Chapter 9

Cosmic Rays

9.1 Composition and energy distribution

Cosmic rays can be broadly defined as the massive particles, photons ( rays, X-rays, ultra-
violet and infrared radiation, ...), neutrinos, and exotics (WIMPS, axions,...) striking the
earth. The primary cosmic rays are those entering the upper atmosphere, the cosmic rays
of the interstellar medium. Secondary cosmic rays are those produced by the interactions of
the primary rays in the atmosphere or in the earth. Also products of cosmic ray interactions
in the interstellar medium (e.g., spallation products from cosmic ray - cosmic ray collisions)
are also labeled as secondary cosmic rays. Cosmic rays can be of either galactic (including
solar) or extragalactic origin.

If we confine ourselves to the particle constituents (protons, nuclei, leptons), their motion in
the galaxy has been roughly randomized by the galactic magnetic field. (We will mention
some exceptions to this below.) Thus they provide very little information about the direction
of the source. The peak of the distribution in energy is in the range of 100 MeV - 1 GeV.
The intensity of cosmic rays of energy 1 Gev/ nucleon or greater is about 1/cm2 sec sr. An
approximate formula is
 
E
IN (E) 1.8 nucleons/cm2 sec sr (1)
GeV
where E is the energy per nucleon (rest and kinetc) and -2.7. The energy density cor-
responding to this is thus about 1 ev/cm3 . This can be compared to the energy density of
stellar light of 0.3 eV/cm3 .

The principal components of the (primary) cosmic rays are shown Figure 1. This abundance
distribution is approximately independent of energy, at least over the dominant energy range
of 10 MeV/nucleon through several GeV/nucleon. By mass about 79% of nucleons in cosmic
rays are free protons, and about 80% of the remaining nucleons are bound in helium. The
composition has been measured by instruments mounted on balloons, satellites, and space-
craft. Figure 2 shows the chemical distribution of the elements in our solar system differs
from that of the cosmic rays in some remarkable ways. The most dramatic difference is an
enormous enrichment in the cosmic ray abundances for the elements Li/Be/B. Note also that
there is enrichment is even Z elements relative to odd Z, when normalized to solar system
abundances. Finally the cosmic rays are enriched in the heaviest elements relative to H and
He.

As can be seen from Figure 2, many elements heavier than the iron group have been mea-
sured with typical abundances of 105 relative to iron. Much of this information was gained
from satellite and spacecraft measurements over the last decade. Some of the conclusions:
1) Abundances of even Z elements with 30 < Z < 60 are in reasonable agreement with solar
system abundances.

1
 20. Cosmic rays 3
10

0.1
He
Differential flux (m2 sr s MeV/nucleon)1

10 2 H
C
10 3

Fe
10 4

10 5

10 6

10 7

10 8

10 9
10 102 103 104 105 106 107
K ine ti c e n er gy (MeV/ nucleon)

Figure 1: Major components of the primary cosmic rays (from Simpson).

Figure 20.1: Major components of2 the primary cosmic radiation (from Ref. 1).

Most measurements are made at ground level or near the top of the atmosphere,
but there are also measurements of muons and electrons from airplanes and balloons.
Fig. 20.3 includes recent measurements of negative muons [3,13,14,15]. Since + ( ) are
2.1. COMPOSITION 11

Figure 2.1:2: Relative

Figure abundance
The abundances of elements
of cosmic rays in
arecosmic rayswith
compared and solar
in theabundances.
solar
system.

3
 Z Element Relative Abundance
1 H 485
2 He 26
3-5 Li-B 0.40
6-8 C-O 2.20
9-10 F-Ne 0.30
11-12 Na-Mg 0.22
13-14 Al-Si 0.19
15-16 P-S 0.03
17-18 Cl-Ar 0.01
19-10 K-Ca 0.02
21-25 Sc-Mn 0.05
26-28 Fe-Ni 0.12

Table 1: Relative abundances of cosmic ray nuclei at 10.6 GeV/nucleon normalized to oxy-
gen (=1). The oxygen flux at kinetic energy 10.6 GeV/nucleon is 3.26 106 /cm2 sec sr
(GeV/nucleon)1 . From Boezio et al. and Engelmann et al., as quoted by Gaisser and
Stanev.

2) In the region 62 <Z < 80, which includes the platinum-lead region, abundances are en-
hanced relative to solar by about a factor of two. This suggests an enhancement in r-process
elements, which dominate this mass region. This is consistent with r-process scenarios we
have discussed (e.g., if the r-process site is core-collapse supernovae, then one would expect
enrichment in r-process nuclei as supernovae are also believed to be the primary accelera-
tion mechanism for lower energy cosmic rays). However, one also needs to consider whether
heavier nuclei with larger charges might be confined in the galaxy for a longer time, another
effect that could differentiate heavy from light nuclei.

Galactic cosmic rays are fully ionized: the acceleration mechanisms fully strip the ions. Cos-
mic rays also have an antimatter component, as measured in the Space Shuttle Discovery
AMS (Alpha Magnetic Spectrometer) experiment. The AMS detected about 200 antipro-
tons above 1 GeV, generally attributed to nuclear collisions of CR particles with interstellar
matter.

The energy distribution of cosmic rays from about 1010 eV to about 1015 eV is smooth, with
a power-law distribution
particles/cm2 sec MeV/n E s (2)
with s between -1.6 and -1.7. However there is a break or knee in the curve at about 1015
eV. The slope sharpens above this knee (see Figure 3), falling with s ranging from -2.0 to
-2.2, eventually steeping to an exponent of above -2.7. The knee is generally is attributed
to the fact that supernova acceleration of cosmic rays is limited to about this energy. This
would argue that the cosmic rays above this energy either have a different origin, or were

4
cr-knee.gif (GIF Image, 468x333 pixels) http://imagine.gsfc.nasa.gov/docs/features/topics/snr_group/images...

Figure 3: Energy spectrum of cosmic rays, showing a kink or knee near 1015 eV.

further accelerated after production. While this break is commonly attributed to the inabil-
ity of supernova shocks to accelerate particles to energies beyond the knee, the sharpness of
the knee has troubled many of the experts: in is difficult to find natural models that produce
such a defined break.

There is an additional, somewhat less distinct feature at an energy of about 1019 eV, termed
the ankle. Characteristics of this high-energy feature will be discussed later.

9.2 Propagation and origin

The most commonly used toy model for galactic cosmic rays is called the leaky box model.
It assumes that the cosmic rays are confined within the galactic disk, where the mass density
is high, but with some gradual leaking out of the disk. The confining force is the galactic
magnetic field, which is on the order of 107 Gauss. A relativistic particle moving in a
magnetic field executes a helical path. Using the relation between the momentum p, field B,
and magnetic radius R,
p (M eV /c) = 3 104 BR(gauss cm) (3)
one sees that a 1014 eV proton would have a radius of 3 1018 cm, or 1pc, which is much

1 of 1 03/10/2004 10:17 AM
less than the distance to the Crab nebulae, a potential accelerator for cosmic rays relatively
near earth. Thus cosmic rays from the knee and below will have no memory of their origin
when they reach earth. A 1018 eV proton corresponds to 10 kpc, about the galactic radius.
Clearly any cosmic ray much above this energy could be presumed to be extragalactic, unless
it can be associated with a local source, which would be possible as the direction of such
cosmic rays would point back to their origin. For example, a cosmic ray of energy 1020 eV
would be minimally perturbed by the galactic magnetic field and thus would point back
to the extragalactic source from which the cosmic ray originated.

The leaky box models cosmic ray production in the galaxy, cosmic ray trapping by magnetic
fields and eventual escape, and cosmic ray interactions in the intragalactic medium. This
model does a good job in explaining the energy-dependence of the life of cosmic rays (more
on this later). But others have argued for other models, including closed models where
cosmic rays are fully confined, then explaining isotope lifetimes (see below) through devices
such as a combination of a few nearby and many distant cosmic ray sources.

