9th ASTRI Collaboraton Meetng Universidade de São Paulo Instituto de Astronomia, Geofisica e Ciencias Atmosferica The quest for PeVatrons with the ASTRI/CTA mini-array Giovanni Morlino INFN/Gran Sasso Science Insttute for the ASTRI Collaboraton & the CTA Consortum 1
OUTLINE Overview of possible candidate between known shell SNRs Theoretical predictions on direct observation: how many PeVatrons we do expect to see? Indirect detection: Molecular clouds illuminated by escaping CRs 2
Leptonic vs. hadronic origin of γ-ra ys Pion decay and IC are competitive mechanisms Above 10 TeV IC spectrum is suppressed because Klein-Nishina efect detection of Eγ >> 10 TeV will establish the hadronic origin Hadronic models Leptonic models Large B >~ 100 μg Low B ~ 10 μg 3
Evidence of magnetc feld amplifcaton Cas A Kepler SNR Bds (μg) Pw,ds(%) Cas A 250-390 3.2-3.6 Kepler 210-340 2.3-2.5 Tycho 240-530 1.8-3.1 SN1006 90-110 4.0-4.2 RCW 86 75-145 1.5-3.8 SN 1006 Tycho Inferred B fields assuming that the thickness of X-ray rims are determined by electron synchrotron losses and using the information from the X-ray frequencies. Inferred B fields are much larger than can be expected from the simple compression of BISM Bobs >> 10 m G x 4 ~ 40 m G Shock simultaneously places a large fraction of shock energy into relativistic particles (e.g. IONS) and amplifies magnetic turbulence so B/B >> 1 4
Look for PeVatrons in known SNRs Summary of shell SNRs emitng TeV gamma rays NAME Cas A Tycho North. em. SN 1006 (NE) SN 1006 (SW) RX J1713.73946 RX J0852 (Vela Jr.) RCW 86 G353.6-0.7 Age [yr] Distance [kpc] Flux(>1TeV) [10-12 cm-2 s-1] Spectral index Evidence of cutoff 330 3.4 0.77±0.11 2.61±0.24 (?) 5 440 3.3 0.19±0.05 1.95±0.6 ΓGeV-TeV=2.2 NO 10 1000 2.2 0.23±0.05 2.36±0.2 (?) 20 '' 0.15±0.05 2.43±0.2 (?) 6 ~1600 1 15.9±0.6 2.32±0.01 YES 80 420-1400 (best ~700) 200 pc 1 kpc 15.2±3.2 2.24±0.15 YES 10 1600 ~2.5 2.34 2.54 (?) 20 ~14000(?) 3.2(?) 6.91±0.75 2.32±0.06 NO 30 '' ΓTeV @10TeV Eγ,max [TeV] 5
Look for PeVatrons in known SNRs Summary of shell SNRs emitng TeV gamma rays NAME Cas A Tycho North. em. SN 1006 (NE) SN 1006 (SW) RX J1713.73946 RX J0852 (Vela Jr.) RCW 86 G353.6-0.7 Age [yr] Distance [kpc] Flux(>1TeV) [10-12 cm-2 s-1] Spectral index Evidence of cutoff 330 3.4 0.77±0.11 2.61±0.24 (?) 5 440 3.3 0.19±0.05 1.95±0.6 ΓGeV-TeV=2.2 NO 10 1000 2.2 0.23±0.05 2.36±0.2 (?) 20 '' 0.15±0.05 2.43±0.2 (?) 6 ~1600 1 15.9±0.6 2.32±0.01 YES 80 420-1400 (best ~700) 200 pc 1 kpc 15.2±3.2 2.24±0.15 YES 10 1600 ~2.5 2.34 2.54 (?) 20 ~14000(?) 3.2(?) 6.91±0.75 2.32±0.06 NO 30 '' ASTRI mini-array9 has sensitivity better than HESS for Eγ> 10 TeV ΓTeV @10TeV Eγ,max [TeV] Maximum detected energy in γ-rays. In case of hadronic model Ep,max ~ 10Eγ,max 6
SNR Cassiopea A Cas A in TeV (VERITAS) Cas A in X-rays (Chandra) Disfavoured by estimated magnetic field from X-ray filaments ~ 0.3 mg N p (E ) E 2.3 7
Tycho's SNR X-rays (Chandra) Fermi TS map 1-100 GeV [Giordano et al. 2011] VERITAS map E > 1 TeV [Acciari et al. 2011] VERITAS does not show a cut-off Eγ,max=10 TeV = Ep,max >100 TeV The best theoretical model predict Ep,max = 500 TeV ( Eknee = 3000 TeV) Synchrotron emission [G.M. & D. Caprioli, 2012] Pion decay CTA sens. 8
Tycho's SNR X-rays (Chandra) Fermi TS map 1-100 GeV [Giordano et al. 2011] VERITAS map E > 1 TeV [Acciari et al. 2011] VERITAS does not show a cut-off Eγ,max=10 TeV = Ep,max >100 TeV The best theoretical model predict Ep,max = 500 TeV ( Eknee = 3000 TeV) [G.M. & D. Caprioli, 2012] 9
