Shell supernova remnants as cosmic accelerators: II Stephen Reynolds, North Carolina State University I. Observational tools II. Radio inferences III. X ray synchrotron emission IV. MeV GeV observations V. Ground based TeV observations VI. Confrontation of observations and theory VII. Future prospects
Observational tools Radio: Interferometers for adequate spatial res. (~ few arsec arcmin) VLA, Australia Telescope, MOST VLA (NRAO/AUI/NSF) Tycho's SNR, 1.4 GHz X rays: Require spatially resolved spectroscopy; not easy before ASCA satellite. Chandra, XMM Newton have angular res. ~ 1 15, energy res. E/ E ~ 10 15 between 0.3 10 kev Chandra (CXC) Tycho's SNR (CXC)
Gamma ray tools MeV GeV: EGRET aboard Compton Gamma Ray Observatory 30 MeV 20 GeV; angular res: ~ few arcmin up to a degree Some low latitude unidentified sources likely to be SNRs
Gamma ray tools, II INTEGRAL: 17 kev 10 MeV Coded mask imaging: pt. sources to 12 (extended sources: not so good) ESA AGILE: 30 MeV 50 GeV, source localization to ~ 15 arcmin (also hard X ray imager, 18 60 kev, resolution ~ few arcmin) ESA Both optimized for transients, point source studies
GLAST Two orders of magnitude higher sensitivity than EGRET; angular resolution ~ arcmin GLAST All Sky Simulation
Ground based TeV observations: Imaging Air Čerenkov Telescopes (IACTs) Incident photons produce air showers; optical Čerenkov light is detected from secondaries. CANGAROO III Japanese Australian telescope Sensitive above ~ 300 GeV; angular resolution depends on elevation but can approach 10' H.E.S.S. telescope (Namibia)
Observational summary Radio: ~260 Galactic SNRs observed; radio spectral slopes X ray: 4 Galactic synchrotron dominated SNRs; ~6 others with synchrotron components (all historical shells) MeV GeV (EGRET): No absolutely certain detections; 20 candidates based on positional coincidence TeV: Detections of 3 of 4 X ray synchrotron dominated SNRs, 1 historical SNR
Inferences from radio observations ~ 260 radio supernova remnants in Milky Way Electrons radiating near 1 GHz have energies in GeV range: E = 14.7 ( (GHz)/B( G))1/2 GeV. Spectral indices ~ 0.5 (but with much scatter) In young SNRs, brightness contrast with diffuse synchrotron background is far too great for electrons, B to be borrowed from surrounding ISM, compressed in shock (with r ~ 4). New electrons and/or B are required. Spectrum of synchrotron background is wrong for borrowing electrons: newly accelerated electrons are required.
X ray synchrotron radiation in SNRs 1. X ray spectra dominated by synchrotron emission: SN 1006: archetype G347.3 0.5 (RX J1713.7 3946) G266.2 1.2 ( Vela Jr. ) G1.9+0.3: new youngest SNR 2. Synchrotron components: thin rims usually Historical shells Kepler, Tycho, G11.2 0.3 (SN 386) Cas A, RCW 86 3. TeV detections of SNRs with X ray synchrotron SN 1006: new! G347.3 0.5, Vela Jr. Cas A, RCW 86
SN 1006 AD Brightest supernova ever seen from Earth (~ quarter moon) 0.5 0.8 kev X rays 2 4.5 kev X rays (Almost) featureless integrated X ray spectrum. Limbs: Hard, featureless. Center: lines. Explanation: synchrotron radiation in limbs. RXTE data XMM images (CEA/ESA); lower right, VLA (NRAO) ASCA GIS integrated spectrum (Dyer et al. 2001)
Spectral fitting in SN 1006 Spectrum from radio to X rays can be fit with simple model of an electron distribution with exponential cutoff due to one of three energy limitations: finite age (or size), change in diffusion, or losses (electrons only): N(E) = KE s exp( E/Emax). Early model: fits radio, X ray data; makes ICCMB prediction. Only fitted parameter: rolloff frequency related to Emax. Combines physical parameters from appropriate cutoff mechanism. Estimate Emax ~ 80 (B/10 G) 1/2 TeV
Fine structure in SN 1006 NE edge (full resolution). Sharp outer edge, thin filaments Chandra (CXC). Red: below 0.7 kev. How can rims be so thin? (1) Electrons disappear (synchrotron losses) (2) Magnetic field disappears (wave damping)
Magnetic field amplification If thickness of rims is due to synchrotron losses, electrons must be depleted rapidly: for observed rim widths w ~ 0.01 pc, need B > 200 (vshock/1000 km/s)2/3 (w/0.01 pc) 2/3 G (without some amplification process, just expect to compress typical interstellar B ~ 3 G by compression ratio r ~ 4). Similar rims seen in other SNRs (but some rims are thin in radio as well!)