The conventional explanation for the most dramatic isotopic anomaly in the cosmic rays,
the enrichment in Li/Be/B by about six orders of magnitude, is that these isotopes are
produced in the interstellar medium when accelerated protons collide with C, N, and O.
We mentioned this process earlier. The enrichment of odd-A nuclei (these also tend to be
relatively rare in their solar distribution since stellar processes tend to favor production of
more stable even A nuclei) is also often attributed to spallation reactions off more abundant
even-A nuclei. These associations immediately lead to some interesting physics conclusions
because, from the known density of cosmic rays (at least in the earths vicinity) and from
known spallation cross sections, one can estimate the amount of material through which a
typical cosmic ray propagates. Although the estimates are model dependent - and probably
not sufficiently interesting to go through in detail - the resulting values for the effective
thickness are typically 4 - 6 g/cm2 . Now the mass density within intragalactic space is
about 1 proton/cm3 , or about 1.7 1024 g/cm3 . Thus taking a velocity of c, we can crudely
estimate the cosmic ray lifetime

1.7 1024 g/cm3 (3 1010 cm/sec) t = (4 6)g/cm2 (4)

So this gives
t 3 106 y (5)
This calculation assumes an average galactic mass density that is not known by direct mea-
surement. Thus it is nice that a more direct estimate of the galactic cosmic ray lifetime is
provided by cosmic ray radioactive isotopes. The right chronometer is one that has a lifetime
in the ballpark of the estimate above. 10 Be, with a lifetime of 1.51 106 y, is thus quite
suitable. It is a cosmic ray spallation product: this guarantees that it is born as a cosmic
ray. Its abundance can be normalized to those of the other, stable Li/Be/B isotopes: the
spallation cross sections are known. Thus the absence of 10 Be in the cosmic ray spectrum
would indicate that the typical cosmic ray lifetime is much larger than 1.51 106 y. The

6
survival probability should also depend on the 10 Be energy, due to time dilation effects. One
observes a reduction in 10 Be to about (0.2-0.3) of its expected instantaneous production,
relative to other Li/Be/B isotopes. From this one concludes

t (2 3) 107 years

This suggests that the mass density estimate used above (in our first calculation) may have
been too high by a factor of 5-10.

In modeling the origin of cosmic rays, the first conclusion, given their richness in metals, is
that must come from highly evolved stars such as those that undergo supernovae. We have
already noted the abundance of r-process nuclei, which could be taken as a smoking gun
of supernova dominance, for those who accept that supernovae are the r-process site.
However this is clearly not the full picture. Studies of the isotopic composition as the knee
is approached shows that the composition changes: the spectrum of protons becomes notice-
ably steeper in energy, while the iron group elements do not show such a dramatic change.
This is qualitatively consistent with the notion that the galactic accelerator producing
cosmic rays tops out at some maximum confining field strength. Because the acceleration
likely scales as ZB, where Z is the charge, the very highest energies should be dominated
by the largest Zs, the nuclei.

Above the knee - at energies above 1016 eV the galactic magnetic field is too weak to ap-
preciably trap particles. Thus it is probable that at these high energies the character of the
cosmic rays changes from primarily galactic to primarily extragalactic: the trapping that
enhances the abundances of lower-energy galactic cosmic rays would not enhance very high
energy ones.

The cosmic rays appear to be approximately isotropic once one gets above about 50 GeV
to escape local magnetic effects. One exception is a small anisotropy measured by AGASA,
excess events at about 1018 eV pointing back to the galactic center - which generated the
speculation that these are neutrons, which because of time dilation can reach the solar system
once they reach 1018 eV. If this speculation were true, it would argue for a central galactic
accelerator that (most likely) is capable of accelerating ions up to 1018 eV per nucleon. Re-
cent results from Pierre Auger find no evidence for the AGASA anomaly.

9.3 Energetics and origin of galactic cosmic rays

We now have the basic information needed to calculate the energetics of galactic cosmic rays.
If we take the galactic radius as 10 kpc, we have a volume of 1068 cm3 . We noted that the
energy density of cosmic rays is about 1 eV/cm3 . Thus the cosmic ray energy content of the
galaxy is about 1068 eV. We have argued above that the lifetime of cosmic rays might be
about 107 years. Thus the energy production in cosmic rays must be about

1068 eV/107 y 3 1054 eV/sec 5 1042 ergs/sec. (6)

7
We mentioned in passing above that there are cosmic ray models that, unlike the leaky box,
confine cosmic rays longer and thus need to contrive localize sources to account for Be,
etc. Thus our energetics calculation could be modified, replacing the leaky box model with
one where the cosmic rays were effectively confined over times up to the age of the galaxy.
This would extend the dwell times by up to three orders of magnitude, and thus reduce the
energetics requirement proportionally. So that in principle would lower the galactic energetic
requirement to about 1040 ergs/sec. The result is a very generous range for the necessary
energetics, requiring sources capable of generating between 1040 and 5 1042 ergs/sec.

One can quickly show that stellar winds e.g., the flares our sun produces are insufficient
energetically to produce energy at this rate, even if integrated over the 1011 potential sources
in the galaxy the integrated energy in stellar flares falls short of the lower bound above by
at least one and perhaps two orders of magnitude. But an interesting source is supernovae
they eject large quantities of material, that material includes both protons and nuclei, and
the shock wave and associated fields are an acceleration mechanism.

We have seen that the explosion energy of a supernova is about 1051 ergs, and the estimated
rate of Type II supernovae is 1/30 years. Additional contributions would come from rarer
classes of core-collapse supernovae, SNIb SNIc. Thus the energy production rate is about
1042 ergs/sec close to that estimated above. There are many arguments based on the chem-
ical patterns of cosmic rays that support core-collapse supernovae as a principal mechanism
for generating and accelerating cosmic rays. Novae are powered by the accretion onto a
white dwarf. The matter is sucked from a large companion star, perhaps a red giant or main
sequence star, under conditions that allow periodic thermonuclear runaway reactions on the
white dwarf surface. Novae are another plausible contributor. Novae outbursts within our
galaxy occur with a frequency of about 100/year with an energy output about 104 that of
the ejecta from a core-collapse supernova. Thus the integrated nova output is similar to that
of galactic supernovae.

9.4 Solar-system, terrestrial, and atmospheric environmental effects

Cosmic rays measured on earth reflect not only the galactic inventory of cosmic rays, but
also the effect of the local environment of our solar system, the earth, and the atmosphere.
The suns heliosphere the region of space altered by the solar wind extends to about 200
AU. The magnetic structure of the heliosphere shields the region against energetic charged
particles. This is apparent experimentally from the correlations between induced cosmic ray
activity measured on earth and the 11-year solar cycle, as shown in Fig. 4. This correla-
tion is reflected in changes in the chemistry of the atmosphere, such as 14 C production, an
important radio-isotope for dating. 14 C is a product of cosmic ray interactions with O in
the atmosphere, and the production is anticorrelated with the shielding, and thus with the
intensity of solar activity. Interesting, 14 C production also appears to correlate with various
long-term climate anomalies of the past 1000 years, such as the Maunder Minimum (14 C
high) and the 12th-Century Maximum (low), leading to some interesting speculations about

8
Figure 4: The top curve is the cosmic ray flux from the neutron monitor in Climax, Colorado
(1953 1996). The middle curve is the annual mean variation in the cosmic ray flux as
measured by ionization chambers (1937 1994). The bottom curve is the relative sunspot
number. While there is a clear solar cycle modulation of the cosmic ray flux, the amplitudes
are not well correlated. From Svensnak.

the solar role in climate change.

There are also more localized effects due to the earths magnetic field, a dipolar field that
extends from the magnetic poles and thus is roughly parallel to the earth surface at the
equator (Fig. 5). If we envision a cosmic ray proton come into the earth at the equation
and use 1000 km as a rough trapping radius, then our previous formula

p (M eV /c) = 3 104 BR(gauss cm)

for B=0.3 Gauss would give a p E 10 GeV as the energy of the trapped proton. Thus
this defines an energy below which charged cosmic rays would not penetrate to the surface,
but instead be sharply deflected. This is the reason that so many cosmic ray balloon exper-
iments are done at the poles, where the magnetic fields are weaker away from the earth and
more perpendicular to the surface: one can then more easily sample cosmic rays character-
istic of the environment away from the immediate terrestrial neighborhood.

The trapping of lower-energy cosmic rays by the earths magnetic field in also part of the

9
Figure 5: Rough representation of the earths magnetic field, which has an average value
near the surface of about 0.3 Gauss.

third effect of the local environment, interactions of the cosmic rays with the atmosphere.
The trapping leads to a larger time in the vicinity of upper atmosphere and thus enhances
nuclear interactions than can produce all sorts of cosmic ray secondaries. Such secondaries
are also produced by higher energy cosmic rays that are not perturbed significantly by the
magnetic field, but do interact in the upper atmosphere. The secondaries include pions,
muons, new nuclei created by spallation, electrons and positrons, neutrinos, and the high-
energy showers produced by very energetic cosmic ray interactions.

