Supernova Remnant Studies with Fermi LAT

Supernova Remnant Studies with Fermi LAT J.W. Hewitt Univ. of North Florida on behalf of the Fermi LAT collaboration TeVPA 2017 Columbus, OH August 7, 2017

Fermi Larger Area Telescope (LAT) Large Area Telescope (LAT) : 20 MeV to >300 2 TeV, GeV (including unexplored region -0 GeV GeV region 0 GeV) 2.4 sr field of view 2.4 scans sr field the of entire viewsky (scans every the ~3 hrs entire sky every ~3 hrs) Gamma-ray Burst Monitor (GBM) : 8 8 kev to 40 40 MeV MeV views Views entire entire unocculted sky sky 4 Aug 2017 marks 9 yr of continuous science data! 2

GLAST LAT Project SSI Jul 29, 2005 SNR: Sites of Hadronic Acceleration? Supernova Remnants: sites of galactic cosmic ray acceleration Non-thermal emission (X-rays) Signs of γ-ray activity EGRET: GeV activity near SNR (3EG J1714-3837) HESS: Recent detection of TeV γ-rays Question: Do γ rays originate from hadronic or leptonic processes? Measurements in the range of 0MeV-0GeV essential ingredient to resolve the origin (p vs e+/-) Uchiyama (2003) Adapted from Aharonian s talk at the Texas Symposium 2004 Supernova remnant RX J1713.7-3946 HESS 04 (color) + ASCA (contours) HESS data π 0 model e +/- model GLAST E. do Couto e Silva 16

Has Fermi reached its promise? (Re: SNRs) The LAT s excellent spatial and energy resolutions will separate the extended shell emission of an SNR from a compact source (pulsar, tiny plerion) inside it. Systematic searches have cataloged XX SNRs with detectable extension, allowing secure identifications It will also spectrally resolve electron and nuclei emission. Pion cutoff detected in 4 SNRs (with potential for more) The LAT will resolve > remnants, to establish the location of cosmic ray production. Pass 8 data allows spectra-morphological studies of bright SNRs, at a resolution comparable to best TeV telescopes 4

Has Fermi reached its promise? (Re: SNRs) The LAT s excellent spatial and energy resolutions will separate the extended shell emission of an SNR from a compact source (pulsar, tiny plerion) inside it. Systematic searches have cataloged XX XX SNRs with detectable extension, allowing secure identifications It will also spectrally resolve electron and nuclei emission. Pion cutoff detected in in 4 4 SNRs (with potential for for more) The LAT will resolve > remnants, to establish the location of cosmic ray production. Pass 8 8 data allows spectra-morphological studies of of bright SNRs, at at a a resolution comparable to to best TeV TeV telescopes 4

Spectrally resolving nuclei emission 4 SNR/MCs show π 0 -decay spectral cutoff at <300 MeV. Spatial correlation between γ-rays and dense MC interaction region IC/Bremsstrahlung models require internal break in electron spectrum not seen in radio W49B W51C 5

Young SNRs Tycho & Cas A Hard spectrum can be explained by π 0 -decay dominated emission Cassiopeia A Tycho s SNR LAT collaboration, in prep. V/c c 6

Limitations for Low Energy Studies Confusion is strong at low energies (average angular separation of 3FGL sources is 2.2 o outside the Galactic plane) Galactic interstellar emission model is not perfect (see Acero et al. 2016, ApJS 223, 26) Small residuals (2-3%) impact sources at the same level as statistical errors over the whole Galactic plane Low energy spectral studies are limited by diffuse systematics 0 - Latitude Longitude 160 80 40-40 -80-160 -5 Sigma units 5 3 7

1SC: First SNR Catalog with LAT https://fermi.gsfc.nasa.gov/ssc/data/access/lat/1st_snr_catalog/ Search 279 Galactic SNRs (Green 2009) for 1-0 GeV γ-rays and account for systematics using 3 yr of Pass 7 data Use spatial overlap to classify 32 likely GeV SNRs (16 extended) /22/2014 8

1SC: First SNR Catalog with LAT https://fermi.gsfc.nasa.gov/ssc/data/access/lat/1st_snr_catalog/ Include multi wavelength information (distance, density, etc.) to explore differing γ-ray SNR types: young TeV shells and evolved interacting SNRs Extended Pointlike Classified Marginal Upper Limits (i=2.5, 99%) ULs, interacting (i=2.5, 99%) ULs, young (i=2.5, 99%) 9

