A source of gamma rays coincident with the shell of the supernova remnant CTB 80
MMNRAS , 1–7 (2020) Preprint 12 January 2021 Compiled using MNRAS L A TEX style file v3.0
A source of gamma rays coincident with the shell of the supernovaremnant CTB 80
M. Araya ID , (cid:63) C. Herrera, Centro de Investigaciones Espaciales and Escuela de F´ısica, Universidad de Costa Rica
Accepted —. Received —; in original form —
ABSTRACT
CTB 80 (G69.0+2.7) is a relatively old (50–80 kyr) supernova remnant (SNR) with acomplex radio morphology showing three extended radio arms and a radio and X-raynebula near the location of the pulsar PSR B1951+32. We report on a study of the GeVemission in the region of CTB 80 with
Fermi -LAT data. An extended source with a sizeof 1.3 ◦ , matching the size of the infrared shell associated to the SNR, was discovered.The GeV emission, detected up to an energy of ∼
20 GeV, is more significant at thelocation of the northern radio arm where previous observations imply that the SNRshock is interacting with ambient material. Both hadronic and leptonic scenarios canreproduce the multiwavelength data reasonably well. The hadronic cosmic ray energydensity required is considerably larger than the local Galactic value and the gamma-ray leptonic emission is mainly due to bremsstrahlung interactions. We conclude thatGeV particles are still trapped or accelerated by the SNR producing the observedhigh-energy emission when interacting with ambient material.
Key words: gamma rays: ISM – ISM: individual (CTB 80) – ISM: supernova remnants
CTB 80 was first suggested to be an SNR based on its struc-ture and strong polarization (Velusamy & Kundu 1974).The radio spectrum of the core was observed to be flatand to steepen away from the core (Angerhofer et al. 1981).Kulkarni et al. (1988) discovered the 39.5 ms pulsar PSRB1951+32 in the region. This pulsar has a spin-down energy-loss rate ˙ E = 10 . erg s − . Fesen et al. (1988) discovereda 1 ◦ -diameter infrared shell centered 30 (cid:48) east of the pulsarwith an average hydrogen density of 3 cm − . The projectedlocation of this shell suggests that the SNR produced boththe shell and the pulsar. Many more observations in radio,the optical and X-rays of the source were carried out follow-ing the first studies (e.g., Junkes et al. 1988; Whitehead et al.1989; Salter et al. 1989; Hester & Kulkarni 1989; Mavro-matakis et al. 2001). Koo et al. (1990) found an H I shellconsistent with an SNR shell with a dynamical age of 77 kyr d , where d is the source distance in units of 2 kpc. Thisshell was found to match the infrared shell. It appears thatthe pulsar has caught up with the SNR shell, which may haveproduced the peculiar radio morphology with its interactionwith the magnetic field in the shell. The H I shell was laterfound to contain clumps with core densities of n H ∼ − surrounded by a more diffuse envelope (with averagedensity n H ∼ − ), and the interstellar medium (ISM) (cid:63) E-mail: [email protected] around CTB 80 was seen to be very inhomogeneous (Kooet al. 1993). Safi-Harb et al. (1995) detected an extendednebula of X-ray emission consistent with synchrotron radia-tion near PSR B1951+32.The distance to CTB 80 is still not very clear. For ex-ample, based on the dispersion measure of PSR B1951+32 avalue of 1.4 kpc has been derived (Kulkarni et al. 1988), andmore recently, a distance of 4 . ± . ±
18 kyr, assuming a source distanceof 2 kpc, according to Migliazzo et al. (2002). Later, moreprecise observations of the pulsar proper motion yielded akinetic age of 51 kyr (assuming a distance of 2 kpc to thesource, Zeiger et al. 2008). Based on these studies we adopta reference distance of 2 kpc for CTB 80 in this work.High resolution radio observations revealed faint exten-sions of the arms of CTB 80 and an excellent radio-infraredcorrespondence along the northern arm (Castelletti et al.2003). The “curling” observed in the northern arm in radioand infrared may have been caused by the shock interactingwith the dense clumps described earlier and this scenario isalso consistent with optical observations (Castelletti et al.2003). A study of radio spectral index variations along thearms of the SNR revealed a consistent softening of the spec-trum away from the pulsar (Castelletti & Dubner 2005).The spectral steepening was found to be smooth