996, Nucl. Instrum. Methods Phys. Res. A 372, 527. Krizmanic, J. F., J. F. Ormes, and R. E. Streitmatter 1998, Eds., Workshop on Observing Giant Cosmic Ray Air Showers from geq 1020 eV Particles from Space, AIP Conf. Proc. No. 433 (AIP, New York). This volume is dedicated to the design of a space-based cosmic-ray observatory. Kronberg, P. P., 1994a, Rep. Prog. Phys. 57, 325. Kronberg, P. P., 1994b, Nature (London) 370, 179. Lawrence, M. A., et al., 1991, J. Phys. G 17, 773. Libby, W. F., 1965, Radio Carbon Dating, 2nd ed. (University of Chicago, Chicago). Linsley, J., 1962, Phys. Rev. Lett. 10, 146. Linsley, J., 1997, in Proceedings of the 25th International Cosmic Ray Conference, Durban, edited by M. S. Potgieter, B. C. Raubenheimer, and D. J. van der Walt (World Scientic, Singapore), Vol. 5, p. 381. Millikan, R. A., and G. H. Cameron, 1928, Phys. Rev. 32, 533. Nagano, M., 1996, Ed., Proceedings of the International Symposium on Extremely High Energy Cosmic Rays: Astrophysics and Future Observations (Institute for Cosmic Ray Research, University of Tokyo, Japan).
S172
Ong., R. A., 1998, Phys. Rep. 305, 93. Sekido, Y., and H. Elliott, 1985, Eds., Early History of Cosmic Ray Studies (Reidel, Dordrecht). Simpson, J., 1995, in The Physical Reviewthe rst hundred years, edited by H. Stroke (AIP, New York), p. 573. Sokolsky, P., P. Sommers, and B. R. Dawson, 1992, Phys. Rep. 217, 225. Swordy, S., 1994, rapporteur talk, in Proceedings of the 23rd International Cosmic Ray Conference, Calgary, edited by R. B. Hicks, D. A. Leahy, and D. Venkatesan (World Scientic, Singapore), p. 243. Takeda, M., et al., 1998, Phys. Rev. Lett. 81, 1163. Telescope Array Collaboration, 1997, in Proceedings of the
25th International Cosmic Ray Conference, Durban, edited by M. S. Potgieter, B. C. Raubenheimer, and D. J. van der Walt (World Scientic, Singapore), Vol. 5, p. 369. Watson, A. A., 1991, Nucl. Phys. B (Proc. Suppl.) 22, 116. Watson, A. A., 1997, University of Leeds, private communication. Weekes, T. C., F. Aharnian, D. J. Fegan, and T. Kifune, 1997, in Proceedings of the Fourth Compton Gamma-Ray Observatory Symposium, AIP Conf. Proc. No. 410, edited by C. D. Dermer, M. Strikman, and J. D. Kurfess (AIP, New York), p. 361. Yoshida, S., et al., 1995, Astropart. Phys. 3, 105. Yoshida, S., and H. Dai, 1998, J. Phys. G 24, 905. Zatsepin, G. T., and V. A. Kuzmin, 1966, JETP Lett. 4, 78.
Astroparticle Physics 17 (2002) 441458 www.elsevier.com/locate/astropart
Distributions of secondary muons at sea level from cosmic gamma rays below 10 TeV
J. Poirier a, S. Roesler b, A. Fasso
a
c,*
Department of Physics, Center for Astrophysics, University of Notre Dame, Notre Dame, IN 46556, USA b Stanford Linear Accelerator Center, MS. 48, 2575 Sand Hill Road, Menlo Park, CA 94025, USA c CERN-EP/AIP, CH-1211 Geneva 23, Switzerland Received 12 May 2001; received in revised form 8 August 2001
Abstract The F L U K A Monte Carlo program is used to predict the distributions of the muons which originate from primary cosmic gamma rays and reach sea level. The main result is the angular distribution of muons produced by vertical gamma rays which is necessary to predict the inherent angular resolution of any instrument utilizing muons to infer properties of gamma ray primaries. Furthermore, various physical eects are discussed which aect these distributions in diering proportions. 2002 Elsevier Science B.V. All rights reserved.
PACS: 95.75.Pq; 98.70.Sa; 98.70.Rz; 13.60.Le Keywords: Air shower simulation; Gamma rays; Muons
1. Introduction Muons detected at ground level arise mainly as decay products of charged mesons. These mesons are created abundantly in hadronic showers, most of which are caused by primary cosmic ray protons and nuclei interacting inelastically with the nuclei of the atmosphere. A small fraction of muons, however, have their rst origin in photonuclear reactions induced by primary or secondary cosmic gamma rays. Several authors have per-
Corresponding author. Tel.: +41-22-7672398; fax: +41-227679480. E-mail address: alberto.fasso@cern.ch (A. Fass o).
formed analytical or Monte Carlo (MC) calculations of the muon ux produced in gamma showers (see Refs. [1020] of [1]). Most of these calculations were one-dimensional and all of them referred to gamma energies much larger than 10 TeV. Despite their relative scarcity compared to the large background of muons from hadron-generated showers, muons originating from primary gamma ray interactions are important for groundbased high statistics cosmic ray experiments which are sensitive to energies 6 10 TeV such as MILAGRO [2] and GRAND [3]. Cosmic gamma rays, unlike charged hadrons, are unaected by the Earths and galactic magnetic elds and their direction points directly to the location of their source. Therefore, an accumulation of muon
0927-6505/02/$ - see front matter 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 7 - 6 5 0 5 ( 0 1 ) 0 0 1 7 4 - 8
442
J. Poirier et al. / Astroparticle Physics 17 (2002) 441458
directions around a particular angle (and also at a particular time in the case of pulsed sources or gamma ray bursts) carries direct information about the location of the source and the neutrality of the primary causing the excess. The ability to distinguish these muons from background muons depends upon statistics, the strength and energy spectrum of the emitting source, the angular resolution of the experiment, and the degradation of the angular resolution due to all the physical processes which occur between the primary and the muon at detection level. This last point is the focus of this calculation. Knowledge of this angular resolution is important in order to determine a detection window which is large enough to include a good fraction of the gamma signal and, at the same time, small enough to minimize the large uniform background of hadronic origin. The information about the primary gamma direction is degraded by multiple physical processes so the nal measured direction of the muon is no longer collinear with the primary which originated the muons ancestors. In particular, the direction resolution is aected by the angle of production of the mesons in the primary interaction (and then in subsequent hadronic interactions of these mesons to create additional mesons), the decay angle of the muon relative to its parent meson, and the lengths of the charged particle paths including that of the detected muon at ground level. Along these path lengths there is Coulomb scattering and deection in the Earths magnetic eld for the trajectories of the mesons and the muon (usually most important for the muon which typically has the longest ight path and the lowest momentum). In addition, these eects depend upon the energy of the particles involved and on the varying properties of the atmosphere (pressure and temperature). The muons are produced by mesons of second, third, or further generations. The number of generations typically increases with increasing primary gamma energy and each new generation involves an additional production angle which tends to increase the angular disparity between muon and primary because of more steps in a random walk type process. An opposing eect at higher energies is that all angular distributions are more forward peaked thus decreasing the angular
disparity; the resulting eect is a combination of the two. The purpose of the present study is to estimate the inuence of the various eects on the ultimate angular resolution which can be obtained utilizing ground level muons. The complexity of the problem requires detailed analyses which can only be obtained by a full three-dimensional MC calculation. In addition, since only a small fraction of all primary photons gives rise to an event of interest, it is desirable to resort to statistical variance reduction techniques to bring the computing task within manageable limits. The MC program F L U K A [47] provides an accurate and well-tested description of the hadronic and electromagnetic interactions with all the relevant physical eects, and a set of eective statistical tools to accelerate convergence of the results. The program has been used to calculate the geometrical properties of secondary muons at sea level arising from primary gamma cosmic rays with energies from 1 GeV to 10 TeV interacting in the Earths atmosphere. In this energy range the cross-section for direct muon pair production is much smaller than that for hadroproduction and has been neglected. This range has been shown in Ref. [1] to be the interval which produces the maximum secondary muon ux at ground level for primary dierential gamma spectra with spectral indices larger than 2.4 (a smaller index would yield a harder spectrum and give a maximum secondary muon ux arising from primary energies beyond those considered here). In this paper, only incident cosmic gamma rays at perpendicular (vertical) incidence on top of the atmosphere are considered. Ground-based experiments often choose angles close to normal for reasons of simplicity and because the detection rate is a maximum, the rate dropping rather rapidly as the angle deviates markedly from normal. For small angles from normal, the angular resolution results will probably not dier signicantly from those presented in this paper. In addition to the angular correlation of secondary muons at ground level, this study also provides additional information such as the distribution of heights at which the muons originate, the muons kinetic energy distribution at the
443
ground and production level, the radial distribution of the muons, and the number of generations which preceeded that of the muon.
2. Air shower calculations with
FLUKA
The 2000 version of the F L U K A MC code was used to simulate the electromagnetic and hadronic particle cascades induced by primary gamma rays in the atmosphere. The powerful biasing capabilities of the code allowed a simulation of the full three-dimensional shower in a single run from the top of the atmosphere down to ground level. The physical models implemented in F L U K A which are relevant for the present study as well as various variance reduction techniques were already discussed in Ref. [1]. There it was emphasized that these models were validated by comprehensive comparisons of F L U K A results with experimental data obtained mostly at accelerators but also in recent cosmic ray studies [811]. Nevertheless, the simulation of an atmospheric shower and, in particular, of its muon component is a relatively new area of application of F L U K A . Additional examples showing that the code allows remarkably accurate predictions in this area are given in Section 2.1. Section 2.2 contains some details of the present gamma ray shower simulations which supplement those discussed in Ref. [1]; this reference contains a more complete description. 2.1. Muon production in air showers induced by hadronic primaries and comparison to experimental data No data exist for the muon component of gamma-ray-induced showers to which F L U K A results could be directly compared. On the other hand, data on muon ux and energy spectra have become available recently for atmospheric showers induced by cosmic ray protons and nuclei. These data were mostly obtained with spectrometers in balloon-borne experiments and cover the whole range of atmospheric depths from the top of the atmosphere down to sea level. Fortunately, the mechanisms and models for muon production and transport in gamma-ray-
induced showers are largely the same as those in pure hadronic air showers. In both cases, the production of the pions which eventually decay into muons is described by the Dual Parton Model (DPM) [12]. The only dierence is that in the former case the photon is assumed to uctuate rst into a quarkantiquark state which then interacts hadronically, identical to a meson. This so-called vector meson dominance model (VMD) for the hadronic photon uctuation is a well-established concept (see Ref. [13] and references therein) and its application together with the DPM to hadronic interactions of photons has been proven to be very successful (see, for example, Refs. [14,15]). A good description of muon production from proton- and nuclei-induced air showers should therefore give condence that the corresponding results obtained with F L U K A for photon-induced air showers are reliable as well. In the following a few examples are given. A critical aspect of muon production in the atmosphere is the dependence of the total muon ux on the height above sea level. This dependence is commonly expressed as a function of depth in the atmosphere (in g/cm2 of air) which is a measure of the amount of air penetrated by the cascade. It has been measured by the CAPRICE experiment which was able to distinguish between positively and negatively charged muons. Results obtained with F L U K A for negative muons are compared with CAPRICE data [16] in Fig. 1. The dierent curves correspond to the muon ux calculated for dierent intervals of the muon momentum. To correspond to the average geometrical response of the CAPRICE apparatus, only muons with a polar angle of less than 9 with respect to the vertical (i.e., vertical muons) were scored. The simulations are based on sophisticated models and methods for the sampling of the identity, energy, and angle of the hadronic primary from the distributions of galactic cosmic rays on top of the atmosphere at the time of the measurements. More details can be found in Refs. [10,17]. The input spectrum which was used for this calculation is based on a large amount of recent balloon and space shuttle measurements which have become available. Solar modulation is treated based on a diusionconvection model whose parameters are determined
444
Fig. 1. Dependence of the negative muon ux produced by hadronic primaries on the depth in the atmosphere shown for dierent intervals of the muon momentum. F L U K A results (solid lines) are compared to data obtained by the CAPRICE experiment [16] (points). The comparison is shown for dierent intervals in muon momentum. The curves and data points for the rst ve momentum intervals were shifted by constant factors as indicated.
Fig. 2. (a) Energy spectra of positive and negative muons at 886 g/cm2 . F L U K A results (histograms) are compared to data obtained by the CAPRICE experiment [18]. plab is the detected muon momentum. (b) Muon charge ratio as a function of kinetic energy measured by the CAPRICE experiment [18] (points) and calculated with F L U K A (solid line).
from the climax neutron monitor count rates. As can be seen from the gure, F L U K A gives a good description of the negative muon ux for all depths. Similar agreement exists for positive muons (not shown here). Energy spectra of muons were recorded by the same experiment at ground (i.e., at an atmospheric depth of 886 g/cm2 ) [18]. F L U K A results for positive and negative muons are compared to these data in Fig. 2a. Again, good agreement between simulation and data is found.
Finally, the ratio of the uxes of positively and negatively charged muons is compared to CAPRICE data [18] in Fig. 2b. Data and calculations are again for vertical muons at a depth of 886 g/ cm2 . The ratios are plotted as a function of the muon energy. The above comparisons of F L U K A results and experimental data together with those published in earlier studies [811] indicate that the application of F L U K A to cosmic gamma ray showers should give reliable predictions as well.
445
