Journal of Cosmology and

Astroparticle Physics

You may also like

- Neutrino physics with JUNO
Neutrino decays over cosmological distances and Fengpeng An, Guangpeng An, Qi An et al.

the implications for neutrino telescopes - Letter of intent for KM3NeT 2.0
S Adrián-Martínez, M Ageron, F Aharonian
et al.
To cite this article: Philipp Baerwald et al JCAP10(2012)020 - Physics at a future Neutrino Factory and
super-beam facility
A Bandyopadhyay, S Choubey, R Gandhi
et al.

View the article online for updates and enhancements.

This content was downloaded from IP address 185.121.174.253 on 10/02/2023 at 06:25

J ournal of Cosmology and Astroparticle Physics
An IOP and SISSA journal

Neutrino decays over cosmological

distances and the implications for

JCAP10(2012)020
neutrino telescopes
Philipp Baerwald, Mauricio Bustamante and Walter Winter
Institut für Theoretische Physik und Astrophysik, Universität Würzburg,
97074 Würzburg, Germany
E-mail: philipp.baerwald@physik.uni-wuerzburg.de,
mauricio.bustamante@physik.uni-wuerzburg.de, winter@physik.uni-wuerzburg.de

Received September 3, 2012

Accepted September 18, 2012
Published October 15, 2012

Abstract. We discuss decays of ultra-relativistic neutrinos over cosmological distances by

solving the decay equation in terms of its redshift dependence. We demonstrate that there
are significant conceptual differences compared to more simplified treatments of neutrino
decay. For instance, the maximum distance the neutrinos have traveled is limited by the
Hubble length, which means that the common belief that longer neutrino lifetimes can be
probed by longer distances does not apply. As a consequence, the neutrino lifetime limit from
supernova 1987A cannot be exceeded by high-energy astrophysical neutrinos. We discuss the
implications for neutrino spectra and flavor ratios from gamma-ray bursts as one example
of extragalactic sources, using up-to-date neutrino flux predictions. If the observation of SN
1987A implies that ν1 is stable and the other mass eigenstates decay with rates much smaller
than their current bounds, the muon track rate can be substantially suppressed compared to
the cascade rate in the region IceCube is most sensitive to. In this scenario, no gamma-ray
burst neutrinos may be found using muon tracks even with the full scale experiment, whereas
reliable information on high-energy astrophysical sources can only be obtained from cascade
measurements. As another consequence, the recently observed two cascade event candidates
at PeV energies will not be accompanied by corresponding muon tracks.

Keywords: cosmological neutrinos, neutrino astronomy, ultra high energy photons and neu-
trinos, gamma ray bursts theory

ArXiv ePrint: 1208.4600

c 2012 IOP Publishing Ltd and Sissa Medialab srl doi:10.1088/1475-7516/2012/10/020

Contents

1 Introduction 1

2 Decay framework 3
2.1 General framework including cosmology 3
2.2 A simplified solution of the decay equation, and a stability paradox 5
2.3 Solution of the redshift-dependent decay equation 6

JCAP10(2012)020
3 Application to GRBs 8
3.1 Benchmark GRBs and method 8
3.2 Impact of decay on neutrino fluxes, and comparison of methods 9
3.3 Fluxes and flavor ratios for ν1 stable 11

4 Consequences for realistic GRB stacking analyses 13

5 Summary and discussion 15

1 Introduction

While the masses and mixings of the neutrinos are already well probed, other neutrino prop-
erties are less understood, such as the electromagnetic properties and the neutrino lifetime.
The most stringent phenomenological bound on neutrino lifetime comes from the observation
of neutrinos from supernova 1987A [1, 2], which were measured in the electron flavor. Given
the uncertainty on the supernova neutrino flux (order 50%) and the neutrino mixing param-
eters, it may apply to the mass eigenstate ν1 or ν2 . For the sake of simplicity, we assume that
the bound applies to ν1 : τ1 /m1 & 105 s/eV.1 The (model-independent) bounds on the other
mass eigenstates are less stringent: bounds on ν2 lifetime are imposed by solar neutrino data,
yielding τ2 /m2 & 10−4 s/eV for decays into invisible daughters [3–5] and τ2 /m2 & 10−3 s/eV
for decay modes with secondary ν̄e appearance [6, 7]. Furthermore, ν3 is constrained from
the analysis of atmospheric and long-baseline neutrino data, τ3 /m3 & 10−10 s/eV [8]. More
stringent bounds can be derived when specific decay models are assumed (see, e.g., refs. [9–
11] for an overview). For example, solar neutrinos strongly limit the possibility of radia-
tive decays [12], while for Majoron decays [13, 14], explicit bounds can be obtained from
neutrino-less double-beta decay and supernovae [15]. Another possibility is the decay into
un-particles [16–19]. In this study, we do not consider specific decay models, but focus on
the phenomenology of neutrino decay, given the bounds on the lifetimes above. Especially
neutrino telescopes [20–23] are sensitive to neutrinos with an average energy and traveled
distance many orders of magnitude larger than present neutrino experiments, and may be
an interesting approach to probe neutrino decay.
Neutrino oscillations are typically assumed to be averaged out over cosmological dis-
tances (see ref. [24] for a discussion), and the usual flavor mixing remains to be taken into
Neutrino lifetime is usually described by κ−1
1
i ≡ τi,0 /mi , where τi,0 is the rest frame lifetime of the
mass eigenstate νi . The origin of the mass dependence is the fact that decays scale as exp[−t/(τi,0 γ)] '
exp[−(Lmi )/(Eτi,0 )], i.e., the rest frame lifetime τi,0 is boosted by γ = E/mi into the observer’s frame, and
L (baseline source-detector) and E (neutrino energy) are quantities related to source and experiment.

–1–
account. As far as decays are concerned, one distinguishes between decays into products
invisible to the detector, such as sterile neutrinos, un-particle states, Majorons, or active
neutrinos strongly degraded in energy, and decays into visible states, i.e., active neutrino
flavors. The decays can be complete, i.e., all unstable mass eigenstates have decayed (see,
e.g., ref. [25]), or incomplete, i.e., the decay spectral signature will be visible. A complete
classification of complete (visible or invisible) decay scenarios has been performed in ref. [26],
while incomplete invisible decay has, for instance, been studied for active galactic nuclei
(AGNs) in ref. [27, 28]. The description of incomplete visible decays is, in general, more
complicated [29, 30], which is why we focus on incomplete invisible decay in this study. The
characteristic energy dependence of neutrino decay may lead to easily observable imprints in

JCAP10(2012)020
the neutrino fluxes in that case [27, 28]. On the other hand, it has been proposed to use the
flavor composition at the detector, such as the ratio between muon tracks and cascades, to
distinguish different decay scenarios [19, 25, 26, 31–33]. We will discuss both options in this
study, where our main focus are the fluxes.
It is often believed that decays are always complete if the sources lie far enough away,
which, in turn, allows for the test of very long neutrino lifetimes of the order of
 
−1 s τ [s] L [Mpc]
κ ≡ ' 102 . (1.1)
eV m [eV] E [TeV]

