High energy emission from supernova
remnants and regions of star formation
Diego F. Torres
dtorres@igpp.ucllnl.org
Lawrence Livermore Lab.
California, 94550, USA
www.angelfire.com/id/dtorres
Summary



SNRs
 EGRET sources and SNRS
 GLAST/MAGIC prospects
 SNRs searched at TeV energies with HESS
Unidentified gamma-ray sources at TeV
: not within this talk.
(Torres et al. ApJ Letters 601,
 Multiwavelength information
75, 2004)
 Possible ideas for an explanation
Star forming galaxies at TeV
 Modeling for Arp 220 with Q-diffuse
Gamma-rays from Supernova Remants
Torres et al. Physics Reports 382, 303, 2003
and references therein + some new results, especially Aharonian et al.’s from HESS
Spectrum of cosmic rays
The bulk of CRs occurs with energies
below the knee and are thought to
come from our own galaxy.
Accelerated electrons, ok. Accelerated protons?
Crab
Cassiopeia A
Synchrotron radiation
g
protons,
electrons
e
p0
Shock front
p
g
Molecular cloud
Since CRs are deflected by the galactic magnetic field, they do not
preserve the information on the location of their source. We must,
consequently, look for electromagnetic signatures produced by the
protons and ions during their acceleration.
Spatially coincident pairs of SNRs and unidentified EGRET sources
Spatially coincident pairs of SNRs and unidentified EGRET sources
Identifying the pairs
Spatially coincident pairs of SNRs and unidentified EGRET sources
Data contained in the3EG
Spatially coincident pairs of SNRs and unidentified EGRET sources
Variability
Variability indices
Spatially coincident pairs of SNRs and unidentified EGRET sources
Variability
Some of the sources are variable. Not related to SNRs
3EG J0535+2610
Big variations in the measured flux
for adjacent viewing periods.
Romero, Torres et al. A&A, 376, 599, 2001
Anchordoqui, Torres, et al. ApJ, 2003  neutrino yield detectable at Earth!
High Resolution radio map of the
nearby star LS 5039 obtained with
VLBA and VLA in phased array mode
(similar to a single dish of 115m) at 6
cm.
The presence of radio jets is the main
evidence supporting its microquasar
nature.
Contours shown go from 6-50 times
0.085 mJy per beam, the rms noise. The
map is centered at the star position. 1
milliarcsec is equivalent to 3AU (~1013
cm) for a distance of 3kpc.
Paredes et al. Science 288, 2340
3EG J1824-1514
Spatially coincident pairs of SNRs and unidentified EGRET sources
Known data for the SNRs
From Torres et al. 2003, Physics Reports
Torres et al. 2004, Adv. In Space Physics
Most plausible cases appear to present broad correlations
The GeV future
W66
RX J2020.3+4026
(Pulsar)
Some recent results from HESS
CANGAROO 1997 – the small single telescope observations
but H.E.S.S. 2004...
SN1006 – the prototype ?
The panorama for SNR G347.3-0.5 (RX J1713.7-3946 )
ROSAT X-ray contours. Emission
from the bulk of the SNR rim
candepicts
be seen
with
particular
Red
the TeV
significance
enhancements
along(5.3 the
contours.
The flux was
±
west/northwest
0.9
[statistical]regions,
± where
1.6
While
electrons
give
rise
to the bulk of the non-thermal radio, X-ray and TeV
bright
non-thermal
radio
[systematic])
x 10-12 photons
in seen.
the
the
CR protons and ions are exposed at GeV energies via
-2 semission
-1 (atis E>1.8
emission
also
The
total
cm
± NW,
0.9
TeV).
their
interactions
radio flux
well AA354,
below
10
Jy, in the dense material of cloud A, leading to pion
Muraishi
etishadronic
al.
L57
gamma-decay GeV emission in the NE.
Slane et al. ApJ 525, 357 (1999)
(2000).
1st SNR at TeV energies !
SNR RX J1713.7-3946 has all the ingredients:
extreme X-ray bright, EGRET source nearby, dense molecular cloud region
Butt et al. 2001
Aharonian et al. Nature October 2004
But...
Interpretation still in doubt...
evidence for hadronic particle acceleration in SNRs still unclear.
Enomoto et al. 2002 (Nature)
IC interpretation in conflict with data
Butt, Torres et al. 2002 (Nature), Reimer & Pohl 2002 (A&A)
p0 interpretation in conflict with data, too !
-> SNR RX J1713.7-3946
Partial Summary