The density profile of the atmosphere is approximately exponential with a scale height of
about 7.6 km,
(r) = 1.205 er/7600m kg/m3 . (7)
Thus the total amount of matter that a cosmic ray encounters on approaching the earth
grows exponentially with decreasing altitude.

9.5 Muons and neutrinos at the earths surface and below.

The muons and neutrinos, the most penetrating components of the cosmic ray secondaries,
result from strong interaction mechanisms that produce pions and kaons through decay
chains such as
+ + +
+ e+ + e + (8)
as we have discussed before. Figure 6 illustrates some important features of these and other
cosmic ray secondaries. The figure was generated by cascade code calculations that begin

10
gure 20.2: Differential spectrum of electrons plus positrons multiplied by
ata summary from Ref. 7). The dashed line shows the proton spectrum mul
0.01.

Altitude (km)
15 10 5 3 2 1 0
10000

1000
[m2 s1 sr1]

_
+
100 + +

10
p+n
Vertical flux

1
e+ + e
+ +
0.1

0.01
0 200 400 600 800 1000
Atmospheric depth [g cm2]

Figure 6: The solid lines show vertical fluxes of cosmic rays in the atmosphere with energy
above 1 GeV, as estimated from parameterized initial nucleon primary fluxes. The points
are from measurements of . From Gaisser and Stanev.
gure 20.3: Vertical fluxes of cosmic rays in the atmosphere with E > 1
imated from the nucleon flux of Eq. (20.2). The points show measuremen
gative muons with E > 1 GeV [3,13,14,15].
11

use muons typically lose almost two GeV in passing through the atmosph
6 20. Cosmic rays
0.2

p2.7dN/dp [cm2 s1 sr1 (GeV/c)1.7] 0.1

0.05

0.02

0.01

0.005

0.002
1 10 100 1000
p (GeV/c)

Figure 7: The spectrum of muons at the earths surface at = 0 degrees (solid points) and
75 degrees (open diamonds). From Gaisser and Stanev.

Figure 20.4: Spectrum of muons at = 0 ( [18],  [20], H [21], N [22], X,+ [23]
and = 75  [24]).
with the primary cosmic ray spectrum at the top of the atmosphere, then propagate that
spectrum and the associated secondaries through the atmosphere, taking into account the
various reactions that produce new particles and degrade their energies. In contrast to the
Figure 20.4
very large shows the
flux/energy muon
losses energyforspectrum
encountered at sea level
other secondaries, for two
the muons angles. At large an
and neutrinos
low generally
energy penetrate
muons decay before well.
the atmosphere reaching the surface
The atmosphere (andand high
earth) energy pions
are transparent to decay be
theyneutrinos,
interact, while muons typically lose 2 GeV in ionization energy while transiting the atmo-
thus the average muon energy increases. An approximate extrapolat
sphere.
formula valid when muon decay is negligible (E > 100/ cos GeV) and the curvatu
the Consequently
Earth can be neglected
muons dominate (the< 70 ) is
charged-particle spectrum at the earths surface. They
are the most easily measured component of the cosmic rays and also the primary background
in a lot of neutrino experiments the reason such neutrino experiments must go deep un-
derground. The effects of energy loss are illustrated
2.7 in Figure 7, where the vertical flux of
dN 0.14 E
muons at the earths surfaceis compared with the much harder spectrum that results for
muons impinging the dE earth 2 s srfrom
cmdegrees GeV
at 75 vertical. The latter must penetrate much
more matter, which then hardens the spectrum.



1 0.054
+ , (2
1 + 1.1E

12
cos
1+
1.1E cos

115 GeV 850 GeV
where the two terms give the contribution of pions and charged kaons. Eq. (20.5) neg
a small contribution from charm and heavier flavors which is negligible except at v
Muons lose energy by ionization, bremsstrahlung, production of e+ e pairs, photonuclear
reactions like Compton production, etc. Their range R in standard rock (Z 11, 2.65
g/cm3 ) is given in the table. R is given in meters-of-water-equivalent (mwe), or 102 g/cm2 .
The energy loss is given in terms of ,
dE
= . (9)
dx
The table shows that the energy loss per g/cm2 increases sharply between 100 GeV and 1
TeV, a result of direct bremsstrahlung, pair production, and nuclear reactions taking over
from ionization losses, which dominate at lower energy. The former grow linearly with the
muon energy. Consistent with the table, a rough rule of thumb is that the net flux of muons
underground decreases by an order of magnitude for every 1500 mwe, or about 500m of
standard rock.

Some of you may know that there has been great interest in establishing underground lab-
oratories in the U.S. for sensitive dark matter, neutrino, and nucleon decay experiments.
While Europe has facilities like Gran Sasso and Frejus, Russia has Baksan, and Japan has
Kamioka, the deepest current site in the US is the Soudan Laboratory, operated by the
University of Minnesota, at a depth of 710 m. Access underground at Soudan is challenging
because of the small lift that provides vertical access to that site.

One wonderful development for North America, however, is SNOLab, an expansion of the
SNO experimental area in the Sudbury Mine, including both additional underground rooms
and a significant surface support facility. This is the worlds deepest underground laboratory,
two kilometers underground and providing 6010 mwe in overburden (more than two kilome-
ters of rock overburden). Furthermore, as the nickel deposits that interest the mines owner
have so far continued to be increasingly rich at depth, it quite conceivable that the mine
could some days reach depths of three kilometers or more. As discussed below, this would
approach the ultimate depth for an underground laboratory, the point at which atmospheric
neutrinos become an irreducible background.

Some next-generation experiments could require 5000 mwe or more of overburden. Examples
include certain next-generation dark matter, solar neutrino, and double beta decay efforts.
Overburden is often important in experiments focused on measuring low-energy events: pen-
etrating muons themselves can often be vetoed by turning off the detector after a muon
is detected. (Kamiokande, for example, has about a 10% deadtime due to this technique
for background suppression.) But muons, in addition to producing prompt secondaries like
knock-out neutrons, can also activate nuclei in the detector that, some time later, decay to
produce a signal. In cases where there is a long delay before the induced activity is detected,
conventional vetoing techniques can lead to an unacceptable deadtime. In such cases there
is no alternative but great depth to remove the initiating muon events.

13
 E (GeV) R (1000 mwe) (106 MeV cm2 /g)

10 50 2.19
100 410 2.74
1000 2450 6.60
10000 6090 46.43

Table 2: From Gaisser and Stanev: the calculated ranges R and energy losses (per gm/cm2 )
are both sharply dependent on the muon energy, and lead to a systematic hardening of
the muon spectrum as one goes underground.

In other cases, backgrounds may be controllable at moderate depth, making great depth
less important than other factors, such as the economies that come from ease of access: the
ability to drive-in large equipment and efficiently operate large experiments may be the
priority. Both Gran Sasso and Kamioka, the worlds premier underground laboratories, are
drive-in facilities at moderate depth.

While Europe has established several underground laboratories (Gran Sasso, Canfranc, Fre-
jus,...) and while Japan has steadily developed Kamioka, the US has lacked a comparable
facility. Experiments were done at Homestake before the mines closing ten years ago, and
the Soudan Mine (Minnesota) and WIPP (New Mexico) both currently host experiments.
But a site offering good and frequently experimental access and depth comparable to our
better than Gran Sasso has been lacking. The NSF headed a ten-year effort o create a Deep
Underground Science and Engineering Laboratory in the US, settling early on Homestake as
the preferred site, but very recently (and oddly) decided the operation of such a facility was
not in its mandate. A DOE-appointed committee is now trying to sort things out, though
the cost of developing Homestake is a major obstacle, given how tight science budgets are
expected to be in 2012 and beyond, and given DOE obligations to existing as well as new,
previously approved projects.

The relevance to the present discussion is the importance of being able to calculate the muon
flux at depth, folding the energy and angular spectrum at the surface with the site topog-
raphy. An example shown here were the investigations of University of Washington physics
grad student Kregg Philpott to determine the effective depth of a tunnel in the Cascades,
which several years ago was a DUSEL candidate. The resulting contours of the muon flux
under the Steven Passs Cowboy and Big Chief Mountains is shown in Fig. 8, for a horizontal
slice at the Pioneer tunnels elevation. This tunnel proves to be at just the right place for
convenient access to the point of greatest overburden, under Cowboy Mt. The resulting flux,
2.07 107 /cm2 sec, is about 90% that at Kamioka (2.28 107 /cm2 sec). .

