Particle Acceleration in Supernova Remnants
Don Ellison, North Carolina State Univ.
Tycho’s Supernova Remnant
Flux
Galactic Cosmic Rays
Solar
modulation
blocks low
energy CRs
1020 eV
1015 eV
http://chandra.harvard.edu/photo/2005/tycho/
Energy [eV]
109 eV
1021 eV
Hillas_Rev_CRs_JPhysG2005.pdf
Don Ellison - Galactic Physics with VERITAS
 A number of young SNRs are observed to emit GeV-TeV gamma rays
 The most likely acceleration mechanism is Diffusive Shock
Acceleration (also known as the first-order Fermi mechanism)
 Collisionless shocks are common in astrophysics on all scales from
solar system to galaxy clusters
 Wherever you see a collisionless shock you see particle acceleration
SNR RX J1713
Tycho’s SNR
Vela Jr
Tanaka et al. 2008
Acciari et al. 2011 ApJL
Don Ellison - Galactic Physics with VERITAS
Aharonian et al. 2007
Shock accel. may be the most important acceleration mechanism in astrophysics because
(1) shocks are common, (2) parameters often favor it (B-field, densities) over reconnection,
stochastic acceleration, etc., and it’s (3) easy to transfer bulk K.E. into individual particle energy
Solar Orbiter Team Collaboration (Mueller, D. et al.) arXiv:1207.4579
Shocks in Galaxy Clusters
Solar flares and CME shocks
http://chandra.harvard.edu/photo/2005/tycho/
Galaxy cluster CIZA J2242.8+5301: van Weeren+ 2010
Range in length scales ~ 15 orders of
magnitude, BUT, expect physics to be
essentially the same !
What we learn from SNRs can be applied to
less accessible systems
Tycho’s Supernova Remnant
Theory of shock acceleration has been around for ~40 yr but
important questions remain for CRs produced at SNRs:
 Overall acceleration efficiency? Need >10% from SNRs to power CRs
 Spectral shape? Compare against CRs observed at earth
Maximum CR energy a given shock can produce?
1) CR “knee” at 1015-16 eV
2) UHECRs above 1019 eV
 Electron/ion injection ratio ?
1) Ions contain most energy but electrons radiate more efficiently.
2) For synchrotron and inverse-Compton emission must know e/p ratio
 Magnetic field amplification ?
1) Shock acceleration depends on self-generated B-field turbulence
2) Evidence for MFA comes from SNR observations and PIC simulations
 Escape of highest energy CRs from shock precursor ?
1)
2)
3)
SNRs interacting with nearby molecular clouds
Contribution of escaping CRs to total galactic CR population
Generation of B-field turbulence (e.g., Bell’s instability)
 Role shock geometry (B-field orientation) plays in acceleration? SN1006
Cosmic rays measured at Earth
Electron spectral shape is
similar when radiation
losses are considered.
104
CR Ions
Note: recent work shows small,
but important, difference in
shapes of H and He spectral
measured at Earth
PAMELA (Adriani et al. 2011)
p / He
q  q p  qHe  0.1
Rigidity (GV), R = pc/(eZ)
10-32
This small difference has meaning
for CR source environment
Figure from P. Boyle & D. Muller via Nakamura et a. 2010
Diffusive shock acceleration is only known mechanism that can
naturally produce such similar spectral shapes and abundances
Don Ellison - Galactic Physics with VERITAS
Key elements of Efficient Diffusive Shock Acceleration theory (non-rel. shocks):
1)
Theory and observations suggest DSA can be efficient (~50% of ram kinetic energy
can go into relativistic CRs in high Mach number shocks)
2)
This energy is almost totally in ions not electrons (99% vs 1%)
3)
Taking this much energy out of thermal gas and putting it into relativistic particles
changes the hydrodynamics of the shocked gas. There is less pressure in the
shocked gas with CR production so the SNR expands more slowly.
4)
With efficient CR production, the shocked density increases and the shocked
temperature decreases from TP case. This changes thermal X-ray line emission which
depends on density, temperature, and ionization fraction
5)
Temperature-shock speed relation depends on CR efficiency.