As a consequence, especially objects with high redshifts may be well suited to test neutrino
lifetime, and may potentially lead to bounds stronger than the one from SN 1987A. An
example, which we are going to use in this work, are gamma-ray bursts (GRBs), with observed
redshifts as high as z ' 6 − 8; see ref. [34] for the redshift and luminosity distribution. This
leads to potentially strong constraints via eq. (1.1). However, note that the relationship
between distance L and redshift z depends for z & 0.1 (or L & 360 Mpc) on the cosmological
distance measure to be used, which we are going to address in this study; see refs. [35, 36]
for a discussion in the context of neutrino oscillations, and refs. [37, 38] in the context of
pseudo-Dirac neutrinos. For the neutrino flux, especially redshifts z ' 1 will dominate [39],
which is a consequence of the convolution of the star formation rate (including some redshift
evolution function) and the contribution to the total flux scaling as 1/d2L , where dL is the
luminosity distance. This value is already significantly beyond the indicated z ' 0.1, which
means that cosmological effects have to be taken into account.
GRBs are not only an interesting class of candidate neutrino sources because they
may originate from high redshifts, but also because the non-observation of neutrinos from
GRBs [40] has started to constrain the gamma-ray-neutrino connection in conventional in-
ternal shock models [41, 42], based on the ideas in ref. [43]. A re-calculation of the predicted
neutrino fluence from the gamma-ray fluence has yielded an order of magnitude lower re-
sult [44] for the same burst parameters (see also ref. [46]), which is a possible explanation
soon going to be tested by new data. We will use this new nominal prediction for the quasi-
diffuse flux [44], calculated with the IC-40 bursts [47], in this study (IC-40 refers to the
40-string configuration of the IceCube detector). Note that model-specific [44, 45] and more
generic [48] astrophysical uncertainties may imply even lower neutrino flux predictions. On
the other hand, another plausible explanation for why nothing has been seen yet may be
that neutrinos have actually (at least partially) decayed between source and detector. We
will investigate the impact of this hypothesis on the predicted fluxes.
One of the very recent puzzles of neutrino astronomy is the potential observation of two
cascade events at IceCube at PeV energies [49]. In particular, assuming that these are of ex-

–2–
tragalactic origin, an interesting question is: Why have these been observed as cascades, and
no muon tracks have been seen? This question may become more prominent with increasing
statistics. In this work, we demonstrate that neutrino decay can provide an answer to that
question, given the current constraints on neutrino lifetime; for alternative explanations, see
refs. [50, 51]. We will demonstrate that muon tracks may be strongly suppressed in the pres-
ence of neutrino decay compared to (especially electromagnetic) cascades, and that cascade
measurements are much more powerful to find astrophysical neutrinos in this case. We will
use GRBs as an example, but our conclusions can be applied to AGNs as well.
This study is organized as follows: We describe the decay framework in section 2, which
consists of the general framework including cosmological distances, a simplified solution,

JCAP10(2012)020
and the proper solution of the redshift-dependent decay equation. We then discuss the
implications for GRBs in section 3, as one possible example of extragalactic neutrino sources.
In that section, we introduce several benchmark GRBs and the analysis techniques, discuss
the impact of decay on the neutrino fluxes, and introduce a phenomenologically viable model
for which we show the implications for fluxes and flavor composition. In section 4 we apply
this model to a realistic stacking analysis of IceCube, and compare the predicted neutrino
fluxes to current and future bounds. Finally, we conclude in section 5. Note that, while we
use GRBs as test case, the results from this study apply to the neutrino propagation from
other extragalactic sources as well, including cosmogenic neutrinos.

2 Decay framework

Here we introduce in the first subsection the general decay framework including the aspect of
cosmological distances. We solve the decay equation in a simplified approach in the second
subsection to study when a cosmological framework has to be used, and we show the proper
solution in the third subsection.

2.1 General framework including cosmology

Neutrino decay of the mass eigenstate νi can be described by the usual differential equation
dNi
= −λi Ni . (2.1)
dt
Here the decay rate (inverse lifetime) λi is given by
mi 1 κi
λi = ≡ , (2.2)
τi,0 E E

with τi,0 the rest frame lifetime boosted by γi = E/mi into the observer’s frame, κi ≡ mi /τi,0 ,
and E the neutrino energy. Since neither the rest frame lifetime τi,0 nor the lifetime τi = γi τi,0
are directly observable, typically κ−1i = τi,0 /mi [s eV−1 ] is quoted as the lifetime of the
neutrino mass eigenstate νi .
The time t is for ultra-relativistic neutrinos typically associated with the baseline L,
which is the distance between source and detector. Let us assume that the neutrinos under
study are produced at a source with a redshift z. Over cosmological distances, we therefore
need to express this distance in terms of redshift, i.e., L(z). In cosmology, however, the
relationship between distance and redshift depends on the cosmological distance measure,
which can be luminosity distance, angular diameter distance, proper distance, or light-travel
distance; see, e.g., ref. [52]. For neutrino decay, the relevant quantity is the time traveled

–3–
 5
10
light-travel distance
10
4 luminosity distance

3
10

2
10

L(z) [Gpc]
LH =3.89 Gpc
1
10

0
10

-1
10

-2
10

JCAP10(2012)020
-3
10 -3 -2 -1 0 1 2 3
10 10 10 10 10 10 10
z

Figure 1. Light-travel distance as a function of redshift in a ΛCDM cosmology (solid black line).
For comparison, we have also included the luminosity distance (dashed curve).

in eq. (2.1), which is to be associated with the light-travel or look-back distance. In other
words, the clock which triggers neutrino decay is directly related to the distance the neutrino
has traveled since its production. By definition,
dz 0
Z z
L (z) = LH 0 0
, (2.3)
0 (1 + z ) h (z )

with LH ≡ c/H0 ≈ 3.89 Gpc the Hubble length [53], h (z) ≡ H (z) /H0 , and H0 the Hubble
constant. The Hubble parameter H(z) is defined as
q
H (z) ≡ H0 Ωm (1 + z)3 + ΩΛ , (2.4)
assuming a flat ΛCDM cosmology. We choose Ωm = 0.27 and ΩΛ = 0.73 in the following [54],
unless explicitly noted otherwise.
The light-travel distance L(z) is shown in figure 1 compared to the luminosity distance.
It is limited by the Hubble length LH ' 3.89 Gpc, which is the horizon beyond which ultra-
relativistic particles cannot be seen. This has interesting implications for neutrino decay: the
maximum distance relevant for the test of decay is limited, and therefore also the maximum
time the ultra-relativistic neutrinos have traveled. From figure 1 it is clear that the notion
that larger distances can probe longer lifetimes only holds for z . 0.1 (or L . 360 Mpc),
where the cosmological distance measures are very similar. On the other hand, for z & 1,
the light-travel distance is limited. For “typical” neutrino peak energies higher than about
102 TeV for GRBs and AGNs, one can estimate that the maximal testable lifetimes are about
κ−1 ∼ 104 s eV−1 from eq. (1.1). Therefore, the lifetime limit from SN 1987A cannot be
exceeded for the electron flavor (or the corresponding mass eigenstates). Although neutrino
telescopes may also probe κ−1 3 using muon tracks (the poorly constrained lifetime of ν3 ), any
future galactic supernova neutrino observation, which may use different flavors and neutral
currents, will likely provide much better bounds because of the lower neutrino energies.
The observable neutrino flux φα (GeV−1 cm−2 s−1 ) including flavor mixing and decay
(neglecting possible Earth attenuation effects) is given by
X
φνα +ν̄α (E0 , z) = Pαβ (E0 , z) φ0νβ +ν̄β , (2.5)
β=e,µ,τ

–4–
where φ0νβ +ν̄β is the neutrino flux without propagation effects and E0 is the observed energy.
Note that we simply add the neutrino and antineutrino fluxes here, since most observables
are not sensitive to the neutrino polarity. The transition probability Pαβ for the transition
να → νβ including flavor mixing and decay is given by
X Ni (E0 , z) X
Pαβ (E0 , z) = |Uαi |2 |Uβi |2 = |Uαi |2 |Uβi |2 Di (E0 , z) , (2.6)
i N̂i (E0 ) i

where Ni (E0 , z) is the number of neutrinos left after traveling over a distance L(z), N̂ (E0 )
is the initial number of neutrinos (as a function of the observed energy), and Di (E0 , z) ≡

JCAP10(2012)020
Ni (E0 , z)/N̂i (E0 ) ≤ 1 is the corresponding damping factor [55]. It is the solution of eq. (2.1),
which we will address in the next subsections. For the mixing angles, we use the best-fit values
(for normal hierachy) from ref. [56], i.e., sin2 θ12 = 0.307, sin2 θ23 = 0.398, sin2 θ13 = 0.0245,
and the CP-violating phase δ = 0.89 · π. As observables, we use the muon neutrino flux
φµ ≡ φνµ +ν̄µ , representing muon tracks, the electron neutrino flux φe ≡ φνe +ν̄e , representing
electromagnetic cascades, and the flavor ratio R ≡ φµ /(φe + φτ ) [57], representing the ratio
between muon tracks and cascades.2 The flavor composition at the source is computed in
a self-consistent way including the cooling of the secondary pions, muons, and kaons, and
the helicity-dependent muon decays; see, e.g., refs. [31, 58–60]. For example, for GRBs, a
transition from φ0e : φ0µ : φ0τ = 1 : 2 : 0 (pion decay source) to 0 : 1 : 0 (muon-damped source)
is expected, which can be predicted as a function of the source parameters.

