Galactic Cosmic Rays - IPAGipag.obs.ujf-grenoble.fr/IMG/pdf/AM-Seminaire-IPAG.pdf · •Spectre...

Galactic Cosmic Rays

Alexandre Marcowith

Laboratoire Univers &Particules de Montpellier

9/14/12 A.Marcowith‐Seminar‐IPAG 2

Outlines• Introduction: Cosmic Rays spectra in 3D

• Historical Supernova remnants– Sources of high energy photons and cosmic rays

• Fermi acceleration– Collisionless shock physics

• Other energetic particle sources– Massive star clusters

– Supernova remnant/molecular clouds interaction

– Low energy cosmic rays

• Conclusions

• Perspectives


100 years of cosmic rayresearch(1912-2012)

• Viktor HESS (1883-1969)– Discovery of cosmic rays:

altitude ionization effect

– Nobel price 1936

• Pierre Auger (1899-1993)– Discovery of electromagnetic

showers (1938): start ofastroparticle physics


Introduction

• Cosmic Rays in 3D– Energy spectrum

– Angular spectrum: anisotropy

– Mass spectrum: composition


⊗


Galactic Cosmic Rays

GCR


Sources of galactic cosmic rays

• Supernova remnants: Hadrons + electrons-positrons

• Massive star clusters: Hadrons+ electrons-positrons

• Pulsars and pulsar wind nebulae: Electrons-positrons (+ Hadrons)

• X-ray binaries: Electron-positron + Hadrons

Drury+01 (SSR,vol.99)


Les restes de supernova• Base du modèle “standard”:

– Argument énergétique:• Densité d’énergie du RC eRC ~ 1.5 eV/cm3

• Temps de résidence tres~10 M ansPcr = Vgal eRC / tr ~ 1041 erg/s soit 10% PSN

– Argument de composition:• Requiert d’accélérer la matière du MIS.

– Argument spectral:• Spectre source proche de celui d’accélération

diffusion par onde de choc (c.f. + loin)


HistoricalSNRs

CassiopeiaA SN ~ 1681Type II (type IIb)Distance 3.4 kpc (+0.3

-0.1)Radius 2.5 pcShock speed ~5000km/s

ChandraChandra Chandra + HST

Tycho SN 1572Type IaDistance 3.8 kpc (+1.5

-0.9)Radius 3.4 pcShock speed ~ 4600km/s

SN 1006Type IaDistance 2.2 kpc (+/-0.1)

Radius 9 pcShock speed ~ 3000km/s

+ Kepler (type Ia) 1604 6.1 kpc+ RCW86 (type II? ) 185(?) 2.8 kpc


Young SN remnant structure

<----------Ejecta --------->

2 shocksInterstellar mediumCassiopaeAbyChandra

Court.ADecourchelle

Analysisrestrictedto“young”SNrwithageafew103yearsFreeexpansion–earlySedovphase.

Green: 4-6 keV


Observations: radio• Non‐thermalradiation:radio(Greencatalog;Green’08):

synchrotron(polarizedemission)F(ν)∝ν‐α;<α>=0.55(0.4‐0.7)s=(1.8‐2.4)

•Inagreementwithdiffusiveshockacceleration(DSA)s=2

Kepler


Young SNR: X-rays

CasA Tycho

SN1006 Kepler

RCW86

Blue:synchrotronnon‐thermalX‐rays@keV(Vink’08)

Filamentsize%Rsh


X - ray Filaments: consequencies• Downstream: within asynchrotron losstimescaletsyn(Eph,B)

- Advection with a speedVav⇒Advection length : ΔRdif

‐Diffusion with acoefficient D⇒Diffusion length : ΔRadv

To be compared with thefilament size ΔRx⇒ConstraintoverB

Shock restframe view


X-ray Filaments: Magnetic field

Parizot,AM,Ballet,Gallant’06

DownstreamMF=>2ordersofmagnitudeabovestandardISMvalues:Amplification

Magnetic field amplification is required


X-ray stripes

Eriksen+11Tycho

• Turbulence pattern? Bykov+11


Supernova remnants: gamma-rays

Detected at one (GeV/TeV) or both GeV-TeV wavebands• Historical SNRs:

– Tycho, SN1006, Cassiopeia A, RCW86

• other SNRs:– RX J1713-3946.5, Vela Jr, HESS 1731-347

• SNR in interaction with molecular clouds:– IC443, W28, W51c, W44, CTB37a, W49b, W30

• Evolved SNRs:– Cygnus loop

+ star forming regions and massive star clusters:– W1, W2, Cygnus X (données Milagro), LMC

GeV: Fermi, Agile TeV: HESS, MAGIC, Veritas, CangarooIII

Par

t-II

Par

t-I


Emission mechanisms• Non-thermal particle

distribution :N(E)=N0E‐s

• Leptonic : Inverse Compton– Source of low energy photons:

Cosmic microwave backgroundbut sometimes IR, UV

– Luminosity∝neE‐(s‐1)/2

(Thompson regime).