SNRs are TeV sources! (Even using Whipple criterion –
accepting observations by CANGAROO on G347.3-0.5)

Thus, evidence suggests that some of them should also be GeV
sources, and all theoretical models for GLAST and MAGIC
energy ranges await testing.

Proton acceleration up to TeV energies yet awaits testing.
Gamma-rays from Luminous and Ultra
Luminous Infrared Galaxies
Torres, Reimer, Domingo, Digel
ApJ Letters, 607, 99-102 (2004)
Torres  Arp 220
ApJ, 617, 966 (2004)
Cillis, Torres & Reimer 2005
ApJ in press
Gamma-ray emission from the interstellar medium


High-energy gamma-rays are produced in cosmic-ray interactions
with interstellar gas and photons
 Cosmic-ray production is associated with regions of massive star
formation (e.g., SNRs, colliding OB stellar winds)
This represents approximately 90% of the high-energy gamma-ray
luminosity of the Milky Way (~106 solar)
~60% of all EGRET gamma-rays were diffuse emission from the Milky Way
Diffuse emission from external galaxies
LMC
EGRET
IRAS
(1.9 ± 0.4) x 10-7 cm-2 s-1
30 Doradus: extensive massive SFR and molecular
clouds


Only one other external galaxy detected in the light of its diffuse emission – LMC
The problem is distance: Milky Way at 1 Mpc would have a flux of about 2.5 x
10-8 cm-2 s-1 (>100 MeV), well below EGRET’s detection limit
[Akyuz et al. 1992, Volk et al. 1996, Paglione et al. 1996, Bloom et al. 1999]
Nearby Starbursts: Upper limits with EGRET data
M82
Average
Flux [photons cm-2 s-1 MeV-1]
NGC 253
Bloom et al. 1999
Energy [MeV]
10 starbursts selected by distance (<10Mpc),
Infrared luminosity (>109 Lsolar) at latitudes |b|>10.
Barnes and Hernquist 1996: merging of gas-rich galaxies
Left: Time-evolution of a
galactic encounter, viewed along
the orbital axis. Here dark halo
matter is shown in red, bulge
stars are yellow, disk stars in
blue, and the gas in green.
Right: showing only gas in both
galaxies
Almost all ULIRGs seems to be double or
interacting
Only one within the 100 Mpc sphere [Arp 220]
And there are tens of LIRGs (luminosities >1011 LSUN):
detectability depends on the combined effect of distance and
starburst activity.
[review on LIRGs and ULIRGs: Sanders and Mirabel, ARA&A, 1996]
Detectability of LIRGs

Gamma-ray detectability is favored in
starburst galaxies (Akyuz, Aharonian,
Volk, Fichtel, etc)
 Large M, with high average gas
density, and enhanced cosmic ray
density

Recent HCN-line survey of Gao &
Solomon (2004) of IR and CO-bright
galaxies, and nearby spirals
 Allows estimate of SFR (from HCN
luminosity) and minimum required k
for detection by LAT and IACTs
(from HCN + CO intensities and
distance)

Several nearby starburst galaxies and a
number of LIRGs and ULIRGs are
plausible candidates for detection
MW
CR Enhancement required for detectability/LAT
Arp 220

The best studied and nearest ULIRG (72 Mpc)