2.2 A simplified solution of the decay equation, and a stability paradox

Here we solve the decay equation in the usual way used in the literature, and apply it to
cosmological distances. Although we will demonstrate that this approach does not work, it is
quite illustrative to understand the problems and in what redshift range simplified approaches
can actually be used.
The decay solution, eq. (2.1), has, for λi = const., the well-known solution

Ni (t) = Ni (t = 0) e−λi t = N̂i e−λi t , (2.7)

where N̂i is the initial number of νi . We can easily modify that equation, using eq. (2.3),
such that3
Ni (z) = N̂i e−λi L(z) , (2.8)
where Ni (z) is the number of remaining neutrinos after traveling L(z). Note that here the
redshift is measured relative to the origin z, whereas in cosmology z = 0 (and quantities
marked with “0”) refer to the current epoch.
We also need to take into account the effect of the adiabatic expansion of the Universe
on the neutrino energy: if E(z) is the energy in the production epoch and E0 is the observed
energy, then
E(z) = (1 + z) E0 . (2.9)
With this, the decay rate in eq. (2.2) acquires a redshift dependence, i.e.,
κi
λi = λi (z) = . (2.10)
E0 (1 + z)
2
For the sake of simplicity, we ignore neutral current cascades here, which would enter as a background.
3
We use natural units, i.e., ~ = c = 1.

–5–
Using that in eq. (2.8), we can therefore write the damping factor needed for eq. (2.6) as
 
κi L (z)
Di (E0 , z) = exp − (2.11)
E0 (1 + z)

with the definition of L(z) from eq. (2.3). In order to isolate the redshift-dependent part of
the evolution, we define the dimensionless function
 
L(z)
Z1 (z) ≡ exp , (2.12)
LH · (1 + z)

such that eq. (2.11) can be re-written as

JCAP10(2012)020
κi L H
−
Di (E0 , z) = [Z1 (z)] E0 . (2.13)

Here the exponent depends on κi and observed energy E0 only, whereas the base depends on
redshift only. One can easily see that for E0 → ∞, the damping factor Di (E0 , z) → 1, which
corresponds to stable neutrinos. For finite E0 and large decay rates κi  E0 /LH , however,
one finds Di (E0 , z) = 1 (stable neutrinos) for Z1 = 1, Di (E0 , z) → 0 (complete decays) for
Z1 > 1, and Di (E0 , z) → ∞ (exploding solution) for Z1 < 1. Because large enough decay
rates correspond to complete decays, only the second case can be a physical solution.
Since Z1 is a function of redshift, we show the functional dependence Z1 (z) in figure 2.
Obviously, Z1 ≥ 1, which means that the number of particles does not explode (third case
excluded). However, for large z, Z1 → 1, which means that the neutrinos from high redshifts
are stable, and the notion of complete decays does not exist. From eq. (2.13), they are
even stable independently of the value of E0 , which is counter-intuitive, since more energetic
neutrinos should live longer than less energetic ones. So what is wrong with this approach?
Obviously, introducing the energy dependence eq. (2.9) after the solution of eq. (2.1) does
not take into account that the neutrinos lose energy continuously, which changes the solution
of the differential equation. That is, the energy assigned to the neutrinos is chosen to be the
one at the production point, which is on average too high, and the neutrinos are hence too
long-lived. While one would naively expect that this effect is small, we have encountered a
severe contradiction here, which we are going to resolve in the next subsection.

2.3 Solution of the redshift-dependent decay equation

The proper solution is, of course, to re-write eq. (2.1) in terms of redshift using the redshift-
dependent decay rate in eq. (2.10):
dNi (E0 , z) κi dL Ni (E0 , z)
=− . (2.14)
dz E0 dz 1 + z
Note that in the following, we choose z = 0 as the current epoch, i.e., we integrate from z to
0. Therefore, from the definition of light-travel distance eq. (2.3), we have
dL LH
=− . (2.15)
dz (1 + z) h (z)
The solution is
κLH z dz 0
 Z  κL
− EH
Di (E0 , z) = exp − = [Z 2 (z)] 0 , (2.16)
E0 0 (1 + z 0 )2 h (z 0 )

–6–
 JCAP10(2012)020
Figure 2. Comparison of the simplified (Z1 ) and proper (Z2 ) redshift dependencies. For z . 0.1 the
approaches are identical. The gray-shaded region illustrates the typical redshift range populated by
GRBs.

where we have isolated the z-dependent part through4

dz 0
Z z
I2 (z) ≡ , (2.17)
0 2 0
0 (1 + z ) h (z )

Z2 (z) ≡ eI2 (z) . (2.18)

The function Z2 (z) is also shown in figure 2. It is now monotonously increasing with
redshift, and larger than unity. This means that for large decay rates κi  E0 /LH , one
finds Di (E0 , z) → 0 (complete decays) for z > 0, as expected in a self-consistent framework.
In addition, since the maximal distance is limited by the Hubble distance, the base Z2
in eq. (2.16) is asymptotically limited. This has interesting implications for the notion of
complete decays, which one can read off from eq. (2.16): The condition for complete decays
is κLH /E0  1, or, equivalently,

4 × 105
E0 [TeV]  . (2.19)
κ−1 [s eV−1 ]

First of all, this confirms our earlier estimates that energies considerably lower than 1 TeV
are needed to test lifetimes comparable to the SN 1987A bounds. Second, it is a condition on
energy, which is independent of distance (or redshift). Therefore, over cosmological distances
(for z & 1), “complete decays” is an energy-dependent concept, and has nothing to do
with distance.
A useful analytical approximation for Z2 is

Z2 (z) ' a + be−cz , (2.20)

with5 
 a ' 1.71
b ' 1 − a = −0.71 . (2.21)
c ' 1.27


4
Note that the function I2 (z) was already encountered in the context of oscillations, and corresponds to
eq. (6) in ref. [35].
5
Actually, a and b are not independent, since Z2 (z = 0) = 1 must be satisfied. This implies that b = 1 − a.

–7–
Hence, the asymptotic value is Z2 → a for z → ∞. It depends on cosmology, as it is
illustrated in figure 2 for different values of Ωm and ΩΛ . Clearly, the higher the cosmological
constant contribution (for a flat universe), the more efficient neutrino decays will be for large
redshifts. In principle, if the neutrino lifetime was known from different sources, one might
even use this effect to independently probe cosmology by comparing low and high redshift
populations of certain objects — although the expected precision is probably not very high.

3 Application to GRBs

We introduce our benchmark GRBs and method in the first subsection. Then we show the

JCAP10(2012)020
impact of decay on the benchmark fluxes and compare the simplified and proper solutions. In
the last subsection, we define a phenomenologically viable model and show the implications
for fluxes and flavor ratios.