along theeastern arm while the northern and southwestern arms show © a r X i v : . [ a s t r o - ph . H E ] J a n Araya & Herrera locally flatter structures, which coincide with optical, radioand infrared enhancements. This feature was interpreted byCastelletti & Dubner (2005) as a result of the combinationof old relativistic electrons injected by PSR B1951+32 andparticles accelerated at the sites where the shock of the SNRencounters the inhomogeneities of the ambient medium.Radio observations by Leahy & Ranasinghe (2012) re-vealed an outer slowly-moving H I shell with a radius of 76arcmin and a velocity of 40 km s − . The shell is consistentwith the cool dense shell expected in the “snowplough” phaseof an SNR. The authors estimate an age for the shell of 60kyr. They also found extended X-ray emission associatedwith CTB 80 over a 1 ◦ . ∼ . ◦ ) impliedby observations of Mavromatakis et al. (2001).The average radio spectral index found by Castelletti& Dubner (2005) is α = − . ± .
02 for the whole SNR( F ν ∝ ν α ), in agreement with the measurements by Man-tovani et al. (1985), while Kothes et al. (2006) found α = − . ± .
03. Recent microwave observations by Planck im-ply the presence of a considerable steepening of the syn-chrotron spectrum at higher frequencies (Planck Collabora-tion et al. 2016) which might be caused by electron cooling.Observations of the CTB 80 region at higher (GeV) en-ergies by the Energetic Gamma Ray Experiment Telescope(EGRET) revealed emission from the pulsar PSR B1951+32(Ramanamurthy et al. 1995), which was later detected bythe
Fermi
Large Area Telescope (LAT, Abdo et al. 2010).Besides the pulsar, the latest catalog of LAT sources, theFermi Large Area Telescope Fourth Source Catalog (4FGL,Abdollahi et al. 2020), shows a point source possibly associ-ated to the SNR CTB 80, labeled 4FGL J1955.1+3321. It isfound ∼ . ◦ north east of PSR B1951+32 at the locationof the northern radio arm of the SNR shell.In this work we report the discovery of GeV emission ex-tending across the SNR CTB 80 using data from the Fermi -LAT. In section 2 we describe the LAT data analysis and insections 3 and 4 we present the multi-wavelength data ob-tained from the literature and the model of the non-thermalemission.
Data gathered from August 2008 to June 2020 were analyzedwith the publicly available software fermitools version 1.2.23and the package fermipy version 0.19.0 (Wood et al. 2017).The instrument response functions P8R3 SOURCE V2 wereused and standard recommended cuts applied. We selectedSOURCE class events in front and back interactions, in thereconstructed energy range 0.3–500 GeV. The maximumzenith angle chosen was 90 ◦ to avoid contamination fromgamma rays produced in the Earth’s limb and time inter-vals were selected when the data quality was good, filteringevents collected while passing the South Atlantic Anomalyand other low-quality events. A spatial binning scale of 0.05 ◦ per pixel and ten logarithmically spaced bins per decade inenergy for exposure calculation were used. Events from aregion of interest (ROI) with a radius of 15 ◦ centred at thecoordinates RA=298 . ◦ , Dec=33 . ◦ (J2000) were includedin the analysis.The model of the region included the sources from the 4FGL catalog located within 20 ◦ of the ROI centre. Thesource 4FGL J1955.1+3321, possibly associated to CTB 80,was removed from the model to carefully study the emissionin the region. The source 4FGL J2005.8+3357, located about2 . ◦ from the ROI centre, was also removed and replacedwith the extended source associated to 2HWC J2006+341,as described by Albert et al. (2020). The Galactic diffuseemission was described by the file gll_iem_v07.fits andthe residual background and extragalactic (isotropic) emis-sion by the file iso_P8R3_SOURCE_V2_v1.txt . The energydispersion correction was applied to all components of themodel except for the isotropic template. The maximum like-lihood technique (Mattox et al. 1996) was used to fit the freeparameters of the model in order to maximize the probabil-ity of the model to account for the data. The significance ofa new source with one free parameter was estimated withthe square root of the test statistic (TS), which is defined as − × log( L / L ), with L and L the values of the maximumlikelihoods for models without the source (the null hypoth-esis) and