Arp 220’s center has two radio-continuum and two IR sources, separated
by ~1 arcsec (e.g., Scoville et al. 1997, Downes et al. 1998, Soifer et al.
1999, Wiedner et al. 2002).
The two radio sources are extended and nonthermal (e.g., Sopp &
Alexander 1991; Condon et al. 1991; Baan & Haschick 1995), and likely
produced by supernovae in the most active starforming regions.
CO line, cm, mm-, and sub-mm continuum (e.g., Downes & Solomon
1998) as well as recent HCN line observations (e.g., Gao & Solomon
2004a,b) are all consistent with these two sources being sites of extreme
star formation and having very high molecular densities.
Other less luminous candidates –if closer- can be detected.
Arp 220
Downes & Solomon 1998,
Gao & Solomon 2004
~350 pc
Arp 220: Geometry
Arp 220: Supernova explosion rates
18 cm VLBI (3 x 8 milliarcsec resolution) continuum imaging of Arp 220 has revealed the
existence of more than a dozen sources with 0.2-1.2 mJy fluxes (Smith et al. 1998), mostly in
the western nucleus. In November 2002, new observations with VLBI revealed 30 supernova
remnants candidates, 20 in the western, and 10 in the eastern nucleus.
All, the previous result, models of
the nuclei using Starburst99
(Shioya et al. 2001); and
relationships between the infrared
luminosity and the rate of
supernova explosions (Van Buren
et al. 1994, Manucci et al. 2003)
suggest that the rate is ~2 yr ( ! )
This rate is ~300 times larger than
the largest of the Local Group
Galaxies (M31: ~0.9 SN/century)
7 hours, 17 telescopes.
Size of the sources ~0.1 pc
No single compact, central core, as in AGNs
High brightness argues for non-thermal origin
Q-DIFFUSSE
High energy emission computed from first principles

Q-DIFFUSE set implements







Spectral computation of secondary and tertiary particles and
their emissions at different frequencies (pions, muons, electrons,
positrons, neutrinos)
Solution of the diffusion-loss equation => steady population of
particles
Radio emission through synchrotron and free emission of
primary and secondary electrons
Free-free absorption processes
IR and FIR spectra through the dust emissivity
Gamma-emission through Brem, IC, of the primary and
secondary populations and pion decay (emission of the steady
distributions)
Minimize the set of assumptions, relate them to observations
Q-DIFFUSSE
Observational measurements
SN rate, mass, density, & IR luminosity
Injection proton and electron spectrum, diffusion/escape timescales
Secondary production: knock-on process,
neutral and charged pion decay
diffusion-losses: steady spectrum of protons & electrons
Synchrotron, IC, Bremsstrahlung, Pion Decay. Absorption
of gamma-rays, opacities, eq. of radiation transport.
Model predictions from radio to IR with emission of
secondaries, FIR with emission of dust. Parameter
fixing: B-field.
Model Predictions at high energy: gamma-rays, cosmicray & neutrino fluences. Comparison with corresponding
sensitivities of RXTE, INTEGRAL, GLAST, ICECUBE,
MAGIC, HESS at each energy band.
The slope for the injection spectrum for
protons and electrons is assumed. The
normalization is defined by the SNR rate.
2nd and 3rd generation of particles is computed with
the steady spectrum of protons. Electron and positron
sources taken into account to define the steady
steady electron distribution
Computed without further assumptions
The model reproduces the FIR emission
with dust emissivity, and uses it + the CMB
for computing losses. The magnetic field is
defined by requiring that the synchrotronfree free emission of the steady population
of electrons matches observations.
IR-FIR luminosity of Arp 220
the FIR emission is modelled by dust,
having an emissivity law proportional to
Blackbody optical
contribution
Non-thermal
radio contribution
Radiation is coming from each of the
components of Arp 220, assuming that it is
radiated with a single temperature and
emissivity law.
The model (sum of the three
contributions) derived to fit the data (s =
1.5, T = 42.2 K)
• simple & conservative model
• avoids model degeneracies by increasing the number
of free parameters
Steady distribution of protons in each
of the components of Arp 220.
Example for a steady distribution of
electrons and positrons in a western-like
starburst (with B = 10 mG). The
contribution to the total steady
distribution of the primary and secondary
electrons and positrons is separately
shown. The horizontal rectangle shows
the region of electron kinetic energies
where the steady distribution of secondary
electrons is larger than that of the primary
electrons. It is in this region of energies
where most of the synchrotron radio
emission is generated.
Radio emission of the steady population of electrons
Infrared
With the magnetic field strength given
in the Table and the relativistic steady
state populations of previous Figures,
only the molecular disk is in magnetic
energy equipartition.
DISK
Lines are not fits to the data but
predictions of the model for a
particular choice of parameters.
WEST
EAST
Now we have the steady population of electrons and positrons and the IR dust emission
that is in agreement with all observations & with that population we compute the
gamma-ray flux
EGRET upper limits
The emissivity of high energy photons is the largest in the western extreme starburst, the most active region of star
formation. The differential flux, shown in the right panel without considering absorption effects, shows the influence of
volume. The disk flux is the largest, and the nuclei are now subdominant. Nevertheless, only the western starburst
provides more than one fourth of the total flux
Opacities to gamma-ray escape in the
different components of Arp 220 as a
function of energy.
At the highest energy, the opacity is
dominated by gg processes, whereas gZ
dominates the opacity at low energies.
Significant opacities are only encountered
above 1 TeV.
The inset shows the total, and the
contributions to the total opacity, in the
case of the western nucleus of Arp 220 for
this range of energy.
The equation of radiation transport is
solved to compute the predicted fluxes
taking into account all absorption
processes.
Results for integrated gamma-ray fluxes are:
This would make Arp 220 observable for GLAST and VERITAS/HESS/MAGIC
telescopes. The latter would need < 100 hours to detect it.
Be aware of cross sections for pion decays above 1 TeV.
Proof of concept beyond ARP 220 detectability itself:
LIRGs well within the 100 Mpc sphere should be TeV sources!
Summary