3.1 Benchmark GRBs and method

In the following, we apply the above decay framework to GRBs, some of which have been
identified with high redshifts. Note, however, that our results can be equally applied to AGNs
or other objects. We choose four benchmark bursts, for which the parameters are given in
table 1. In these cases, the observed photon spectrum is assumed to be a broken power law
with spectral indices α and β below and above the break at γ,break , respectively. The photon
fluence is Fγ in the energy range 1 keV − 10 MeV. The estimated boost factor is Γ, the time
variability is tv , and the burst duration (where 90% of the counts are observed) is T90 . As
far as the selection of bursts is concerned, the standard burst “SB” is a burst for which most
parameters, such as α, β, z, and Liso γ , are frequently used in the literature. In addition,
three different recent Fermi-measured GRBs have been chosen as examples: GRB 080916C,
GRB 090902B, and GRB 091024. GRB 080916C has been selected, because it is one of the
brightest bursts ever seen, although at a large redshift, and one of the best studied Fermi-
LAT bursts. The gamma-ray spectrum of GRB 090902B has a relatively steep cutoff. GRB
091024 can be regarded as a typical example representative for many Fermi-GBM bursts [63],
except for the long duration. Note that the middle two bursts have an exceptionally large
Γ & 1000, whereas Γ ' 200 for the last burst. All three observed bursts have in common that
the required parameters for the neutrino flux computation can be taken from the literature;
see the figure caption for the references.
Once the parameters in table 1 are given, the neutrino fluxes can be predicted, using
methods such as in refs. [41, 42, 44]. We use the most updated numerical method from ref. [44]
(see ref. [39] for details), which has corrected earlier flux predictions down by about one order
of magnitude by taking into account the full spectral information, the energy dependence of
the mean free path of the protons, and the cooling of secondaries including adiabatic energy
losses. In addition, t-channel and high-energy photo-pion production processes, neutrinos
from muon, pion, kaon, and neutron decays, the helicity dependence of the muon decays, and
flavor mixing are included. As one of the additional assumptions of the method, we assume
that the protons carry ten times more energy than the electrons, and that there is energy
equipartition between electrons and magnetic field energy, i.e., we use the same parameters
as in refs. [40, 42, 47]. With these parameters it is possible to estimate the maximal proton
energy by balancing the acceleration rate of protons (acceleration efficiency 0.1 used) with the
faster of the synchrotron loss and adiabatic loss rates, which are assumed to be the dominant
energy loss rates.

–8–
 SB GRB080916c GRB090902b GRB091024
α 1 0.91 0.61 1.01
β 2 2.08 3.80 2.17
γ,break [MeV] 1.556 0.167 0.613 0.081
Γ 102.5 1090 1000 195
tv [s] 0.0045 0.1 0.053 0.032
T90 [s] 30 66 22 196
z 2 4.35 1.822 1.09
Fγ [erg cm−2 ] 1 · 10−5 1.6 · 10−4 3.3 · 10−4 5.1 · 10−5

JCAP10(2012)020
Liso −1 1052 4.9 · 1053 3.6 · 1053 1.7 · 1051
γ [erg s ]

Table 1. Properties of the four bursts discussed in the following, see ref. [61] for SB (“Standard
Burst”, similar to refs. [43, 62]), refs. [63, 64] for GRB080916c, refs. [63, 65] for GRB090902b and
refs. [63, 66] for GRB091024. The luminosity is calculated with Liso 2
γ = 4πdL · Fγ /T90 with Fγ the
fluence in the energy range 1 keV − 10 MeV. Adopted from ref. [67].

3.2 Impact of decay on neutrino fluxes, and comparison of methods

In order to show the impact of neutrino decay on the neutrino fluxes, let us first of all choose a
(hypothetical) model where all mass eigenstates have the same lifetimes. We show in figure 3
the predicted muon neutrino flux6 as a function of neutrino energy for the four benchmark
bursts in table 1 for this model. The different lifetimes used for the different curves are
shown in the plot legend. Thick curves correspond to the proper solution in section 2.3, and
thin curves to the simplified approach in section 2.2 for comparison. Note that this model
is hypothetical in the sense that the lifetimes are chosen such that the SN 1987A bound for
one of the mass eigenstates must be violated. However, it is the simplest model to illustrate
the implications of neutrino decay.
The actual shapes of the different predictions in the stable case (“no decay”) depend on
the burst parameters. For large enough magnetic fields, the characteristic wiggles from the
secondary (pion, muon, kaon) cooling can be seen. For instance, for “SB” (upper left panel),
the three peaks correspond, from left to right, to muon decays, pion decays, and kaon decays,
respectively. In some cases, the peaks are less pronounced, especially if adiabatic cooling on
protons and secondaries is important (such as for GRB 080916C); cf., figure 13 in ref. [60].
As far as the impact of neutrino lifetime is concerned, note that all bursts in table 1
have redshifts larger than one. As a consequence, we can use eq. (2.19) to estimate where
the decays will be complete. In fact, the energy E0 , shown in the plot, below which the flux
is practically switched off, can be computed with this equation. For instance, in the upper
left panel, the flux is suppressed below 4 000 TeV for κ−1 = 102 s eV−1 . This condition is
the more accurate, the higher the redshift of the burst is. For instance, for GRB 091024
(z ' 1.1), the suppression is somewhat weaker (shifted to smaller energies) because the base
Z2 in figure 2 has not yet reached its saturation value.
Let us now compare the simplified approach in section 2.2 (thin curves) with the proper
solution in section 2.3 (thick curves): Obviously the suppression in the proper treatment
is stronger, which is expected from Z2  Z1 for large redshifts. The difference between
6
Here the neutrino flux is just the neutrino fluence divided by T90 .

–9–
 NeuCosmA 2012 NeuCosmA 2012
SB GRB 080916C
10-5 10-5
No decay
Κ-1 =104 seV
Κ-1 =102 seV
E2 Φ Μ HGeV×s-1 ×cm-2 L 10-6 Κ-1 =1 seV 10-6

E2 Φ Μ HGeV×s-1 ×cm-2 L
10-7 10-7

10-8 10-8

10-9 10-9

JCAP10(2012)020
102 103 104 105 106 107 108 109 1010 102 103 104 105 106 107 108 109 1010
E0 GeV E0 GeV

NeuCosmA 2012 NeuCosmA 2012

GRB 090902B GRB 091024
10-5 10-5

10-6 10-6
E2 Φ Μ HGeV×s-1 ×cm-2 L

E2 Φ Μ HGeV×s-1 ×cm-2 L

10-7 10-7

10-8 10-8

10-9 10-9

102 103 104 105 106 107 108 109 1010 102 103 104 105 106 107 108 109 1010
E0 GeV E0 GeV

Figure 3. Predicted muon neutrino flux as a function of neutrino energy for the four benchmark
bursts in table 1 (panels). The different curves in each panel correspond to no decay (stable neutrinos)
and three different lifetimes, as given in the legend. Thick curves correspond to the proper solution
in section 2.3, and thin curves to the simplified approach in section 2.2 for comparison. Here we
(hypothetically) assume that all mass eigenstates have the same lifetime, as indicated in the legend.

the curves also increases with redshift, and can be quite significant for large redshifts (cf.,
GRB 080916C). Now one may argue that for the individual burst, one would measure a too
short lifetime in the simplified approach, which just needs to be corrected. This argument,
however, breaks down if different bursts at different redshifts are compared. Let us illustrate
this with one example. For the diffuse neutrino flux, it is expected that the peak contribution
comes from a redshift z ' 1 [39] (figure 3), which means that low redshift (z < 1) and high
redshift (z > 1) bursts contribute almost equally. In a diffuse (or a stacked) neutrino flux, the
fluxes from these individual bursts are summed over. For z  1, the simplified and proper
approaches are identical, whereas for z & 1, the properly calculated fluxes are more strongly
suppressed for a certain lifetime. Hence, if the simplified approach is used, there will be a
contradiction implied.
It is, in fact, also interesting to study if z ' 1 as the dominant contribution actually
holds in the presence of decay. We have tested this for different values of E0 . For instance, we
choose E0 ' 106 GeV, which is the peak for the SB spectrum, where the neutrino telescopes
typically have good sensitivity. Eq. (2.19) can then be used to estimate that the flux at this

– 10 –
energy is suppressed for κ−1 . 400 s eV−1 . From the convolution of the GRB rate (following
the star formation rate, possibly with some correction), the 1/d2L drop of E 2 φ, and the decay
damping, one can show that at this energy, the main contribution does not come from z ' 1
anymore, but from very low z (peaking at z = 0) if κ−1 . 400 s eV−1 . This is expected, since
the neutrinos from the low redshift bursts will not have decayed yet. If, on the other hand,
κ−1 & 400 s eV−1 , the bursts from z ' 1 will dominate, as for stable neutrinos.
Finally, comparing the lifetimes used for figure 3 with the SN 1987A bound, one can
easily see that no strong decay suppression can be obtained if this bound is applied to all
mass eigenstates. Therefore, the neutrino lifetimes of the different mass eigenstates have to be
different for a phenomenologically allowed scenario, where the most plausible case, satisfying

JCAP10(2012)020
the SN 1987A constraint, might be that ν1 is stable. In fact, this is for the discussed GRB
fluxes equivalent to saturating the SN 1987A bound on ν1 . Therefore, we only discuss this
case in the following, and we only use the proper solution of the decay equation.