with the additional source, respectively. The TSvalue was also used to choose the best description of a sourcespectrum from a set of fits with nested functions.The analysis consisted of a morphological and a spec-tral characterization of the emission. For the morphologicalstudies, only events with energies above 1 GeV were used inorder to take advantage of the improved point spread func-tion of the LAT at higher energies. Only the spectral normal-ization of the 4FGL sources located within 10 ◦ of the ROIcentre and the normalizations of the diffuse and isotropiccomponents were left free in the fits, while the other spectralparameters were fixed to the values reported in the 4FGLcatalog. The null hypothesis was optimized first by search-ing for new point source candidates in the ROI having aTS >
25 with the find_sources algorithm of fermipy , andusing a power-law spectrum. The normalization and spectralindex of the new sources were also fitted initially. Differentspatial models were used to find the best morphology of theemission in the region of CTB 80, as described below.Once the morphology of the emission is found the bestspectral description of the source was searched using eventswith energies above 300 MeV. We repeated the search fornew point source candidates having a TS >
25 in this energyrange and optimized their normalizations and spectral in-dices. In the fits, besides the normalizations of the sourceslocated within 10 ◦ of the ROI centre and those of the diffuseand isotropic components, also the other spectral parame-ters of the sources located within 2 ◦ from the centre of theROI, were left free to vary. The effect of the uncertainty in the model of the Galactic dif-fuse emission on the source parameters was estimated by re-peating the fits using the eight alternative models developedby the LAT Collaboration in their search for high-energyemission from supernova remnants (Acero et al. 2016). Thefiles released by the LAT Collaboration were scaled to ac-count for differences in energy dispersion between Pass 7 These files are distributed as part of the fermitools athttps://github.com/fermi-lat/fermitools-data .MNRAS , 1–7 (2020) source of gamma rays coincident with CTB 80 reprocessed data and Pass 8 data . The uncertainties wereestimated as in Acero et al. (2016) for both the spectral pa-rameters of the global fit to CTB 80 as well as for the individ-ual spectral energy distribution (SED) flux points obtained.For the SED points these systematic errors were comparableto the statistical errors below an energy of 4 GeV, while thestatistical uncertainties dominated above this energy. Theuncertainties in the LAT effective area were propagated ontothe global spectral parameters and SED flux normalizationsusing a set of bracketing response functions as recommendedby Ackermann et al. (2012). In the fits with the bracketingresponse functions the pivot energy value, estimated withthe covariance error matrix of the global fit, was used. The optimized model for the null hypothesis using eventswith energy above 1 GeV was used to create a significancemap of the ROI which shows the residual emission. A close-up image of the map in the CTB 80 region is shown in Fig.1. To make the image the TS value of a test point sourcewith a spectral index of 2.0 was evaluated in each spatialbin. The position of 4FGL J1952.9+3252 (included in themodel) is marked in the map and corresponds to the pul-sar PSR B1951+32, associated to CTB 80. The radio con-tours plotted in the figure were taken from the 4850 MHzGB6/PMN survey (Condon et al. 1994) and show the threecharacteristic arms of the SNR. The map clearly shows thepresence of significant emission at GeV energies, mainly atthe location of the northern radio arm but also extendingtowards the south of CTB 80.In order to characterize the morphology of the emissionseveral hypotheses were tested in the fits: a point source, a2D Gaussian template, a uniform disk template and a mapof the radio emission. For the radio map, data from the GB6survey at 4850 MHz was used (Condon et al. 1994). The ra-dio emission from the pulsar wind nebula (PWN) associatedto PSR B1951+32 was removed with a 8 (cid:48) × (cid:48) mask, a sizegiven by Castelletti et al. (2003), as well as the emissionfrom a point source that is clearly visible at the coordinatesRA = 299 ◦ .