While we have discussed going deep underground to find an environment sufficiently clean
to do neutrino physics, there are some absolute limits to what can be achieved. Atmospheric

14
Figure 8: The cosmic ray muon flux under Cowboy Mt., Stevens Pass, at the 743 m elevation
of the Pioneer Tunnel. Calculation done by Kregg Philpott.

15
neutrinos provide a underground source of high-energy muons and electrons that does not
attenuate: these are the muons that the energetic atmospheric neutrinos produce as secon-
daries. As Figure 9 shows, at a depth of about 10 km mwe (3.5 km of rock) these become a
constant background that is not reducible by going further underground. It also follows that
upward going muons (or electrons, though they have a shorter range) in deep detectors thus
can be used as a signal of neutrino reactions in the rock below. Figure 10 is a reminder of
our earlier discussion describing how these neutrino fluxes were exploited to probe neutrino
oscillations, through the muons and electrons they produce in underground detectors.

Given that we have discussed so many aspects of the terrestrial neutrino flux, including the
changes induced by neutrino oscillations, it is amusing to display the local neutrino spectrum
in its entirety in Figure 11. The atmospheric neutrinos, despite their importance due to their
high energies and larger cross sections in matter, do not appear on the graph because their
fluxes are so low. Two sources we have not discussed in detail are the geoneutrinos, produced
from terrestrial radioactivity and perhaps recently seen in KamLAND, and the thermal flux
( keV) of solar neutrinos of all flavors, generated in our sun from neutral current processes,
analogous to those we discussed (at higher energies) in the context of supernovae cooling.

9.6 The highest energy cosmic rays, the GZK cutoff, and Pierre Auger
Some of the most curious observations in cosmic ray physics have to do with the highest
energy cosmic rays. A number instruments, such as the Flys Eye and AGASA detectors,
and most recently the Pierre Auger observatory, are designed with sensitivity to ultrahigh-
energy cosmic rays. For example, the Flys Eye uses air fluorescence to detect UHE cosmic
rays. An extensive air shower is generated when a primary cosmic ray interacts with the
atmosphere. This is imaged using the fluorescence light produced by excitation of the ni-
trogen molecules by the secondaries in the extensive air shower. The shape of the shower
allows the experimenters to reconstruct the energy of the primary. It also may provide some
information on composition. The Flys Eye can provide stereo information on the developing
shower because of detector arrays located 3.4 km apart.

The rare high energy events show some structure, particularly around 1019 eV, where the
spectrum flattens from a slope of about -3.0 to one of about -2.6. Some years ago there
were claims that events were been seen above the so-called Greisen-Zatsepin-Kuzmin cutoff
of about 1020 eV. At one time there were tens of such events extending up to about 1021
eV )(see Figs. 12 and 13). The GZK cutoff is explained below, the new results from Pierre
Auger are described.

The cosmic medium is filled by background radiation of relic photons, left over from the
back bang and noninteracting since the time electrons and nuclei recombined to form atoms.
Their typical energy is about 103 eV. We consider the propagation of a high energy proton
through this medium.

16
 20. Cosmic ray
10 6
Vertical intensity (cm2 sr1 s1 )

10 8

10 10

10 12

10 14
1 10 100
Depth (km water equivalent)

Figure 9: Vertical muon intensity as a function of depth (in km.w.e.) The shaded region
at depth represents neutrino-induced muons above 2 GeV. The upper line is for horizontal-
induced muons, the lower one for vertically upward muons. From Gaisser and Stanev.

Figure 20.5: Vertical muon intensity vs depth (1 km.w.e. = 105 g cm2 of standa
rock). The experimental data are from: : the compilations of Crouch [32], :
Baksan [33], f : LVD [34], v : MACRO [35], : Frejus [36]. The shaded area a
large depths represents neutrino induced muons of energy above 2 GeV. The upp
17
line is for horizontal neutrino-induced muons, the lower one for vertically upwar
muons.
 400 sub-GeV e-like 500 sub-GeV -like

400
300
300
200
200
Number of Events

100
100

0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1

140
multi-GeV e-like 300 multi-GeV -like + PC
120
250
100
80 200
60 150
40 100
20 50
0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
cosine of zenith angle

Figure 1: The SuperKamiokande atmospheric neutrino results showing excel-

Figure
lent 10: The SuperKamiokande
agreement atmospheric
between the predicted neutrino
(blue lines)results which exploited
and observed the fact
electron-like
that penetrating neutrinos produce a irreducible flux of electrons and muons at depth.
events, but a sharp depletion in the muon-like events for neutrinos coming
from below, through the earth. The results are fit very well by the assump-
tion of oscillations with maximal mixing (red lines).

18

4
 Figure 11: The fluence of neutrinos at earth, ranging from cosmic background neutrinos to
those produced by high-energy cosmic ray interactions in the atmosphere. From Haxton and
FIG.
Lin.4. Natural neutrino sources. The terrestrial e flux and continuous flux of extragalact
supernova neutrinos of all flavors are from Krauss et al. [12]. The solar (fusion) e flux is th
standard solar result of Bahcall et al. [11]. The thermal solar neutrinos are for a single flavor.

spectrum contains a great deal of information on the temperature distribution within th

sun.
These remarks are made because the most likely opportunity for measuring the therm
neutrino spectrum is a process that depends on flux density, not on total flux, and whic
samples that flux at a precise energy, the19resonant reaction
e + e + (A, Z) (A, Z 1). (2
This reaction has been discussed previously in connection with terrestrial e sources [12
14 20. Cosmic rays
10
E 2.7dN/dE [cm2 s1 sr1 GeV 1.7]

0.1

1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021
E [eV/nucleus]
16 20. Cosmic rays
1026

Figure 20.9: The all-particle spectrum:  [52], N [54], H [55], O [56], 4 [57],
E 3dN/dE [cm2 s1 sr1 eV 2]

+ [58], X [59],  [60]. References for the high energy portion of the spectrum are
102520.10.
given in Fig.

of the muons ( ) is larger and depends on the transverse momenta of the muons at
production as well as multiple scattering.
24
There are 10
large fluctuations in development from shower to shower, even for showers of
the same energy and primary massespecially for small showers, which are usually well
past maximum development when observed at the ground. Thus the shower size Ne and
primary energy E0 are only related in an average sense, and even this relation depends
1023atmosphere. One estimate of the relation is [52]
on depth in the
1018 1019 1020 1021
E0 106
3.9E[eV] 6 0.9
GeV (Ne /10 ) (20.13)
for vertical showers with 1014 < E < 1017 eV at 920 g cm2 (965 m above sea level).
Figure 12: From about ten years ago, when an all-particle cosmic-ray spectrum derived from
Because existing
of fluctuations,
detectors N e as aapproximately
showed function of 20 E0cosmic
is notray
theevents
inverse
withofenergies
Eq. (20.13).
betweenAs E0
1019
increases
Figure the
and 10shower
21
20.10: eV. maximum
Expanded
An (on
of average)
viewspectrum
expanded alsomoves
the ishighest down
energy
shown. into the
portion
From ofand
Gaisser atmosphere
the cosmic-rayand the
Stanev.
relation between
spectrum: o [71]Ne (stereo),  [71] (monocular)
and E0 changes.  [72], Hof[73].
At the maximum shower development, there are
approximately 2/3 particles per GeV of primary energy.
Detailed simulations and cross-calibrations 20 between different types of detectors are
necessary to establish the primary energy spectrum from air-shower experiments [52,53].
Figure 20.9 shows the all-particle spectrum. In establishing this spectrum, efforts have
been made to minimize the dependence of the analysis on the primary composition. In
the energy range above 1017 eV, the Flys Eye technique [71] is particularly useful because
Figure 13: A comparison of the HIRES and AGASA high-energy cosmic ray events. The
latter data set tends to show more events at or above 1020 eV.