3.3 Fluxes and flavor ratios for ν1 stable

As discussed above, we focus in the following only on the case of ν1 stable, which satisfies
the SN 1987A constraint. Recall that the reason for this constraint is that only the electron
flavor has been measured by the SN 1987A observation, which consists mainly of ν1 . On the
other hand, we assume that ν2 and ν3 decay with the same lifetime, for the sake of simplicity,
with lifetimes satisfying the ν2 (and, consequently, ν3 ) bound. The mass eigenstate ν2 is
about an equal mixture among νe , νµ , and ντ , whereas ν3 is a mixture between νµ and ντ
with a small contribution of νe . The decay of ν2 will affect the electron flavor somewhat,
which however can be absorbed in the uncertainties of current supernova flux predictions.
The decay of ν3 will be hardly observable in the electron flavor.
We show the electron, muon, and tau neutrino fluxes for this model in figure 4 for the
standard burst SB and different values of the lifetime (panels). In the stable case (upper left
panel) at the peak, where the flavor composition νe : νµ : ντ is approximately 1 : 2 : 0 at
the source, flavor equilibration at the detector is roughly achieved. Even for larger energies,
the fluxes of all flavors are roughly comparable on a logarithmic scale — we will discuss the
flavor ratio below. In the most extreme decay case (lower right panel), both ν2 and ν3 decay
with a lifetime corresponding to the current ν2 bound. In this case, the electron neutrino flux
survives (because ν1 is stable), whereas the muon and tau neutrino fluxes are suppressed.
The asymmetry between the muon and tau neutrino fluxes mainly comes from the large value
of θ13 , the chosen δCP , and the non-maximal θ23 . As a consequence:
• Muon tracks are suppressed by about one order of magnitude compared to the stable
case.
• Electromagnetic cascades are observable at about the same level as in the stable case.
• Hadronic cascades are also strongly suppressed. The contributions of the leptonic tau
decay channel into leptons can be neglected as well, since they are doubly suppressed
by tau neutrino flux and the tau branching ratios.
• Neutral current cascades are reduced to about 1/3 of the stable case.
• Glashow resonances should occur at roughly the expected level as in the stable case.
Note, however, that the right energy has to be hit (6.3 PeV), and that Glashow reso-
nant events are expected to be suppressed for neutrino production from photohadronic
interactions, see, e.g., ref. [59] for a detailed (quantitative) discussion.

– 11 –
 Κ-1 = 100 seV
NeuCosmA 2012 NeuCosmA 2012
stable
10-5 10-5

Νe +Νe
Ν Μ +Ν Μ
-6
E2 ΦΝ HGeV×s-1 ×cm-2 L 10 ΝΤ +ΝΤ 10-6

E2 ΦΝ HGeV×s-1 ×cm-2 L
10-7 10-7

10-8 10-8

10-9 10-9

JCAP10(2012)020
102 103 104 105 106 107 108 109 1010 102 103 104 105 106 107 108 109 1010
E0 GeV E0 GeV

Κ-1 = 1 seV Κ-1 = 10-4 seV

NeuCosmA 2012 NeuCosmA 2012

10-5 10-5

10-6 10-6
E2 ΦΝ HGeV×s-1 ×cm-2 L

E2 ΦΝ HGeV×s-1 ×cm-2 L

10-7 10-7

10-8 10-8

10-9 10-9

102 103 104 105 106 107 108 109 1010 102 103 104 105 106 107 108 109 1010
E0 GeV E0 GeV

Figure 4. Predicted muon neutrino flux as a function of neutrino energy for the benchmark burst SB
in table 1. The different panels correspond to different lifetimes, as given in the panels. The different
curves correspond to different flavors (after flavor mixing), as given in the legend. Thin dashed curves
shown the stable case (upper left panel) for reference. As decay model, ν1 is assumed to be stable,
whereas ν2 and ν3 decay with the indicated lifetime, i.e., all of the panels are consistent with the SN
1987A bound.

It is therefore clear that this scenario can easily explain why only cascades have been recently
observed at PeV energies, and will be dominating with increasing statistics, in spite of the
effective areas comparable between muon tracks and cascades [49]. The PeV energies are
exactly at the peak expected from the standard burst (SB), which means that the events
may come from the diffuse GRB flux.
From the lower left panel of figure 4, one can read off that the qualitative conclusions
do not change even if the ν2 and ν3 lifetimes are increased by four orders of magnitude. In
this case, only the rightmost peak (neutrino production from kaon decays) is not affected
by the decays. A very interesting case are intermediate lifetimes, such as κ−1 ∼ 100 s eV−1
in the upper right panel. Here the muon (and tau) neutrino flux dominates above about
1 PeV, whereas below 1 PeV the electron neutrino flux dominates, i.e., there is a spectral
swap induced by neutrino decay.
The implications for the flavor composition can be better seen on a linear scale. We
therefore show in figure 5 the flavor ratio R (muon tracks to cascades) as a function of

– 12 –
 1.0 1.0
NeuCosmA 2012 NeuCosmA 2012
SB GRB080916C
No decay
Κ-1 =104 seV
0.8 Κ-1 =102 seV 0.8
Κ-1 =1 seV
muon damped muon damped
0.6 0.6
Φe + ΦΤ

Φe + ΦΤ
ΦΜ

ΦΜ
pion beam pion beam
R=

R=
0.4 0.4

JCAP10(2012)020
0.2 0.2
all decayed Hexcept Ν1 L all decayed Hexcept Ν1 L

0.0 0.0
103 104 105 106 107 108 109 1010 103 104 105 106 107 108 109 1010
E0 GeV E0 GeV

Figure 5. Predicted flavor ratio R (muon tracks to cascades) as a function of neutrino energy for
the first two benchmark bursts in table 1 (panels). The different curves in each panel correspond to
no decay (stable neutrinos) and three different lifetimes, as given in the legend. As decay model, ν1
is assumed to be stable, whereas ν2 and ν3 decay with the indicated lifetime, i.e., all of the lifetimes
are consistent with the SN 1987A bound.

neutrino energy for two different bursts (panels) and different lifetimes (curves). One can
easily see the characteristic transition from a pion beam source (initial flavor composition
νe :νµ :ντ of 1 : 2 : 0, yielding R ≈ 0.52) to a muon damped source (initial flavor composition
of 0 : 1 : 0, yielding R ≈ 0.67) at higher energies, which comes from the energy losses of
the muons in the magnetic field. It depends, apart from particle physics parameters, on the
value of B 0 , the magnitude of the magnetic field in the shock rest frame; see ref. [60] for a
review. The different panels show two different cases for this transition.
It is easy to show that if only one mass eigenstate is stable, an asymptotic value of the
ratio R is reached for complete decays which does not depend on the initial flavor composition
(see, e.g., refs. [26, 68]). This asymptotic value R ' 0.14 is marked in figure 5, and it is always
reached for low enough energies in the presence of decay. The low value simply reflects that
muon tracks are strongly suppressed compared to the cascades. The transition towards this
asymptotic curve depends on the lifetime, of course. Depending on the chosen lifetime and
magnetic field in the source, the pion beam flavor ratio may be actually observable in a
certain energy range, such as for κ−1 = 104 s eV−1 in the left panel, or not, such as for
κ−1 = 102 s eV−1 in the left panel. The presence of different flavor ratios may be useful to
distinguish different decay scenarios; see refs. [26, 33].