05, Dec = 33 ◦ .
31 using a mask with a radiusthat is somewhat larger than the point-source response in-dicated in the survey ( ∼ . (cid:48) ). Changing the sizes of themasks has no effects on the results shown here. The spectralmodel used in the fits was a power law whose normaliza-tion and spectral index were free to vary. The choice of thisspectral shape is justified below. The spatial and spectralparameters of the different models were fitted and Table 1shows the results. The Akaike Information Criterion (AIC,Akaike 1974), defined as AIC = 2 k − L ) where k is thenumber of parameters and L the maximum likelihood, wascalculated to identify the best model. In the table the valueof ∆AIC ≡ AIC k − AIC min is given, which is the differenceof the AIC of each model k and the one that minimizes theAIC (∆AIC = 0 for the best available model).The Gaussian and disk morphologies provide similarly See https://fermi.gsfc.nasa.gov/ssc/data/access/lat/Model details/Pass8 rescaled model.html.
Table 1.
Results of the morphological analysis of
Fermi -LAT dataabove 1 GeV.
Morphology Size a ( ◦ ) RA ( ◦ ) Dec ( ◦ ) ∆ AIC
Point source - 298 . ± .
03 33 . ± .
03 36.9Disk 0 . ± .
04 298 . ± .
04 32 . ± .
04 0Gaussian 0 . ± .
04 298 . ± .
05 32 . ± .
05 0.4Radio template - - - 36.4 a Radius for the disk and σ for the gaussian. Right ascension ( ) D ec li n a t i o n ( ) J2000 J Figure 1.
TS map of the gamma-ray emission above 1 GeV oncethe background sources and diffuse emission are subtracted. Thecolor scale is in units of TS. The cross indicates the location ofthe source 4FGL J1952.9+3252, associated to the pulsar PSRB1951+32. The contours represent the radio emission of CTB80 in six equally-spaced intervals from 0.002 to 0.202 Jy/beam(taken from a GB6 survey, Condon et al. 1994). The dashed (solid)circle represents the 68% containment radius and its statisticaluncertainty of the best-fit disk (gaussian) template of the GeVsource while the X marks the corresponding centroid. good representations of the data. The 68%-containment sizesfor both the best-fit disk and Gaussian are shown in Fig. 1,as well as their corresponding 1 σ -statistical uncertainties.Even though these spatial shapes provide the best descrip-tion of the GeV emission among the models compared, theactual photon distribution is clearly more complex, as canbe seen in Fig. 1. Although the residual significance mapsobtained after adding the source, for both the disk andGaussian morphologies, were the most satisfactory amongthe tested models, a 3 . σ excess was still seen at the loca-tion of the peak of the emission in Fig. 1 in both cases. Amore detailed study of the morphology should be carriedout in the future. The likelihood ratio between the best-fitpoint source (ps) and the best-fit disk (ext) hypotheses wasTSext ≡ × log( L ext /L ps) = 88. The threshold to preferthe extended source hypothesis over a single point sourcehas been set as TSext >
16 (Ackermann et al. 2017), basedon simulations with LAT data showing that the cumulativedensity of TSext follows a χ distribution with one degreeof freedom (Lande et al. 2012). The emission found is statis-tically significantly extended. The radius of the best-fit disk MNRAS000
16 (Ackermann et al. 2017), basedon simulations with LAT data showing that the cumulativedensity of TSext follows a χ distribution with one degreeof freedom (Lande et al. 2012). The emission found is statis-tically significantly extended. The radius of the best-fit disk MNRAS000 , 1–7 (2020)
Araya & Herrera found is 0 ◦ . ± ◦ .