Kepler's SNR Kepler's SNR is very similar to Tycho, but is far away not observed in γ-rays at the moment Gamma emission should be detectable by CTA Synchrotron emission Pion decay IC on CMB + Galactic light + IR from dust 10
SN 1006 SN 1006 in TeV (HESS) SN 1006 in X-rays (Chandra) Size 30 arcmin Bdown ~ 80-120 μg HESS data (130 hrs of observation) Total gamma-ray flux <~ 1% Crab 11
SN 1006 leptonic Model fit parameters from Aharonian et al. (2014), arxiv:1004.2124 Leptonic model (1 zone): Explain the integrated gamma-ray fux hadronic Fails to explain the steep spectrum Requires low B, contrary to what inferred from observed thin X-ray rim (B~120 μg) Hadronic model (1 zone): Requires efciency ~ 30% 1) Steep spectrum E-2.3 with Ecut>>100 TeV mixed 2) hard spectrum E-2 with Ecut~80 TeV How to distinguish between the two scenarios? At high energies X-rays come from downstream while IC photons come from upstream 1' resolution will be able to detect a displacement between X-rays and γ-rays Extending the detection to E>10 TeV will reveal the presence of a cutof 12
HESS J1713.7-3946 The remnant RX J1713.7-3946 has been considered the most promising candidate to prove the existence of accelerated hadrons FermiLAT data seem to favor a probable leptonic origin BUT... Hadronic model(s): 0 Leptonic model(s): inverse Compton scattering 13
HESS J1713.7-3946 Curves from T. Tanaka et al., ApJ 685 (2008) Hadronic Both leptonic and hadronic models have problems in ftting Ge-TeV emission. Leptonic model (1 zone): Problems in ftting the highest energy points Need a IR background 30 > Gal. average Leptonic on (unknown) IR radiation 14
HESS J1713.7-3946 Curves from T. Tanaka et al., ApJ 685 (2008) Hadronic Leptonic on (unknown) IR radiation Both leptonic and hadronic models have problems in ftting Ge-TeV emission. Leptonic model (1 zone): Problems in ftting the highest energy points Need a IR background 30 > Gal. average Hadronic model in clumpy medium: Reasonable ft with hard spectrum E-1.72 and with Ep,cut~250 TeV Clumpy CSM How to produce hard spectrum? Expansion in circumstellar medium with low average density but with high density clumps: High en. particles penetrate inside the clumps Low en. particles do not penetrate we get a hard spectrum shock 15
HESS J0852-4622 (Vela Jr.) Remnant size ~ 120' γ-ray emission well correlate with Radio and X-ray emission Main uncertainty due to distance 200pc < d < 1kpc Both hadronic and leptonic model can fit the data Lept. model favored for spectral shape but need B ~ 6 μg X-ray filaments require B~100 μg Issue in fitting the shell in γ-rays A better morphological study in γ-rays will help in distinguish between L. and H. Contours:X-rays (ROSAT) FermiLAT HESS Hadronic Leptonic ASTRI mini-array9 res.(@10tev) = 4'-5' 16
Looking for PeVatrons When we do expect production of PeV particles? To produce PeV particles we need magnetic field ~mg downstream of the shock. All possible mechanisms for magnetic amplification (upstream of the shock) require the presence of CRs. The most invoked ones are: 1) Resonant streaming instability 2) Non resonant amplification 17
How to get magnetc feld amplifcaton 1) Resonant streaming instability Particles of a certain momentum amplify Alfvén waves with a wavenumber equal to the inverse gyroradius of the particle. works for medium Mach number [e.g. Skilling (1975), Bell & Lucek (2001), Amato & Blasi (2006), Blasi (2014) arxiv:1412.8430] When the growth rate is fast : but this condition imply ( ) ncr 3 ξcr V sh = ni γmin Λ c 2 2 n va When CR ni V sh c the growth rate is slow: In both situations we get ( ) δb B0 2 few 18