6)
Shock structure must adjust to CR production  predictions for the spectral shape of
CRs (i.e., concave). BUT, not so simple in evolving SNR where spectra evolve with
adiabatic losses & B-field changes
7)
Shock acceleration only works if there is magnetic turbulence. B must be
self-generated by CRs (ISM turb. too weak and need wide dynamic range)
8)
Magnetic Field Amplification : CRs produce
kTs 
B 2  BISM
3
m p vsk2
16
This is
Modified
 CR Max. En. &
Synchrotron emission
Several Coupled Nonlinear Processes  Complex models
Have developed a computer simulation to model evolving SNRs
and cosmic-ray production :
CR-hydro-NEI
Main group: Don Ellison, Pat Slane, Dan Patnaude, Herman Lee
With help from: Andrei Bykov, Daniel Castro, Hiro Nagataki, John
Raymond, Jack Hughes & Kris Eriksen
Early work with: Anne Decourchelle & Jean Ballet (2000,2004)
Latest papers in a long series (check references for details and approximations):
Slane et al., 2014, ApJ, V. 783, p. 33
“A CR-hydro-NEI Model of the Structure and Broadband Emission from
Tycho's Supernova Remnant”
Lee et al., ApJ, submitted
“Reverse and forward shock X-ray emission in an evolutionary model of
supernova remnants undergoing efficient diffusive shock acceleration”
Don Ellison - Galactic Physics with VERITAS
SNR Model (CR-hydro-NEI code)
SNR hydrodynamics, Nonlinear Shock Acceleration of CRs,
Broadband continuum radiation, Thermal X-ray line emission
1) VH-1 code for hydrodynamics of evolving SNR (e.g., J. Blondin)
2) Semi-analytic, nonlinear DSA model (P. Blasi and co-workers)
3) Non-equilibrium ionization for X-ray line emission at forward and reverse shocks
4) Ejecta composition from core-collapse and Type Ia SNe  New
5) Nnonlinear shock acceleration coupled to SNR hydrodynamics
6) Magnetic field amplification (so far, resonant instability only)
7) Electron and Ion distributions from thermal to relativistic energies
8) Continuum photon emission from radio to TeV
9) Simple model of escaping CRs propagating beyond SNR
Apply model to individual SNRs: e.g., RX J1713, CTB 109, Vela Jr., Tycho
Don Ellison - Galactic Physics with VERITAS
Shocked ISM
material :
1-D: Model Type Ia or core-collapse SN with Pre-SN wind
Calculate
X-ray emission
from this region
Forward Shock
1) CR electrons and ions
accelerated at FS (and RS?)
a)
CD
Protons give pion-decay
-rays
b) Electrons  synchrotron, IC,
& non-thermal brems.
c) High-energy CRs escape from
shock precursor & interact
with external mass
Reverse
Shock
2) Evolution of shock-heated
plasma between RS and FS
Escaping
CRs
Shocked Ejecta
material: X-ray lines
depend on SN Type
Extent of shock
precursor
a)
Electron & ion temperatures,
density, charge states of
heavy elements tracked
b) Include adiabatic losses &
radiation losses
3) X-ray emission lines
a)
Spherically symmetric: We do not yet
model clumpy structure
Don Ellison - Galactic Physics with VERITAS
Forward shock (heavy element
composition from ISM)
b) Reverse shock (ejecta
composition from SN
explosion models)
Shock wave
Diffusive Shock Acceleration: Shocks set up
converging flows of ionized plasma
Unshocked ISM,
cool, at rest
SN
explosion
VDS
Vsk = u0
Particles make nearly elastic
collisions with background
plasma
 gain energy when cross
shock
Post-shock gas  Hot,
compressed, dragged along
with speed VDS < Vsk
charged particle
moving through
turbulent B-field
 bulk kinetic energy of
converging flows put into
individual particle energy
 This is why heavy
particles get more energy
(unlike electric potential)
Don Ellison - Galactic Physics with VERITAS
Particle distributions from NL DSA
e’s
continuum emission
p4 f ( p)
p’s
synch
Kep
pion
IC
A number of parameters
needed for modeling !!
