Testing the origin of the UHECRs
with neutrinos
Walter Winter
DESY, Zeuthen, Germany
Kavli Institute for Theoretical Physics
(KITP), Santa Barbara, CA, USA
UHECR 2014,Springdale, UT, USA
Oct. 12-15, 2014
Contents
> Introduction
> Can the observed neutrinos come from the same sources as the
UHECRs?
> GRBs as test case for the UHECR-neutrino connection
> Summary
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 2
Cosmic messengers
Physics of astrophysical
neutrino sources = physics of
cosmic ray sources
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 3
3
2014: 37 neutrinos in the TeV-PeV range
Where do these come from?
Prompt atmospherics?
Directional information: Clustering?
Isotropic/from Galactic plane/Galactic center?
Why no events > few PeV?
Can these come from the sources of the ultra-high
energy cosmic rays?
Which source class? More than one?
Flavor composition?
 Requires more statistics
Science 342 (2013) 1242856; update by Gary Hill @ Neutrino 2014
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 4
Connection with primary nuclei?
> In pp and pg interactions, the secondary pions take about 20% of the
proton energy, the neutrinos about 5% (per flavor)
> PeV neutrinos must come from 20-500 PeV nuclei (depending on comp.)
> Observed cosmic ray composition
non-trivial function of energy (at Earth!)
> Simple
example:
Neutrinos from
cosmic ray
interactions
with hydrogen
in the
Milky Way
[O(0.1-1) event]
Joshi, Winter, Gupta,
MNRAS, 2014
n
primaries UHECRs
Gaisser, Stanev, Tilav, 2013
> Connection with UHECR sources requires extrapolation over
several orders of magnitude both in spectrum and composition
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 5
Fitting the observed neutrino spectrum
> Simplest possible
model: Ap (or AA)
interactions in sources;
SFR evolution
> Possible fits to data:
Protons, a=2.5
[Problem:
Fermi diffuse
g-ray bound
Murase, Ahlers,
Lacki, PRD 2013]
Protons
a=2
B ~ 104 G
(magnetic field effects on
sec. pions, muons, kaons)
Nuclei
a=2, Emax=1010.1 GeV
Composition at source
Protons
a=2
Emax=107.5 GeV
with b=0.4
WW, arXiv:1407.7536
(PRD, accepted)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 6
Connection to UHECRs?
Yes, but: Energy input per decade very different
in neutrino-relevant and UHECR energy ranges
(Energetics seem to favor a~2, see e.g.
B. Katz, E. Waxman, T. Thompson, and A. Loeb (2013),
1311.0287)  will come up again later!
Protons, a=2.5
[Problem:
Fermi diffuse
g-ray bound
Protons
a=2
B ~ 104 G
Murase, Ahlers,
Lacki, PRD 2013]
Nuclei
a=2, Emax=1010.1 GeV
Composition at source
Protons
a=2
Emax=107.5 GeV
Yes, but: Need energydependent escape timescale
leading to break/cutoff within
source (diff. from ejection!)
see e.g. Liu et al, PRD, 2004;
arXiv:1310.1263
Yes, but: Synchrotron losses
limit maximal proton energies
as well. Need large Doppler
factors (e. g. GRBs)
with b=0.4
WW, arXiv:1407.7536
(PRD, accepted)
Yes, but: A(E) change somewhat too
shallow to match observation;
difference source-observation from
propagation?
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 7
GRBs as a test case
> Idea: Use timing and directional information to suppress atm. BGs
Coincidence!
Neutrino
observations
(e.g. IceCube, …)
(Source: IceCube)
(Source: NASA)
GRB gamma-ray observations
(e.g. Fermi, Swift, etc)
> Stacking limit exceeds
observed neutrino flux
(~10-8) by one order of
magnitude; interesting to
test specific models
Nature 484 (2012) 351
> Prediction (One zone model.
based on fixed collision radius
models) almost reached
(some recent corrections!)
(Hümmer,
Baerwald, Winter,
PRL 108 (2012)
231101;
method based on
Guetta et al, 2004;
Waxman, Bahcall
1997)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 8
GRB - Internal shock model
(Source: SWIFT)
Engine
(intermittent)
G ~ 200-1000
“Isotropic
equivalent
energy“
Observable:
Light curves
(Simulation by
M. Bustamante)
Prompt phase
Collision of
shells
 Shocks
 Particle acc.
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 9
UHECR-neutrino connection: escape mechanisms?