21
Let p = (, p~) be the photon four-momentum. Then |~p| = . Let P = (, P~ ) be the proton
four-momentum. We evaluate in the center-of-mass

(P + p )(P + p ) = ( + )2 = 2CM (10)

which we recognize as the square of the center-of-mass energy. Note that if CM exceeds

m + MN (11)

then clearly the reaction

+N +N (12)
can occur, degrading the nucleon energy. But the center-of-mass energy is a Lorentz invariant
quantity, so it can be evaluated in the laboratory frame

2CM = MN2 + 2 2P~ p~ (13)

As the cosmic background photons are moving in all directions, we are free to maximize the
RHS by taking cos 1. Noting that the incident nucleon is highly relativitistic

(max 2 2
CM ) MN + 4 (14)

Thus the requirement for photoproduction is max

CM > m + MN
m2 + 2MN m MN m
>

4 ! 2
2.7K
3 1020 eV (15)
E

This results in a mean free path for protons of about 108 light years for protons at 1020
eV or higher energies, a distance substantially smaller than the horizon. Thus if the origin
of such cosmic rays is all of extragalactic space, there should be a very sharp cutoff in the
cosmic ray flux at about this energy. This is called the Greisen-Zatsepin-Kuzmin cutoff, and
is shown in the figure. Yet the very high energy events from AGASA show no evidence for
any such cutoff.

The AGASA events created a lot of controversy. Fairly mundane solutions were offered, such
that these high energy events are exclusively nuclear (CNO elements and then iron), which
can achieve higher energies at lower velocities. It has been suggested that the events are
secondaries from the collisions of still higher energy neutrinos. However this requires new
physics in that the Standard Model predicts a declining neutrino cross section above the Z
mass that would not be sufficient to produce the needed event rate, it is thought. It was
pointed out that ultrahigh energy cosmic rays both above and below the GZK cutoff could
be of relatively local origin, thus evading the cutoff but also raising new questions of what
confines the cosmic rays if they are extragalactic but somehow produced primarily in a local

22
region about us. Perhaps we are somehow near some remarkable local source or sources: our
position in the cosmos is special.

In fact the expect GZK cutoff involves a lot of physics. Pion production may be the key issue
for protons, but in nuclei additional processes have to be considered. At a center-of-mass
energy considerably below pion production theshold, nuclei can absorb (in their rest frame)
a photon of energy 20 MeV, resulting in photodistintegration. This leads to a GZK cutoff
for iron nuclei of 5 1020 eV.

Very high energy cosmic ray interactions with atmospheric nuclei lead to an extensive air
shower of secondary particles that, by the time the shower reaches the earths surface, may
number in the billions and cover an area as large as 25 square kilometers. The event rate
of very high energy collisions, above 1019 eV, is about 1 per 100 square kilometers per year.
Because of this daunting number, the recent construction of the Pierre Auger Observatory
was a major step forward for the field. This detector is sited in western Argentina and its
surface sites are spread over an area of about 4000 square kilometers. The detector is a
hybrid, with two detection schemes. One scheme uses 1600 water-tank detectors that form
a grid with spacings of about 1.5 km. Energetic particles scattering in the detector produce
Cerenkov light that is recorded by phototubes mounted on the tanks. So a shower covering
up to 25 square kilometers might be recording simultaneously in 5-10 tanks. The energy of
the incident particles can be estimated from the amount of light produce in the tanks, and
then this energy can be extrapolated over the entire detection area (that is, one corrects for
the small fraction of the surface that is covered by tanks) to determine the total shower en-
ergy. The second scheme uses the upper atmosphere as the detection medium. The charged
particles in the shower interact with atmospheric nitrogen, leading to atomic excitation and
subsequent emission of ultraviolet light. The resulting N fluorescence is recorded in Pierre
Augers optical detectors a grid of focusing mirrors to collect light and cameras that can
then view the light at distances up to 15 km. The cameras can follow the path of high
energy cosmic ray showers as the shower progresses through the atmosphere. Since all of the
shower is viewed, a more accurate estimate of total shower energy can be made provided
atmospheric conditions allow the shower to be seen.

The Pierre Auger Observatory began operations in 2004. It has gathered significant in-
formation on events with energies in the vicinity of the GZK cutoff. Shower development
depends on the nature of the collision, with the depth to which the shower develops sensitive
to whether the incident cosmic ray was a proton or a nucleus like Fe. Variations in observed
showers are compared to simulations a tricky business because the most energetic cosmic
ray collisions correspond to center-of-mass energies two orders of magnitude beyond those
directly probed at the LHC/FermiLab and RHIC. Because one has the shower energy and a
test of composition, one may be able to tell whether cosmic ray composition is changing with
energy. Cross checks are possible. For example, shower-to-shower variations are predicted
to decrease as the typical mass of the incident nucleus increases.

23
 <X max> [g/cm2]

RMS(Xmax) [g/cm2]
QGSJET01 80
on
850
QGSJETII
Sibyll2.1 prot 70 proton
EPOSv1.99
60
800
50

750 40

30
700 iron
20

10
iron
650

18 19
0 18 19
10 10 10 10
E [eV] E [eV]

FIGURE 3. hXmax i and RMS(Xmax ) compared with air shower simulations [22] using different hadronic interaction models [23].
Figure 14: Pierre Auger data for shower depth (in g/cm2 , left panel) and in the RMS
variation in shower depth (right panel) both show trends that suggest the primary cosmic
range. For this purpose the effective upper and lower field FD stations within the quoted uncertainties.
ray spectrum is trending away from primarily protons at energies around 1018 eV to nuclei at
of view limits are calculated for each event and hXmax i is The measured values are displayed in Fig. 3. A fit of
energies
measurednear the GZK
as a function cutoff.
of these From
limits. M. Unger
An example of and the
hXmax PAwith
i data Collaboration, arXiv:1103.5857.
a constant elongation rate does not de-
the hXmax i dependence on the lower FOV limit is shown scribe our data ( 2 /Ndf=34.9/11), but using two slopes
in Fig. 2 for data and simulated events. As can be seen, yields a satisfactory fit ( 2 /Ndf=9.7/9) with an elonga-
the hXmax i is asymptotically unbiased for events with a tion rate of (106+35 2
21) g/cm /decade below 10
18.240.05 eV

The Pierre deep

sufficiently Auger FOVdatalimit, of
butFig. 14 show a too
it is systematically trend and
from protons 2 to nuclei as one approaches
(243) g/cm /decade above this energy. If the prop-
theshallow
GZKwhen cutoff.the FOV
Thisboundary starts cutting
is reminiscent into similar
of the the trend
erties that was
of hadronic seen near
interactions do notthe
change knee that
significantly
tails of the Xmax distribution. Obviously, the unbiased re- over less than two orders of magnitude in primary energy
is gion
usually attributed to the cutoff for galactic supernova cosmic ray acceleration, with SN-
depends on the distribution itself and can thus not (< factor 10 in center of mass energy), this change of
associated
be determined acceleration
by simulations. mechanisms
Instead, we likefit theshock
data waves
D10 =(82 able+35 to accelerate
2 nuclei to somewhat
21 ) g/cm /decade would imply a change in
higher
with theenergies
mean of athan protons.
one-sided truncated Thusnormalnuclei tend to
distribu- the dominate
energy dependenceat theof very end of the
the composition aroundenergy
the an-
tion (shownThus
spectrum. as solidsomething
lines in Fig. 2) and reject
similar mightall events
be happening in thetheextragalactic
kle, supporting accelerators
hypothesis of a transition re-
from galac-
that have a FOV limit for which the measured hXmax i de- tic to extragalactic cosmic rays in this region.
sponsible for the highest energy cosmic rays.
parts by more than 5 g/cm2 from its asymptotic value. The shower-to-shower fluctuations, RMS(Xmax ), are
After all cuts, 3754 events are selected for the Xmax obtained by subtracting the detector resolution in quadra-
What about
analysis. The Xsuper-GZK
max resolution events?
as a function Pierre Auger
of energy for finds
tureafromsuppression
the width of of the
theobserved
event rate Xmax setting in
distributions
these events is
19 estimated using a detailed simulation of 6 2 . As can be seen in
above 3 10 eV, consistent with the GZK cutoff. It does not find evidence for super-GZK
resulting in a correction of g/cm
the FD and the atmosphere. The resolution, defined by the right panel of Fig. 3, we observe a decrease in the
events
the fullstandard
suggesting that perhaps earlier
2 levelfirst-generation experiments may not have fully un-
deviation, is at the 20 g/cm above fluctuations with energy from 19 about 55 to 26 g/cm2 as
derstood
a few EeV. theirTheenergy
difference calibrations
between the (see Fig. 15). At
reconstructed energies above 5.5
the energy increases. Assuming again 10 eV athat correlation
the hadronic
with
X max the distribution of nearby extragalactic objects
values in events that had a sufficiently high energy interaction is found,
propertiesincluding
do not change anmuchexcesswithin inthethe
ob-
to be detected
direction independentlyA,
of Centaurus by two theornearest
more FD radio-loud
stations served
active energy range, these
galaxy. decreasingAfluctuations
Centaurus is aboutare15an
is used to cross-check these findings and as it was shown independent signature of an increasing average mass of
million
in [13], light years from
the simulations earth,
reproduce theand
data has
well. a prominent relativistic jet that is thought to extract
the primary particles.
energy from the vicinity of a supermassive black hole at the
For the galaxysof center.
interpretation the absolute values of hXmax i
and RMS(Xmax ) a comparison to air shower simulations
9.7 High energy neutrinos, RESULTS GZK Neutrinos, and Beyond is needed. GZK As can be seen in Fig. 3, there are considerable
Neutrinos
differences between the results of calculations using dif-
TheThe field of very
measured hXmaxhigh energymaxneutrino
i and RMS(X ) are measuredastronomy
in is in its infancy. The Amanda neutrino
ferent hadronic interaction models. These differences are
telescope
energy bins wasof algprototype
E = 0.1 below south Poleanddetector,
10 EeV lg E = and notnow a much
necessarily larger since
exhaustive, second-generation
the hadronic interactionde-
19.4 eV,
tector, IceCube, is operating. Neutrinos penetrating the ice can be detected via the muons
0.2 above that energy. The last bin starts at 10 models do not cover the full range of possible extrapo-
integrating
they produce. up to Thethe highest
muons energy event (E
produce = (59
Cherenkov lationsin
light of low
theenergy accelerator
ice, which candata.
beIf,detected
however, takenby
8) EeV). The systematic uncertainty of the FD energy at face value, the comparison of the data and simulations
photomultipliers
scale is 22% [24]. place on detector
Uncertainties stringsat-placed
of the calibration, in the deep ice from the surface. Deep
leads to the same conclusions as above, namely a gradual
icemospheric
- ice a conditions,
mile below the surface - is exceptional
reconstruction and event selection increaseand
pure of thetransparent,
average mass of with light
cosmic raysattentua-
with energy
13 2 for
tion lengths of 1002 meters or more. Thus a relatively
give rise to a systematic uncertainty of g/cm up to 59 spare
EeV. array of photomultiplier strings,
hXmax i and 6 g/cm for the RMS. The results were It is illustrative to compare the data with predictions
found to be independent of zenith angle, time periods and for a simple two-component proton/iron model using