Energetics under the SNR Paradigm

Evaluating Diffuse Systematics https://fermi.gsfc.nasa.gov/ssc/data/access/lat/1st_snr_catalog/ Developed 8 alternative Interstellar Emission Models using GALPROP to vary CR source distribution (SNR/Lorimer), halo height (4/ kpc) and HI spin temperature (150K/optically thin) Warning: Models do not span the complete systematic uncertainty! They are hard, soft, point-like (x) and extended (o) sources and they are located in regions with different intensities of the IEM. 11

Evaluating Diffuse Systematics https://fermi.gsfc.nasa.gov/ssc/data/access/lat/1st_snr_catalog/ No particular correlation with the sky position is found for ratio of the alternative IEM systematic and statistical errors on flux Need better knowledge of diffuse backgrounds to improve studies of SNRs at ~1 GeV (and lower) energies 12

SNRs searches with Pass 8 data High energy searches for extended sources (SNRs, PWNe, etc) are motivated by lower diffuse backgrounds and improved PSF. LAT PSF 68% Containment Angle 13

LAT now a higher-energy γ-ray telescope Pass8 data Low energy spectral studies limited by: Knowledge of diffuse background (5% precision) Source density extrapolated from measured logn-logs Hard sources limited by count rate (improves faster than t -1/2 ) 14

Fermi s Hard Source Catalogs Preliminary 2FHL Adaptively smoothed 360 sources at E>50 GeV in 80 months of Fermi-LAT data (~61,000 photons) Ackermann et al. (2016) 15

Fermi s Hard Source Catalogs 3FHL See P. Saz Parkinson s talk Gamma rays, Aug, 15:00 Adaptively smoothed 1,556 sources at E> GeV in 84 months of Fermi-LAT data (~700,000 photons) Ajello et al. (2017), submitted to ApJS 16

Extended Source Catalog Fermi-LAT Galactic Extended Source Catalog FGES contains 46 extended sources Contact Authors: J. Cohen, M.H. Grondin, E.A. Hays, M. Lemoine-Goumard 16 new GeV extension detec6ons 17 previous detec6ons diﬀer in FGES detects 46 extended sources morphology (Ackermann, et al. 2017, ApJ, 843, 2) 16 newly measured GeV extensions Preliminary Disk sizes 0.12 to 2.5 (radius) Preliminary 17 previous detections differ in morphology Integrated into 3FHL 13 SNR + 17 snr + 9 spp 7 August 2017 TeVPA 2017: Galactic Sources 5 Cygnus Cocoon SNR PWN 17

FGES Analysis Method 80 months of Pass 8 data; GeV to 2 TeV energy γ rays Scan the Galactic plane (±5 o ) using overlapping regions and two independent analysis pipelines as a cross-check main pipeline cross-check Test candidates for position, extension, alternative hypotheses (2 pt. sources vs 1 ext. source) and spectral curvature Extended sources are those with: TS > 25; TSext > 16; TS2pts < TSext TS 2pts = 2 log(l 2pts /L ps ) TS ext = 2 log(l ext /L ps ) 18

Revealing energy-dependent morphology Example: SNR G8.7-0.1 (W30 region) Previous lower-energy LAT measurements (Ajello, et al. 2012) match to radio extent while higher-energies correspond to TeV (HESS J1804-216; Aharonian, et al. 2006) Very large for PWN analog of Vela X? SED TS Map + Source Model TS map + MW FGES model Previous model (3FGL) TeV contours 19

Motivating new TeV observations Example: CTB 9 Age : 8800 14000 yrs; Central magnetar. First detection of GeV extension (pointlike in Castro et al. 2012) Good agreement with X-ray/Radio size: 0.25 ± 0.02 stat SED TS Map + Source Model TS map + MW FGES model Previous model (3FGL) ROSAT X-rays ~0.3% Crab at 1 TeV 20

To find interesting upper limits Puppis A is γ-ray SNR with PL index of 2.1 (Hewitt, et al. 2012) Radio spectrum shows break in e - synchrotron ~40 GHz HESS observations show high energy TeV cutoff 21

Puppis A: proto-interacting SNR/MC <4,000 yr old SNR with extended GeV γ-rays and interacting with dusty ISM (n~4 cm -3 ) near a molecular cloud. Not unlike SNR/MCs, but younger with far lower density. GeV diameter = 0.76±0.08 from Arendt, et al. (20) IR emission from shocked dust as Puppis A expands towards a molecular cloud (CO contours). Interaction with a dense clump is indicated. 22