Baerwald, Bustamante, Winter, Astrophys. J. 768 (2013) 186
Optically thin
(to neutron escape)
n
Optically thick
(to neutron escape)
n
n
n
p
n
n
n
n
 One neutrino per
cosmic ray
 Protons magnetically
confined
p
n
n
n
p
n
p
n
p
n
p
n
n
p
Direct proton escape
(UHECR leakage)
n
p
p
l‘ ~ c t‘pg
 Neutron escape limited
to edge of shells
 Neutrino prod.
relatively enhanced
l‘ ~ R‘L
 pg interaction rate
relatively low
 Protons leaking from
edges dominate
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 10
An example (before propagation)
(only adiabatic energy losses)
 For high enough acceleration
efficiencies:
R‘L can reach shell thickness
at highest energies
(if E‘p,max determined by t‘dyn)
 Hard spectrum, aka “high pass
filter“ (Globus et al, 2014)
 Relative importance depends
on optical thickness to pg
interactions
Neutron spectrum
harder than E-2
proton spectrum
(from: Baerwald, Bustamante, Winter,
Astrophys. J. 768 (2013) 186)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 11
Combined source-propagation model: Ankle transition
(ap=2, fit range 1010 ... 1012 GeV)
> Neutron-dominated cases can be constrained by neutrino emission
> Baryonic loading fe-1 (energy protons to photons) typically somewhat larger
than IceCube assume, to fit UHECR data (here Liso=1052 erg s-1, Eiso=3 1052 erg)
G=300
G=800
(Baerwald, Bustamante, Winter, Astropart. Phys. 62 (2015) 66; figures with TA data)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 12
Combined source-propagation model: Dip transition
(ap=2.5 with SFR evolution, fit range 109 ... 1012 GeV)
> Neutron-dominated cases even more extreme
> Required baryonic loading fe-1 extremely large; implication of unequal
energy output per decade (bolometric correction)
G=300
G=600 1050.5 erg/s
(Baerwald, Bustamante, Winter, Astropart. Phys. 62 (2015) 66; figures with TA data)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 13
Parameter space constraints (ankle model, fit to TA data)
Example:
Moderate acc.
efficiency, escape
by Bohm-like
diffusion, SFR
evolution of
sources,
ankle transition
log10 fe-1
(baryonic loading)
obtained from fit
Direct
escape
Optically
thick pg
IceCube
expectation
(15yr)
Current
IceCube
limit
Best-fit (shaded
contours: TA
UHECR fit)
(Baerwald, Bustamante, Winter, Astropart. Phys. 62 (2015) 66; figures with TA data)
… but - maybe - assigning one parameter set
to all shells is too simple?
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 14
The future: more dynamical collision models
> Set out a number of shells with a
Lorentz factor distribution
> Shells collide, merge and cool by
radiation of energy
> Light curve predictable (see below)
> Efficient energy dissipation (e. g.
into gamma-rays) requires broad
Lorentz factor distribution
(Bustamante, Baerwald, Murase, Winter, 2014;
based on collision model Kobayashi, Piran, Sari,
1997; see Globus et al, 2014 for a similar approach)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 15
Consequences for different messengers
> Collision radii reach from below
photosphere to circumburst medium
> UHECR escape as neutrons (red) and
directly (blue) at intermediate radii
> Energy output ~ no of collsions x energy
per collision (counting important!)
> The burst looks different in different
messengers!
(Bustamante, Baerwald, Murase, Winter, 2014)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 16
Consequences for neutrino production
> Neutrino flux comes from a
few collisions at photosphere
Eiso=1053 erg per GRB
> Photospheric radius and
photohadronic interactions
both depend on particle
densities (scale at same way)
> Super-photospheric
(minimal?) prediction hardly
depends on baryonic loading,
G (different from earlier works!)
> Testable in high-energy
extension of IceCube?
> Sub-photospheric
contribution could be much
larger. However: photons
from below photosphere not
observable
(Bustamante, Baerwald, Murase, Winter, 2014)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 17
Summary
> Neutrino observations open new window to cosmic ray source
identification; data (discovery and constraints) become meaningful
> UHECR connection somewhat more challenging, as several orders of
magnitude in energy between UHECRs and primaries leading to
observed neutrino flux
> GRBs are an interesting test case, as
 The constraints are strongest on GRBs because of timing cuts
 Well-motivated models for gamma-ray emission exist
 IceCube data already test the parameter space
> Different messengers are produced in different regions of a GRB.