– IC/synchrotron => B in a one-zone model

+ Bremsstrahlung (NT electrons)– Luminosity ∝nenISME‐s

•Hadronic : Neutral pion production– Cross section increases only inlog(E): gamma-ray and hadronindices similar above 1 GeV.

– Luminosity∝∫nISMnCRdV


TeV

GeV

TychoCassiopeia A SN 1006

Acciari+11

Uchiyama+11Abdo+10

Acero+10

x

Albert+07

+Chandra & CO data

+XMM newton

+XMM newton


Tycho

CasA

Abdo+10

Hadronic model

sGeV =2.3 may be up to VHEECR consistent with 10% ofESN for next=0.3 cm-3

Fermi data more consistent withhadronic model.

Hadronic modelsGeV=2.1 + cut off at 10 TeV (blue)sGeV =2.3 no cut off (red)

ECR =3.2 x 1049 erg = 1/30 ESNfor next = 10 cm-3

Density possibly smaller.Leptonic (Inverse Compton) modelalso possible.

Uchiyama+11


Other SNRs

Vela junior SNR

Type Ic ? Iyudin+10

Age ~ 450-900 yrs ?

Distance ~ 150-1000 pc

Radius 13 pc (D=750 pc)

RXJ 1713.7-3946

Type ?

Age ~ 1600 yrs (SN 393 ?)

Distance ~ 1 kpc

Radius 20 pc

Aharonian+07 Aharonian+07

Both powerful gamma-ray objects and show X-ray filaments

+ASCA

+ROSAT


Spectra

Abdo+11

• RX J1713: data more consistent withleptonic models: index sGeV =2 and amean MF = 10mG <=> X-ray filaments ?• Inclusion of hadrons in a mixedmodel: ECR < 30% ESN but a hardspectrum is required.

• Cut off beyond 10 TeV: if hadronsare present: several hundred TeVparticles.

• Vela Jr: both models are possiblebut hadronic model requires a lot ofenergy into hadrons (50% ESN) andleptonic model difficult to reconcilewith X-ray filaments

Hadronic/Mixed

Leptonic

RXJ1713


Short summary: Young SNRand cosmic rays

• Several issues:– Most of historical SNR are weak gamma-ray sources but emit

non-thermal X-rays => particle accelerators.

– Cosmic ray content in question• ECR is < 10% (CasA) but 10% Tycho.

• Emax-CR < knee (max: RXJ 1713, Tycho)

– Leptonic origin difficult to reconcile with high MF deducedfrom X-ray filaments

• MF relaxation ? (Rettig & Pohl’12, A.M. & Casse’10, Pohl+05)

• Time dependent turbulent features ? (Bykov+08)

– Contribution from the reverse shock (CasA) ? Thought to beweakly magnetized.

– Injection dependence wrt the mean magnetic field direction(SN1006) ?

• No definite observational proofs that historicalyoung isolated SNR are the sources of GCR YET!

• What about theory ?


l’accélération de Fermi (1erordre)

Accélération diffusive par ondede choc (ADOC ou DSA)

• Cas linéaire

• Cas non-linéaire



Non-linear DSA in young SNR

• (too ?) Efficient mechanism (Drury & Voelk’81)

– PCR ~ [0.1-0.5] ρush2

– continuity conditions (Rankine-Hugoniot) haveto include PCR => modification of the shockprofile


CR retro-action


MFA feed-back• MFA mainly has a negative feed-back

- MFA in the shock precursor => increase of the Alfvén velocity

Btot=(δB2+〈B〉2)1/2,ūa=Btot/(4πρ)1/2⇒ Decrease scattering center velocity Vsh-ua => smaller compression⇒ Stationary solution closer to the test particle one : s=4 or even

harder (AM, Lemoine & Pelletier’06)- A part of the turbulence energy => pre-heating (but not too

much) => smaller compressibility

Caprioli+09

Va (B0): no MFATrans: Ua in thekinetic CR Eq.Trans+Ampl: Ua

in the kineticCR Eq.+growthrate

Fluidvelocity

Particledistribution

Spectralindex

Magneticenergydensity


Collisionless shockmicrophysics

• How to simultaneously explain:1/ High energy CR production including

the non-linear effects

2/ MF amplification

3/ High energy emission


Multiple species and scales• Still quite challenging

– Completeness:• Need to include: plasma

waves, magnetized fluidand energetic particles +radiation.