Finding new types of γ-ray SNRs? SNR G150.3+4.5 is a high-latitude, large diameter SNR Identified in 2FHL, but only with increased statistics is there agreement with radio size. Large size suggest old, nearby remnant but hard spectrum is not seen for other such older SNRs (no energetic PSR to power PWN) SED TS Map + Source Model TS map + MW FGES model 3FGL 2FHL model Radio (Gao & Han, 2014) Large SNRs (such as G150.3+4.5) are difficult to detect at TeV 23

SNR G150.3+4.5: An unusual γ-ray SNR HI spectra shows clouds at d = 0.38 to 5.6 kpc (Perseus arm, d = 2 kpc) SNR/MCs Cas A Puppis A Tycho RX J1713 G150.3+4.5? Low γ-ray luminosity similar to shell TeV SNRs? Similar candidate γ-ray SNR detected by Fermi High-Latitude Extended Source Catalog (FHES) with large angular extent (~3 o ) See M. Meyer s FHES talk Gamma rays: 9 Aug, 14:45 and hard spectral index (Γ~1.8) 24

Joint LAT / TeV / HAWC Studies HAWC is ideal to find SNRs like G150.3+4.5 (large, hard spectrum) 2HWC found 19 sources more than 0.5 o away from any known TeV counterpart See Nahee Park s talk Galactic, Aug, 16:30 GeV/TeV follow-up (VERITAS, MAGIC) -2 Energy (TeV) s -1 Integrated flux (cm ) 2 11 VERITAS energy range Fermi energy range HAWC 75% central energy range HAWC w/ point source assumption HAWC w/ larger radius VERITAS, point analysis VERITAS, extended analysis 1 12 1 13 2 J0700+143 J0700+143 J40+308 J40+308 J1852+013* J1852+013* J1902+048* J1902+048* J1907+084* J1907+084* J1914+117* J1914+117* J1928+177 J1928+177 J1938+238 J1938+238 J1949+244 J1949+244 J1953+294 J1953+294 J1955+285 J1955+285 J2006+341 J2006+341 25

New detection of PWN G54.1+0.3 Not detected by 3FHL, but updated LAT analysis (8.5 yr) detects source matching VERITAS and HAWC TeV sources γ ray data well-matched to leptonic models of the PWN. s -1 ) -2 dn/de (erg cm 2 11 Fermi - this study VERITAS - this study 2HWC J1930+188, PL 2HWC J1930+188, PL + Cut-off E s -1 ) -2 dn/de (erg cm 2 11 Li et al (20), lepton model C Li et al (20), lepton + hadron Tanaka et al (2011), model 2 Torres et al (2014) Gelfand et al (2015) E 12 12 13 13 11 12 13 Energy (ev) 14 Energy (ev) 26 11 12 13 14

Composite SNR G326.3-1.8 38 arcmin SNR shell PWN PSR Sedan phase SNR with energetic PSR+PWN (modeled by Temim, et al. 2013) LAT Pass 8 data can spatially separate hardspectrum PWN from soft-spectrum SNR shell Requires spatial resolution, 3-decade energy reach and multi-wavelength templates work by J. Devin, F. Acero, J. Schmid, J. Ballet, on behalf of the Fermi LAT collaboration (in prep.) 27

Young TeV Shell SNR RCW 86 IR + shock + NT X-rays Fermi LAT significance map RCW 86 (One-zone model) HESS excess counts Synchrotron Inverse Compton (CMB) Bremsstrahlung Pion decay TOTAL RCW 86 (~1830 yr) is only fit by leptonic models due to hardness of GeV emission Two-zone model constrain WCR,p < 4x 49 erg; n = 1 cm -3 SW is TeV bright / GeV faint RCW 86 (Two-zone model) s -1 ) -2 dn/de (erg cm 2 E 11 12 13 Ajello, et al. 2016, ApJ, 819, 98 Synchrotron Inverse Compton (CMB) Bremsstrahlung Pion decay TOTAL 14 6 4 2 2 4 6 8 12 14 1 Energy (ev) 28 Fig. 9. Spectral Energy Distribution of RCW 86 with the best-fit leptonic one-zone (top) and two-zone (bottom) models. Th points are from Molonglo at 408 MHz and Parkes at 5 GHz (Caswell etal.1975;lemoine-goumardetal.2012)andlowerlimi 15