04. We note that the emission centroidsof the best-fit disk and, particularly, the Gaussian, are notlocated at the position of the PWN (the statistical uncer-tainties in the position are smaller than the marker sizes inFig. 1), and there are no other known pulsars in the region.We also note that the disk size and location match well thosecorresponding to the SNR shell seen in the infrared (see alsoFig. 3 below). We note, however, that a detailed model of thepulsar emission during its lifetime should be explored andcould perhaps explain the GeV morphology as originatingfrom leptons from the pulsar which are now located in theshell of the SNR. In what follows we use the disk templateto represent the GeV emission from CTB 80.
Using the best-fit disk found in section 2.2.1, fits with eventsin the entire energy range considered (0.3–500 GeV) weredone with different spectral shape models. Two models werecompared, a power law spectrum, dNdE = N ( EE ) − Γ , and alog-parabola, given by dNdE = N ( EE b ) − ( α + β log( E/E b )) , where E and E b are fixed scales. The difference in log L betweenthe two models is ∼ .
6, and therefore the power law is cho-sen as the final spectral shape for the GeV emission fromCTB 80, since the log-parabola does not improve signifi-cantly the fit with respect to the power law. This justifiesthe use of this spectral shape in the search for the best-fitmorphological template above. The overall TS value of thebest-fit disk is 349.7, corresponding to a detection signifi-cance of 18 σ above 300 MeV.Setting E = 10 MeV, the values of the best-fit spectralindex and normalization are Γ = 2 . ± . stat ± . sys ,and N = (1 . ± . stat ± . sys ) × − MeV − cm − s − . The SED of the new extended GeV source in the CTB80 region was obtained by dividing the energy range in 10logarithmically spaced bins and fitting the normalization ofthe emission in each bin using the best-fit morphologicalmodel obtained. The spectral index value of the disk repre-senting CTB 80 was fixed to 2 in each energy bin and thenormalization of the sources located within 2 ◦ of the cen-tre of the ROI (as well as the normalization of the disk andthe diffuse and isotropic components) were free to vary. In asimilar fashion, the SED points of the pulsar PSR B1951+32were also obtained for comparison. The resulting SEDs areseen in Fig. 2. The source associated to CTB 80 is not signif-icantly detected above ∼
20 GeV and 95%-confidence levelupper limits on the fluxes were calculated in the bins wherea TS was lower than 4. A TS value of 6.6 was obtained forCTB 80 in the energy bin 113-237 GeV but it is not clearif this is a statistical fluctuation or indication of spectralhardening, and a more detailed study on this is left for thefuture.The SED flux points from CTB 80 were calculated inan identical manner replacing the standard diffuse emissionmodel with the eight alternative emission models describedearlier and the systematic error on each point was calcu-lated. These errors were added in quadrature to the errorresulting from propagation of the effective area uncertain-ties in the normalization, estimated to be ∼ E (MeV) − − − E d N d E ( M e V / c m s ) PSR B1951+32CTB 80
Figure 2.
Gamma-ray SEDs of the SNR CTB 80 and the pulsarPSR B1951+32 obtained in this work. The circles represent thefluxes from CTB 80 while the stars those from the pulsar. Theshaded region represents the propagated 1 σ -statistical error bandof the global fit for CTB 80. The flux points show statistical andsystematic errors added in quadrature. resultant errors were also added in quadrature to the statis-tical uncertainties for each SED point. Fig. 3 shows an infrared image of the CTB 80 region ob-tained with data from the AKARI far-infrared all-sky sur-vey (Doi et al. 2015; Takita et al. 2015). The image partlyreveals the infrared shell associated to the SNR (Fesen et al.1988). The gamma-ray (TS) contours obtained in this workare plotted as well as the radio contours shown previously.The peak of the GeV emission is seen in the northern armof the SNR at the location where there is enhancement ofthe infrared emission (near the coordinates RA =298 . ◦ ,Dec=33 . ◦ ). This infrared enhancement has been noticed toperfectly match the radio emission and is consistent withshock-heated dust with a temperature ∼
26 K (Castellettiet al. 2003). The interaction of the shock with dense gascould also be responsible for the curving of the northern ra-dio arm. It is also worth noting that the infrared shell partlyseen in the image is consistent with the location and size ofthe gamma-ray disk found here (the circle in Fig. 3).We used the radio fluxes from CTB 80 from the liter-ature as listed by Castelletti et al. (2003) and added themore recent Planck measurements in the microwave (PlanckCollaboration et al. 2016) which are useful to constrain thehighest energy in the electrons. We excluded the radio fluxesbelow a frequency of 200 MHz from the model, as they couldbe affected by free-free thermal absorption (Castelletti &Dubner 2005).