– Scale description:• Requires to compute

acceleration processfrom thermal to ultra-relativistic particlescales.

• Intrinsic linear & non-linear process (scaleback-reaction).

Plasma fluctuations Energeticparticles

Magnetized thermalparticles

Hea

ting,

dam

ping

Generation, scattering,

stochastic acceleration

Com

pression, adiabatic

losses

Photons

Radiation, cooling


Streaming instability

• Principle

• Recent developments


JCR

F(x,p)

Non‐resonantinstabilityλ<rL=E/ZeB,left‐handcircularlypolarized

Resonantinstabilityλ>rL=E/ZeB,right‐handcircularlypolarized

Γ(k=1/λ)=ūa(kkc)1/2;kc=Jcr/B0

Γ(k)=Χ0|k//|ūa,Χ0=Pcr/Ubtot

Btot=(δB2+〈B〉2)1/2,ūa=Btot/(4πρ)1/2Vch

Shock precursor

MIS

Skilling’75,McKenzie&Völk’82,Bell&Lucek’01,Bell’04,Pelletier,Lemoine,AM‘06,AM,Pelletier,Lemoine’06,Amato&Blasi’09…Bykov+12(review)

PrincipleShock moving upstream


MF amplification• Non-resonant instability:

– Fastest especially at highshock velocity (Pelletier,Lemoine, AM’06)

– Expected MF amplitude atthe shock front (B ∝ Vsh

3/2)

– But:• only small scales : Issue for

confinement of high-energyparticles

Bturb

BISM

!

" #

$

% &

2

= Ma

2 ush

c

!

"

$

%

PCR

'ush

2

!

" #

$

% & = 500( )

2(1/50)(1/5) = 1000

Voelk’05


Different numericalapproaches

Particle‐In‐Cell(PIC):allspecieskinetic:Smallscalesl~rthi:instabilitiesthatmediatethe

shockformation‐injectionproblem.

“Hybrid”(electronasfluid,ionsaskinetic):Dominantinstabilityforparticleacceleration‐

backreactionovertheCRcurrentKinetic‐magneto‐hydrodynamic(MHD)(electron+ionfluid,

energeticparticlesaskinetic):Largescalesl~rCR‐longtermevolutionofthedominant

Instability‐CRtransportandescape.

Microscopic M

esoscopic Macroscopic


Développements• Instability studies

– Shock formation– Particle injection at supra-thermal energies

– Acceleration process– High energy cosmic rays tansport and instabilities

=> versus: ISM properties (magnetization, temperature, ionizationdegree), shock properties (MF obliquity, shock velocity)

1. Fluctuations produciton• PIC:Hybrid Riquelme & Spitkovsky’10, Gagrgaté & Spitkovsky’12• MHD/kinetic Bykov+12, Reville & Bell’12

2. Turbulence properties• Upstream (Pelletier+06) downstream (AM & Casse’10)

3. Particle transport• Kinetic/MHD Reville+08, AM & Casse’10

4. High shock speeds studies• Very young SNR Renaud,AM+ in prep

• Gamma-ray burst Lemoine & Pelletier’11


Conclusions: shockmicrophysics

• MFA likely connected to energetic particles assource of free energy:– Streaming

– Pressure gradient => Sonic waves

• Debates:– Master instability depending on the ISM and shock

properties

– Saturation process and turbulence properties

– Simulations:• Injection• Fermi acceleration at work

• Role of high CRs controling the turbulence up- anddownstream => Maximum energy

• Not clear answer about SNR as sources of CRs maybe valuable to look after alternative sources …


Other sources of energeticparticles/photons

• Massive star clusters• SNR/Molecular clouds interaction


Massive starclusters

• Cygnus X cocoon by Fermi– Extended hard gamma emission

Tibaldo+11

Cygnus X cocoonspectrum: index close to 2.(calorimeter?)

Hadrons seem to bemandatory => E > 3 eV/cm-3

Photon residual map 10-100 GeV

Fermi MSX 8 micron

Ackerman+11


CR acceleration: collectiveprocesses

• In massive star clusters (Bykov’01,Parizot, AM+04, Ferrand & AM’10…)– Strong SNR shocks– Multiple weak shocks– Super-sonic/alfvenic turbulence: second

order Fermi acceleration

=> Efficient CR accelerators up to 100 PeVbut theory difficult due to strong non-linear feed-backs.