2.2. The Monte Carlo simulation of cosmic gamma ray showers In order to study the dependence of the shower properties on the primary photon energy, the showers were calculated for monoenergetic photons impinging vertically on top of the atmosphere (taken to be 80 km above sea level). In total nine sets of simulations were performed for the following primary energies: 1, 3, 10, 30, 100, 300, 1000, 3000, and 10,000 GeV. As in the earlier study [1] the atmosphere was approximated by 50 layers with a constant density in each layer and with layer densities decreasing exponentially with increasing altitude. It has been veried [1] that this approximation does not aect the conclusions drawn in this study. The geometry is described in a right-handed orthogonal system with its origin at the intersection of the shower axis with sea level elevation, z-points to the center of the Earth and x- and y-point North and East, respectively. Most of the results discussed throughout the paper were obtained without the eect of the magnetic eld of the Earth included. However, for two primary energies (10 GeV and 1 TeV) additional simulations were performed which included the magnetic eld of three dierent geographical locations (see Section 3.4). The use of several variance reduction (biasing) techniques was essential to obtain results with reasonable statistical signicance. They included leading particle biasing at each electromagnetic interaction, biasing of the photon mean free path with respect to photonuclear interactions, biasing of the decay length of charged mesons, and particle splitting at the boundaries of dierent air layers. Each of these biasing techniques alters the statistical weight of a particle in the cascade. A muon as produced and transported in the MC simulation and which carries a certain statistical weight (hereafter called MC muon) contributes to distributions, yields, etc., of actual, i.e., measurable muons with a probability equal to its weight. The use of biasing techniques would not be appropriate to study uctuations within the same shower which, however, is not the aim of this study. More details on the biasing techniques can be found in Ref. [1].
For each MC muon reaching sea level (detection level) the following information was recorded in a le for later analysis: (1) Information on the muon at detection level: muon charge, lateral coordinates x; y with respect to shower axis, direction cosines with respect to the x- and y-axes, kinetic energy, statistical weight of the muon, and number of the event (i.e., primary cosmic ray photon) which produced this muon. The muon arrival time was not recorded in this series of calculations. (2) Information on the muon at its production vertex (i.e., meson decay vertex): lateral coordinates x; y with respect to shower axis, direction cosines with respect to the x- and y-axes, height z, kinetic energy, statistical weight of the muon, and identity of the decaying meson. (3) Information on the grandparent at the parent production vertex: particle identity, lateral coordinates x; y with respect to shower axis, direction cosines with respect to the x- and y-axes, height z, kinetic energy, statistical weight of the particle, and generation number. (4) Information on the photonuclear interaction vertex preceding the hadronic cascade in which the muon at detection level has been created: lateral coordinates x; y with respect to shower axis, direction cosines with respect to the x- and y-axes, height z, energy, statistical weight of the photon, and generation number of the photon. The generation of a particle increases with each sampled discrete interaction (electromagnetic or hadronic), i.e., the primary photon is generation 1, the generation of the electron and positron after the rst pair production process would be 2, etc. Delta ray production and Coulomb scattering are not considered to increase the generation number. A summary of the simulated particles cascades for each primary gamma ray energy is given in Table 1. In the following, all results refer to the sum of positive and negative muons. The number of histories (number of primary gamma rays), Nc , calculated for each primary energy decreases with energy in order to obtain uniform statistical signicance on the muons which reach ground level for the dierent primary energies. At the lowest
446
Table 1 Summary of the simulated particle cascades Ec (GeV) 1 3 10 30 100 300 1000 3000 10,000 Nc 4,542,000 26,000,000 2,250,000 1,260,000 800,000 324,445 143,131 27,040 18,000 Nl 3:78 103 117.7 1191 3843 11,351 17,259 31,709 21,954 58,874 Nl =Nc 8:32 0:46 1010 4:53 0:23 106 5:29 0:03 104 3:05 0:01 103 1:42 0:01 102 5:32 0:02 102 2:22 0:01 101 8:12 0:05 101 3:27 0:03
Nc is the number of histories calculated for each primary energy (Ec ) and Nl is the total number of muons (i.e., the sum of the weights) scored at sea level. In the last column the average muon multiplicity per primary gamma ray is given. The errors quoted in the last column represent the statistical uncertainties of the calculations.
energy (1 GeV) the muon ux per gamma at sea level is very small so that even with a signicant amount of biasing in the simulations it is dicult to reach adequate statistics. Less emphasis was therefore put on the simulation at 1 GeV. The column Nl gives the total number of muons reaching detection level for the given number of primary photons Nc . The predictions for the muon multiplicity at sea level per incident primary gamma ray Nl =Nc are listed in the last column. Similar values were already reported in Ref. [1], however with smaller statistical signicance than those in Table 1.
Fig. 3. Angular distribution of muons at sea level. Hxz is the angle of the detected muon projected onto the North-down or xz-plane. The distributions are given for dierent primary photon energies and are normalized to unit area. Eects of the Earths magnetic eld are not included in this gure. Units of angle are in degrees.
3. Angular correlations 3.1. Angular distribution of muons at sea level In this section the eect of the Earths magnetic eld is not included; Section 3.4 has a separate discussion of this added eect which depends on the geographical location of interest. Fig. 3 summarizes the deviations of the angle of the muons which reach sea level from the primary gamma direction (vertical angle). It shows the distribution of the angle Hxz which is the muons angle with respect to the z-axis projected onto the xz-plane (North-down plane). This angle is dened as Hxz arctancos hx = cos hz where cos hx and cos hz are the direction cosines of the detected muon. Negative projections have been reected upon the positive values for Hxz since positive and negative
angles would be symmetric in the absence of the magnetic eld. Also, the distributions in the yzplane (East-down plane) are similar to those shown in Fig. 3. Since the angle of the primary gamma with respect to the z-axis is zero (vertically incident primaries), any deviation of the muon direction from zero degrees is due to the various interaction and transport processes occurring in the cascade between the primary gamma ray and the muon which reaches sea level. Thus the results shown in Fig. 3 represent the correlation between the direction of
447
the detected muon and the direction of the primary gamma ray. All distributions in Fig. 3 are normalized to unit area to compare their shapes. The shape of the distributions for primary energies above about 30 GeV varies rather slowly. Therefore, only two histograms (100 GeV and 10 TeV) are shown for this energy range. At lower energies the shape of the distributions changes signicantly as can be seen from the histograms for 3 and 10 GeV. The distributions become wider with decreasing primary energy. At low primary gamma ray energies, the muons have, on average, lower energies which cause larger angular deviations from the primary since, at low energy, scattering dominates the angle. At the higher energies, the slow increase in muon angular widths may be due to the increased number of progenitors each of which involves an additional production angle for the mesons in that generation. This is illustrated in Fig. 4 for primary energies of 10 GeV and 10 TeV. An arbitrary cut of 2 GeV was applied to the angular distributions so that contributions from muons with energies below and above this cut could be seen separately along with the total. Two measures of the widths of these distributions as a function of the energy of the incident primary cosmic gamma ray are given in Table 2. The width parameters are: the half-width at halfmaximum-height (HWHM) and the angular halfwidth containing 68% of the muons, i.e., the same number of events that a 1r cut would include if the distributions had a Gaussian shape (which they do not have, because of the long tails). Both of these parameters provide a measure of the distribution widths which minimizes the distorting eect of a tail. In addition, the values for various cuts on the kinetic energy of the detected muons (1, 2, and 4 GeV) are presented. The cuto energy results for primary energies of less than 10 GeV were omitted due to lack of statistics. For the HWHM calculation, the maximum height is assumed to be the value in the rst bin of the corresponding histogram, i.e., the average value in the interval jHxz j < 0:25. The x- and y-rectangular coordinate system was chosen to represent the results of the muons angular distribution as the corrections for the additional deection due to the Earths mag-
Fig. 4. (a) Angular distribution of muons at sea level for 10 GeV and (b) 10 TeV primary gamma ray energy. The distributions labeled total are the same as shown in Fig. 3. In addition, the contributions from muons with kinetic energies below 2 GeV and above 2 GeV are presented. Units of angle are in degrees.
netic eld dier in these two directions. In addition x; y is the natural coordinate system for GRAND. For some experiments a space angle is 1=2 more natural. If we dene R x2 y 2 , then the space angle resolution (dHR (HWHM)) can be estimated directly from the numbers in Table p 2 by multiplying the dHxz (HWHM) values by 2 due to the symmetry of x and y in the absence of a magnetic eld. As can be seen in this table: (i) as the primary energy decreases below 10 GeV, the width of the angular distribution increases dramatically, (ii) as
448
Table 2 (dHxz (HWHM)) Half-width at half maximum height of the angular distributions and angular half-width containing 68% of the muons (dHxz (68%)) Ec (GeV) dHxz (HWHM) 1 3 10 30 100 300 1000 3000 10,000 dHxz (68%) 1 3 10 30 100 300 1000 3000 10,000 El > 0 31.0 8.20 3.23 1.87 1.12 0.92 0.91 0.98 1.05 28.3 7.36 3.83 3.07 2.87 3.04 3.32 3.62 4.05 El > 1 GeV 2.95 1.73 1.04 0.85 0.82 0.89 0.94 3.06 2.40 2.15 2.18 2.31 2.42 2.57 El > 2 GeV 2.57 1.59 0.95 0.77 0.74 0.78 0.82 2.46 1.95 1.69 1.68 1.75 1.81 1.89 El > 4 GeV 1.87 1.35 0.81 0.66 0.63 0.66 0.67 1.66 1.45 1.21 1.16 1.19 1.23 1.27
For both quantities the values are given for all muon energies and for muon kinetic energies above 1, 2, and 4 GeV. Units of the widths are in degrees.
the energy rises above 300 GeV, there is only a small, gradual increase in the width of the angular distribution, and (iii) the width of the distribution narrows as the muons cuto energy is raised; i.e., the angular resolution becomes better by eliminating the lower energy muons which have, on average, poorer angular resolution. 3.2. Correlation of muon angle with distance from the shower axis In the preceding section, angular distributions were given for all muons regardless of their lateral distance from the shower axis. However, as the distance from the shower axis increases, the angle is systematically biased away from the shower axis. Fig. 5 shows the correlation between the projected angle Hxz and the x-coordinate (i.e., the northward distance from the shower core) of the muon position at sea level for primary gamma ray energies of 10 GeV and 1 TeV. The angular distributions are given for dierent intervals in x and each distri-
bution is normalized to unit area. Here, negative values of Hxz have not been reected onto positive values and are explicitly shown. As expected, the peak of the Hxz distributions are shifted toward positive values as x becomes more positive and the distributions become wider as x increases. Both eects depend on the energy of the primary gamma ray and are more pronounced at higher energies. Hence, the angular distributions narrow significantly if they are limited to muons within a certain interval around the shower axis. Information on this correlation between the angle and the lateral distance of the muon from the shower axis allows experiments to improve the angular resolution by applying cuts to the data, e.g., in x. If the available statistics demand using all available muons, their angle at large xs y s could be corrected by the most probable angle at that x (or y) to obtain a better estimate of the primarys direction. This correlation can also be used to infer the height from which the muon originated, a parameter dicult to obtain experimentally. If one
449
gin would yield, upon extrapolation, an innite height of origin. This problem can be circumvented by nding the mean x within an interval of x and dividing this mean by the corresponding mean of tan Hxz . The averaging serves another purpose as well: the randomness of the various physical processes, such as scattering, tend to cancel in the averaging. The averaged muon height obtained in this fashion could be useful in the case of hadronic primaries in order to study the composition of the hadrons. A new MC study is now underway to study the question of hadronic primaries and how they might dier from the present case of gamma ray primaries. 3.3. Angular resolution for a spectrum of primary gamma ray energies The results of the preceding sections can be combined to obtain the expected angular resolution for a spectrum of primary gamma rays. Various steps in the calculation are shown in Fig. 6a. Here, a dierential energy spectrum dUc =dEc / a Ec with a spectral index of a 2:41 is assumed; this a corresponds to the average of the spectral indices reported in the third EGRET catalog [19]. Folding dUc =dEc with the number of muons reaching sea level per primary photon, Nl =Nc , yields the number of muons at sea level as function of the primary gamma ray energy, dUl =dEc , with Ul Nl =Nc Uc . Note that the dierential spectra in Fig. 6a are multiplied by Ec . The atness of the dierential muon ux per gamma at sea level (Ec dUl =dEc ) above 10 GeV signies that the muons originate rather uniformly from a broad range of primary energies. Below 10 GeV, the muon ux decreases steeply as Nl =Nc rapidly approaches zero. Harder spectral indices (a < 2:41) would enhance the muon ux from higher primary energies and, conversely, softer indices would enhance lower energies. Furthermore, in Fig. 6a the dHxz (HWHM) (given in Table 2 in the El > 0 column) are plotted. These widths are multiplied by the dierential muon ux which is also shown in Fig. 6a. As can be seen, the larger angular widths at the lower energies (Ec < 3 GeV) do not contribute due to the small sea level muon ux at these energies.
Fig. 5. Angular distributions of muons at sea level given for dierent intervals in the x-coordinate. Results are shown for primary gamma ray energies of (a) 10 GeV and (b) 1 TeV. All distributions are normalized to unit area. Units of angle are in degrees.