24
esults from the Auger Observatory Esteban Roulet

log (E/eV)
10
18 18.5 19 19.5 20 20.5
E 3 J(E) [km-2 yr-1 sr -1 eV 2]

sys(E)=22%
1038

HiRes
1037
Auger
power laws
power laws + smooth function

1018 1019 1020

Energy [eV]

1: Spectrum measured above 1 EeV (solid dots) and power law fits with breaks (dotted line). For
Figure 15: The solid dots are the Pierre Auger measurements at the GZK cutoff. Also
ison also the HiRes measurements are displayed (open dots).
shown are data from a second observatory, HiRes. From Roulet and the PA Collaboration,
arXiv:1101.1825.

he physical origin of the ankle is still uncertain, being the main candidate scenarios to explain
ture those relating it to the transition from a dying galactic component to a harder extragalac-
ponent becoming dominant, or alternatively the so-called dip-scenario [3], in which cosmic
e assumed to be extragalactic protons down to energies below 1 EeV and the concave shape
ed arises from the effect of energy losses by pair creation with cosmic microwave back-
(CMB) photons. To properly fit the observed spectral shape this last scenario requires soft
at the sources ( at the source being typically g 2.4-2.7) and/or strong evolution of
rces with redshift, what makes the distant sources intrinsically brighter (or more abundant)
a larger fraction of the observed protons come from 25 far away and are hence more affected
ractions with CMB photons. Also an upward shift of the energy scale of Auger by 40%
be required in this scenario to fit the location of the dip in the spectrum.
nisotropy measurements will help to distinguish among the two scenarios, because the galac-
placed in the ice by using hot water to drill very deep holes in the ice, can view vast masses
of ice efficiently. IceCube is the worlds largest high energy neutrino detector, viewing one
cubic kilometer of deep South Pole ice at depths of 1.4-2.4 kilometers via detectors placed on
86 strings. LBNL played an important role in the design of the detectors and the associated
data acquisition system. IceCube construction was completed at the end of 2010. While
many muons penetrate the ice, almost all of these come from above the detector, the sec-
ondaries from cosmic ray interactions. But because only neutrinos can penetrate the earth,
upward going muons must be the daughters of upward going neutrinos that interact in or
near the detector. Thus by studying upward going muons, the experimenters can determine
the rate of neutrinos penetrating the earth from below.

The IceCube mass is sufficient to detect neutrinos up to about 1016 eV. Beyond this point,
the event rate is too low: the km3 volume is insufficient. One of IceCubes important scien-
tific justifications is using neutrinos to located possible specific point sources so that we
can connect very energetic events with candidate astrophysical accelerators. Since neutrinos
are unaffected by the CMB or by magnetic field, they point back to their sources.

But neutrino astronomy with still higher energy neutrinos is clearly important. One of the
important consequences of the Pierre Auger observation of the GZK cutoff is the associ-
ated prediction that very high energy neutrinos, produced for example by + +
+ e+ + + e . In addition, it is quite plausible that the astrophysical accelerators produc-
ing cosmic rays in the vicinity of 1019 1020 eV are in fact capable of accelerating protons
or nuclei to still higher energies though we would not see such energetic primary cosmic
rays because they cannot penetrate the CMB. Nevertheless, in the vicinity of the accelera-
tor, these cosmic rays would be converted to neutrinos, which would reach earth. The flux
of GZK neutrinos and the flux of those neutrinos associated with cosmic rays accelerated
beyond the GZK cutoff can be estimated, if one assumes a spectrum of primary cosmic rays
(say a proton flux declining as E ); a maximum proton energy (1023 eV was chosen for
one of the calculations used in defining the band in Fig. 16), and a cosmological model
(effectively the z at which the typical neutrino was produced, so that redshift effects can be
accounted for). In this way, reasonable assumptions yield curves within the band of Fig. 16.
The bulk of these neutrinos are beyond the reach of IceCube: the low flux of the highest
energy neutrinos requires detection volumes of 100 km3 or more, not 1 km3 .

Yet these neutrinos are important: they may be our only tool for investigating the universe
at asymptotically high energies and large distances. How can they be detected? First, plau-
sible estimates of fluxes quickly lead to the conclusion that the detector must be huge, on
the order of 100 km3 or more. Clearly such a detector cannot be a conventional one. Second,
one needs some kind of neutrino interaction that can be detected in such a large volume, at
reasonable cost.

There is an attractive possibility, based on the Askaryan effect, proposed by Gurgen Askaryan

26
 o environment at the South Pole.
ct, radio frequency

the ultra-
above 1017
n-Zatsepin-
the interac-
background
nteed flux
of the flux
n, UHECR
tors. Mea-
t only shed
he UHECR
vidual neu-

oherent ra-
y described
etical work
mental work
many of the
sion in the
Figure 1: World ultra-high energy cosmic ray and predicted cos-
[4] has cur- Figure 16: neutrino
mogenic The hashed band as
spectrum is the predicted
of early 2007, spectrum of high
including data fromenergy neutrinos resulting
xes. Direct from
the the interaction
Yakutsk of primary
[5], Haverah cosmic
Park [6], rays with
the Flys the AGASA
Eye [7], CMB. The[8], data points are various
n ice target measurements
HiRes [1], and
of Auger [9], collaborations.
the hadronic Data points
cosmic ray spectrum, represent
including dif-from Pierre Auger. Graph
data
ferential flux dI(E)/dE, multiplied by E 2 . Error bars are statistical
from Allison et al.
only. GZK neutrino models are from Protheroe & Johnson [10] and
Kalashev et al. [11].
lley)
in 1962. A very energetic particle traveling through matter can produce a shower of sec-
ondary particles that carry a net charge and that emit a cone of coherent Cherenkov radiation
at radio or microwave frequencies. Indeed, high energyApril
neutrino reactions in matter produce
8, 2009
an electron/ shower containing perhaps 20% more electrons than positrons: positrons and
electrons annihilate into s, but s can Compton scatter off electrons in the matter to pro-
duce high energy electrons. The charges move faster than the local speed of light in matter,
c/n where n is the index of refraction. At long wavelengths radio or microwaves the
moving front of charges which may have a thickness on the order of 10cm can produce
Cherenkov microwave or radio waves coherently. (The wavelength is large compared to the
thickness of the charge wave.) The Askaryan effect has been observed in several materials,
such as silica, salt, and ice: several of the key experiments were done at SLAC. There are
several detector ideas being developed around this idea, most based on instrumenting a very
large volume of South Pole ice with surface radio detectors.