e.g., e/p ratio, Kep
brems
In addition, emission lines in thermal
X-rays. Depends on electron Temp.
equilibration model
In nonlinear DSA, Thermal & Non-thermal emission coupled
 big help in constraining parameters
At GeV-TeV, pion decay from p-p collisions competes with IC from electrons
Don Ellison - Galactic Physics with VERITAS
Thermal & Non-thermal Emission in SNR RX J1713
Important question for
SNR RX J1713 and other
SNRs
Are highest energy
photons produced by
Ions (p-p collisions and
pion decay) ?
or
Suzaku image
HESS contours
Tanaka et al. 2008
Electrons (inverseCompton off background
photons) ?
(or some combination) ?
For J1713, reasonable fits possible to continuum only with either
pion-decay or inverse-Compton dominating GeV-TeV emission
Leptonic
Hadronic
Suzaku
Fermi LAT
pion
IC
Hadron model parameters:
np = 0.2 cm-3
e/p = Kep = 5 10-4
B2 = 45 µG
Uniform ISM model: Ellison, Patnaude,
Slane & Raymond ApJ 2010
Lepton model parameters:
np = 0.05 cm-3
e/p = Kep = 0.02
B2 = 10 µG
Only CMB photons for IC emission
Don Ellison - Galactic Physics with VERITAS
When X-rays are calculated self-consistently, force lower density and higher
Kep ~ 0.02, eliminates pion-decay fit for SNR J1713 (at least in simplest scenario)
Hadronic
Well above
Suzaku limits
Leptonic
Fermi LAT
pion
IC
Hadron model parameters:
np = 0.2 cm-3
e/p = Kep = 5 10-4
B2 = 45 µG
Uniform ISM model: Ellison, Patnaude,
Slane & Raymond ApJ 2010
Lepton model parameters:
np = 0.05 cm-3
e/p = Kep = 0.02
B2 = 10 µG
Only CMB photons for IC emission
Don Ellison - Galactic Physics with VERITAS
Core-collapse model improves fit for SNR J1713 (Lee, Ellison &
Nagataki 2012; Ellison, Slane, Patnaude, Bykov 2012)
SN explodes in a 1/r2 pre-SN wind
Inverse-Compton emission off
CMB dominates GeV-TeV
emission
synch
IC
Good fit to highest energy HESS
observations with IC
p-p
brems
Large majority of CR energy is still
in ions even with IC dominating the
radiation
 SNRs produce CR ions
With pre-SN wind, B-field lower than ISM  Can have MFA and still have B-field low enough
to have high electron energy. For J1713, we predict average shocked B ~ 10 µG
Low wind density disfavors p-p. As long as FS is in the wind, IC should dominate
Don Ellison - Galactic Physics with VERITAS
Vela Jr. Similar to J1713
Vela Jr. Images from Iyudin et al. 2007
Don Ellison - Galactic Physics with VERITAS
Vela Jr. Similar to J1713. Parameters that give p-p fit to GeV-TeV
over-produce X-ray lines. Inverse-Compton works fine (Lee et al. 2013)
p-p
IC
p-p
IC
Without X-ray line model, cannot strongly distinguish p-p from IC fits
Consistent X-ray line emission eliminates pion-decay fit (in homogeneous models)
IC & p-p fits give very different parameters for SN, SNR environment, & DSA
99% of energy in CRs ions not electrons
This is a core-collapse, pre-SN wind model
Don Ellison - Galactic Physics with VERITAS
VERITAS discovery of TeV gamma-rays from Tycho’s SNR
Acciari et al. 2011 ApJL
Tycho’s SNR : Example where pion-decay dominates GeV-TeV
emission (Morlino & Caprioli 2012)
Morlino &
Caprioli 2012
Tycho’s SNR
p-p
pion-decay
emission
InverseCompton
Fitting parameters: Kep = 1.6 10-3 , Shocked B2 ~ 300 G,
Shock acceleration efficiency : ~12% of explosion energy in CRs ions
In this pion-decay fit, soft underlying proton spectrum is required.