4 Consequences for realistic GRB stacking analyses

We now apply the scenario in section 3.3 to a realistic sample of GRBs, as it has been actually
used in a stacking analysis. Remember that from the SN 1987A constraint, we have been
forced to choose ν1 stable for all practical purposes, whereas we have assumed that ν2 and
ν3 decay with equal rates.
We show in figure 6 the prediction for the quasi-diffuse muon (left panel) and electron
(right panel) neutrino fluxes from the 117 bursts in the IC-40 stacking sample as curves

– 13 –
 JCAP10(2012)020
Figure 6. Prediction of the quasi-diffuse muon (left panel) and electron (right panel) neutrino flux
from the bursts used in the IceCube stacking analysis. The different curves in each panel correspond
to no decay (stable neutrinos) and three different lifetimes, as given in the legend, which are applied
to each burst in the stacking sample individually. As decay model, ν1 is assumed to be stable, whereas
ν2 and ν3 decay with the indicated lifetime, i.e., all of the lifetimes are consistent with their current
bounds. The limit from IC-40+59 is shown [40], as well as an extrapolated one for IC-86 (ten years,
AIC86
eff ' 3 × AIC40
eff ; see, e.g., ref. [69]). For the prediction, the 117 bursts in the IC-40 sample have
been used [47], the “no decay” corresponds to the nominal prediction in figure 3 of ref. [44].

labeled “no decay”. The left curve corresponds to the nominal prediction in figure 3 of
ref. [44], updated with the current values of the mixing angles. It uses exactly the same bursts
and parameters as the original IceCube analysis, but the numerical (updated) method for the
flux prediction from ref. [44]. Note that the quasi-diffuse flux normalization hardly depends
on the burst sample used, but only the shape (cf., dashed curves in figure 1 of ref. [40]).
This means that the shape will be affected by the actual bursts in the sample, whereas the
normalization will remain approximately constant — modulo statistical fluctuations from
the quasi-diffuse flux extrapolation; see refs. [39, 70]. We also show in figure 6, left panel,
the current stacking limit from muon tracks, as well as an extrapolation for the full-scale
experiment. While the current limit does not yet exceed our nominal prediction, the full
scale experiment will finally exceed it for the chosen standard values of the astrophysical
parameters. Note that in the original IceCube analysis, z = 2 has been assigned to a burst if
the redshift has not been measured, while the pion production efficiency has been computed
(for long bursts) with Liso 52 −1
γ = 10 erg s . Since we stick to these rules, most bursts will have
a redshift z > 1, and cosmological effects will be important for decay.
In figure 6, we also show the flux predictions for different neutrino lifetimes. In the
extreme cases, κ−1 . 1 s eV−1 , the muon neutrino predicted flux is suppressed to below the
expected limit from IC-86 over ten years. However, in these cases, a substantial electron
neutrino flux is expected; see right panel. It can be read off from the figure, that gamma-ray
burst neutrinos should be detected with the full scale experiment in cascades if the effective
area for cascades is at least 25% of the one for muon tracks at 1 PeV.7 On the other hand, no
7
Unfortunately, there are no GRB cascade analyses available up to this point, which means that reliable
limits in the right panel of figure 6 cannot be shown.

– 14 –
reliable information on astrophysical neutrino sources, such as GRBs, can be obtained from
muon tracks only. In particular, it is not clear if no neutrinos are found because they decay,
or because the baryonic loading in GRBs is smaller than anticipated.
A very interesting case is the intermediate value κ−1 ' 100 s eV−1 and its implications
for muon tracks; see left panel. In this case, only the low energy component of the flux is
damped, and only neutrinos above a certain threshold energy are observable. It is expected
that the full scale experiment may detect neutrinos in such a case, but the interplay with the
(energy-dependent) detector response becomes important (see ref. [71] for a more detailed
discussion). For instance, better information may be obtained from down-going than up-
going muon tracks, because these exhibit a better sensitivity at higher energies (cf., figure 3

JCAP10(2012)020
in ref. [71]). As a consequence, the sensitivity may be dominated by a different viewing
window and therefore different bursts compared to the “no decay” case, which may have
different characteristics. One has to keep that in mind if one compares the sensitivities for
these cases.

5 Summary and discussion

We have discussed neutrino decays of ultra-relativistic neutrinos over cosmological distances,

i.e., for redshifts z & 1. These neutrinos may originate from extragalactic cosmic acceler-
ators, such as AGNs or GRBs. As class of decay models, we have considered decays into
secondaries invisible to the detector, such as Majorons, unparticles, sterile neutrinos, or neu-
trinos strongly degraded in energy. The decays have been allowed to be incomplete, i.e., the
spectral signature of the decay may be seen.
We have demonstrated that simplified approaches to neutrino decay cannot be used
over such distances, and that the differential decay equation has to be carefully solved as a
function of redshift. We have also given a simple parameterization for the solution, which
can be easily used. From a comparison of different approaches, we have demonstrated that
conventional approaches to neutrino decay can be safely used for z . 0.1 (or distances
L . 360 Mpc), whereas the light-travel (or look-back) distance triggers the clock for neutrino
decays over cosmological distances. As a consequence, the horizon is limited by the Hubble
length ' 3.89 Gpc, which is the maximal distance scale testable by decay. This has several
interesting consequences: First of all, the common notion “the longer the distance, the longer
lifetimes can be tested” does not hold anymore. Second, the concept of complete decays (all
neutrinos have decayed) cannot be assigned to long enough distances, but only to low enough
energies, i.e., it is an energy-dependent and not distance-dependent statement. And third,
the neutrino lifetime bound from SN 1987A, which emitted neutrinos at much lower energies,
cannot be exceeded by high-energy astrophysical neutrinos from AGNs and GRBs tested in
neutrino telescopes. The same argument applies to the cosmogenic neutrinos from cosmic
microwave (and infrared) background interactions, which are at even higher energies. Note
that the SN 1987A neutrinos have only been measured in the electron flavor, which means
that this limit does not apply to all mass eigenstates. However, it is expected that the
neutrinos from any future galactic supernova explosion will be measured in different flavors
and neutral currents, which means that in long terms the best lifetime bounds are expected
to come from supernova neutrinos.
While our results can be easily applied to AGNs as well, we have used GRBs as test case.
Because of the SN 1987A constraint, we have been forced to choose ν1 stable and ν2 and ν3

– 15 –
decaying with equal rates.8 We have shown that, safely within the current bounds for ν2 and
ν3 , the muon neutrino flux can be substantially reduced by about one order of magnitude,
whereas the electron neutrino flux is hardly affected. This has very interesting implications
for neutrino telescopes: First, improved bounds on the neutrino flux from muon tracks can
be interpreted in terms of the astrophysical parameters or the possibility that neutrinos are
unstable. On the other hand, the electromagnetic cascade bounds are hardly affected by
the neutrino lifetime, which means that reliable conclusions on the astrophysical objects
can only come from cascades, and highlights the importance of dedicated cascade analyses
for GRBs and other classes of objects. This would be different, of course, if muon tracks
from astrophysical neutrinos were actually observed, since their discovery would constrain

JCAP10(2012)020
this possibility.
For GRBs, in particular, we have used a realistic sample of IceCube GRBs and the
state-of-the-art technology for the prediction of the quasi-diffuse neutrino flux to illustrate
the consequences. We have demonstrated that, for the discussed decay scenarios and our
nominal flux prediction using the same parameters as IceCube, even the full scale IceCube
experiment may not find neutrinos from GRBs in muon tracks after ten years, whereas the
perspectives for a detection in cascades are actually very good.
Finally, note that recently two cascade neutrino event candidates at PeV energies have
been found. Our scenario provides a plausible explanation why only cascades are seen,
whatever the origin of the neutrinos is, whereas there are no accompanying muon tracks in
spite of comparable effective areas. For GRBs, in particular, the PeV energies are exactly
where we predict the maximum of the quasi-diffuse flux; see figure 6, right panel. On the
other hand, for the relevant search time window, we only expect 0.07 electromagnetic cascade
events for the nominal (“no decay”) prediction from GRBs. The observed two events might in
this case come from a strong statistical fluctuation, or a significant deviation of astrophysical
parameters from their assumed mean values.