The radio to GeV non-thermal fluxes were modeled usingone-zone leptonic or hadronic scenarios with the naima pack-
MNRAS , 1–7 (2020) source of gamma rays coincident with CTB 80 Right ascension ( ) D ec li n a t i o n ( ) J J Figure 3.
Infrared image of CTB 80 at 65 µ m from the AKARI far-infrared survey. The color scale is in MJy/sr. The radio contoursshown in Fig. 1 are also shown here (solid, cyan), the (dashed)magenta contours represent the TS values in Fig. 1 at the levelsof 8, 25, 42, 59 and 76, while the solid (blue) circle representsthe disk found at GeV energies. The peak of the GeV emission islocated in the northern arm of the SNR where an enhancement ofthe IR emission is also seen. The circular infrared shell associatedto the SNR can be partly seen in this image at the location of itsnorthern radio arm. age (Zabalza 2015). For the hadrons we used a particle dis-tribution which is a power law, which is enough to explainthe data, while the lepton distribution is described by apower law with an exponential cutoff, where the cutoff de-fines the maximum attainable energy of the leptons. The lep-tonic gamma-ray emission is produced by the inverse Comp-ton scattering (IC) and bremsstrahlung processes. For theIC calculation, we used three photon background fields, thecosmic microwave background (CMB) a far-infrared (FIR)photon field and stellar optical and near-infrared (NIR) pho-tons. From the bolometric infrared luminosity of the SNR,2 . × erg s − (Fesen et al. 1988), an estimation of theenergy density in the FIR photon field at the surface of theshell yields ∼ .
01 eV cm − . This value is negligible com-pared to the Galactic FIR energy density at a Galactocentricdistance of 8 kpc, ∼ .
35 eV cm − (Shibata et al. 2011). Weadopt an FIR component produced by a modified black bodywith a temperature of 26 K and an energy density of 0 .
35 eVcm − . The NIR photon field is similarly described with anenergy density of 0.7 eV cm − and a temperature of 2000 K,comparable to the local (Solar System) values (Shibata et al.2011). As will be seen, the results do not depend strongly onthese parameters. To calculate the bremsstrahlung and piondecay fluxes, we fixed the ambient target density to 3 cm − ,consistent with the average hydrogen density measured inthe infrared shell (Fesen et al. 1988), part of which is seento coincide with the location of the peak of the gamma raysin Fig. 3. We fixed the particle spectral index to 1.72 as expected fromradio observations (Castelletti & Dubner 2005) and adjusted − − − − − − − E (MeV) − − − E d N d E ( M e V / c m s ) IC (CMB)IC (FIR)IC (NIR)SynchrotronBremsstrahlungTotal gamma-ray flux
Figure 4.
Leptonic model for the non-thermal emission from theSNR CTB 80. The gamma-ray data include both statistical andsystematic errors. the other parameters. The resulting model and the data areshown in Fig. 4. The required total energy for electrons withenergies above 1 GeV is 2 × erg, for a source distanceof 2 kpc. Although this distance is uncertain, this energy isonly ∼ .