Competing models for Interacting SNRs Surrounding environment determines CR acceleration efficiency, escape and γ-ray emission interpretation Two scenarios: 1. Crushed clouds: CRs + MC compressed Re-acceleration of GCRs e.g., Uchiyama+ (20), Tang & Chevalier (2014) 2. Illuminated clouds: CRs escape, passively interact with cloud e.g., Gabici+ (2007), Casanova+ (20) 29

Competing models for Interacting SNRs Spectra of 11 SNR/MCs show striking similarity expect more variation if illuminated by escaping CRs? Interaction with dense gas leads to acceleration of existing ambient CRs, instead of thermal seed particles in young SNRs from Tang, arxiv:1707.00958 TeV-bright SNR/MCs may require reaccel of freshly accelerated CRs, or harder ambient CR spectrum than locally Fresh DSA requires ~33% shock energy, while re-accel needs only ~1%(Lee et al. 2015) 30

Molecular line observations indicate SNR shock interacting w 4 M Interacting 443(e.g. in depth southern ridge Lee+ 2008) sun The~1.1x Astrophysical Journal, 749:34along (14pp), 2012 April SNR IC (a) Schematic of IC 443 from Lee, et al. (2008) γ-ray spectral turnover matches what 23 00 is expected from π0-decay ck Abdo, et al. 2014, ApJ, 786, 145 Radio Continuum Halo Declination (J2000) n io C A molecular shock B Bremsstrahlung π0-decay 22 40 IC 443 HCO + ic o sh Clear detection o (b) 22 20 CR acceleration Turnoverefficiency matches1-4% what is e (23 pc) 22 00 (uncertainty in mass of clouds) h m 6 19m Bremsstrahlung 6h 18m 6h 17m 6h 16 Best-ﬁt mod ~200 GeV/c but VHE shows breakright Ascension (J2000) decline Figure 1. (a) Schematic showing the overall morphology of IC 443 adopted from LEE08, with the nomenclatures for shells overlaid. The asterisk indi of the pulsar (Olbert et al. 2001). (b) 21 cm radio continuum image of IC 443 (LEE08) with the area covered by the FCRAO observations: thin solid p August 2017 polygon for HCO+ and HCN. TheTeVPA 2017:are Galactic Sources of shocked 12 CO (see Figure 4 for details). The shock31 contours distribution clumps (J = 1 0) and 7thick dotted by Denoyer (1979a) and Dickman et al. (1992) are indicated.

Molecular line observations indicate SNR shock interacting w IC 443: very well-studied SNR 4 IC 443: A well-studied SNR Interacting SNR IC 443 in depth Msun southern ridge (e.g. Lee+ 2008) The~1.1x Astrophysical Journal, 749:34along (14pp), 2012 April (a) Clear detection o (b) γ-ray spectral turnover matches what Evolved (radiative) SNR interacting with a molecula Middle-aged (20,000 is yr?) SNR from interacting 23 00 expected π0-decaywith molecu Schematic of IC 443 from Lee, et al. (2008) ck Abdo, et al. 2014, ApJ, 786, 145 n io Bremsstrahlung 3-color π0-decay 22 40 A molecular shock WISE IR image of shocked dust Declination (J2000) Radio Continuum Halo C B IC 443 HCO + ic o sh 22 20 CR acceleration Turnoverefficiency matches1-4% what is e (23 pc) 22 00 (uncertainty in mass of clouds) h m 6 19m Bremsstrahlung 6h 18m 6h 17m 6h 16 Best-ﬁt mod ~200 GeV/c but VHE shows breakright Ascension (J2000) Radio decline Figure 1. (a) Schematic showing the overall morphology of IC 443 adopted from LEE08, with the nomenclatures for shells overlaid. The asterisk indi 20cm of the pulsar (Olbert et al. 2001). (b) 21 cm radio continuum image of IC 443 (LEE08) with the area covered by the FCRAO observations: thin solid p August 2017 polygon for HCO+ and HCN. TheTeVPA 2017:are Galactic Sources of shocked 12 CO (see Figure 4 for details). The shock31 contours distribution (J = 1 0) and 7thick dotted from Troja et al. (2006) clumps by Denoyer (1979a) and Dickman et al. (1992) are indicated.