CR solutions

F(p) distributionSpectral indexCase of a cluster with N=100starsFerrand & AM’10

Test-particle solutions Non-linear solutions

Ferrand+08, Ferrand & AMin prep.

MFI+FII FI


Massive star sample

θ* = tFII/tesc and ransition hard-soft at p*=1/θ* GeV


SNR in interaction withmolecular clouds

Uchiyama+10

Green: VLA data, ellipses CO cloud (a),black/white crosses OH masers

Aharonian+08a/b, Acciari+09, Fiasson+09

Triangle(c)/open(b) crosses: OH masers, stars: HIIregions(d)/PWN(c) White(a,b)/black contours(c): COdata

CTB 37a

TeVGeV


IC443

Abdo+10, + Agile Tavani+10

Age: 3-30 kyrsDistance ~ 1.5 kpc?Size ~ 19 pc

• Shock/dense materials interaction(OH masers).

• Good fit provided by a hadronic modelwith broken power-law.

• ECR~0.6-2.2x1049 erg

Also W44:

=> Neutral pion decay nicely reproducedcombining Agile and Fermi dataGiuliani+11

FermiEgretVeritasMagic


Shock/cloudinteraction

• MF Amplification due toturbulent medium (shockrippling) that is shocked– B grows due to velocity

shear along mean B– B => few hundred microG

• Network secondary shocks

– M < 2 (M=√5 in the densecloud limit)

– Behind the blast wave =>propagate in an ionizedmedium.

Ino

ue+0

9+1

0

2D MHD simulations (perpstrong shock)


SNR/MC• Several explanations:

– Hadrons accelerated from the thermal pool(Drury+96, Bykov & Uvarov’00, Inoue+10, Malkov& Diamond’11 W44).

• Single shock acceleration + Break due to loss cone

• transition between single strong shock and multipleweak shock re-acceleration + High energy spectrumsofter spectrum due to secondary shocks

– Hadrons from the cloud re-accelerated at theshock front (Blandford & Cowie’82,Uchiyama+10)

– Hadrons escaping the SNR and interacting withMCs (Gabici+07+09, Torres+08, Casanova+10)


Illuminated clouds ?• Probe of the diffusion coefficient around CR sourceGabici+09, Ohira+10:⇒ One can expect soft GeV and hard TeV spectra especiallyfor young SNR and/or close MCs.• If we know SNR-MC distance and the CR released time =>probe the diffusion coefficient around the sources.

• W28 case: if 1801 & 1800a and b are within a diffusivelength: Diffusion coefficient (HESS1801, and HESS 1800a/b)~ 6% ISM values Gabici+10• See also Torres+08 in the case of IC443 ~ % ISM values.

CR spectrum at a MC atdifferent times (Gabici+09)

32000 yr 8000 yr 2000 yr 500 yr

But OH masersdetected Hewitt& Yusef-Zadeh’09


Ion radicals measurements:dense gas

• Probe CR ionization rates in different environments usinglines produced by different ion radicals

Dense shocked gas:

1. W51c: Ionization in dense region probed by [DCO+/HCO+]Ceccarelli+11 ξ~10-15s => 2 orders of magnitude above standardvalues

2. IC443 [H3+] Indriolo+10 ξ> 10-15s

IC443

Ionization rates


Some challenges• Inducing LE-CR spectrum

– Local deconvolution from the solar wind modulationeffect likely ≠ Around SNR/MC shocks.

– MeV protons and keV electrons: interface betweenthermal and non-thermal components.

• Enhanced LE-CRs around sources:– LE CR may be released differently depending on the

upstream region: shock aging effect

– Transport to diffuse clouds (where enhanced ionizationhas been observed too).

• NO yet detailed modeling either for sourcespectrum or for transport– Only reference : heliospheric shock


Importance of LE CRs

• LE CRs important for:– Ionization and ISM heating through wave production.

– Chemistry.

– Spallation => MeV Astrophysics.

– Dynamical effects => pressure gradients largelyoverlooked.

Influence over ISM

– Spatial: A source has influence on its local environmentover 100-200 pc.

– Timescales: 104 => 107 yrs

– Sources are usually in clusters


Conclusions• Young (historical) SNRs: CR spectrum

– HE particle acceleration + MF amplification likelyconnected to non-resonant streaming instability.

– HE particles dominate EP pressure and largelycontrol the Fermi process and escape.

– No yet observational/theoretical proof the SNRare the sources of GCR

• Other sources of EP and likely CRs– Massive star clusters => promising candidates to

be tested– SNR/MC

• Aged shocks but still (re)accelerate particles =>most ofFermi SNRs

• Evidences of LE-CR ionization


Perspectives• Theory: numerical calculations

– Injection: PIC-Hybrid effects of obliquity andISM

– Fermi process: Hybrid-MHD parametric studiesand role of HE CRs

– Escape: go towards 3D simulations– Shock/cloud interaction: impact of neutrals,

decide among different scenarii.