LIRGs and ULIRGs, following simple population analysis, are to be
detected as gamma-ray sources
 Starburst activity – cosmic ray populations – difussion
Detailed analysis for ULIRG Arp 220 confirms this. Many other
LIRGs (several tens) may appear in the forthcoming catalogs
 first multiwavelength analysis of Arp 220 (the strongest site of
star formation known, the nearest ULIRG)
 first estimation of the magnetic field – compatible with Zeeman
splitting measurements in Galactic active star forming sites
 observations with GLAST & Cherenkov telescopes are possible


IC hard X-ray emission in the model were also computed and found in
agreement with OSSE and RXTE upper limits
EBL do not affect photon propagation once gamma-rays leave the galaxy
(very low redshift)
Thank you.
Scoville et al. 1997: Arcsec imaging of CO emission
Integrated intensity map for the brightest CO (2-1) components peaks at the
positions of the near-infrared nuclei, indicated by the + symbols.
D. F. Torres 2004
Losses for protons and electrons: example
D. F. Torres 2004
Arp 220: Geometry
1.6 GHz OH emission. The grayscale is continuum and the boxes are line observations
done with MERLIN (Rovilos et al. 2003)
D. F. Torres 2004
Context: The evolution of the gamma-ray sky
1975-1982, COS-B, orbit resulted in a large and
variable background of charged particles,
~200,000 γ-rays.
1991-2000, EGRET, large effective area, good
PSF, long mission life, excellent background
rejection, and >1.4 × 106 γ-rays
Unidentified sources: 120 at high latitudes, 80
at low (|b|<10) latitudes. 66 Possible gammaray active galactic nuclei. --- 6 pulsars.
D. F. Torres 2004
Future surveys (with GLAST)
Simulated LAT maps (>100 MeV, >1 GeV, 1 yr).
More than 10000 point sources. Simulations by Seth Digel
D. F. Torres 2004
The multi-messenger context & the evolution of the sensitivity
From Torres & Anchordoqui 2004
Complementary Capabilities
From S. Ritz
D. F. Torres 2004
Perhaps the main discovery in the EGRET era…
Diversity of high-energy gamma-ray sources
Variability: the more direct way to acknowledge the
existence of several different gamma-ray sources
*Clearly defined variable and non-variable sources
*No correlation with sky position
Possible Galactic Sources:
-Pulsars, Plerions and SNRs (NV)
-Isolated Black holes, X-ray binaries, microquasars (V)
-Stars (?)
Possible Extragalactic Sources:
-AGNs (V), Radiogalaxies (?)
-Clusters of galaxies (NV)
-Regions of star formation, starbursts and ULIGS (NV)
Gamma-ray emission from the interstellar medium