extrapolates the muon direction backward (upward), the intersection point with the centroid of the shower (which can be obtained from the more numerous electrons in the same shower), is an estimator for the height where the muon was created. However, deviations of the muon production vertex from the centroid and the physical processes which alter the muon track between the production vertex and sea level obscure this information somewhat. For this reason it is not possible to study this question on a muon-by-muon basis. For example, tracks with projected angles close to zero which exist at almost all x-distances from the ori-
450
The average angular resolution (i.e., average angular width) in the North-down plane, hdHxz HWHMi, for a dierential gamma ray spectrum with an index a is obtained from R a d Ec dHxz Nl =Nc Ec R hdHxz HWHMi a Nl =Nc Ec dEc 1 and shown in Fig. 6b. As can be seen, the resolution is about constant (1) up to a spectral index of 2 and then rises for larger indices due to the increasing importance of the wider angular widths at low energy. 3.4. Eect of the magnetic eld The magnetic eld of the Earth deects positively (negatively) charged particles primarily eastward (westward) causing the angular distributions in the projected angle Hyz (i.e., projected onto the East-down plane) to become wider. The size of the eect depends on the strength and direction of the eld and, therefore, varies with geographic location. To obtain a measure of the magnitude of the added angular widths and how it depends on geographic location, three dierent locations are considered and summarized in Table 3. In order to estimate the maximum possible deection, a location of maximum northward eld component was chosen. Due to the oset of the magnetic eld axis from the Earths center, this location is at about 9 N, 100 E (below referred to as maximum). In addition, two locations with intermediate magnetic elds were chosen: 42 N, 86 W (GRAND) and 36 N, 106 W (MILAGRO). The magnetic eld components for these locations were obtained from the latest revision of the International Geomagnetic Reference Field (IGRF) [20] and are listed in Table 3. Here, the components of the magnetic eld, B, are given in nanotesla (nT) and the components Bx , By , Bz are directed North, East, and toward the center of the Earth, respectively. For uniformity of comparison, all values are for sea level elevation. F L U K A is able to transport particles in arbitrary magnetic elds. However, since the variation of the eld in the atmosphere with height above sea level and with time is only minor (about 4%), for
Fig. 6. (a) The gure shows the following quantities as a function of the primary gamma ray energy: the energy spectrum of primary gamma rays (dUc =dEc ) with a shape given by a dierential spectral index a 2:41 and arbitrary normalization, the multiplicity of muons at sea level per primary gamma ray (Nl =Nc ), the result of folding the multiplicity with the primary spectrum (dUl =dEc Nl =Nc dUc =dEc ; thin solid line with circles), the HWHM for all muon energies (dHxz (HWHM)) in degrees, and the result of folding these widths with dUl =dEc (dot-dashed lines with diamond points). The calculated values are joined by lines to guide the eye. Note that all dierential uxes are multiplied by Ec . (b) Average HWHM for the angle projected onto the North-down plane as function of the spectral index of the primary spectrum; the ordinate is in units of degrees. Values p for the space angle resolution (dHR (HWHM)) would be 2 times larger.
451
Table 3 Earths magnetic eld components Bx (North), By (East), and Bz (vertically downward) in nanotesla for the three considered locations Location GRAND, 42 N, 86 W MILAGRO, 36 N, 106 W Maximum, 9 N, 100 E Bx (nT) 18,592 22,751 41,407 By (nT) 1361 4141 211 Bz (nT) 52,547 46,241 1326
The eld has been obtained from the IGRF 2000 model [20]. For a given location the eld components vary only by 6 4% for heights of 080 km. Here, the values are given for sea level and a date in February, 2001.
simplicity the magnetic eld used in the calculation for a certain location was assumed to be constant throughout the whole geometry. In addition, although the MILAGRO experiment is located at a height of 2170 m above sea level, all results reported for that location refer to sea level as this allows a direct comparison to the results at the other locations. The angular distributions at sea level for the three magnetic eld conditions and for primary gamma ray energies of 10 GeV and 1 TeV are shown in Fig. 7. In addition, the distributions obtained without magnetic eld are given. Comparing these curves gives the eect of the magnetic eld upon the muons angle. As can be seen, for a 10 GeV gamma primary, the eect of the Earths magnetic eld is considerable and changes with geographic location; as the northward component of the magnetic eld increases, the width increases. However, at 1 TeV, there is only a slight widening of the distributions with increasing the North component of the eld. Since we are dealing with a gamma ray primary, the magnetic eld only aects the secondaries; the main eect is upon the muon which has, on average, the lowest momentum and the longest path length. To within the statistical uncertainties of the calculations, the total number of muons which reach sea level per primary gamma ray is independent of the eect of the Earths magnetic eld for the two energies studied in this section (10 GeV and 1 TeV). The half-widths of the angular distributions in the North-down (Hxz ) and East-down planes (Hyz ) containing 68% of the muons are presented for the dierent locations in Table 4. Again, the widths are given for distributions containing only muons above certain kinetic energy thresholds (including zero threshold, or no cut). In addition, the corre-
Fig. 7. Muon angular distribution at sea level including the eect of the Earths magnetic eld for (a) 10 GeV and (b) 1 TeV primary gamma ray energy. The distributions are shown as a function of the muons angle projected onto the yz (Eastdown)-plane, Hyz , and for three dierent geographic locations: (1) 42 N, 86 W (GRAND), (2) 36 N, 106 W (MILAGRO), and (3) 9 N, 100 E (maximum). The magnetic deection for Hxz (North-down) is minimal and not shown. In addition, the distribution obtained without magnetic eld (labeled no eld) is given. Units of angle are in degrees.
452
Table 4 The half-widths of the muons sea level angular distributions which contain 68% of the muons and their dependence on the magnetic eld strength Ec (GeV) n.f. (1) (2) (3) 10 1000 10 1000 10 1000 10 1000 El > 0 (GeV) N-d. 3.83 3.32 3.85 3.31 3.89 3.33 3.86 3.32 E-d. 3.84 3.32 4.41 3.48 4.73 3.61 6.44 4.19 E-d.(b) 3.41 3.03 3.38 3.0 3.40 3.02 3.38 3.02 El > 1 (GeV) N-d. 3.06 2.31 3.06 2.30 3.09 2.32 3.05 2.31 E-d. 3.06 2.30 3.53 2.42 3.83 2.52 5.28 2.93 El > 2 (GeV) N-d. 2.46 1.75 2.47 1.74 2.48 1.75 2.44 1.74 E-d. 2.45 1.74 2.88 1.84 3.15 1.91 4.44 2.24 El > 4 (GeV) N-d. 1.66 1.19 1.67 1.19 1.67 1.19 1.64 1.20 E-d. 1.64 1.19 2.03 1.26 2.25 1.30 3.34 1.56
Results are presented for three dierent geographic locations: (1) 42 N, 86 W (GRAND), (2) 36 N, 106 W (MILAGRO), and (3) 9 N, 100 E (maximum) in order of increasing values of Bx . For each location the widths are given for the projected angles onto the North-down (labeled N-d.) and East-down (labeled E-d.) planes. The reference values from the calculations with no magnetic eld (labeled n.f.) are also shown; here, the small dierences between North-down and East-down are only statistical uctuations. Widths are listed for muon kinetic energies above 0 (all), 1, 2, and 4 GeV for primary gamma energies of 10 and 1000 GeV. The additional column for El > 0 gives the widths in the East-down plane of the muons at their production vertex (birth, labeled b). Units of the widths are in degrees.
sponding values from the calculations without the eect of the magnetic eld are given. The widths in the North-down plane are much less aected than in the East-down plane which contains the dominant eect of the magnetic deection. This eect increases with increasing values of Bx or decreasing values of muon momentum. In addition to the widths of the muons sea level angular distributions, Table 4 also shows the values at their production vertex (birth, E-d.(b)). Interestingly, these values are almost constant for the dierent eld conditions. The eect of the magnetic eld on the angular distributions is therefore mainly the deection of the muon over its longer path and not the deection of the parent pions or other charged particles preceding in the cascade due to their shorter path lengths and higher momenta as compared to the muons. In addition, the dominant factor in the nal width (except for the highest magnetic eld value) is the muons angular distribution at its production (birth) point.
4. The eect of pressure and temperature variations Air pressure and temperature changes cause small variations in the density prole of the at-
mosphere and thus aect the muon ux at sea level. In order to obtain a quantitative estimate of these two eects, the following cases were studied: (i) the air density in each layer was increased by 3% while all other variables (geometry, cutos, etc.) were kept constant and (ii) the height above sea level of each slab boundary was raised by 5% and the corresponding density decreased in the same proportion such that the total thickness of air (in g/cm2 ) above the surface of the Earth remained constant. The former modication simulates an air pressure increase (p) whereas the latter simulates an overall 5% increase in absolute temperature (T) of the atmosphere. Since muons are produced at atmospheric heights far above those which determine the weather at the surface of the Earth, the mean T which is involved in this calculation has little correlation with the temperature at the Earths surface. Thus the temperature eect calculated here is not expected to correlate with the Earths surface temperature but would require temperature measurements of the Earths atmosphere averaged from 0 to 20 km above the Earths surface as obtained, for example, in weather balloon measurements. The results presented in this section allow an investigation of, or corrections for, the muon ux due to small variations in the
453
Table 5 Summary of the eect of pressure and temperature variations on the average muon multiplicity per primary gamma ray at detection level, Nl =Nc Ec (GeV) 10 1000 Muon multiplicity per primary gamma ray, Nl =Nc Reference atmosphere 5:29 0:03 104 0:221 0:001 3% Density increase 5:08 0:02 104 0:214 0:001 5% Temperature increase 4:97 0:02 104 0:213 0:001
The column labeled reference atmosphere repeats the values listed in Table 1 for the reference problem. The 5% temperature increase corresponds to an average increase in the absolute temperature prole from 0 to 80 km.
atmospheric pressure and height-averaged temperature. The results for the total muon yield per primary gamma ray are summarized for two energies (10 GeV and 1 TeV) in Table 5. In all cases the muon yields decreased by 36%, the decrease being more pronounced at the lower energy. As for increased pressure, a higher density increases the interaction probability of the pions and thus decreases the probability of decay, i.e., fewer muons are produced. As well, the added thickness of air which the muons must traverse is a barrier which some will fail to overcome. On the other hand, raising the boundaries of the layers and reducing their density causes two counterbalancing eects: (i) the pions propagate in a less dense medium and thus have a greater probability to decay rather than interact thus producing more muons, and (ii) the muons must propagate longer distances resulting in a greater probability that they will decay before reaching ground level. The calculated temperature dependence of the muon multiplicity ratio suggests
that the enhanced muon decay probability dominates at these energies and is the more dominant at the lower energy.
5. General properties of muon production in gamma ray showers Information on the muons at sea level and on their ancestors (i.e., the parent, grandparent, etc., see Section 2.2) allows a more detailed study of the properties of these ancestors. 5.1. The ancestors of the muons Table 6 shows which parent contributes through its decay to the detected muons for the dierent primary gamma ray energies. At the lowest energy (1 GeV) about 89% of the muons are produced in decays of negative pions and 11% in decays of positive pions. This asymmetry can be explained by the dierent interaction cross sections
Table 6 Fractional contributions to the parents of the muons which reach sea level Ec (GeV) 1 3 10 30 100 300 1000 3000 10,000 p 0.106 0.495 0.492 0.482 0.478 0.477 0.475 0.476 0.474 p 0.894 0.485 0.489 0.482 0.477 0.476 0.476 0.475 0.477 K 0.0 0.020 0.011 0.019 0.022 0.023 0.024 0.025 0.024 K 0.0 0.0 0.007 0.014 0.018 0.019 0.019 0.020 0.020 Neutral kaons 0.0 1:7 104 9:8 104 3:1 103 4:4 103 4:7 103 5:2 103 5:1 103 5:2 103
Values are given for dierent energies of the primary cosmic gamma ray.
454
at low energy of pions of either charge and by the fact that these pions are mainly produced in secondary interactions of neutrons (see below). Correspondingly, the probability for positive pions to interact instead of decay is larger than for negative pions. Above 3 GeV this asymmetry disappears and decaying kaons begin to contribute to the sea level muon ux. The latter contribution increases with energy and amounts to about 5% at 10 TeV. It is interesting to note that at all energies F L U K A predicts the relative contribution of a positive kaon parent to be always larger than that of a negative kaon by about 20%. This eect is due to the properties of the DPM describing inelastic hadronic interactions within F L U K A . In particular, it is a feature of the Reggeon contribution which describes particle production by one quarkdiquark string stretched between a valence quark of the uctuating photon and a diquark of a target nucleon. In this picture, kaon production involves the creation of a s s quarkantiquark pair and, in case of negative kaons, also the creation of an uu pair. On the other hand, positive kaons can be readily formed also by a u-quark of the uctuating photon leading to the observed asymmetry. The fractional contribution to the muons grandparent is given in Table 7. The calculations for 1 GeV show distinctly dierent features. About 97% of the muons at sea level originate from mesons produced in interactions of neutrons in the close vicinity of the detector. At all other energies photoproduction dominates the picture with a contribution decreasing with energy from 99% at 3
GeV to 60% at the highest energy; the remaining fraction is mainly from pions. 5.2. Distributions of the number of generations The distributions of the generation number of the grandparent of the sea level muon is shown in Fig. 8. The lower the energy of the primary gamma ray the smaller is the atmospheric shower resulting in a relatively narrow distribution. This distribution is peaked at generation one for energies below a few hundred GeV where the parent is mainly produced in a photoproduction process of the primary photon (see also Table 7). At higher energies secondary hadron interactions contributes signicantly to the production of the parent. These secondary hadrons are mainly of third or fourth generation and cause a shift of the peak of the total distribution at TeV energies. The large tail at these energies which extends up to more than 100 generations reects photoproduction processes of secondary photons in the large electromagnetic shower. 5.3. The muon production heights The distributions of heights above sea level at which the detected muons were produced are shown for four primary energies in Fig. 9. As expected, at high energy there is a decrease in height with increasing energy of the primary gamma ray because of the logarithmic increase of the shower length. However, as the energy of the primary gamma ray decreases below 10 GeV, the energy of