Caveat: thin radio rims High resolution VLA image of SN 1006 (12" x 7"); Moffett et al. 1993 High pass filtered image: shows edges and filaments Right: full image. Note thin feature on profile
Thin radio filaments in other remnants too: Tycho (Reynoso et al. 1997: VLA, 20 cm, resolution 1.4") N rim: width < 10 pixels (4") Most of periphery shows thin radio rim at location of thin X ray rim Synchrotron losses can't affect lower energy electrons; damping of magnetic turbulence downstream? (Pohl, Yan, & Lazarian 2005)
G347.3 0.5 (RX J1713.7 3946) Chandra closeup of NW (Lazendić et al. 2004) Featureless X ray spectrum, ~ 2.4. Severe upper limits on thermal gas (n < 0.1 cm 3) ROSAT image (Slane et al. 1999)
TeV detection and modeling H.E.S.S. ACT mapping and spectroscopy (Aharonian et al. 2007)
Broadband modeling of G347.3 0.5 Hadronic process: evidence for relativistic protons (Berezhko & Völk 2006) (but does required thermal gas exceed X ray limits?) Model with ICCMB: but require very small magnetic field filling factor (~0.01) (Lazendić et al. 2004)
G266.2 1.2 (Vela Jr.) ASCA (Slane et al. 2001) Featureless X ray spectrum ( ~ 2.6; Slane et al. 2001); thermal gas density < 0.03 cm 3 H.E.S.S. (color); X rays (white contours) (Aharonian et al. 2007)
Broadband modeling of Vela Jr. Uncertain distance makes modeling difficult (200 pc? 1 kpc?) Larger distance: remnant age is large, shocks slow, probably can't produce enough CR's unless SNR expanded rapidly in a bubble first (but now encounters denser material). Similar problem to G347: either 0 decay or ICCMB can describe observed broadband spectrum with effort. Again, may be too few target thermal protons where TeV emission is observed (no thermal X rays) but problems with ICCMB parameters as well.
G1.9+0.3: the youngest Galactic SNR Radio (VLA 1985) Chandra (2007) (platelet smoothed) Angular diameter ~ 100''. Radio flux (1 GHz) ~ 0.9 Jy; ~ 0.65. Discovered in search for young SNRs (Green & Gull 1984).
Radio X ray Expansion 1985 radio + 2007 X ray 1985 scaled radio + 2007 X ray Infer expansion by (16 ± 3)% in 22 yr age 140 m yr (R t m) (Cas A: outer blast wave has m ~ 0.7; here that gives age ~ 100 yr)
New radio observations 1985 (20 cm) 2008 (6 cm) Expansion: (15 ± 2)%, consistent with X rays. 20 cm flux: Increase by factor 1.25 ± 0.09 in 13 yr
X ray spectroscopy Lineless spectrum. High absorption: NH = (5.5 ± 0.3)x 1022 cm 2. Background modeled, not subtracted. Model: srcut (synchrotron from power law with exponential cutoff). Rolloff h roll = 5.8 (2.9, 14.5) kev, highest reported for integrated spectrum. (Caveat: high absorption will require better modeling of dust scattering) Fourth Galactic X ray synchrotron dominated SNR
Other historical shells Kepler (1604): some thin rims, diffuse regions, blobs, mainly on outer periphery. Rolloff frequencies give Emax ~ 40 120 TeV Red: Total. Green: nonthermal continuum. (Reynolds et al. 2007) Tycho: thin rims have featureless spectra also; similar Emax Blue: hard continuum (Warren et al. 2005)
More on historical shells Integrated spectra of Kepler, Tycho, Cas A, SN 1006 all show power law continua above 8 kev (Allen et al. 1999) (Could be nonthermal bremsstrahlung, above common line energies) Cas A: also thin rims (outer blast wave?) Chandra (Stage et al. 2006)
RCW 86: SN 185 AD? SW corner: combination of thermal, synchrotron NE: almost pure synchrotron H.E.S.S. detection (Hoppe et al. 2007); requires large energy density in cosmic ray ions for 0 decay explanation. XMM + Chandra (orange: lower energy X rays) `
Magnetic field amplification 1. Upper limits on > 100 MeV flux from Cas A bound electron population from above (bremsstrahlung), so require stronger B to account for observed (bright!) synchrotron flux. B > 800 G (Cowsik & Sarkar 1980) 2. Integrated X ray flux from SN 1006 (synchrotron or nonthermal) requires bending of spectrum from radio; if due to losses, B > 200 G (Tycho: B > 400 G) (R. & Chevalier 1981) 3. X ray synchrotron thin rims require at least local B > 100 G or so (but is B in waves which damp, or does it become non zero mean?) 4. TeV detections or upper limits all bound electron density from above (ICCMB) so bound B from below to explain synchrotron. Get (larger scale volume averaged) B > 30 G (SN 1006), > 100 G (Tycho), > 15 G (G347.3 0.5), > 6 G (Vela Jr.) 5. Hadronic models for TeV emission tend to require high B