2% of the typical kinetic energy available in SNRs.Therefore, the total energy required would still be reason-able for a wide range of reasonable distances. The particleenergy cutoff found is 15 GeV, and the resulting magneticfield is 13 µ G. As can be seen in Fig. 4, the synchrotronmeasurements at the highest energies are in tension withthe predicted fluxes, but we note that this could be causedby the simplicity of the one-zone model used. It is also clearthat the dominant contribution to the gamma rays comesfrom bremsstrahlung emission. For a typical particle energyof 10 GeV and a magnetic field of 13 µ G, the synchrotronloss time is of the order of 10 yr (e.g., Schure et al. 2010).Since this time is much larger than the age of the system,our choice of the particle distribution for the electrons isjustified, as no synchrotron cooling break is expected. The fit to the multiwavelength data from CTB 80 wherethe GeV emission is dominated by hadronic interactions isshown in Fig. 5. The proton energy distribution used is apower law and the resulting particle spectral index is 2 . n =3 cm − (Fesen et al. 1988), a total energy content in thehadrons of W p = 2 . × erg is required, which is ∼ . × erg and21 µ G, respectively. The ambient particle density used forcalculation of the bremsstrahlung fluxes was also 3 cm − .Considering a volume V in space occupied by the high-energy particles equal to the volume of the SNR (approxi-mated as a sphere of radius 20 pc), the average energy den-sity in these particles would be W p /V ≈
16 eV cm − , whichis much greater than the local energy density in Galactic cos-mic rays. This implies that an enhancement of high-energyhadrons above the ISM numbers would be present at the MNRAS000
16 eV cm − , whichis much greater than the local energy density in Galactic cos-mic rays. This implies that an enhancement of high-energyhadrons above the ISM numbers would be present at the MNRAS000 , 1–7 (2020)
Araya & Herrera − − − − − − − E (MeV) − − E d N d E ( M e V / c m s ) IC (CMB)IC (FIR)IC (NIR)SynchrotronBremsstrahlungPion decayTotal gamma-ray flux
Figure 5.
Hadronic-dominated model for the GeV emission fromthe SNR CTB 80. The gamma-ray data include both statisticaland systematic errors.
SNR, which is consistent with the expected scenario wherethe SNR accelerates (or traps) cosmic rays.As with the leptonic scenario explained above, wherethe gamma rays are mostly attributed to bremsstrahlungemission, a scenario where hadronic processes contributesubstantially to the gamma-ray emission is also consistentwith the fact that this emission is more significant at thelocation of the northern arm of the SNR. We recall that theIR enhancement to the north of the SNR, which is seen inFig. 3, perfectly matches the radio morphology. The radiospectral variations and morphology of this region indicatethat there is interaction of the shock with denser ambientmedium (Castelletti & Dubner 2005), which would then pro-duce substantial gamma rays, and this explains the GeVmorphology seen in Fig. 1. The shock interaction could alsoproduce compression amplifying the magnetic field and in-creasing the cosmic ray energy density.
We discovered a new extended GeV source matching the lo-cation and size of the SNR CTB 80 using data from the
Fermi -LAT. The spectrum of the source is best describedby a power law above an energy of 300 MeV, with a spec-tral index of 2 . ± . stat ± . sys . We have shown thatthe GeV emission is enhanced in the northern radio arm ofthe SNR where the shock is believed to be interacting withmaterial which is seen in the infrared. This fact is consis-tent with our spectral modeling which attributes the GeVemission to either pion decay, resulting from interactions ofprotons with matter, or to bremsstrahlung from high-energyelectrons which also interact with matter. A more realisticscenario may include contributions from both types of par-ticles, as shown in Fig. 5, since the presence of synchrotron-emitting electrons and target material inevitably results inemission of bremsstrahlung at gamma-ray energies. The re-quired energy density in the cosmic rays is greater than theenergy density in Galactic cosmic rays seen locally at Earth. ACKNOWLEDGEMENTS
We thank the anonymous referee for the valuable commentsand for a thorough revision of the manuscript. This work waspossible due to funding by Universidad de Costa Rica and itsEscuela de F´ısica under grant number B8267. This researchis based on observations made with NASA’s Fermi Gamma-Ray Space Telescope, developed in collaboration with theU.S. Department of Energy, along with important contri-butions from academic institutions and partners in France,Germany, Italy, Japan, Sweden and the U.S.
DATA AVAILABILITY.
The data underlying this article are available in the FermiScience Support Center, at https://fermi.gsfc.nasa.gov/ssc/.The derived data generated in this research will be sharedon request to the corresponding author.