Acknowledgments

We would like to thank Markus Ahlers, Amol Dighe, Francis Halzen, Svenja Hümmer,
Poonam Mehta, Irina Mocioiu, Sandip Pakvasa, Eli Waxman, and Nathan Whitehorn for
useful discussions on aspects regarding this work.
We would like to acknowledge support from DFG grants WI 2639/3-1 and WI 2639/4-1.
MB and PB would like to acknowledge support from the GRK 1147 “Theoretical Astrophysics
and Particle Physics”. WW would like to thank GGI Florence for hospitality during their
stay within the “What’s ν?” program. This work has also been supported by the FP7
Invisibles network (Marie Curie Actions, PITN-GA-2011-289442) and the “Helmholtz Al-
liance for Astroparticle Physics HAP”, funded by the Initiative and Networking fund of the
Helmholtz association.

References

[1] KAMIOKANDE-II collaboration, K. Hirata et al., Observation of a Neutrino Burst from the
Supernova SN 1987a, Phys. Rev. Lett. 58 (1987) 1490 [INSPIRE].
8
We have demonstrated that saturating the SN 1987A bound for the ν1 lifetime does not change the results,
which means that it is equivalent to the studied model for the purpose considered.

– 16 –
 [2] R. Bionta et al., Observation of a Neutrino Burst in Coincidence with Supernova SN 1987a in
the Large Magellanic Cloud, Phys. Rev. Lett. 58 (1987) 1494 [INSPIRE].
[3] A.S. Joshipura, E. Masso and S. Mohanty, Constraints on decay plus oscillation solutions of the
solar neutrino problem, Phys. Rev. D 66 (2002) 113008 [hep-ph/0203181] [INSPIRE].
[4] A. Bandyopadhyay, S. Choubey and S. Goswami, Neutrino decay confronts the SNO data,
Phys. Lett. B 555 (2003) 33 [hep-ph/0204173] [INSPIRE].
[5] J.F. Beacom and N.F. Bell, Do solar neutrinos decay?, Phys. Rev. D 65 (2002) 113009
[hep-ph/0204111] [INSPIRE].
[6] KamLAND collaboration, K. Eguchi et al., A High sensitivity search for ν̄e ’s from the sun and
other sources at KamLAND, Phys. Rev. Lett. 92 (2004) 071301 [hep-ex/0310047] [INSPIRE].

JCAP10(2012)020
[7] SNO collaboration, B. Aharmim et al., Electron antineutrino search at the Sudbury Neutrino
Observatory, Phys. Rev. D 70 (2004) 093014 [hep-ex/0407029] [INSPIRE].
[8] M. Gonzalez-Garcia and M. Maltoni, Status of Oscillation plus Decay of Atmospheric and
Long-Baseline Neutrinos, Phys. Lett. B 663 (2008) 405 [arXiv:0802.3699] [INSPIRE].
[9] S. Pakvasa, Neutrino decays and neutrino telescopes, hep-ph/0305317 [INSPIRE].
[10] Particle Data Group collaboration, W. Yao et al., Review of Particle Physics, J. Phys. G
33 (2006) 1 [INSPIRE].
[11] S. Pakvasa, Neutrino Flavor Goniometry by High Energy Astrophysical Beams, Mod. Phys.
Lett. A 23 (2008) 1313 [arXiv:0803.1701] [INSPIRE].
[12] G. Raffelt, Neutrino radiative lifetime limits from the absence of solar gamma rays, Phys. Rev.
D 31 (1985) 3002 [INSPIRE].
[13] G. Gelmini and M. Roncadelli, Left-Handed Neutrino Mass Scale and Spontaneously Broken
Lepton Number, Phys. Lett. B 99 (1981) 411 [INSPIRE].
[14] Y. Chikashige, R.N. Mohapatra and R. Peccei, Spontaneously Broken Lepton Number and
Cosmological Constraints on the Neutrino Mass Spectrum, Phys. Rev. Lett. 45 (1980) 1926
[INSPIRE].
[15] R. Tomas, H. Pas and J. Valle, Generalized bounds on Majoro-neutrino couplings, Phys. Rev. D
64 (2001) 095005 [hep-ph/0103017] [INSPIRE].
[16] S.-L. Chen, X.-G. He and H.-C. Tsai, Constraints on unparticle interactions from invisible
decays of Z, quarkonia and neutrinos, JHEP 11 (2007) 010 [arXiv:0707.0187] [INSPIRE].
[17] S. Zhou, Neutrino Decays and Neutrino Electron Elastic Scattering in Unparticle Physics,
Phys. Lett. B 659 (2008) 336 [arXiv:0706.0302] [INSPIRE].
[18] X.-Q. Li, Y. Liu and Z.-T. Wei, Neutrino decay as a possible interpretation to the MiniBooNE
observation with unparticle scenario, Eur. Phys. J. C 56 (2008) 97 [arXiv:0707.2285]
[INSPIRE].
[19] D. Majumdar, Unparticle decay of neutrinos and it’s effect on ultra high energy neutrinos,
arXiv:0708.3485 [INSPIRE].
[20] ANTARES collaboration, E. Aslanides et al., A deep sea telescope for high-energy neutrinos,
astro-ph/9907432 [INSPIRE].
[21] IceCube collaboration, J. Ahrens et al., IceCube – the next generation neutrino telescope at
the south pole, Nucl. Phys. Proc. Suppl. 118 (2003) 388 [astro-ph/0209556] [INSPIRE].
[22] NESTOR collaboration, S. Tzamarias, NESTOR: A deep-sea neutrino telescope, Nucl.
Instrum. Meth. A 502 (2003) 150 [INSPIRE].
[23] NEMO collaboration, P. Piattelli, The Neutrino Mediterranean Observatory project, Nucl.
Phys. Proc. Suppl. 143 (2005) 359 [INSPIRE].

– 17 –
[24] Y. Farzan and A.Y. Smirnov, Coherence and oscillations of cosmic neutrinos, Nucl. Phys. B
805 (2008) 356 [arXiv:0803.0495] [INSPIRE].
[25] J.F. Beacom, N.F. Bell, D. Hooper, S. Pakvasa and T.J. Weiler, Decay of high-energy
astrophysical neutrinos, Phys. Rev. Lett. 90 (2003) 181301 [hep-ph/0211305] [INSPIRE].
[26] M. Maltoni and W. Winter, Testing neutrino oscillations plus decay with neutrino telescopes,
JHEP 07 (2008) 064 [arXiv:0803.2050] [INSPIRE].
[27] A. Bhattacharya, S. Choubey, R. Gandhi and A. Watanabe, Diffuse Ultra-High Energy
Neutrino Fluxes and Physics Beyond the Standard Model, Phys. Lett. B 690 (2010) 42
[arXiv:0910.4396] [INSPIRE].
[28] A. Bhattacharya, S. Choubey, R. Gandhi and A. Watanabe, Ultra-high neutrino fluxes as a