How to get magnetc feld amplifcaton 2) Non resonant amplification [Bell's instability, Bell (2004)] The diffusive electric current of CRs amplifies almost purely growing waves with wave-numbers much greater than the inverse particle gyroradius. works for very high shock velocity (initial phase of SNR expansion) CR current escaping from the SNR j Amplified magnetic field lines 19
How to get magnetc feld amplifcaton 2) Non resonant amplification [Bell's instability, Bell (2004)] The diffusive electric current of CRs amplifies almost purely growing waves with wave-numbers much greater than the inverse particle gyroradius. works for very high shock velocity (initial phase of SNR expansion) Type I SNR (expanding into a uniform medium) At the beginning of the ST phase: 3 nism =1cm M ej =1M Sol 51 E SN =10 erg A factor 10 below the knee Type II SNR (expanding into a red supergiant wind) 5 M =10 M Sol / yr vw =10km / s M ej=1m Sol 51 E SN =410 erg 6 E M 2 10 GeV Right number, but this last only ~50 yr! 20
How many PeVatrons? A statstcal approach [Cristofari, Gabici et al., MNRAS (2013)] Assume that SNRs are the source of CRs: how many TeV sources we do expect? HESS sample l<40 ; b <3 ; fux(>1tev)>1.5% Crab 35 obj. - 3 shell SNRs - 3 SNR/MC system - 12 pulsar - 17 unidentified 3<#SNRs< 20 21
How many PeVatrons? [Cristofari, Gabici et al., MNRAS (2013)] Assume that SNRs are the source of CRs: how many TeV sources we do expect? HESS sample l<40 ; b <3 ; 35 obj. fux(>1tev)>1.5% Crab - 3 shell SNRs - 3 SNR/MC system - 12 pulsar - 17 unidentified 3<#SNRs< 20 Model assumptions: CR luminosity = 1041 erg/s SN explosion rate= 3/century ηcr~10% SNR composition [Ptuskin et al.(2010)]: Ia(32%), IIP(44%),Ib/c(22%), IIb(2%) Spectrum slope α=4.1 4.4 Kep= 10-2 10-5 Magnetic feld amplifcation 22
How many PeVatrons? [Cristofari, Gabici et al., MNRAS (2013)] Assume that SNR are the source of CRs: how many TeV sources we do expect? HESS sample l<40 ; b <3 ; 35 obj. fux(>1tev)>1.5% Crab - 3 shell SNRs - 3 SNR/MC system - 12 pulsar - 17 unidentified Model assumptions: 3<#SNRs< 20 Obtained with 1000 Montecarlo realizations CR luminosity = 1041 erg/s SN explosion rate= 3/century ηcr~10% SNR composition [Ptuskin et al.(2010)]: Ia(32%), IIP(44%),Ib/c(22%), IIb(2%) Spectrum slope α=4.1 4.4 Kep= 10-2 10-5 Magnetic feld amplifcation 17<#SNR< 22 4<#SNR< 7 23
How many PeVatrons? [Cristofari, Gabici et al., MNRAS (2013)] Assume that SNR are the source of CRs: how many TeV sources we do expect? HESS sample l<40 ; b <3 ; 35 obj. fux(>1tev)>1.5% Crab - 3 shell SNRs - 3 SNR/MC system - 12 pulsar - 17 unidentified Model assumptions: 3<#SNRs< 20 Obtained with 1000 Montecarlo realizations CR luminosity = 1041 erg/s SN explosion rate= 3/century ηcr~10% SNR composition [Ptuskin et al.(2010)]: Ia(32%), IIP(44%),Ib/c(22%), IIb(2%) Spectrum slope α=4.1 4.4 Kep= 10-2 10-5 Magnetic feld amplifcation 17<#SNR< 22 4<#SNR< 7 These SNRs are PeVatrons only for age < few 100 yrs Number of observable PeVatrons in the same sample compatible with 0 in the worst case 24
SN 1987 A (Type IIp SN) Example of very young SNR compatible with theoretical requirement to produce PeV protons HESS upper limit after 210 hr of observation: F(>1TeV) < 5 10-14 cm-2 s-1 Wpp < 9 1 0 48 erg ~ 1% ESN Berezhko et al.(2011) predict F(>1TeV) ~ 8 10-14 cm-2 s-1 In the next decade the SNR should impact the equatorial rind whose density is estimated n~103 104 cm-3 Dwarkadas (2013) predict a γ-fux: F(>1TeV) ~ 3 10-12 cm-2 s-1 for f(e) E -2 F(>1TeV) ~ 3 10-14 cm-2 s-1 for f(e) E -2.6 From Berezhko et al.(2011) 25