Table 7 Fractional contributions to the grandparent of the detected muon at sea level Ec (GeV) 1 3 10 30 100 300 1000 3000 10,000 c 1:7 10 0.99 0.99 0.97 0.91 0.83 0.73 0.65 0.60
3
p and p 0.023 3:9 104 8:2 104 4:1 103 0.014 0.025 0.037 0.049 0.054
n and n 0.97 5:0 103 1:9 103 5:3 103 0.015 0.026 0.041 0.049 0.057
p 7:8 10 2:3 103 2:6 103 9:3 103 0.026 0.056 0.091 0.12 0.13
4
p 4:9 103 2:1 103 2:7 103 9:2 103 0.027 0.056 0.090 0.12 0.14
455
Fig. 8. Distribution in the number of generations in the shower (see text) carried by the particles creating the parent meson. F L U K A results are shown for dierent energies of the primary photon.
Fig. 10. Radial distribution of the muon production vertices (i.e. meson decay vertices, dotted histograms) and of the muons at sea level (solid histograms). Results are given for primary photon energies of 3, 10, 100, 1000, and 10,000 GeV, respectively (from the bottom curve to the top). Values have been normalized per primary photon.
resents the excess contribution from very low energy muons. 5.4. Radial distributions The radial distributions of the muons at their production vertices and at sea level are shown in Fig. 10 for ve dierent gamma ray energies. The radial distance, R, is dened with respect to the shower axis (z-axis). The quantity dN =dA denotes the number of muons per unit area and per primary gamma ray. The radial distributions of the production vertices are labeled production. All distributions extend to more than 10 km. Whereas the sea level distributions have a relatively at shape below R 2 km the production vertex distributions are increasing toward the shower axis and exhibit a change in slope or discontinuity at about 500 m which is most pronounced at low primary energy. This discontinuity indicates that two components with dierent shapes contribute. In order to further investigate this feature the muon production vertices were scored separately for those muons which originated from mesons produced in photoproduction processes and those from mesons produced by interactions of nucleons or other hadrons. The
Fig. 9. Distribution of the muon production heights. Results are given for primary photon energies of 3, 10, 100, and 1000 GeV, respectively, and are normalized per primary photon. The top of the atmosphere is at 80 km in the calculations.
the muon parent is close to the minimum energy required for the muon to penetrate the blanket of air between its production and sea level. Hence, the most probable height of muon production begins to decrease with energies decreasing below 10 GeV. The blip-up near zero height, most pronounced in the 3 and 10 GeV distributions, rep-
456
contributions from these components are shown for a primary energy of 10 GeV in Fig. 11. As can
Fig. 12. Kinetic energy spectra of muons at their production vertices (dotted histograms) and at sea level (solid histograms). Values are normalized per primary photon.
be clearly seen in the production vertex distributions (Fig. 11a) the photoproduction component dominates at small radii and the component due to interacting nucleons constitutes the tail at large R. The dierence is smaller at sea level (Fig. 11b) due to multiple scattering of the muons in air. 5.5. Energy distribution of the muons The kinetic energy distributions of muons at sea level and at the production vertex of these muons are shown in Fig. 12. Note, that the distributions are multiplied by the muon energy El in order to enhance possible spectral structures at the higher energies. Whereas the sea level spectra are a smooth function with energy, the distributions of the muon energies at their birth exhibit a twocomponent structure. This two-component structure is again due to the superposition of muons from mesons generated in photoproduction processes and those from interacting hadrons. The two contributions are plotted separately in Fig. 13 for 10 GeV primary photon energy. It is clear that muons from photoproduction processes have a higher average energy. 6. Conclusion Air showers caused by cosmic gamma rays with energies below 10 TeV were simulated using the
Fig. 11. Radial distribution of the muon production vertices for 10 GeV primary photons at normal incidence. The distributions are given for (a) the muon production vertex and (b) for sea level. In addition to the total distribution, the contributions from dierent parents of the decaying meson are given. For example, the histograms labeled photons show the radial distributions of muons from mesons which were produced by photonuclear interactions. Distributions are normalized per primary photon.
457
Fig. 13. Kinetic energy spectra of muons from 10 GeV primary photons at normal incidence. The spectra are shown for (a) the production vertex and (b) for sea level. In addition to the total spectra which are identical to those shown in Fig. 12, the contributions from dierent parents of the decaying meson are given (similar to Fig. 11).
MC code F L U K A . The primary gamma rays were assumed to enter the atmosphere vertically. The reliability of F L U K A predictions in this energy re-
gion was explored by comparing its predictions for proton- and nuclei-induced air showers with experimental data; good agreement was found. Many general properties of muon production in atmospheric gamma ray showers were studied. As expected, most of the muons which reach sea level are decay products of charged pions; the contribution from kaons increases as the primary energy increases. Below 100 GeV primary energy, the decaying mesons are produced mainly in photoproduction processes. At higher energies, hadronic interactions of nucleons and pions contribute signicantly. Variations in atmospheric pressure and temperature were investigated: A 3.0% increase in atmospheric pressure decreases the muon yield at sea level by 4.0% for 10 GeV primaries (3.2% at 1 TeV). Similarly, a 5.0% increase in the heightaveraged temperature decreases the muon yield by 6.0% (3.6% at 1 TeV). The main goal of the study was to determine the angular resolution for primary cosmic gamma rays from measurements of secondary muon angles at ground level. The angular distribution of muons at sea level narrows as the muon energy increases since low energy muons constitute the tails of the angular distributions. Thus, the angular resolution can be improved by eliminating the lower energy muons. The HWHM of the muons projected angles are 6 1.1 above 100 GeV but rise to a value of 3:2 as the primary gamma ray energy is lowered to 10 GeV in the absence of the Earths magnetic eld deection. Primary energies below 10 GeV essentially do not contribute muons at ground level. The eects of magnetic bending were considered separately as they depend on several factors; for example, they are more important in the East-down plane, for low primary energies, low muon energies, and larger North-components of the Earths magnetic eld. The width of the angular distribution can be narrowed if the correlation of Hxz (angle projected onto the North-down plane) with x is taken into consideration (assuming the location of the core of the shower is measured). This narrowing can be accomplished either by eliminating large x and y values from the data or by compensating the measured angles knowing their expected mean values versus x and y as calculated from F L U K A .
458
J. Poirier et al. / Astroparticle Physics 17 (2002) 441458 Environments, CERN 1995, CERN Report TIS-RP/97-05, 1997, p. 158. et al., in: H. Hirayama (Ed.), Proceedings of the A. Fasso third Workshop on Simulating Accelerator Radiation Environments, KEK 1997, KEK Proceedings 97-5, 1997, p. 32. , A. Ferrari, P.R. Sala, Electronphoton transport A. Fasso in F L U K A , in: A. Kling et al. (Eds.), Proceedings of the International Conference on Advanced Monte Carlo for Radiation Physics, Particle Transport Simulation and Applications, Monte Carlo 2000, Lisbon, 2326 October 2000, Springer, Berlin, 2001, pp. 159164. et al., F L U K A : status and perspectives for A. Fasso hadronic applications, in: A. Kling et al. (Eds.), Proceedings of the International Conference on Advanced Monte Carlo for Radiation Physics, Particle Transport Simulation and Applications, Monte Carlo 2000, Lisbon, 2326 October 2000, Springer, Berlin, 2001, pp. 955960. V. Patera et al., Nucl. Instr. Meth., Phys. Res. A 356 (1995) 514. A. Ferrari, T. Rancati, P.R. Sala, in: H. Hirayama (Ed.), Proceedings of the third Workshop on Simulating Accelerator Radiation Environments, KEK 1997, KEK Proceedings 97-5, 1997, p. 165. S. Roesler, W. Heinrich, H. Schraube, Rad. Res. 149 (1998) 87. G. Battistoni et al., Astropart. Phys. 12 (2000) 315. A. Capella et al., Phys. Rep. 236 (1994) 225. T.H. Bauer, R.D. Spital, D.R. Yennie, Rev. Mod. Phys. 50 (1978) 261. R. Engel, Z. Phys. C 66 (1995) 203. S. Roesler, R. Engel, J. Ranft, Phys. Rev. D 57 (1998) 2889. M. Boezio et al., Phys. Rev. D 62 (2000) 032007. S. Roesler, W. Heinrich, H. Schraube, Monte Carlo simulation of the radiation eld at aircraft altitudes, SLAC-PUB-8968, August 2001. J. Kremer et al., Phys. Rev. Lett. 83 (1999) 4241. R.C. Hartman et al., Astrophys. J. Suppl. Ser. 123 (1999) 79. International Association of Geomagnetism and Aeronomy (IAGA), Division V, Working Group 8, International Geomagnetic Reference FieldEpoch 2000, Revision of the IGRF for 20002005, http://www.ngdc.noaa.gov/ IAGA/wg8/wg8.html.
Finally, the average muon angular correlation was calculated for a spectrum of primary gamma rays characterized by a spectral index. For a soft spectrum, e.g. a dierential spectral index a 3:0, the average projected angular width (HWHM) is 2:6; for a mean value, a 2:41, the width is 1:6. If the spectrum is harder, e.g., a 6 2:0, the angular width improves to a constant value of 1:0. Corresponding values for pa space angle resolution (dHR (HWHM)) are 2 larger. As mentioned above, if data with larger x and y values and/or the lower muon energies are removed, this angular correlation can be improved. This precise calculation of the angular correlations provides experiments which study the angular location of primary gamma ray sources by measuring muon angles with information on the angular resolution which is dicult to obtain experimentally.
[5]
[6]
[7]
[8] [9]
Acknowledgements Part of this work was supported by the Department of Energy under contract DE-AC0376SF00515. Project GRAND is funded through grants from the University of Notre Dame and private donations.
[10] [11] [12] [13] [14] [15] [16] [17]
References
, J. Poirier, Phys. Rev. D 63 (2001) 036002. [1] A. Fasso [2] The MILAGRO Collaboration, R. Atkins et al., Nucl. Instr. Meth., Phys. Res. A 449 (2000) 478. [3] J. Poirier et al., in: Proceedings of the 26th International Cosmic Ray Conference (ICRC), vol. 5, 1999, p. 304, http://www.nd.edu/$grand. et al., in: G.R. Stevenson (Ed.), Proceedings of [4] A. Fasso the second Workshop on Simulating Accelerator Radiation
[18] [19] [20]
ARTICLE IN PRESS
Nuclear Instruments and Methods in Physics Research A 507 (2003) 657669
Development of a two-fold segmented detection system for near horizontally cosmic-ray muons to probe the internal structure of a volcano
H. Tanakaa,d,*, K. Nagaminea,b, N. Kawamurab, S.N. Nakamurac, K. Ishidaa, K. Shimomurab
a b
Muon Science laboratory, RIKEN (The Institute of Physical and Chemical Research), Wako, Saitama 351-0198, Japan Meson Science Laboratory, High Energy Accelerator Research Organization, Oho 1-1, Tsukuba, Ibaraki 305-0801, Japan c Department of Physics, Tohoku University, Sendai 980-8578, Japan d Department of Earth and Planetary Science, Nagoya University, Furo-cho, Chikusa, Nagoya, Aichi 464-8602, Japan Received 12 December 2002; accepted 18 March 2003
Abstract Very high-energy cosmic-ray muons penetrating through a mountain enable us to probe internal structure of volcanoes. An improved cosmic-ray muon detection system comprising two segmented detectors with multiplicity cut of the soft-component background of cosmic ray was developed. By applying to the measurement on internal structure of the volcano Mt. Asama, we proved that the volume occupancy in the region of a crater is less than 30%. r 2003 Elsevier B.V. All rights reserved.
PACS: 29.90.+r; 91.35.Gf; 91.40.k; 96.40.Tv Keywords: Cosmic-ray muon; Segmented detection system; Multiplicity-cut analysis
1. Introduction The intensity of cosmic-ray muons penetrating through a material is uniquely determined from the zenith-angle dependence of the muon energy spectrum and the rangeenergy relation. This is
*Corresponding author. Muon Science Laboratory, RIKEN (The Institute of Physical and Chemical Research), Wako, Saitama 351-0198, Japan. Tel.: +81-48-467-9355; fax: +81-48462-4648. E-mail address: ht@riken.go.jp (H. Tanaka).
applicable to observations of the time-dependent variation of the density distribution of volcanoes, which may contribute to a database for the prediction of an eruption of a volcano and to clarify the dynamics of volcanic activity. We have promoted a project to measure the ux of cosmic-ray muons passing through a mountain since 1994 [1]. In 1994, an analog system was adopted. In this case, a straight line connecting the intersecting points of cosmic-ray muons at three counters determined the path, where these points were
0168-9002/03/$ - see front matter r 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0168-9002(03)01372-X
ARTICLE IN PRESS
658 H. Tanaka et al. / Nuclear Instruments and Methods in Physics Research A 507 (2003) 657669
determined by the difference in the arrival time of scintillation light among four phototubes attached to the corner of the scintillators. Soft components (electrons and positrons) were rejected by requesting a straight-line trajectory through three analog counters for the cosmic-ray muons. However, because the spatial determination by the analog system was affected by the gain-drift of the phototubes, this detection system was not suitable for long-term use. In the present work, a segmented detection system was adopted. The reasons why we chose this system are as follows: 1. We claried the denition of the arriving angles of cosmic-ray muons. 2. This segmented detection system enables us to identify the counter segment causing the entire problem by the real-time monitoring of each phototube so that a stable long-term operation can be easily realized. In order to realize long-term and good S=N observations, a detection system comprising two segmented detectors with the multiplicity cut analysis was developed. The multiplicity method, which eliminates soft components by using cascade shower generation through an intermediate iron absorber, enables us to reduce the number of detectors from three to two. We describe a measurement for the volcano Mt. Asama lying across the NaganoGunma border in Japan. A crater is located at the top region with a diameter of 350 m and a depth of 228 m:
yn is Nm Em ; yn dEm AWm Em DEm