• Observations: CTA– Improved sensitivity : spectral studies– Important role of angular resolution =>

population studiesAcero+12


Back-up

1. Simulations of turbulence in SNRs

2. Neutral damping effects


Numerical highlights

PICRiquelme&Spitkovsky’09

HybridGargaté&Spitkovsky’12

Non-resonant instability saturationby CR feed-back

Non-thermal acceleration bythe Fermi process for differentAlfvén Mach numbers (// shock)

Solid blue: 3D simulationstransvers MFDotted: longitudinal MF


Draw-backs• PIC/Hybrid methods cannot catch large

scales and are limited in time evolution.

• Simulation regime not in correspondencewith SNR conditions– Vsh > 0.5c– Biases in PIC mp/me < 1836– Low Ma (except Gargaté & Spitkovsky’12)

• Cover a part of the parameter space– Upstream magnetization + MF Obliquity (but

Gargaté & Spitkovsky’12)– Ionization degree: shock/molecular cloud regions.


PIC-MHD simulations• particles propagate into a MHD solution.• Statistical reconstruction of the transport

properties => diffusion coefficient.

• Only done in test-particle case (but see Reville &Bell’12)

Turbulent Magnetic field

Mean square displacement at different particle energies vs time:Sampling over a large amount of particles implementedrandomly in space and in pitch-angle

Reville+08


MHD-Kinetic calculationsKepler:MFadvecteddownstreamX(synchrotron)andg‐ray(InverseCompton)profilesproducedbyelectrons

Kepler:MFrelaxingdownstream

AM&Casse’10

=>Neutralpion(Acero+inprep)=>CTA

SDE+HD1Dspherical


Ion-neutral damping• Case of resonant Alfvén waves (weak turbulence limit; δB < B)

(Drury et al’96)– Typical energy where ion-neutral damping is the strongest: E1

for waves with ω=kVa>Γ– Maximum energy of particles if acceleration is limited by ion-

neutral damping: E2

1. If E2 < E1 then acceleration highly reduced compared to theneutral free case.

2. If E2 > E1 then acceleration slightly reduced compared to theneutral free case.

E1

= 8GeVT

104K

!

"

#

$

%0.4ni

1cm%3

!

"

#

$

%3 / 2B

1µG

!

" &

#

$ '

2

E2

= 1TeVU

sh

103km /s

!

"

#

$

3

P

0.1

!

"

#

$

T

104K

!

"

#

$

%0.4ni

1cm%3

!

"

#

$

0.5

nn

1cm%3

!

"

#

$

%1

• High speed shocks ~1000 km/s are more likely in case 2.• Low speed shocks ~100 km/s: E2 < 1 GeV and likely in case 1.


High MFA limit (δB2/B2>1)

Ion-neutraldamping

Ptuskin & Zirakashvili’03

SNR/MC Young SNR


Particle acceleration performancesHomogeneous-ionised-’low

density’ medium• Adiabatic high speed phases last

longer.

• Streaming instability likely togenerate strong magnetic fields

• Internal injection from the heatedthermal plasma unlesspropagating in massive starclusters regions

• Acceleration up to very highenergies (PeV and more).

• TeV particles observed from X andGamma-rays.

Inhomogeneous-partiallyneutral-’dense’ medium

• Faster shock aging. Lowervelocities.

• Alfvèn waves damped by ion-neutraldamping.

• Magnetic field compression ? Theclouds can have magnetic fields >10 µGauss.

• External energetic particlesinjection likely important

• Importance of re-acceleration/Coulomb losses at lowenergy (keV-MeV).

• Acceleration performances to betested versus wave damping and lowshock velocities.


Break & spectral indices

• Spectral break:– Strong shock: balance tFI=tdamping => E~1-10 GeV

Ptuskin & Zirakashvili’03

• Index at low energy:– Strong shock: sGeV = 2

• Index at high energy:– Multiple weak shocks: sTeV > 2

– E.g. dense cloud case M=√5 gives s=2(M2+1)/(M2-1) => 3

Inoue+10

Date post:	04-Feb-2018
Category:	Documents
Upload:	dangdung
View:	217 times
Download:	0 times

Galactic Cosmic Rays - IPAGipag.obs.ujf-grenoble.fr/IMG/pdf/AM-Seminaire-IPAG.pdf · •Spectre...

Documents