Is this a problem for efficient shock acceleration?
Don Ellison - Galactic Physics with VERITAS
Tycho’s SNR using CR-hydro-NEI model (Slane et al. 2014)
 Include feedback between NL
shock acceleration and SNR hydro
Thermal X-ray
emission lines
Suzaku
VERITAS
Fermi LAT
 Include constraints from
Thermal X-ray emission.
 Our “best-fit” model shows
near-equal contributions from IC
and p-p at GeV energies
radio
Pion decay from Ions
IC from electrons
This is a Type Ia, uniform ISM model
Don Ellison - Galactic Physics with VERITAS
Broadband Tycho SNR modeling:
Fermi LAT
Total
VERITAS
This is a Type Ia, uniform ISM model
Our “best-fit” model shows
near-equal contributions from
IC and p-p at GeV energies.
Pion-decay dominates at TeV
Pion decay from Ions
Inverse-Compton
from electrons
 IC : p-p combination  “soft” total gamma-ray spectrum.
Tycho’s SNR
region 2
background
region 1
region 1:
at edge
region 2:
inside
Slane et al., constrain parameters
with
Radial profiles &
Detailed study of thermal X-rays at SNR
edge
Don Ellison - Galactic Physics with VERITAS
Particle spectra & parameters for Tycho’s SNR models
Slane et al. 2014
Morlino & Caprioli 2012
protons
Electrons x 1/Kep
Log [p/mpc]
Log [ p4 f(p) ]
protons
electrons
Log [ p/mpc ]
Protons show slight concave curvature in both models
Soft proton spectra stems from finite scattering center speed in
self-generated magnetic turbulence
Electron spectra : Self-consistent model (Slane+14) includes radiation losses,
ionization losses, and evolutionary effects in shocked plasma
M&C:
Kep= 1.6 10-3 , B2 ~ 300 G,
Slane+: Kep= 3 10-3 ,
DSA efficiency ~12% of SN explosion en. in CRs ions
B2 ~ 180 G, DSA efficiency ~16% of SN explosion en. in CRs ions
Ackermann et al. Science 2013
Fermi LAT observations of SNRs
interacting with dense material
(molecular clouds)
VERITAS &
MAGIC
p-p kinematic feature at ~100 MeV:
Direct evidence for pion-decay -rays
Direct evidence for SNRs being a
primary source of galactic CR ions
Other reasons to believe SNRs are
primary source of the bulk of CRs
below the “knee”
Energy budget and Ionic composition
of CRs are most compelling reasons
Don Ellison - Galactic Physics with VERITAS
Warning: many parameters and approximations in CR-hydro-NEI model
(NL shock acceleration is complicated) but :
For spherically symmetric models of Core-collapse SNRs J1713 & Vela Jr.:
Inverse-Compton is best explanation for GeV-TeV
Other SNRs can certainly be Hadronic or mixed, e.g. Tycho’s SNR & CTB 109
Important: For DSA, ~99% of CR energy (~20% of ESN) is in ions even with IC
dominating the radiation
 All nonlinear shock models show that SNRs produce CR ions !!!
Besides question of CR origin: SNRs can provide constraints on critical
parameters for shock acceleration:
a)
b)
c)
d)
e)
f)
Shape and normalization of CR ions from particular SNRs
electron/proton injection ratio
Acceleration efficiency
Magnetic Field Amplification
Properties of escaping CRs
Geometry effects in SNRs such as SN1006
A critical open question for VERITAS is maximum CR
energy: 1013 eV -rays  ~1014 eV protons: still below
CR knee at 1015-16 eV
Is there a cutoff ? Cannot fit higher energies with IC !
Emax Cutoff as critical as p-p threshold
Fermi LAT
Tycho SNR
VERITAS
IC
Cutoff?