g
n 1 rg p Bp sec y Em DEm Bp sec yn # 1 n rg B sec y K K 0:36br 2:1 dEm Em DEm BK sec yn
"
where Wm is the muon survival probability [1]. The ratio of the muon momentum to the momentum of the parent pion rp is 0.78, and that of the parent Kaon rK is 0.52. The decay length of the mesons which have an energy of Bp 90 GeV and BK 442 GeV correspond to the thickness of the atmospheric layer. The branching ratio of the Km2 decay mode is br 0:635; and DEm is the muon energy loss in the atmosphere from the top to sea level. 2.2. Range of cosmic-ray muons through rock A very high-energy muon loses energy while passing though matter in several ways. The mean energy loss of a muon through matter is expressed in a function of E as dE =dx kE bb E E bp E E bn E : 2:2
The terms bb E E ; bp E E and bn E are proportional to losses to bremsstrahlung, pair production, and nuclear interactions, respectively. Their values have been evaluated [3]. Therefore, a unique relationship between x and the intensity of penetrating muons Nm Ec ; yn is determined from Eqs. (2.1) and (2.2), where Ec is the minimum energy of the cosmic-ray muons arriving at the zenith angle yn which can penetrate through rock. The relationship is shown in Fig. 1. 2.3. Rejection of an electron shower background
2. Principle 2.1. Zenith-angle dependence of the cosmic-ray muon ux The Thompson and Whalley expression [2] assumes that the muon parents are pions and kaons with production spectra of the form AE g with gD2:70: The differential muon spectrum Nm at sea level arriving at zenith angle
The typical energy of the soft components (electrons, positrons, and photons) spans from 0.1 to 2:0 GeV=c at the ground level of an atmospheric depth of 945 g=cm2 [4]. The neutral pion, the secondary particle as a result of collision of the primary cosmic ray with atmospheric nucleus, decays immediately into two photons.
ARTICLE IN PRESS
H. Tanaka et al. / Nuclear Instruments and Methods in Physics Research A 507 (2003) 657669 659
Fig. 1. Integrated ux of cosmic-ray muons at various angles penetrating through a given thickness of rock m with a density of 2:5 g=cm3 ; as obtained with Eqs. (1) and (2). The vertical axis of Ref. [1] was corrected.
They also create new particles, an electron and a positron, by the pair-creation process. Electrons and positrons, in turn, produce more gamma-rays by the bremsstrahlung process, which initiate a large cascade air shower. A muon also radiates a photon of high energy, mainly through the muondecay process and the knock-on process in a material; the photon initiates a large cascade shower. Such soft components are the background noise for our measurements. In the detection of cosmic-ray muon passing through a mountain, it is also necessary to take account of the contributions of the soft components. In order to reject an electron shower background, we utilized the cascade shower generation of electrons through iron plates. When an electron passes through a material, it causes a cascade shower, and the number and spatial extent of e and e increases. Concerning the produced secondary particles in the iron plates, it was assumed that two or more particles with a common origin hit mutually separated segments in the detector plane within 20 ns (multiplicity event). If the energy of the primary soft components is very high, the total number of particles in a shower is very large, and it
may extend for several meters. We can reduce the soft-component shower background by rejecting such multiplicity events. In order to estimate the amount of multiplicity from the soft components, by using Geant 3 [5], a Monte-Carlo simulation was carried out for electrons/positrons with energies of up to 2 GeV based upon the above-mentioned soft component energy spectrum, as shown in Fig. 2. An iron plate with a thickness of one radiation length 1:7 cm hardly absorbs low-energy electrons B100 MeV: On the other hand, an iron plate with a thickness of three radiation lengths 5:1 cm well absorbs low-energy electrons and produces many secondary electrons for higher incident energies. Fig. 3 shows that the electron/positron showers as a result of cascade generation spread over a large area in the detector, and also shows that the arrays of the counters operated for the multiplicity analysis method are usefull for rejecting such events. These Monte-Carlo simulation results indicate electrons/positrons with primary energy E0 > 0:1 GeV well spread through an iron plate with a thickness of more than three radiation lengths. 2.4. Two-dimensional map of the density length to be obtained In the measurement procedure, various muon paths X yn ; f are recorded. Since, in most cases, the size of the counter is much smaller than the spacial resolution of the vertex point at the objective substance, the path of the cosmic-ray muon can be represented by the azimuthal and zenith angles with reference to a line perpendicular to the detector plane y; f; y 90 yn ; as shown in Fig. 4. Thus, a histogram of the Nm events as a function of yn ; f can be produced. Nm yn ; f can be normalized by the cosmic-ray muons passing through nothing in order to obtain the ratio nyn ; f: As a result, the dependences of the absolute intensity of cosmic-ray muons, solid angles, and the counter efciency are canceled. Using the data in the form of either Nm yn ; f or nyn ; f with an angular resolution of Dyn and Df; corresponding to L=l Dy and L=l Dx;
ARTICLE IN PRESS
Fig. 2. Calculated number of multiplicity from electron/positron scattering: (a) 0:1 GeV incident electrons; (b) 0:5 GeV incident electrons, and (c) 2:0 GeV incident electrons as obtained with the Geant Monte-Carlo code [5].
respectively, we can obtain the values of X yn ; f with steps of Dyn and Df:
3. Description of the apparatus 3.1. A segmented detection system In order to identify the path of cosmic-ray muons, two segmented detectors were employed in this experiment. The segmented detector consisted of x and y planes. Each plane consisted of an array of counters. Each counter with a size of 100-cm length 10-cm width 3-cm thickness was composed of a plastic scintillator (Bicron, BC-408) and a photo-multiplier tube (Hamamatsu, H7195). Twenty counters were arranged in the x and y planes, and were segmented by a 10-cm square. As
shown in Fig. 4, the path of a muon can be determined by the combination of two signals from the Nos. 1 and 2 segmented detectors. The two segmented detectors were placed at a distance of 1:5 m in order to achieve 766-mrad angular resolution. This angular resolution corresponds to a 7200-m spatial resolution at 3000 m apart. The effective vertical angle is 0o yo588 mrady 90 y * ; and the effective azimuth angle is 0ojfjo588 mrad: To reject cosmic-ray soft components without using cosmic-ray trajectory information, we needed an additional trigger condition to reject the soft components. High-energy cosmic-ray muons which can penetrate a mountain cause a single-hit event. However, in some cases the soft component of a cascade shower causes a multi-hit event. For this purpose, two 5-cm-thick (around
ARTICLE IN PRESS
Fig. 3. Electron/positron showers as a result of the cascade generation spread over a large area in the detector. These Monte-Carlo simulation results for 0.1, 0.5, 1.0, and 2:0 GeV indicate electons/positrons with primary energy E0 > 0:1 GeV well spread through an iron plate with a size of more than 3X0 (radiation lengths).
three radiation lengths) iron plates were placed between the two counter planes. The triggering requirement for the soft-component rejection analysis is a coincidence within 20 ns of at least two out of 10 counters within the same plane (multiplicity cut analysis method). A high voltage was applied to each photo multiplyer tube (PMT). The gain of each PMT used in this observation was adjusted so that the
difference of a single counting rate of counters would be within 710% by measuring each PMTs pulse height. 3.2. Electronics and data taking A schematic diagram of the data-taking electronics is shown in Fig. 5. Time to Digital Converter (TDC; Kaizu, KC3781A 12 bit=25 ps) circuits
ARTICLE IN PRESS
Fig. 4. Segmented detection system. Two segmented detectors were placed with a distance of 1:5 m to achieve a 766-mrad angular resolution. The effective vertical angle is 0oyo588 mrad; the effective azimuthal angle is 0ojfjo588 mrad:
narrow enough for a cosmic-ray event rate of B100 Hz: Accidental backgrounds are almost rejected by requesting the signal coincidence of each layer. The data from each TDC, that is, cosmicray muon events Nm t are accumulated in a number of bins representing the arriving angles of cosmic-ray muons, are transferred to the computer controlled by the data acquisition system, EXP95 [6], where the most important correction of rejecting multi-hit events was carried out. The operation of two 1-m2 cosmic-ray muon detection systems started on February 2001, as shown in Fig. 6. Forty counters were protected by aluminum housing in order to prevent thermal and mechanical damage caused by long-term use. Sufcient holes in this housing exist in order to maintain air circulation around the PMTs. An air conditioning system kept the room temperature at B20225 C in order to protect the detection system from thermal damage. We considered investing in added protection for our data acquisition with an Uninterruptible Power Supply (UPS) unit. Two detection systems were placed in a 6-m container (standard carrying container used for oceangoing cargo), which made it possible for us to move it in response to needs. 3.3. Analysis The data from the 40 TDCs were converted into two-dimensional histograms y; f on the online monitor. In order to determine the arriving angle of a cosmic-ray muon, a multiplicity analysis was carried out by the following procedures: (A) Because a cosmic-ray muon makes an event trigger, a real muon signal is observed at a certain time in the TDC time spectrum. (B) The spatial positions of a cosmic-ray muon hitting on the two segmented detectors were determined from a combination of the coincident TDC timings. (C) From the straight line connecting the positions on the two segmented detectors, the arriving angles were determined.
Fig. 5. Diagram of the electronics and computer system used for the present system.
measure all timings of the 40 PMTs: The start signal to TDC and the event trigger are generated by a coincidence of signals of any PMTs of the x and y planes of the Nos. 1 and 2 detectors in a certain time gate. The gate width for muon data capture is 100 ns: It is wide enough for a muontransverse time of 5 ns between two detectors, and
ARTICLE IN PRESS
Fig. 6. Schematic view of the experimental setup.
(E) The relative timing from two detectors, which was used to identify the direction of the muon path, provided Nm Em ; yn for forward and for backward directions. The zenith-angle dependence of cosmic-ray muons, as obtained with and without a multiplicity cut, is shown in Fig. 7. The data eliminating multiplicity events are consistent with the estimation in Fig. 1 with a rock thickness of zero.
4. Measurement on Mt. Asama; results and discussion As the rst-step application of our detection system, we performed a measurement of the internal structure of the volcano Mt. Asama. Fig. 8 shows the geometrical arrangement for the detection system versus Mt. Asama taken in the present measurement together with the location of Mt. Asama on the main island of Japan, Honshu. A detection system was placed at the foot of
Fig. 7. Zenith-angle dependence of the cosmic-ray muon intensity, as obtained with and without a multiplicity cut.
(D) When more than two signals from the same layer coincided in the time gate, such an event was discarded as soft component (multiplicity cut).
ARTICLE IN PRESS
Fig. 8. Geometrical arrangement for the detection system versus Mt. Asama taken in the present measurement: (upper) conceptual three-dimensional view; (middle) The location of Mt. Asama in the main island, Honshu; (lower) a topological map.
Mt. Asama (Asama-en), located in Gunma Prefecture, Japan; its elevation is 1400 m from sea level and its horizontal distance is 3750 m from the crater, which is 300 m in diameter and 228 m in depth.
4.1. Monte-Carlo simulation of cosmic-ray muon events passing through Mt.Asama To examine Nm y; the following Monte-Carlo simulation was carried out. In the present
ARTICLE IN PRESS
simulation, the process was reversed in order to save simulation time. Namely, the cosmic-ray muons were initiated at the detection system. In the course of the computer analysis, a calculation was made to determine the density length average density path length for each path of muons. The path lengths were determined from the contours of the mountain through a topological map published by Geographical Survey Institute. The process of cosmic-ray event generation consisted of ve steps, as follows: 1. Generation and determination of cosmic-ray muons Nm E ; yz with an energy of E for a zenith angle of yn and for an azimuthal angle of f from the proper distribution function, Eq. (1). 2. Obtaining the path lengths of muons for zenith angles of yn and for azimuthal angles of f from a topological map of Mt. Asama. 3. Determination of the density length Ly; fr along the muon paths for a certain density r of a mountain. 4. Determination of the cosmic-ray muon intensity decrease as a result of passing through a mountain, from Eq. (2). 5. Return the momentum and intersection position on the detection system. Muons with momenta of less than 10 GeV=c were simulated to conrm their behavior. None of them penetrates a mountain. Therefore, the low cut-off energy was set to 10 GeV: For the high-momentum region, the probability decreases rapidly, we may thus cut off the very highmomentum region. We set the high cut-off energy to 10;000 GeV; almost 100% of the entire muon. After all, this contributes to saving simulation time. Since our detection system was segmented, the angular resolution would be strictly determined. A Monte-Carlo simulation for several crater-ll conditions of Mt Asama predicted the expected experimental results. 4.2. Vertical and the azimuthal-angle distribution of muons in Mt. Asama First, a comparison between the zenith angular dependence of cosmic-ray muons for forward and
backward was performed. The observation time was 3 months (from February 2002 to April 2002). Fig. 9 plots the variation in the cosmic-ray muon intensity through the mountain with the azimuthal angle (compass direction) in bins of 66 mrad as the ratio nyn f of the expected events from the sky. nyn f is normalized by the Monte-Carlo simulation results, as obtained with assuming a uniform density of 2:5 g=cm3 : Other normalized MonteCarlo simulation results for 0.5, 1.0, 1.5, and 2:0 g=cm3 are also plotted. These data are consistent with rock densities deduced from the Bouguer anomaly map published by GSJ (Geological Survey of Japan). We have conrmed the following important features in cosmic-ray muons passing through Mt. Asama: (a) The intensity of cosmic-ray muons which penetrate the mountain in the vicinity of the crater region reects the density change of the internal structure of the mountain, suggesting a picture of the existance of a crater. (b) The intensity of cosmic-ray muons which penetrate the mountain under the crater region intimates the density change of the internal structure of a mountain, suggesting a picture of the existance of a conduit. But we can still say that no change occurs in this region within the limits of statistical error. These results demonstrate that cosmic-ray muons are promising probe to study the internal structure of a mountain. We can use cosmic-ray muon data in order to determine the location of a crater with respect to the exterior features of the mountain. The difference in the relative change of the counting rate from the east to the west gives information about the position of the crater. 4.3. Interpolation of the data Multi-dimensional spatial interpolation from scattered data as a method for predicting and representing of multi-variate elds are suitable for visualization. The justication of the interpolation for our data is based on a continuous variation of the mountain shape. Initially, the trends within the data sets will be analyzed to determine which is the