Multi-messenger connections are more model-dependent than
previously anticipated
> Heavy nuclei are anticipated to escape from larger radii than protons,
as disintegration is to be avoided – but they can survive
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 18
BACKUP
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 19
Neutrino production
Dashed arrows: kinetic equations include cooling and escape
B‘
Input  Object-dependent:
Q(E) [GeV-1 cm-3 s-1]
per time frame
N(E) [GeV-1 cm-3]
steady spectrum
Optically
thin
to neutrons
from:
Baerwald, Hümmer, Winter,
Astropart. Phys. 35
(2012) 508
Walter Winter | UHECR 2014 |
Oct. 13-15, 2014 | Page 20
Kinetic equations (steady state, one zone)
> Energy losses in continuous limit:
Injection
Energy losses
Escape
b(E)=-E t-1loss
Q(E,t) [GeV-1 cm-3 s-1] injection per time frame (from sep. acc. zone)
N(E,t) [GeV-1 cm-3] particle spectrum including spectral effects
NB: Need N(E) to compute particle interactions
> Simple case: No energy losses b=0:
> Special cases:
 tesc ~ R/c (leaky box)
 tesc ~ E-a . Consequence: N(E) ~ Qinj(E) E-a, Escape: Qesc(E) = N(E)/tesc~ Qinj
(Neutrino spectrum from N(E) can have a break which is not present in
escaping primaries Qesc(E))
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 21
Peculiarity for neutrinos: Secondary cooling
> Secondary spectra (m, p, K) losssteepend above critical energy
Example: GRB
Decay/cooling: charged m, p, K
nm
Pile-up effect
 Flavor ratio!
E‘c depends on particle physics
only (m, t0), and B‘
E‘c
E‘c
Leads to characteristic flavor
composition and shape
E‘c
Spectral
split
Adiabatic
Decouples maximal neutrino and
proton energies
Baerwald, Hümmer, Winter, Astropart. Phys. 35 (2012) 508;
also: Kashti, Waxman, 2005; Lipari et al, 2007
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 22
22
From the source to the detector: UHECR transport
> Kinetic equation for co-moving number density:
Expansion of
Universe
Pair production
Blumenthal, 1970
[here b=-dE/dt=E t-1loss]
Photohadronics
Hümmer, Rüger,
Spanier, Winter, 2010
CR inj.
z-dep!
GZK cutoff
> Energy losses
 UHECR must from
from our local
environment
(~ 1 Gpc at 1010 GeV,
~ 50 Mpc at 1011 GeV)
(M. Bustamante)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 23
Cosmogenic neutrinos
Cosmogenic neutrinos
EeV
> Prediction depends on
maximal proton energy,
spectral index g, source
evolution, composition
> Can test UHECR beyond the
local environment
Protons
> Can test UHECR injection
independent of CR
production model
 constraints on UHECR
escape
(courtesy M. Bustamante; see also Kotera, Allard, Olinto, JCAP 1010 (2010) 013)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 24
UHECR transition models
> Transition between Galactic (?) and extragalactic cosmic rays at
different energies:
> Ankle model:
Extragalactic
 Injection index g ~ 2
possible
( Fermi shock acc.)
 Transition at > 4 EeV
> Dip model:
 Injection index
g ~ 2.5-2.7 (how?)
 Transition at ~ 1 EeV
 Characteristic shape
by pair production dip
Figure courtesy M. Bustamante; for a recent review, see Berezinsky, arXiv:1307.4043
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 25
More details: Gamma-ray observables?
> Redshift distribution
~
(1+z)a
> Can be integrated over.
SFR
Total number of bursts in the
observable universe
Threshold correction
 Can be directly determined
(counted)!
 Order 1000 yr-1
(Kistler et al, Astrophys.J. 705 (2009) L104)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 26
26
Consequence: Local GRB rate
> The local GRB rate can be written as
where fz is a cosmological correction factor:
(for 1000
observable
GRBs per
year and 30%
of all bursts
seen)
(Baerwald, Bustamante, Winter, arXiv:1401.1820)
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 27
Required baryonic loading (analytical)
> Required energy ejected in UHECR per burst:
> In terms of
g-ray energy:
~1.5 to fit UHECR
observations
Fraction of energy
in CR production?
~5-25
How much energy
in UHECR?
Energy in protons
vs. electrons (IceCube def.)
> Baryonic loading fe-1~50-100 for E-2 inj. spectrum (fbol ~ 0.2),
Eg,iso ~ 1053 erg, neutron model (fCR ~ 0.4)
[IceCube standard assumption: fe-1~10]
Walter Winter | UHECR 2014 | Oct. 13-15, 2014 | Page 28