JCAP10(2012)020
probe for non-standard physics, JCAP 09 (2010) 009 [arXiv:1006.3082] [INSPIRE].
[29] M. Lindner, T. Ohlsson and W. Winter, A combined treatment of neutrino decay and neutrino
oscillations, Nucl. Phys. B 607 (2001) 326 [hep-ph/0103170] [INSPIRE].
[30] M. Lindner, T. Ohlsson and W. Winter, Decays of supernova neutrinos, Nucl. Phys. B 622
(2002) 429 [astro-ph/0105309] [INSPIRE].
[31] P. Lipari, M. Lusignoli and D. Meloni, Flavor Composition and Energy Spectrum of
Astrophysical Neutrinos, Phys. Rev. D 75 (2007) 123005 [arXiv:0704.0718] [INSPIRE].
[32] M. Bustamante, A. Gago and C. Pena-Garay, Energy-independent new physics in the flavour
ratios of high-energy astrophysical neutrinos, JHEP 04 (2010) 066 [arXiv:1001.4878]
[INSPIRE].
[33] P. Mehta and W. Winter, Interplay of energy dependent astrophysical neutrino flavor ratios
and new physics effects, JCAP 03 (2011) 041 [arXiv:1101.2673] [INSPIRE].
[34] M.D. Kistler, H. Yuksel, J.F. Beacom, A.M. Hopkins and J.S.B. Wyithe, The Star Formation
Rate in the Reionization Era as Indicated by Gamma-ray Bursts, Astrophys. J. 705 (2009)
L104 [arXiv:0906.0590] [INSPIRE].
[35] D. Wagner and T. Weiler, Neutrino Oscillations form Cosmic Sources: a Nu Window to
Cosmology, Mod. Phys. Lett. A 12 (1997) 2497 [arXiv:1101.5677] [INSPIRE].
[36] T.J. Weiler, W. Simmons, S. Pakvasa and J. Learned, Gamma-ray bursters, neutrinos and
cosmology, hep-ph/9411432 [INSPIRE].
[37] J.F. Beacom et al., PseudoDirac neutrinos: A Challenge for neutrino telescopes, Phys. Rev.
Lett. 92 (2004) 011101 [hep-ph/0307151] [INSPIRE].
[38] A. Esmaili and Y. Farzan, Implications of the Pseudo-Dirac Scenario for Ultra High Energy
Neutrinos from GRBs, arXiv:1208.6012 [INSPIRE].
[39] P. Baerwald, S. Hummer and W. Winter, Systematics in the Interpretation of Aggregated
Neutrino Flux Limits and Flavor Ratios from Gamma-Ray Bursts, Astropart. Phys. 35 (2012)
508 [arXiv:1107.5583] [INSPIRE].
[40] IceCube collaboration, R. Abbasi et al., An absence of neutrinos associated with cosmic-ray
acceleration in γ-ray bursts, Nature 484 (2012) 351 [arXiv:1204.4219] [INSPIRE].
[41] D. Guetta, D. Hooper, J. Alvarez-Muniz, F. Halzen and E. Reuveni, Neutrinos from individual
gamma-ray bursts in the BATSE catalog, Astropart. Phys. 20 (2004) 429 [astro-ph/0302524]
[INSPIRE].
[42] IceCube collaboration, R. Abbasi et al., Search for muon neutrinos from Gamma-Ray Bursts
with the IceCube neutrino telescope, Astrophys. J. 710 (2010) 346 [arXiv:0907.2227]
[INSPIRE].
[43] E. Waxman and J.N. Bahcall, High-energy neutrinos from cosmological gamma-ray burst
fireballs, Phys. Rev. Lett. 78 (1997) 2292 [astro-ph/9701231] [INSPIRE].

– 18 –
[44] S. Hummer, P. Baerwald and W. Winter, Neutrino Emission from Gamma-Ray Burst
Fireballs, Revised, Phys. Rev. Lett. 108 (2012) 231101 [arXiv:1112.1076] [INSPIRE].
[45] J. Barranco, O.G. Miranda, C.A. Moura and A. Parada, A reduction in the UHE neutrino flux
due to neutrino spin precession, arXiv:1205.4285 [INSPIRE].
[46] Z. Li, Note on the Normalization of Predicted GRB Neutrino Flux, Phys. Rev. D 85 (2012)
027301 [arXiv:1112.2240] [INSPIRE].
[47] IceCube collaboration, R. Abbasi et al., Limits on Neutrino Emission from Gamma-Ray
Bursts with the 40 String IceCube Detector, Phys. Rev. Lett. 106 (2011) 141101
[arXiv:1101.1448] [INSPIRE].
[48] H.-N. He et al., IceCube non-detection of GRBs: Constraints on the fireball properties,

JCAP10(2012)020
Astrophys. J. 752 (2012) 29 [arXiv:1204.0857] [INSPIRE].
[49] A. Ishihara, Icecube: Ultra-high energy neutrinos, talk at Neutrino 2012, Kyoto Japan, June
3–9 2012
[50] V. Barger, J. Learned and S. Pakvasa, IceCube Neutrino Initiated Cascade Events: PeV
Electron-antineutrinos at Glashow Resonance, arXiv:1207.4571 [INSPIRE].
[51] D. Fargion, D. D’Armiento and P. Paggi, UHECR bending, clustering and decaying feeding
gamma anisotropy, arXiv:1208.2471 [INSPIRE].
[52] D.W. Hogg, Distance measures in cosmology, astro-ph/9905116 [INSPIRE].
[53] Particle Data Group collaboration, K. Nakamura et al., Review of Particle Physics, J.
Phys. G 37 (2010) 075021 [INSPIRE].
[54] WMAP collaboration, E. Komatsu et al., Seven-Year Wilkinson Microwave Anisotropy Probe
(WMAP) Observations: Cosmological Interpretation, Astrophys. J. Suppl. 192 (2011) 18
[arXiv:1001.4538] [INSPIRE].
[55] M. Blennow, T. Ohlsson and W. Winter, Damping signatures in future neutrino oscillation
experiments, JHEP 06 (2005) 049 [hep-ph/0502147] [INSPIRE].
[56] G. Fogli et al., Global analysis of neutrino masses, mixings and phases: entering the era of
leptonic CP-violation searches, Phys. Rev. D 86 (2012) 013012 [arXiv:1205.5254] [INSPIRE].
[57] P. Serpico and M. Kachelriess, Measuring the 13-mixing angle and the CP phase with neutrino
telescopes, Phys. Rev. Lett. 94 (2005) 211102 [hep-ph/0502088] [INSPIRE].
[58] M. Kachelriess, S. Ostapchenko and R. Tomas, High energy neutrino yields from astrophysical
sources. 2. Magnetized sources, Phys. Rev. D 77 (2008) 023007 [arXiv:0708.3047] [INSPIRE].
[59] S. Hummer, M. Maltoni, W. Winter and C. Yaguna, Energy dependent neutrino flavor ratios
from cosmic accelerators on the Hillas plot, Astropart. Phys. 34 (2010) 205 [arXiv:1007.0006]
[INSPIRE].
[60] W. Winter, Neutrinos from Cosmic Accelerators Including Magnetic Field and Flavor Effects,
Adv. High Energy Phys. 2012 (2012) 586413 [arXiv:1201.5462] [INSPIRE].
[61] P. Baerwald, S. Hummer and W. Winter, Magnetic Field and Flavor Effects on the
Gamma-Ray Burst Neutrino Flux, Phys. Rev. D 83 (2011) 067303 [arXiv:1009.4010]
[INSPIRE].
[62] E. Waxman and J.N. Bahcall, High-energy neutrinos from astrophysical sources: An upper
bound, Phys. Rev. D 59 (1999) 023002 [hep-ph/9807282] [INSPIRE].
[63] L. Nava, G. Ghirlanda, G. Ghisellini and A. Celotti, Spectral properties of 438 GRBs detected
by Fermi/GBM, Astron. Astrophys. 530 (2011) A21 [arXiv:1012.2863] [INSPIRE].
[64] J. Greiner et al., The redshift and afterglow of the extremely energetic gamma-ray burst GRB
080916C, Astron. Astrophys. 498 (2009) 89 [arXiv:0902.0761] [INSPIRE].

– 19 –
[65] Fermi/GBM collaboration, A. Abdo et al., Fermi Observations of GRB 090902B: A Distinct
Spectral Component in the Prompt and Delayed Emission, Astrophys. J. 706 (2009) L138
[arXiv:0909.2470] [INSPIRE].
[66] D. Gruber et al., Fermi/GBM observations of the ultra-long GRB 091024: A burst with an
optical flash, Astron. Astrophys. 528 (2011) A15 [arXiv:1101.1099] [INSPIRE].
[67] S. Hümmer, Neutrinos aus photohadronischen Wechselwirkungen in kosmischen
Beschleunigern, in preparation, Ph.D. thesis, Würzburg University, (2012).
[68] S. Pakvasa, Charged lepton oscillations, Lett. Nuovo Cim. 31 (1981) 497 [INSPIRE].
[69] IceCube collaboration, A. Karle, IceCube, arXiv:1003.5715 [INSPIRE].
[70] F. Halzen and D. Hooper, Neutrino event rates from gamma-ray bursts, Astrophys. J. 527

JCAP10(2012)020
(1999) L93 [astro-ph/9908138] [INSPIRE].
[71] W. Winter, Interpretation of neutrino flux limits from neutrino telescopes on the Hillas plot,
Phys. Rev. D 85 (2012) 023013 [arXiv:1103.4266] [INSPIRE].

– 20 –