ARTICLE IN PRESS
Fig. 9. (Top). Observation region. The location of the crater is indicated at the top region. The results of the measurement (with a zenith angle of 330767 mrad of the variation of the cosmic-ray intensity with the azimuthal angle from the backward (middle, left; cosmic-ray muons passing through nothing) and forward (A, B, and C) directions are plotted with error bars. Normalized MonteCarlo simulation results for 0.5, 1.0, 1.5, 2.0, and 2:5 g=cm3 are also plotted.
most appropriate interpolation technique to apply to the datum. Then, the cubic spline-interpolations method is employed. Although the spline-inter-
polation method may lead to a function which uctuates wildly in order to track the data, rather than to a smooth function, the smoothing curve
ARTICLE IN PRESS
deviates from the expected data points if the uctuation is large. The application of the spline function on the data set will be carried out numerous times so as to obtain the optimal parameters for the representative function. Once the nature of the function and the appropriate parameters to apply to the corresponding function have been determined, these results will then be rewritten and further optimized. However, if the data are not known to be completely accurate, it may not make much sense to force an approximating function to pass through all of the data points precisely. This kind of interpolation was adopted because large variations in the cosmic-ray counting rate through EW was expected, whereas the error bars of muon events were considerably small enough. A spline surface was created by interpolating over the set of data points using the basic parameters of the number of points (15) and of the cell size 766 mrad in order to reproduce the surface of the cell size of 720 mrad: An extreme change from the discrete data directly obtained from our detection system could be seen. The interpolated cosmic-ray muon event decrease reects the outer prole of the mountain. 4.4. Visualization of a crater The most distinctive feature of Mt. Asama is the existance of a crater near the top. Since the crater effect in the cosmic-ray muon counting rate is small, we then turned to a more detailed numerical analysis of the data, namely a comparison between the experimental data and a Monte-Carlo simulation. When the lled-crater was properly calculated in the simulation, the experimentally obtained counts in this region were systematically higher than predicted by the computer, as shown in Fig. 10. In order to clarify the position of the crater, the plots in this gure were produced as follows: 1. Preparation of two interpolated data: (A) the calculated data for the lled crater condition; (B) the obtained data.
Fig. 10. Difference between the actual and simulated data. The concave nature in the vicinity of the top regions indicates the position of a crater.
2. Subtraction B A was carried out in the case that the difference between a data point and the next nearest data point was bigger than 10%, so that the margin would reect the shape of the mountain. Signicant increases of events at the central region seen in the difference between the actual and the simulated data constitutes strong prove that we detected the crater. In the absence of rock on the top of the mountain, the expected values in detail (several crater conditions) were obtained from the simulation program. The generally good agreement of the data points with the simulation for a vacant-crater condition (Fig. 11) are shown. Fig. 10 indicates that the vicinity of the top regions with a concave nature corresponds to the position of a crater which is highlighted by the yellow to red bins. Values of some typical data points are shown in Table 1. 4.5. Conclusion of a test experiment on Mt. Asama We concluded from our study of a measurement of the crater effect on Mt. Asama that a crater at the top region of the mountain was discernible.
ARTICLE IN PRESS
Fig. 11. Comparison between the actual and simulated data. A Monte-Carlo simulation was performed for Mt. Asama with a given density distribution, and was compared with the obtained data from the present system. In order to emphasize the effect, the differences from the simulations for the full-lled-crater condition are shown. The histograms shown in this gure are the data obtained (left), and the Monte-Carlo simulations for several conditions (0%, 30%, and 100%) (right). The number of events increases at the top region, and it indicates the existance of a crater. The absolute value of the data corresponds to the Monte-Carlo simulation in the case that the crater is vacant. Table 1 Values of typical data points (3 months) Horizontal angle (mrad) 169 169 169 Vertical angle (mrad) 185 323 453 Data Simulation Difference
rmed the following important properties of the present method: (a) The stability in terms of the intensity of the cosmic-ray muons from the sky, for all the 40 counters, was kept within 72% during the period of the 3-month observation. (b) An event increase in the vicinity of the top region of the mountain was suggested a cosmic-ray muon event increases because of the short density length in the vicinity of a crater. This area corresponds to the position of the crater. (c) Thus, we conrmed that the crater part of Mt. Asama is vacant with a volume occupancy of less than 30%. A segmented detection system as an effective cosmic-ray observation device with a multiplicity cut method was developed. The multiplicity cut method, which eliminates soft components by using cascade shower generation through an intermediate iron absorber, enables us to reduce the number of detectors from three to two. We observed the effect of the crater of Mt. Asama in the cosmic-ray muon ux. This leads to the possibility of probing the detailed internal structure of a volcano and its time evolution.
2143 29 840 54 351
0 25 470 54 048
2143 4370 303
Fig. 11 is a matrix showing the interpolated total number of events recorded in each of the 2500 21:12 9:24 mrad bins. Fig. 11 also includes one of the nal simulation runs for several crater conditions with a rock occupation of (0%, 30%, and 100%). In order to emphasize the effect, the deviations from the lled-crater condition are shown. Although detection of the crater was much more difcult than imaging the outer prole of the mountain, a comparison between the experimental data and the Monte-Carlo simulation enabled us to detect the position of the crater; together, this proof of the method suggested us that we could have seen any previously unknown conduit or vacancy that might exist in our eld of view. In our relatively long-term observation, we have con-
ARTICLE IN PRESS
Acknowledgements The authors are deeply indebted to late Professor M. Oda for his encouragement and valuable suggestions. Special funding arrangements by Professor S. Kobayashi and related people of RIKEN (The institute of physical and chemical research) and MEXT (Ministry of Education, Culture, Sports, Science and Technology) are gratefully acknowledged. We also acknowledge to Mitsubishi Foundation for the founding at the early stage of this work. We are also gratefull indebted to Mr. Y. Ichimura and Asama-en staff for their hospitalities. Valuable discussions with Dr. Honda and Professors Y. Ida, J Nishimura, Y Totsuka and H. Wakita are acknowledged.
Early stage collaboration with Professor M. Iwasaki is also acknowledged.
References
[1] K. Nagamine, et al., Nucl. Instr. and Meth. A 356 (1995) 585. [2] M.G. Thompson, M.R. Whalley, J. Phys. G 1 (1975) 48. [3] R.K. Adair, H. Kasha, Muon physics, in: V.W. Hughes, C.S. Wu (Eds.), Vol. 1, Academic Press, New York, 1976, p. 323. [4] R.L. Golden, et al., J. Geophys. Res. 100 (A12) (1995) 23,515. [5] GEANT3 Manual, CERN Program Library Long Writeup, W5013, 1994. [6] S.N. Nakamura, et al., Nucl. Instr. and Meth. A 388 (1997) 220.
Secrets of thedead
Can a cosmic ray detector help to uncover the history of an ancient civilisation? Gonzalo Argandoa investigates
WHEN Arturo Menchaca was studying physics at the University of Oxford, he never imagined his passion for subatomic particles would lead him into archaeology. But next month, Menchaca, who heads the Institute of Physics at the National Autonomous University of Mexico (UNAM), will venture under tons of mud and rock in search of a kingstomb that vanished centuries ago. And there he will install a cosmic ray detector a bizarre device even for this era of high tech archaeology. Menchaca believes that the high energy particles that bombard Earth from space will help unearth the secrets of the 2000-year-old Pyramid of the Sun at Teotihuacan, near Mexico City. Teotihuacans origins stretch as far back as 800 BC, when the first inhabitants settled in the area. It was the first large metropolis in the Americas, the capital of a complex civilisation that thrived for longer than Rome did. But that seems to be about all we know. Nobody knows, for instance, whether the city had absolute monarchs like most ancient cultures or whether it was governed by a coalition of the most influential families. Even the real name of the city has been lost. For some unknown reason, the metropolis was abandoned around the 7th century. Later on, when the Aztecs found it completely desolate, they thought the whole area was inhabited by supernatural beings. It was the Aztecs who named it: Teotihuacan means Place of the Gods in their Nahuatl native tongue. Unfortunately, Teotihuacan has no history. We do not have any known language or written
sources to decipher the past of the city, says Linda Manzanilla, an archaeologist at UNAMs Institute of Anthropological Research. One of the most important of the unanswered questions about Teotihuacan concerns its origins: who founded and ruled this sacred city? Archaeologists believe that royal chambers, hidden under the city, might contain the answer, but decades of excavations have so far unearthed no conclusive evidence. So is the answer about to come to them from space? Menchaca thinks it might. After all, he points out, it has been done once before. In the 1960s, a Nobel prizewinner, the late Luis Alvarez of the University of California, Berkeley, used a cosmic ray detector to analyse the Pyramid of Kephren, located 10 kilometres from Cairo, Egypt. Although one of the many wars in the area prevented a complete exploration, Alvarez still managed to report in the journal Science that he had found no undiscovered rooms. Years later, as a postdoc Menchaca met Alvarez when he was a professor at the University of California, Berkeley. They talked about the Kephren experiment, but Menchaca forgot all about the idea until his return to Mexico which was when he met Manzanilla. Together they agreed that recreating Alvarezs Egyptian experiment under the Mexican pyramid might just give the archaeologists the breakthrough they so desperately needed. When cosmic rays reach Earths atmosphere there is a violent but invisible collision that creates millions of muons, tiny particles that live for only a millionth of a second. During that brief time, the muons travel for dozens of kilometres, passing almost unhindered through most kinds of material almost but not quite. Every obstacle the muons pass through will absorb a tiny fraction of them. The denser the object, the bigger the absorption, so a cosmic ray detector installed below a structure like the Pyramid of the Sun will provide an indirect measure of the density of what is above it. If we detect more particles than expected coming from one part of the building, it means that there must be a hole in that direction, explains Menchaca. The detector looks like a cube with sides measuring 1 metre. The top and bottom of the
cube are covered with plastic scintillators, which produce light and an electrical signal when they are hit by high energy particles. Since muons take less than a nanosecond to go across the detector, if both plastic layers detect a signal within this time-window then there is definite proof of a muon passing by. According to Menchacas estimates, if the Pyramid of the Sun has no hidden compartments the detector should count 100 particles every second. If this count rate turns out to be higher, the researchers think that will indicate the presence of a hidden chamber.
The Temple of the Sun at Teotihuacan has evaded archaeologists best attempts to unearth its secrets
34 | NewScientist | 26 July 2003
www.newscientist.com
But detection is only half of the job; the researchers also need to know which direction the muons came from. This is done by 1200 electrical wires stretched across the cube at varying orientations. As muons pass by, they ionise the air within the detector, and the resulting electric field affects the currents flowing in the wires. Since the effect is strongest at the wire closest to the muons trajectory, the researchers can use the location of the varying currents to pin down the muons trajectory and trace back their route through the temple. If they see an unexpectedly high number of muons from a particular direction, the researchers will infer the presence of a hidden chamber. In theory, the system sounds like a winner. But theres an obvious problem. How do you get a cosmic ray detector under the city? Digging out a chamber is just too invasive, as well as expensive, and could destroy the archaeological heritage. Here, the Teotihuacan researchers have struck lucky. The original inhabitants of the city excavated a tunnel that almost reaches the heart of the Pyramid of the Sun. The entrance to the gallery is pretty small, but once inside there is plenty of space. The corridor winds for more than 100 metres until it
reaches a big cave that opens out into a number of chambers. The place is just perfect for the experiment, says Menchaca. He installed a prototype detector last year in the middle of the cave. After all the hardware and software tests turned out fine, Menchaca and his UNAM colleagues built the final version of the machine. Next month the detector at Teotihuacan will be left in the shadows beneath the pyramid, and will stay there for a year, counting millions of muons. Gigabytes of data will travel continuously to a computer located 40 kilometres away at UNAM, where researchers will crunch numbers in search of hidden rooms, with a precision of about half a metre. Once they spot a good candidate for a royal chamber, then comes the hardest part of the job: digging towards the 3D coordinates obtained from the cosmic ray data. But since they will know the precise location, the excavation will have a minimal impact compared to a blind exploration. Should the researchers finally locate and enter a royal chamber, what might they find? Gold and emeralds consecrated to the glory of ancient kings? Or just dirty bones? Based on numerous paintings found at Teotihuacan none of which obviously depict
Should the researchers finally locate and enter a royal chamber, what might they find? Gold and emeralds consecrated to the glory of ancient kings? Or just dirty bones?
monarchs Manzanilla suspects the original inhabitants of the city did not glorify their authorities. Indeed, she believes they operated under a collective leadership. The whole city was divided in four main sectors. The same number, four, is a recurrent leitmotif in Teotihuacan art, as well as in the main plaza of a palace which I am excavating now near the Pyramid of the Sun. All this evidence may be an indicator of a stable political coalition formed by four rulers, she suggests. By this time next year cosmic rays will have played their part and we should know if she is right. G
Gonzalo Argandoa is a science writer with Cabala Productions, based in Santiago, Chile
www.newscientist.com
26 July 2003 | NewScientist | 35
CHARLES AND JOSETTE LENARS/CORBIS
brief communications
Radiographic imaging with cosmic-ray muons