IC 443 Resolved by LAT and VERITAS 7 yr Fermi LAT Pass 8 data 178 hr of VERITAS data PRELIMINARY PRELIMINARY Counts Map >5 GeV (PSF23) Significance Map 32

IC 443 Resolved by LAT and VERITAS 7 yr Fermi LAT Pass 8 data 178 hr of VERITAS data PRELIMINARY PRELIMINARY shocked HCO+ gas Counts Map >5 GeV (PSF23) Significance Map Excellent correlation observed for GeV, TeV and shocked gas is inconsistent with escaping CR scenario 32

IC 443 Resolved by LAT and VERITAS Broken PL fits for all 4 regions: Γ1 ~ 2.3, Γ2 ~ 2.9, Eb ~ 60 GeV PRELIMINARY 178 hr of VERITAS data PRELIMINARY 4 3 1 2 No clear differences in spectral shape for distinct emission regions (e.g. dense cloud in region 1 vs. fast atomic shock in region 4) Observing the same CR spectra across different environments is inconsistent with re-acceleration in crushed clouds 33

Conclusions LAT provides the best γ-ray census of Galactic SNRs New catalog searches: 3FHL, FGES, FHES Pass 8 + new diffuse tools will allow low-energy spectral studies for more than just the few brightest SNRs Spatially resolved spectral studies at high energies (+TeV obs) Differentiate SNR emission from PWN / PSR. Discriminate b/w interaction & escape models for SNR-MCs Resolve varying physical conditions in young TeV shells Fermi LAT continues to identify new targets for TeV telescopes Joint studies with HAWC, HESS, MAGIC, VERITAS CTA? 34

Supernova Remnant Studies in 2018? From S. Funk (2016) HEAD 15 proc. 35

HESS Galactic Plane Survey Donath et al, TeVPA 2016 3FHL (preliminary, > GeV) Dominguez et al, TeVPA 2016 Better resolution and higher stats in HESS Galactic Plane Survey Most sources extended, 60% do not have one obvious MW counterpart PWN brighter at TeV energies, old SNRs brighter at GeV energies 37 15

Resolving the γ-ray morphology of IC 443 Multi-wavelength comparison shows the GeV/TeV γ rays match the distributio of shocked gas in IC 443 LAT morphology compared to TeV, radio, ambient CO, shocked HCO+ PRELIMINARY shocked atomic gas in North? (a) negative ND N1 Jy/Beam km/s N2 Figure 8. (a) Distribution of shocked H i in the NE part of IC 443 for HI absorption 40±4 Msun (Castelletti, et al. 2011) Counts Map >5 GeV (PSF23) H + gas has ne=-00 cm -3 (Rho, et al. 2001) 38

Two nearby molecular clouds Foreground molecular cloud cuts across SNR. RGB image shows vlsr = -2,-4,-6 km/s against Radio contours +5 km/s cloud ends at TeV peak 23 00 23 00 22 45 22 45 δ2000 22 30 δ2000 22 30 22 15 22 15 6 h 19 m 6 h 18 m 6 h 17 m α 2000 6 h 16 m 6 h 15 m 6 h 19 m 6 h 18 m 6 h 17 m α 2000 6 h 16 m 6 h 15 m 12 Figures from Lee+ 2008 Figure 15. Far-IR 90 µm image taken with the AKARI satellite shown in gray scale. The green contours show the distribution of +5 km s 1 clouds (the gray scale in Figure 9). The blue contours show locations of SCs. The solid and dashed circles represent the location of γ -ray sources detected by MAGIC and VERITAS, respectively. 39

IC 443: SNR interacting with Molecular Clouds Molecular line observations show shock interaction with ~1.1x 4 Msun along southern ridge (e.g. Lee, et al. 2008) (a) (b) 23 00 C ionic shock Radio Continuum Halo A molecular shock B Declination (J2000) 22 40 22 20 HCO + 12 CO 22 00 (23 pc) 6 h 19 m 6 h 18 m 6 h 17 m 6 h 16 m Right Ascension (J2000) 6 h 15 m 40

Catching Escaping Cosmic Rays? SNR W44 Uchiyama+ 2012, ApJ 749, L35 Subtracting the SNR (radio model) reveals emission at 2-0 GeV (SRC-1,2) coincident with nearby CO complex 41

Catching Escaping Cosmic Rays? SNR W44 Uchiyama+ 2012, ApJ 749, L35 Subtracting the SNR (radio model) reveals emission at 2-0 GeV (SRC-1,2) coincident with nearby CO complex CR diffusion on scales ~0 pc? => 3x more WCR in MC than in SNR Or uncertainties in Galactic diffuse model? 41