Don Ellison - Galactic Physics with VERITAS
Conclusions
1) Need to be wary going from observed photon spectra to underlying
particle spectra
a) Particle spectra evolve in RS-CD-FS region
b) Radiation losses have strong effect on electron spectrum
2) Difference between core-collapse & Type Ia SNRs :
a) If FS in pre-SN wind, low B and low density favors IC over p-p
b) For Type Ia, can have high enough ambient density for p-p to
dominate
3) Hard to get IC production to extremely high energy.
a) Shocks need high B-field to get to high energies but this produces
strong radiation losses for electrons
b) Extending observations of Tycho gives direct, unambiguous
information of Maximum CR ion energy
4) What we learn about shock acceleration from SNRs can be applied to less
accessible systems
Extra Slides
Don Ellison - Galactic Physics with VERITAS
If you want clumpy:
Don Warren & John Blondin 2013
No DSA
RS-CD-FS
positions
3D hydro simulations
showing positions of
forward shock, reverse
shock and contact
discontinuity.
Includes a
phenomenological
model of NL DSA
Efficient
DSA
Efficient DSA causes
CD-FS separation to
decrease
Rayleigh-Taylor
instabilities alone can
allow ejecta knots to
move ahead of FS
RS-CD-FS
positions
Don Ellison - Galactic Physics with VERITAS
Slane+ 2014
Don Ellison - Galactic Physics with VERITAS
Slane+ 2014
Don Ellison - Galactic Physics with VERITAS
SNR CTB 109 (Daniel Castro et al 2012)
Fermi LAT
Fermi LAT
1 TeV
MeV
XMM-Newton
XMM-Newton
keV
Don Ellison - Galactic Physics with VERITAS
Detailed analysis of X-ray emission with CR-hydro-NEI
model  Mixed Hadronic-Leptonic model fits best
Hadronic
Leptonic
Mixed
Coupling between NL DSA and X-ray line emission
critical for interpreting GeV observations.
Cannot determine nature of emission without nonlinear
model coupling thermal and non-thermal emission
Don Ellison - Galactic Physics with VERITAS
Evidence for High (amplified) B-fields in SNRs
Sharp synch. X-ray edges
Cassam-Chenai et al. 2007
Radial cuts
magnetically
limited rim
synch loss
limited rim
magnetically
limited rim
synch loss
limited rim
radio
X-ray
Chandra observations of Tycho’s SNR
(Warren et al. 2005)
If emission drops from B-field decay
instead of radiation losses, expect synch
radio and synch X-rays to fall off
together.
Thin structures: evidence for radiation losses
and, therefore, large, amplified B-fields.
Higher than expected from compression
Self-generated turbulence at weak
Interplanetary shock
Indirect evidence for strong
turbulence produced by CRs at
strong SNR shocks
B/B
B/B
Baring et al ApJ 1997
B/B
Tycho’s SNR
Sharp X-ray synch edges
Don Ellison - Galactic Physics with VERITAS
JGR, 2013
IBEX satellite
observations at
quasi-parallel
Earth bow shock
Spacecraft observations of particle escape from a Q-parallel shock
Don Ellison - Galactic Physics with VERITAS
IBEX satellite Observations:
Escaping ions
Box Shock
Bn
Can define gradual “Free Escape
Boundary”
Trattner et a; (2013): “Somewhere in the upstream region of a quasi-parallel shock, the shockaccelerated diffuse ions decouple from the acceleration region and stream away, which lets them
escape from the region where Fermi acceleration occurs.”
Spacecraft observations of particle escape from a Q-parallel shock
Don Ellison - Galactic Physics with VERITAS
Peaks in turbulence spectrum may be important for explaining X-ray strips in
Tycho’s SNR (see Bykov et al., ApJL 2011)
Eriksen etal 2011
Chandra 4-6 keV X-rays
Don Ellison - Galactic Physics with VERITAS
p4 f(p) [f(p) is phase space distr.]
Temperature
If acceleration is efficient, shock becomes
smooth from backpressure of CRs
p4 f(p)
test particle shock
Flow speed
Lose universal
power law
subshock
X
NL
TP: f(p) 
► Concave spectrum
p-4
► Compression ratio, rtot > 4
► Low shocked temp. rsub < 4
In efficient acceleration, entire particle spectrum must be described
consistently, including escaping particles  much harder mathematically
BUT, connects photon emission across spectrum from radio to -rays
Particle spectra calculated with semi-analytic code of Blasi, Gabici and co-workers
Hadron
Coulomb Eq.