Natural background particles could be exploited to detect concealed nuclear materials.
espite its enormous success, X-ray radiography1 has its limitations: an inability to penetrate dense objects, the need for multiple projections to resolve three-dimensional structure, and health risks from radiation. Here we show that natural background muons, which are generated by cosmic rays and are highly penetrating, can be used for radiographic imaging of medium-to-large, dense objects, without these limitations and with a reasonably short exposure time. This inexpensive and harmless technique may offer a useful alternative for detecting dense materials for example, a block of uranium concealed inside a truck full of sheep. In X-ray radiography, the intensity of an image pixel is determined by the attenuation of the incident beam caused by absorption and scattering the maximum mean free path for photons is about 25 g cm2 for all materials, corresponding to less than 2 cm of lead. For thicker objects, it is better to use a different type of radiography that is based on the interaction of charged particles with matter by multiple Coulomb scattering. The many small interactions add up to yield an angular deviation that roughly follows a gaussian distribution,
Frequency (arbitary units) 0.1 0.05 0.05 0 plane (rad)
0.1
Experiment
Simulation
Figure 1 Radiographic imaging with muons of a test object (left) and the reconstructed image of its Monte Carlo simulation (right). The test object is a tungsten cylinder (radius, 5.5 cm; height, 5.7 cm) on a plastic (35601 cm3) plate with two steel support rails. The tungsten cylinder and the iron in the rails are clearly visible in both the experiment and simulation reconstructions. Inset, the widths of the scattering distributions for tracks passing through the tungsten target are very similar for the experimental and simulated data.
dN 1 e dx 20
x2 202
with the width, 0, related to the scattering material through its radiation length, L0, as follows: 13.6 [1 L +0.038ln(L/L )] 0 0 cp L0 where p is the particles momentum in MeV c1 and c is its velocity2. The radiation length decreases rapidly as the atomic number of a material increases, and 0 increases accordingly: in a layer 10 cm thick, a 3-GeV muon will scatter with an angle of 2.3 milliradians in water, 11 milliradians in iron and 20 milliradians in lead. By tracking the scattering angles of individual particles, the scattering material can be mapped. Our new technique relies on the scattering of atmospheric muons produced by primary cosmic rays. Muons are the most numerous cosmic-ray particles at sea level, moving at a rate of about 10,000 m2 min1 in horizontal detectors3. These particles are highly penetrating: a typical cosmic-ray
muon of energy 3 GeV will penetrate more than 1,000 g cm2 (10 m of water, for example). To demonstrate the concept of muon radiography, we developed a small-scale experimental system with four drift chamber detectors4 spaced 27 cm apart. Each detector has an active area of 6060 cm2 and records particle tracks at two positions in each of two orthogonal coordinates. The upper pair of detectors records the tracks of incident muons, and the lower pair records the scattered tracks. A tungsten cylinder was used as a test object, supported by a plastic plate and steel support beams. The tungsten is clearly visible in the reconstructed image, and the steel support beams are also evident (Fig. 1, left). We also developed a Monte Carlo simulation code that generated cosmic-ray muons and propagated them through a test volume. The reconstructed images are indistinguishable from those obtained experimentally (Fig. 1), and the scatter angles of the simulated muons from the different materials (tungsten, lexan and steel) are consistent with the measured angles. Simulation of larger, more complex objects demonstrates that we can reliably detect a 101010 cm3 uranium object inside a large metal container full of sheep in 1 min of exposure. We conclude that cosmic-ray muons show promise as an inexpensive, harmless probe for radiography of medium-to-large objects, such as commercial trucks, passenger cars or sea containers. Our experimental results and simulations demonstrate the ability to reconstruct complex objects and to detect dense material of high atomic number hidden in a much larger volume of material of low atomic number, using only the natural
2003 Nature Publishing Group
flux of muons. This method is suitable for a range of practical applications in which radiography of dense objects with low radiation dose is required for example, in surveillance for cross-border transport of nuclear materials.
Konstantin N. Borozdin, Gary E. Hogan, Christopher Morris, William C. Priedhorsky, Alexander Saunders, Larry J. Schultz, Margaret E. Teasdale
Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA e-mail: kbor@lanl.gov
1. Roentgen, W. C. Nature 53, 274 (1896). 2. Hagiwara, K. et al. Phys. Rev. D 66, 01001 (2002). 3. Grieder, P.K.F. Cosmic Rays at Earth (Elsevier Science, Amsterdam, 2001). 4. Atencio, L. G., Amann, J. F., Boudrie, R. L. & Morris, C. L. Nucl. Instr. Methods Phys. Res. 187, 381386 (1981). Competing financial interests: declared none.
Palaeo-oceanography
Deepwater variability in the Holocene epoch