Rapid X ray variability Small features seen to brighten or fade in ~ 1 yr in Cas A (Patnaude & Fesen 2007), G347.3 0.5 (Uchiyama et al. 2007) If this is timescale of particle acceleration, need high B: accel /(vshock)2 where is diffusion coefficient, 1/B Get B ~ 1 mg (Uchiyama et al. 2007) If fading is due to synchrotron losses, similar result Chandra observations of Cas A (Patnaude & Fesen 2007)
Summary of deductions from observations 1. Observation of synchrotron X rays confirms the presence of electrons at energies up to 100 TeV but spectrum must steepen from radio value below 100 TeV. Losses? Escape?
Summary of deductions from observations 1. Observation of synchrotron X rays confirms the presence of electrons at energies up to 100 TeV but spectrum must steepen from radio value below 100 TeV. Losses? Escape? 2. Diffusive shock acceleration can easily explain observed electron energies with observed shock properties. (2nd order may be too slow)
Summary of deductions from observations 1. Observation of synchrotron X rays confirms the presence of electrons at energies up to 100 TeV but spectrum must steepen from radio value below 100 TeV. Losses? Escape? 2. Diffusive shock acceleration can easily explain observed electron energies with observed shock properties. (2nd order may be too slow) 3. No direct evidence for cosmic ray ions yet. Leptonic models can (so far) explain all TeV detections (with some difficulty), but hadronic models can as well
Summary of deductions from observations 1. Observation of synchrotron X rays confirms the presence of electrons at energies up to 100 TeV but spectrum must steepen from radio value below 100 TeV. Losses? Escape? 2. Diffusive shock acceleration can easily explain observed electron energies with observed shock properties. (2nd order may be too slow) 3. No direct evidence for cosmic ray ions yet. Leptonic models can (so far) explain all TeV detections (with some difficulty), but hadronic models can as well 4. Magnetic field amplification of some kind seems required. Not clear if high B persists downstream. High B means fewer electrons required (trouble explaining cosmic ray electrons?).
Summary of deductions from observations 1. Observation of synchrotron X rays confirms the presence of electrons at energies up to 100 TeV but spectrum must steepen from radio value below 100 TeV. Losses? Escape? 2. Diffusive shock acceleration can easily explain observed electron energies with observed shock properties. (2nd order may be too slow) 3. No direct evidence for cosmic ray ions yet. Leptonic models can (so far) explain all TeV detections (with some difficulty), but hadronic models can as well 4. Magnetic field amplification of some kind seems required. Not clear if high B persists downstream. High B means fewer electrons required (trouble explaining cosmic ray electrons?). 5. Observed broadband spectra can be fit without extreme requirements for acceleration efficiency, but can't yet determine it (or e/p ratio)
Future prospects: GLAST to the rescue GLAST energy range Leptonic model for TeV emission Hadronic model Different models for G347.3 0.5 make different predictions in GLAST energy range (sensitive to bremsstrahlung, ICCMB, 0 decay) (Aharonian et al. 2007)
More modeling is required! Observations can be explained by DSA, but predictions are still scarce! 1. What is the efficiency of shock acceleration? e/p ratio? 2. How does the process depend on shock speed, shock obliquity, neutral fraction...etc.? 3. How does magnetic field amplification work? How does that process depend on shock speed,... etc.? Even GLAST won't solve all problems. To model synchrotron emission, need magnetic field information; to model bremsstrahlung and 0 decay emission, need detailed information on thermal material (behind or, for 0 decay, even in front of shock wave)
More data are required too! Lower energy thresholds for IACTs will increase sensitivity X ray polarimetry! Learn about orientation of B; find synchrotron X rays among thermal emission Monitoring of X ray emission at Chandra resolution: how common are small scale changes? New missions for hard X ray/soft gamma ray region: bremsstrahlung should be directly observable (but need capability to image diffuse sources!)