Suzaku
CR-hydro-NEI code includes thermal X-ray lines:
► Non-equilibrium ionization calculation of
heavy element ionization and X-ray line emission
► Compare Hadronic & Leptonic fits
Hadron
Instant equilibration
► Include range of electron temperature
equilibration models
► Include evolutionary effects
► Find: The high ambient densities
needed for pion-decay to dominate at TeV
energies result in strong X-ray lines
Lepton model
Coulomb Eq.
► Suzaku would have seen these lines
 When combined with self-consistent
broadband emission model, Hadronic
models excluded, at least for uniform ISM
environments
Ellison, Patnaude, Slane & Raymond ApJ (2007, 2010)
Don Ellison - Galactic Physics with VERITAS
Different shape for H and He spectra &
Hint of curvature in CR spectra seen at Earth !?
He
C
Protons (open)
Oxy
Helium (solid)
CREAM data from Ahn et al 2010
Si
Concave curvature?
iron
Don Ellison - Galactic Physics with VERITAS
Quasi-parallel Earth Bow Shock
Ellison, Mobius & Paschmann 90
H+
He2+
CNO6+
DS
Critical range for injection
Modeling suggests
nonlinear effects
important
UpS
AMPTE / IRM
observations of
diffuse ions at Qparallel Earth bow
shock
H+, He2+, & CNO6+
Observed during
time when solar
wind magnetic field
was nearly radial.
DS
Data shows high A/Q solar wind ions
injected and accelerated preferentially.
These observations are consistent with
A/Q enhancement in nonlinear DSA
(Eichler 1979)
Don Ellison - Galactic Physics with VERITAS
Word on observations of rapid time variability in SNR synchrotron
emission
RX J1713 Nature 2007
1-2.5 keV X-rays
Interpreted by
Uchiyama et al. as
time-scale for
synchrotron losses
in B > 1 mG fields !
We predict much
lower B-fields
ApJL 2008
2000
2002
2004
2000
2002
2004
4-6 keV X-rays
Don Ellison - Galactic Physics with VERITAS
Alternative explanation that doesn’t set time scale of variations by
radiation losses (Bykov, Uvarov & Ellison 2008)
 Combine turbulent magnetic field with steep electron distribution
 For given synchrotron emission energy, local regions with high B have
many more electrons to radiate than regions of low B
5 keV
20 keV
High B
Log
Ne
50 keV
 Local high-B regions dominate line-of-sight
projection
 Varying magnetic turbulence produces
intermittent, clumpy emission
 Time scales consistent with SNR
observations
Low B
 No need for ~ 1000 µG magnetic fields
Log Electron Energy
Don Ellison - Galactic Physics with VERITAS
Solar flare and coronal mass ejection (CME) shock
Solar Orbiter Team Collaboration (Mueller, D. et al.) arXiv:1207.4579
Magnetic
reconnection and
stochastic
acceleration
important at
flare site
Shock
Solar energetic particles, i.e., solar
cosmic rays, accelerated at CME shock
Can produce >GeV particles
length scales: <<AU
time scales: seconds-hours
Don Ellison - Galactic Physics with VERITAS
Tycho’s Supernova Remnant
Chandra X-ray image
Global information for strong shock
Filamentary blue is
nonthermal X-ray
synchrotron from > 10TeV
electrons
Blast wave shock
Shock acceleration of
ions to ~100 TeV
length scales: ~ pc
time scales: ~ 1000s yr
http://chandra.harvard.edu/photo/2005/tycho/
Galaxy cluster CIZA J2242.8+5301: van Weeren+ 2010
Radio emission:
shock accelerated
electrons.
length scales: Mpc
time scales: 106-8 yr
UHECRs above
1019 eV ??
Inferred Mach # ~4.5
Wherever you see collisionless
shocks you see accelerated particles !
Polarization, E vector
Don Ellison - Galactic Physics with VERITAS