he conversion of surface water to deep water in the North Atlantic results in the release of heat from the ocean to the atmosphere, which may have amplified millennial-scale climate variability during glacial times1 and could even have contributed to the past 11,700 years of relatively mild climate (known as the Holocene epoch)24. Here we investigate changes in the carbon-isotope composition of benthic foraminifera throughout the Holocene and find that deep-water production varied on a centennialmillennial timescale. These variations may be linked to surface and atmospheric events that hint at a contribution to climate change over this period.
NATURE | VOL 422 | 20 MARCH 2003 | www.nature.com/nature
277

Approfondimenti Cosmici

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Approfondimenti Cosmici

Transféré par

Droits d'auteur :

Formats disponibles

Cosmic rays: the most energetic particles in the universe

James W. Cronin: Cosmic rays

James W. Cronin: Cosmic rays

James W. Cronin: Cosmic rays

VII. NATURES DIAGNOSTIC TOOLS

James W. Cronin: Cosmic rays

James W. Cronin: Cosmic rays

James W. Cronin: Cosmic rays

James W. Cronin: Cosmic rays

Rev. Mod. Phys., Vol. 71, No. 2, Centenary 1999

Astroparticle Physics 17 (2002) 441458 www.elsevier.com/locate/astropart

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

2. Air shower calculations with

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

J. Poirier et al. / Astroparticle Physics 17 (2002) 441458

[10] [11] [12] [13] [14] [15] [16] [17]

[18] [19] [20]

Nuclear Instruments and Methods in Physics Research A 507 (2003) 657669

yn is Nm Em ; yn dEm AWm Em DEm

Fig. 6. Schematic view of the experimental setup.

2143 29 840 54 351

2143 4370 303

Early stage collaboration with Professor M. Iwasaki is also acknowledged.

34 | NewScientist | 26 July 2003

26 July 2003 | NewScientist | 35

CHARLES AND JOSETTE LENARS/CORBIS

Radiographic imaging with cosmic-ray muons

Deepwater variability in the Holocene epoch

NATURE | VOL 422 | 20 MARCH 2003 | www.nature.com/nature

Vous aimerez peut-être aussi