REFERENCES
Abdo A. A., et al., 2010, ApJ, 720, 26Abdollahi S., et al., 2020, ApJS, 247, 33Acero F., et al., 2016, ApJS, 224, 8Ackermann M., et al., 2012, ApJS, 203, 4Ackermann M., et al., 2017, ApJ, 843, 139Akaike H., 1974, IEEE Transactions on Automatic Control, 19,716Albert A., et al., 2020, The Astrophysical Journal, 903, L14Angerhofer P. E., Strom R. G., Velusamy T., Kundu M. R., 1981,A&A, 94, 313Castelletti G., Dubner G., 2005, A&A, 440, 171Castelletti G., Dubner G., Golap K., Goss W. M., Vel´azquez P. F.,Holdaway M., Rao A. P., 2003, AJ, 126, 2114Condon J. J., Broderick J. J., Seielstad G. A., Douglas K., Gre-gory P. C., 1994, AJ, 107, 1829Doi Y., et al., 2015, PASJ, 67, 50Fesen R. A., Shull J. M., Saken J. M., 1988, Nature, 334, 229Hester J. J., Kulkarni S. R., 1989, ApJ, 340, 362Junkes N., F¨urst E., Reich W., 1988, Detection of Four SupernovaRemnants in the Polarized Emission from the Effelsberg 11-CENTIMETER Survey. p. 134, doi:10.1007/3-540-50435-4 18Koo B.-C., Reach W. T., Heiles C., Fesen R. A., Shull J. M., 1990,ApJ, 364, 178Koo B.-C., Yun M.-S., Ho P. T. P., Lee Y., 1993, ApJ, 417, 196Kothes R., Fedotov K., Foster T. J., Uyanıker B., 2006, A&A,457, 1081Kulkarni S. R., Clifton T. C., Backer D. C., Foster R. S., FruchterA. S., 1988, Nature, 331, 50Lande J., et al., 2012, ApJ, 756, 5Leahy D. A., Ranasinghe S., 2012, MNRAS, 423, 718Mantovani F., Reich W., Salter C. J., Tomasi P., 1985, A&A, 145,50Mattox J. R., et al., 1996, ApJ, 461, 396Mavromatakis F., Ventura J., Paleologou E. V., PapamastorakisJ., 2001, A&A, 371, 300Migliazzo J. M., Gaensler B. M., Backer D. C., Stappers B. W.,van der Swaluw E., Strom R. G., 2002, ApJ, 567, L141Planck Collaboration et al., 2016, A&A, 586, A134Ramanamurthy P. V., et al., 1995, ApJ, 447, L109Safi-Harb S., Ogelman H., Finley J. P., 1995, ApJ, 439, 722Salter C. J., Reynolds S. P., Hogg D. E., Payne J. M., RhodesP. J., 1989, ApJ, 338, 171Schure K. M., Achterberg A., Keppens R., Vink J., 2010, MN-RAS, 406, 2633 MNRAS , 1–7 (2020) source of gamma rays coincident with CTB 80 Shan S. S., Zhu H., Tian W. W., Zhang M. F., Zhang H. Y., WuD., Yang A. Y., 2018, ApJS, 238, 35Shibata T., Ishikawa T., Sekiguchi S., 2011, ApJ, 727, 38Takita S., et al., 2015, PASJ, 67, 51Velusamy T., Kundu M. R., 1974, A&A, 32, 375Whitehead M. J., Meaburn J., Clayton C. A., 1989, MNRAS, 237,1109Wood M., Caputo R., Charles E., Di Mauro M., Magill J.,Perkins J. S., Fermi-LAT Collaboration 2017, in 35th In-ternational Cosmic Ray Conference (ICRC2017). p. 824( arXiv:1707.09551 )Zabalza V., 2015, Proc. of International Cosmic Ray Conference2015, p. 922Zeiger B. R., Brisken W. F., Chatterjee S., Goss W. M., 2008,ApJ, 674, 271This paper has been typeset from a TEX/L A TEX file prepared bythe author.MNRAS000