γ-ray emission from SNR/MC system Free-escape CC SN born inside star forming region boundary expected connection between SNRs and MCs Forward Shock 1) Crushed Cloud model Invokes a shocked MC into which a radiative shock is driven by SNR's blast wave. valid for middle age SNRs (W44, IC443, W28) we expect very steep spectra low shock speed cutoff ~ 1-10 TeV low acceleration efficiency Runaway CRs NOT GOOD FOR PEVATRONS Crushed Cloud 26
γ-ray emission from SNR/MC system Free-escape CC SN born inside star forming region boundary expected connection between SNRs and MCs Forward Shock 1) Crushed Cloud model Invokes a shocked MC into which a radiative shock is driven by SNR's blast wave. valid for middle age SNRs (W44, IC443, W28) we expect very steep spectra low shock speed cutoff ~ 1-10 TeV low acceleration efficiency Runaway CRs NOT GOOD FOR PEVATRONS 2) Runaway CR model Considers γ-ray emission from MCs illuminated by CRs escaping from the accelerator. Crushed Cloud Valid also for very young SNRs 27
γ-ray emission from SNR/MC system Free-escape boundary For a typical SNR at 1 kpc distance and a MC mass of 104 M detectable level of TeV emission if nsource,cr > ngal,cr this happen when the cloud is located at d <~ 100 pc from the SNR (for 3D diffusion model) Forward Shock Runaway CRs 28
γ-ray emission from SNR/MC system Free-escape boundary For a typical SNR at 1 kpc distance and a MC mass of 104 M detectable level of TeV emission if nsource,cr > ngal,cr this happen when the cloud is located at d <~ 100 pc from the SNR (for 3D diffusion model) Forward Shock Runaway CRs the distance can be enhanced to d <~500 pc if we consider the 1-D propagation along magnetic field line the source can be observable for ~ 104 yr B Simulation from Nava & Gabici(2012) 29
γ-ray emission from SNR/MC system Free-escape boundary For a typical SNR at 1 kpc distance and a MC mass of 104 M detectable level of TeV emission if nsource,cr > ngal,cr this happen when the cloud is located at d <~ 100 pc from the SNR (for 3D diffusion model) Forward Shock Runaway CRs the distance can be enhanced to d <~500 pc if we consider the 1-D propagation along magnetic field line the source can be observable for ~ 104 yr B The predicted spectrum is not universal, can be very different according to the moment of observation: 1) if the propagation distance is inside the cloud 2) if propagation distance > Rcloud Rc 0 D τ 0 Q s (E ) 4 π r dr Q s( E ) τ 4 πr D 2 Qs ( E ) Qs ( E ) 4π r dr Rc 4πr D D( E ) 2 30
SUMMARY Few shell-type SNRs already detected in TeV gamma-rays are good candidates to be PeVatrons (or close to) The ASTRImini-array already has the capability to investigate these cases PeVatrons are statistically difcult to observe: assuming that SNRs are indeed the source of Galactic CRs up to the knee, the number of possible observable PeVatrons with current instrument can be very low (even compatible with 0 in the worst scenario) From the theoretical point of view the most favorable candidate for PeVatrons are remnant originated from type II SN exploding in red supergiant wind during the frst 50-100 yrs. This support the difculty of detection. A possible way to enhance the probability to detect a PeVatron is the indirect observation in γ-ra ysfrom molecular cloud illuminated by escaping PeV particles. These sources could be observable for a much longer time (up to ~104 yr) if MCs are located in the vicinity of a sources at < 500 pc 31