27
9.8 Gamma rays: general issues, GRO
The properties of the earths atmosphere divides gamma ray astronomy into two halves.
From the ultraviolet to gamma rays of energy 20 GeV the atmosphere is opaque. Thus
observations in this energy range must be done with instruments mounted in satellites or
carried by balloons. The ease of detecting the radiation generally goes up with energy, but
the strengths of typical astrophysical sources go down. The net result is that one can do a
lot over this range: this motivated the design of the four instruments on board the Gamma
Ray Observatory. For gamma rays above 20 GeV, interactions in the atmosphere produce
showers that can be observed either by the Cerenkov light produced by the secondaries or,
at high altitude, by direct detection of the secondaries. Of course, observations can also be
(and are) made by space-bound detectors, too.

Many of the concerns of this field are driven by instrumental issues. Some of the central
issues are clear:
Producing detectors with greater sensitivity. The dynamic range of existing detectors -
the gap between the brightest nearby sources such as the Crab and the faintest detectable
sources - is typically about a factor of 100. Thus the situation is equivalent to being able to
see no stars fainter than the 5th magnitude. Improvements in sensitivity can thus greatly
extend the horizon of our observations, while also allowing much more detailed spectral stud-
ies of known bright sources. Greater sensitivity can be achieved with great collection areas
and by reducing ambient backgrounds. The push towards greater size is clearly an expensive
challenge given the necessity for observing outside the atmosphere.
Enhancing spectral resolution. This includes both enhanced spectral resolution and en-
hanced spatial resolution, the latter required to better identify sources.
Enhanced temporal and spatial coverage and resolution.

The technology in the field has advanced greatly over he past 20 years. The second of
NASAs Great Observatory projects (following the Hubble Space Telescope) was the Comp-
ton Gamma Ray Observatory. Launched in 1991 from the shuttle Atlantis, the GRO oper-
ated for over nine years. Its instruments included BATSE, OSSE, COMPTEL, and EGRET.
BATSE - the Burst and Transient Source Experiment - was an all-sky monitor with sharp
timing capability: its NaI detectors were designed to detect gamma rays with energies be-
tween 20 and about 600 keV. This proved decisive in demonstrating that gamma ray bursts
are distributed approximately isotropically. This detector also led to the discovery of new
pulsars, x-ray novae, and the identification of one soft gamma ray repeater. OSSE - the
Oriented Scintillation Spectrometer Experiment - employed NaI and CsI detectors to look
for gammas between 50 keV and 10 MeV, and could point to sources. COMPTEL - the
Imagining Compton Telescope detected gamma between 0.75 and 30 MeV, with an angu-
lar resolution of 1 degree. EGRET - the Energetic Gamma Ray Experiment - was able to
correlate high energy gammas (20 MeV-30 GeV) with the lower energy bursts detected by
BATSE. It also measured high energy gammas from active galaxies.

28
9.9 Nuclear Gamma Rays
We have touched on this theme before, but here Id like to gather together several examples
of what is becoming possible. One of these examples is 26 Al, which 720,000 y lifetime for
decay to 26 Mg. The decay of the 5+ ground state populates the first two excited states of
26
Mg, which are 2+ states with energies of 2.938 and 1.809 MeV. The latter state is populated
97% of the time, so the primary signature is a 1.809 MeV . The 2.938 MeV state decays
to the 1.809 MeV level, so a small number of 1.129 MeV s are also produced.

The primary site for producing 26 Al is thought to be Type II and IIb supernovae. Addi-
tional aluminum may come from the winds of very massive stars. COMPTEL has produced
a galactic map of the 1.809 MeV s. That map differs from higher energy ( 100 MeV) maps
in that there is marked clumpiness to the production, including intense sources associated
with Cygnus, Vela, etc. This map is interpreted as an indicator of recent supernova activity.
The conclusions drawn from such a map are clearly model dependent because one obtains
an angular distribution but no spatial depth information. But under the assumption that
the entire galaxy is contributing in the expected way, one deduces a recent supernova rate
of 3.4 2.8/century. The model dependence also includes the uncertainty in the aluminum
production per event. The attribution of the total flux to the Al injection rate of massive
stars yields an upper bound on the recent star formation rate of 5 4 M per year. This
calculation, of course, requires not only a model of the Al production per supernova, but
also a model of the range of stellar masses that undergo core collapse and a model for the
distribution of stars with mass (populations roughly decline exponentially with an exponent
of about -2.35).

In a similar way, 44 Ti decay proceeds with a 60 year half life to the ground state of 44 Sc,
which in turn electron captures to 44 Ca. The order of the states in Sc is 2+ (gs), 1 (68
keV), and 0 (146 keV). The decay feeds the second excited state 98% of the time, which
then decays through the 1 state to the ground state, producing s of 78 and 68 keV. The
subsequent decay to 44 Ca has a 4 hour lifetime and produces a 1.157 MeV , as the 2+ first
excited state of 44 Ca is populated 99% of the time. The 100 keV line for the the source
Cas A is within the detection abilities of OSSE, while the 1.157 MeV line can be seen by
COMPTEL.

Various types of supernovae are thought to produce 44 Ti, including both types I and II. As
in the case of the 26 Al line, the galaxy is effectively transparent to the produced ray, so
the detection provides a measure of the very recent supernova rate free from worries about
obscuration. With COMPTEL one had an instrument sensitive to nrearby supernovae occur-
ring within the past 1000 years, or 15 half lives. Given a supernova rate of some several per
century, it is clear that the distribution should be from quite localized sources, representing
recent events. Typical productions of 44 Ti from supernovae are, according to modelers, on
the order of 104 M per event.

29
Figure 17: COMPELs map of the galactic plane in the 1.809 MeV gamma ray emitted
by 26Al. The distribution is not uniform. It is thought to represent over 100,000 events,
supernovae or novae, producing something on the order of 1-3 solar masses of 26Al.

The youngest known galactic supernova remnant is Cas A, noted optically by John Flam-
steed in 1680. COMPTEL reported the observation of 44 Ti s from Cas A in 1994. The
results of a survey extending over a six-year period beginning in 1991 found only one ad-
ditional source, identified with a young supernova remnant not observed either optically or
in the radio. The source is located in the direction of the Vela constellation. Constraints
on the doppler broadening of the 1.16 MeV line limits the velocity of the ejecta to no more
than 19000 km/sec. The COMPTEL results are confirmed by ROSAT x-ray data of the
vela region, which found a shell-type supernova remnant at the same location. The shell
temperature is on the order of several keV, which also indicates a young remnant. Finally,
COMPTEL previously saw 26 Al in this region, attributing this to the known Vela supernova
remnant. However the centroid of that distribution has been argued to better fit the new
supernova remnant. Modelers are currently engaged in arguments about the nature of the
progenitor, using both the ejection velocity bound and the 44 Ti yield to bound models. One
possibility is a core-collapse supernova of a massive star that had previously lost its hydrogen
envelope. It has been claimed that a supernova at this distance (100-300 pc, derived from
the Ti flux and age estimates of 600-1100 years based on the expansion velocity) could have
been as bright as the moon, leaving an interesting question as to why it was not observed.

Finally, there have been recent papers suggesting that future generations of gamma ray de-
tectors might see s in the energy range of 100-700 keV associated with elements specific to
the r-process. Establishing a correlation between such s and known supernova remnants
could thus establish the r-process site and, potential, constrain the total r-process production
per site. One of the candidates, 126 Sb, has a 144,000 year lifetime, easily long enough to
allow a galaxy survey. It produces lines at 415, 666, and 695 keV. The proposed detector
ATHENA possibly could detect 126 Sb lines from Vela.