High-energy gamma-rays are produced in cosmic-ray interactions
with interstellar gas and photons
 Cosmic-ray production is associated with regions of massive star
formation (e.g., SNRs, colliding OB stellar winds)
This represents approximately 90% of the high-energy gamma-ray
luminosity of the Milky Way (~106 solar)
~60% of all EGRET gamma-rays were diffuse emission from the Milky Way
Diffuse emission from external galaxies
LMC
EGRET
IRAS
(1.9 ± 0.4) x 10-7 cm-2 s-1
30 Doradus: extensive massive SFR and molecular
clouds


Only one other external galaxy detected in the light of its diffuse emission – LMC
The problem is distance: Milky Way at 1 Mpc would have a flux of about 2.5 x
10-8 cm-2 s-1 (>100 MeV), well below EGRET’s detection limit
The most interesting case?
G347.3-0.5



Positional coincidence of the non-variable EGRET
gamma-ray source, 3EG J1714-3857, with a very
massive (~3×105 solar masses) and dense (~500
nucleons cm-3) molecular cloud…
This molecular cloud is interacting with the X-ray and
TeV gamma-ray emitting SNR G347.3-0.5…
The cloud region is near the shell of the SNR, and shines
at GeV, but it is of low radio and X-ray brightness…
Butt et al. ApJ Letters, 562, 167
Butt et al., Nature 418, 499
Enomoto et al., Nature 416, 823
Reimer & Pohl, A&A 390, L43
Torres et al. Phys. Rept. 2003
Molecular environment of the SNR G347.3-0.5
Total molecular column
density over a wide
section of the fourth
Galactic quadrant around
G347.3-0.5. The lowest
contour is well above the
instrumental noise (9s) to
emphasize the relatively
low molecular column
density toward the SNR.
Slane et al. ApJ, 1999
The clouds that
to interact with
pushed away
cause of the
wave shock
seem
it are
as a
blast
Slane et al. ApJ 525, 357 (1999)
More precise indication of interaction with molecular material
The distribution of 781 line intensity
ratios, R={CO(J=21)/CO(J=10)},
measured every 15′ in the region from
l=346.5348.5; b= -0.5+0.5, and
averaged over 5km/sec bins of velocity
between vlsr= -150 km/sec  +50
km/sec.
The mean of the distribution, ~0.72, agrees with the average unexcited value in the Galactic
plane. The cloud however, show values 3s above that.
Top 0.5% of all values measured.
All other bins with high R are well outside the 3EG field
The complete panorama for SNR G347.3-0.5
ROSAT X-ray contours. Emission
from the bulk of the SNR rim
candepicts
be seen
with
particular
Red
the TeV
significance
enhancements
along(5.3 the
contours.
The flux was
±
west/northwest
0.9
[statistical]regions,
± where
1.6
While
electrons
give
rise
to the bulk of the non-thermal radio, X-ray and TeV
bright
non-thermal
radio
[systematic])
x 10-12 photons
in seen.
the
the
CR protons and ions are exposed at GeV energies via
-2 semission
-1 (atis E>1.8
emission
also
The
total
cm
± NW,
0.9
TeV).
their
interactions
radio flux
well AA354,
below
10
Jy, in the dense material of cloud A, leading to pion
Muraishi
etishadronic
al.
L57
gamma-decay GeV emission in the NE.
Slane et al. ApJ 525, 357 (1999)
(2000).
The gamma-ray luminosity
The expected g-ray flux at Earth coming from the SNR is (Drury et al.
1994),
ESN is the energy of the SN in ergs, q is the fraction of the total energy of
the explosion converted into CR energy, and n and d are the number
density and distance. In most cases, this flux is far too low to be detected
by EGRET, but the existence of massive clouds in the neighborhood can
enhance the emission
9
2
F ( E  100MeV )  10 M 3 Dkpc kqg 0
Here M is the mass of the cloud in thousands of solar masses, k is the
CR enhancement out of the usual emissivity (~2.2 10-25 s-1 H-atom-1).
The gamma-ray spectrum
The spectrum of the EGRET source
Schlickheiser 1982
However: deviation
from other points is less
than 3s. Must yet be
Confirmed.
The single power-law fit (G=-2.3)
through all points (solid black line)
is not at ease with the
enhancement at 50-70 MeV.
This feature is consistent with the
long-sought SNR neutral pion
gamma-decay resonance centered
at 67.5 MeV.
The red curve is an expected
spectrum due to hadronic CR
interactions.
As Schlickeiser has pointed out,
the
bremsstrahlung
from
secondary electrons due to the
decay of hadronically produced
charged pions, p± ’s, will contribute
significantly at energies lower than
~70 MeV.
GeV emission is not leptonic
The electron flux needed to explain the GeV emission via ebremsstrahlung in the cloud material should also produce an enhanced
synchrotron radio emission. The expected ratio of GeV bremsstrahlung
flux to radio synchrotron flux is:
F ( E  100 MeV ) 4.3  10 21
(1 p ) / 2
R