30
9.10 Astrophysics of Intermediate Energy Gammas
EGRET conducted the first all-sky search for gamma sources above 100 MeV, succeeding
in identifying 271 high-energy gamma ray sources, most of which are associated with active
galaxies and characterized by nonthermal spectra. 170 these were unknown prior to the
survey. Such blazars are highly variable and are bright radio sources. The radio structure
consists of knotty jets moving outward at high velocities. The luminosity in gamma rays
can exceed that from other wavelengths by up to two orders of magnitude. The variability
of the emission can be fast, less than a week. The density of high energy gammas at the
source are sufficient that photon-photon pair production would keep them trapped, unless
the gammas are highly beamed, as in a relativistic jet. Thus the hope is that the gamma
ray spectrum can yield information on the nature of the jet and of the acceleration processes
occurring there. It is thought that the gamma rays may originate from a region of the jet
closer to the central engine, than in the case of the radio emission.

A few blazars have been observed producing very high energy gamma rays, up to TeV scales,
which then can be readily observed from earth. One, Markarian 421, was measured in the
TeV range by the Whipple Observatory, which detects Cerenkov radiation from air showers.
It was not seen at GeV energies by EGRET. The high energy flare had a rise time of about
two days. One day after the flare began this source was seen in the X-ray. Both of these
signals differ from typical blazars in their higher energies: most blazars have lower energy
gamma and low-energy radiation that does not extend into the x-ray. The interpretation is
that blazars accelerate electrons to high energies, which then radiate soft synchrotron radi-
ation and hard gammas by inverse Compton scattering. Markarian 421 is thus exceptional
in the energies to which it accelerates electrons, accounting for the higher energy of its x-
ray/gamma emission.

Markarian 421, which was first observed in May, 1994, is not unique. The second closest
blazar of this type, Markarian 501 (z=0.034), was seen in 1997. It flared to become the
brightest TeV source in the sky, outshining the Crab Nebulae by an order of magnitude.
The periods of flaring lasted a few days. The elevated activity spanned a period from about
March through June. The high energy spectrum (observed by Whipple) was flat from below
a TeV to at least 10 TeV. The x-ray cutoff in Markarian 421 is about 1 keV; in Markarian
501 it is above 100 keV. As in 421, it is assumed that this spectrum comes from relativistic
acceleration of electrons along a jet closely aligned with our line of sight.

What does this tell us about the electron energies? First, the plasma in the jet is moving
towards us, boosting the energy of the emitted gammas, relative to the jet rest frame. The
Doppler factor is q
(1 v 2 /c2 )
D= (16)
(1 + z)(1 v cos /c)
where is the observation angle relative to the jet axis. This is the standard relativistic

31
Doppler shift corrected by the redshift. The observed energy of the gamma rays is

Emax Dme c2 (17)

where me c2 is the maximum energy of the electrons in the jet rest frame. The Doppler
factor also appears in the calculation of the photon density in the blob, and thus of the blobs
opacity to high energy s. The argument goes as follows: if tobs is the fastest observed TeV
gamma ray flare variability, then the radius of the blob emitting the photons must be less
than cDtobs . Thus a larger D means a lower photon density. From this one concludes
D> 30. Since the highest energy gammas from Markarian 501 is about 20 TeV, one derives
for the maximum energy of electrons in the jet frame

mc2 0.65TeV (18)

Since the maximum electron velocity is now known, one can deduce the magnetic field in the
jet required to produce synchrontron radiation with a maximum energy of 200 keV. That
yields
Emax DB 2 B 0.7G (19)
where B is the field in the jet rest frame. Other aspects of the blazar dynamics - such as the
physics responsible for the short timescale of the flares - is less clear.

Another exciting result involving high energy s are the gamma ray burst observations of
EGRET. While the vast majority of the bursts are seen by BATSE, on the order of 10%
of the events produce spectra that extend into EGRETs range of above 30 MeV, typically
producing about five counts. A great deal more about such events is now known because of
the Fermi Telescope.

These high energy tails place a lot of constraints on gamma ray burst models. Since high
energy gammas were also seen early in the burst, there must be high-energy particle accel-
eration simultaneous to the keV emission. The lack of attenuation of high energy s from
pair production off lower energy s place a particularly strong constraint on the source
and on models involving beaming.

Finally, a third interesting source of high energy s are both isolated and binary pulsars.
Gamma ray emission can be very strong, representing 10% or more of the spin-down energy
of some pulsars. There is relatively little consensus on the mechanism or even the precise
site of the gamma ray production (neutron star surface? accretion disk? etc?)

Centaurus X-3 is a well-studied high-mass accreting X-ray binary. EGRET measured a

smoothly declining spectrum that extended from 100 MeV to its detection limit. Results
have been reported from the Durham Mark 6 gamma ray telescope in the vicinity of a TeV:
the flux is consistent with a linear extrapolation of the EGRET flux. Centaurus X-3 contains
a 4.8s pulsar in a 2.1 day orbit about an O-type supergiant V779 Centaurus. The pulsar

32
period has been shortening since its discovery almost 30 years ago, which is attributed to
spin-up from matter accreting on the neutron star from the more rapidly rotating inner edge
of its accretion disk. The EGRET GeV burst observations were pulsed in agreement with the
X-ray period. Initial TeV gamma ray observations also indicated pulsation near the pulsar
period and localized in But later measurements at high energy indicated unpulsed emission
that one would then associated with radiation over an extended volume encompassing the
orbit.

9.11 Fermi Telescope

A new high-energy gamma ray telescope the Fermi Telescope, or the telescope-formerly-
known-as-GLAST was launched into space in June of 2008. Its purpose is to probe in-
termediate energy gamma ray sources, such as active galactic nuclei, supernova remnants,
and pulsars, in order to catalog these sources and determine the nature of these accelerators.
The telescopes can detect gammas with energies between 20 MeV and 300 GeV so it can
be viewed as somewhat of a successor to EGRET. The telescope looks for transient events
as well as exotic phenomena, such as dark matter annihilation into high energy s. One
instrument, the Gamma Ray Burst Monitor, is designed to view the entire sky not occulted
by earth, to look for and identify the location of gamma ray bursts. It is sensitive to gam-
mas between 8 keV and 40 MeV. The second instrument, the Large Area Telescope, has an
aperture that can view 20% of the sky at one time. It can detect gammas from 20 MeV
to 300 GeV. In its first year, Fermi extended the EGRET catalogue of 271 point sources to
1451 sources. Among Fermi discoveries to date are the observation of a gamma-ray-emitting
pulsar but emits in no other band; the observation of the most energetic gamma-ray burst
yet recorded; and the observation of very energetic cosmic rays from the vicinity of supernova
remnants. It has also placed important constraints on the diffuse gamma ray background,
an isotropic gamma ray component of unknown origin, and on gamma-ray decay channels
of dark matter candidates.

9.12 Gamma ray bursts

Gamma ray burst have been mentioned several times. Gamma ray bursts are short-lived
burst lasting from a few milliseconds to several minutes. They are detected roughly once
per day, from all directions in the sky. They represent tremendous energy, several hundred
times brighter than a typical supernova (were they radiated in 4). They were originally
detected in the 1960s by military satellites monitoring nuclear testing. They are not local,
but originate at cosmological distances. In the past few years it has become clear they are
associated (at least some of the time) with certain supernovae.

Recently fast-response telescopes - like NASAs High-Energy Transient Explorer or the

Japanese Automated Response Telescope - have been able to view the gamma-ray burst
area within a couple of minutes of the burst. The afterglows have been observed for on
the order of an hour to a day after the burst representing a lot of additional energy.

33
Figure 18: The Fermi Telescope, prior to launch.

34
Figure 19: The Fermi Telescopes map of gamma rays and gamma-ray point sources in the
plane of the galaxy. Compare to Figure 17.

General ideas exist that gamma ray bursts are associated with fireball phenomena where
tremendous energy and high velocities are produced in a region relatively free of baryons
to keep the gamma rays from thermalizing. The motion leads to a beaming of the burst,
reducing the energy requirements but increasing the frequency of events (many of which are
not beamed at earth and thus are not observed).

A specific idea is the hypernova. A massive stars core collapses into a black hole. The black
holes spin or magnetic fields may act like a slingshot, flinging material outward as a beamed
blast wave or jet. The gamma rays are created when the blast wave collides with stellar
material still inside the star. The gamma rays burst out of the star just in front of the blast
wave. Behind the gamma rays, the blast wave pushes the stellar material outward.

The blast wave then sweeps through space, colliding with gas and dust, producing additional
radiation progressing from less energetic gammas, to x-rays, to visible light, and to radio.
This forms the afterglow, lasting perhaps up to days.

35
Figure 20: A long-duration gamma ray burst event.

36