ncm3 BμG
 Hz ( p1) / 2 Jy -1 cm -2 s -1 ,
F(ν) Jy
c( p)
c(p)  105(1 p ) (3.2 1015 )( p 1) / 2 ( p  1) 2 a( p),
Measured from TeV obs..
Measured from CO obs.
Observed by EGRET
Frequency of observations
This is what we want: radio flux
prediction if the flux is leptonic
Spectral index
Violates the observed upper limit by a factor of 20 at 843 MHz.
No other plausible candidate in the 3EG field




There are two recently discovered pulsars within the
95% confidence location contours of 3EG J1714-3857:
PSR J1715-3903 and PSR J1713-3844
Their spin down luminosity is such that they cannot
contribute significantly to the observed gamma-ray
emission.
Two other SNRs within the EGRET 95% contours:
CTB37A&B. They can both be ruled out as strong
gamma-ray emitters because of both: their large
distance (11.3 kpc) and the low density medium around
them
No WR or Of massive stars in the field, no X-ray binaries
or black hole candidates
Torres, Butt, Camilo, ApJ Letters, 560, 155
In summary:
strong hints that the blast wave shock of SNR G347.30.5 is a site of hadronic cosmic ray acceleration
TeV cosmic ray electrons are accelerated in this SNR;
the abutting cloud material is extremely excited;
the cloud region is of low radio and X-ray brightness;
the GeV flux is non-variable and in agreement with that
expected from po gamma-decays;
• the spectral index is as expected for an hadronic CR source
population (but the desired confidence level not yet reached)
• there are no other candidate GeV sources within the 95%
location contours of 3EGJ1714-3857
•
•
•
•
This is probably the best existing evidence for a connection
between EGRET unidentified sources and supernova remnants.
There are other very similar cases though: W66, W28, etc.
Recent claims:
TeV detection of RXJ 1713.7-3946 by CANGAROO II:
Enomoto et al. 2002, Nature 416, 823
Source location:
NW rim of the SNR
(basically the same
position found by
Muraishi et al. 2000,
coincident with Xray maximum).
Against their own
previous claims:
CANGAROO favor a
hadronic origin for
the measured TeV
emission.
Power law with
steep index -2.8.
Can the TeV emission be hadronic as well?
Problems with Enomoto’s fit
Furthermore:
Matter density in the NW rim: less than
0.1 cm-3, not 100 cm-3, as CANGAROO
assumed to plot their curve.
Reimer & Pohl, astro-ph/0205256;
Butt, Torres, et al., Nature 418, 499 (2002)
Finally, EGRET and TeV source are not
•The source
is not !sub-GeV!
even spatially
coincident
But, there is
still the
possibility
for a
hadronic
origin if
the TeV
photons
detected
are actually
coming
from
Cloud B!
D. F. Torres 2004
EGRET stacking
searches
Cillis, Torres, Reimer 2004. Submitted to ApJ