GeV emission in the region of the supernova remnant G51.26+0.09
AAstronomy & Astrophysics manuscript no. g51 © ESO 2021February 18, 2021
GeV emission in the region of the supernova remnant G51.26+0.09
M. Araya ID Escuela de Física, Universidad de Costa Rica, San Pedro de Montes de Oca, 11501-2060 San José, Costa Ricae-mail: [email protected]
Received , 2021; accepted , 2021
ABSTRACT
Context.
The study of the gamma-ray emission in the GeV and TeV energy range allows us to understand the processes of particleacceleration and the origin of cosmic rays. When associated to supernova remnants (SNRs), the study of this radiation, complementedwith observations across the electromagnetic spectrum, is also important to estimate physical parameters of these sources, which areusually poorly known. As an example, the SNR G51.26 + Aims.
We study the GeV emission seen in the direction of G51.26 + + Methods.
Using data from the
Fermi
Large Area Telescope we perform a detailed study of the GeV emission in the direction ofG51.26 + CO emission andneutral hydrogen emission to probe the environment, searching for possible morphological features associated to the gamma rays andSNR.
Results.
We rule out the star-forming region G051.010 + + + + + Key words.
ISM – cosmic rays – supernova remnants – gamma rays
1. Introduction
Supernova remnants (SNRs) play an important role in theGalaxy. They heat up the interstellar medium (ISM), influencestar formation, distribute heavy elements and accelerate cos-mic rays. Given the supernova rate expected and seen in theGalaxy (e.g., van den Bergh & McClure 1994) several thousandSNRs should be observed and yet only a few hundred are known(Green 2019). Selection e ff ects such as the lack of sensitive radiocontinuum observations are thought to be the cause of the dis-crepancy. New candidate SNRs have been discovered recentlywith the increasing number of observations across the electro-magnetic spectrum, for example The HI, OH, Recombinationline survey of the Milky Way (THOR, Anderson et al. 2017),has found 76 candidate SNRs.SNRs accelerate particles to relativistic energies, which canthen be detected through their gamma radiation. At high (GeV)and very high energies (TeV) source catalogs by the Fermi
LargeArea Telescope (LAT, Acero et al. 2016; Abdollahi et al. 2020),the High Energy Stereoscopic System (H. E. S. S. Collaborationet al. 2018b) and the High-Altitude Water Cherenkov Observa-tory (Albert et al. 2020) reveal SNRs as well as several unasso-ciated sources which could be previously unknown SNRs. Thestudy of the gamma-ray emission from SNRs is important to un- derstand the properties of the particle populations that they ac-celerate but also to constrain SNR parameters such as distanceand ambient density, which are usually poorly known.G51.21 + ∼ (cid:48) and a fluxdensity of 24 . ± .
10 Jy at a frequency of 1.4 GHz. Followup radio observations by Driessen et al. (2018) revealed a com-plex morphology and a radio spectral index α = − . ± .
21 forG51.21 + F ν ∝ ν α where ν is the fre-quency), making it a good SNR candidate. They were not ableto confirm its nature due to the lack of data at other wavelengthsthat are relevant for SNR studies, particularly in the X-ray band.Using radio observations at lower frequencies, Supan et al.(2018) claimed that a small southern portion of G51.21 + ff erent SNR, which they labeled G51.04 + + . ± . + + Article number, page 1 of 10 a r X i v : . [ a s t r o - ph . H E ] F e b & A proofs: manuscript no. g51 of it at the location of the originally proposed SNR candidateG51.21 + + + . (cid:48) . There are no observations of the regionin the X-ray band.The Fermi
Large Area Telescope Fourth Source Catalog andits associated Data Release 2 (4FGL, Abdollahi et al. 2020; Bal-let et al. 2020) show two point sources having no association inthe region of G51.26 + + + Fermi in this region to obtain its mor-phology and spectrum. The possible connection of the gammarays to any SNR, known or otherwise, the star forming regionG051.010 + CO (J = →
0) observations from The Boston University-Five Col-lege Radio Astronomy Observatory Galactic Ring Survey (GRS,Jackson et al. 2006) to trace the molecular hydrogen, and atomichydrogen (HI) data from the HI4PI survey (HI4PI Collaborationet al. 2016). We model the gamma-ray spectrum using physi-cal processes to derive the parameters of the relativistic particlesresponsible for the emission under the one-zone approximationand use all this information to constrain its origin. Fermi -LAT observations and data analysis
The LAT onboard the
Fermi satellite continually scans the skyand detects photons in the energy range from about 20 MeVto more than 300 GeV. We used data collected between the be-ginning of the mission, August 2008, to December 2020. Weincluded events in the energy range from 200 MeV to 500GeV reconstructed within 15 ◦ of the center of our region ofinterest (ROI), located at the coordinates RA = . ◦ , Dec = . ◦ (J2000). This is approximately the reported position ofthe source 4FGL J1925.4 + fermitools version2.0.0, a publicly available software for treating Fermi data, andthe open-source
PYTHON package fermipy version 1.0.0, to per-form the analysis. We applied the recommended quality cuts for
PASS8 data analysis, including a zenith angle cut of 90 ◦ to avoidgamma-ray contamination from Earth’s limb, DATA_QUAL>0 and
LAT_CONFIG==1 . The analysis used the
P8R3_SOURCE_V3 in-strument response functions and we combined back and front-converted events in the
SOURCE class (using the parameters evtype=3 and evclass=128 ). We adopted the binned maxi-mum likelihood method (Mattox et al. 1996) to derive the spec-tral and morphological parameters of the sources. The signifi-cance of a new source with one additional parameter with respectto the model without the source (known as the null hypothesis)can be estimated with the square root of the test statistic, TS = − × ln( L / L ), where L and L are the values of the max-imum likelihoods for the null hypothesis and for a model withthe additional source, respectively.To model the background we included the sources in the ROIlisted in the 4FGL catalog along with the recommended Galac-tic and isotropic di ff use emission components, described by thefiles gll_iem_v07.fits and iso_P8R3_SOURCE_V3_v1.txt ,respectively, and provided with the fermitools . The en-ergy dispersion correction was applied as recommended by the LAT team . The sources 4FGL J1925.4 + + In order to take advantage of the improved spatial resolu-tion of the LAT at higher energies, we performed a study ofthe morphology of the gamma-ray emission in the region ofG51.26 + ◦ of the ROI center simultaneously aswell as all the spectral parameters of the sources found within 3 ◦ of the ROI center. As a second step, we searched for additionallocal TS maxima likely associated to new uncatalogued sourcesin the ROI with the find_sources routine of fermipy .A TS map of the null hypothesis was constructed by fit-ting the normalization of a test point source with a power lawspectral model with an index of 2 in each position in the mapand the result is seen in Fig. 1. The image shows the loca-tions and sizes of the SNR G51.26 + + + ff erent morphological models were fitted to the GeV emis-sion to choose the best available representation. In order to dothis, the Akaike Information Criterion (AIC, Akaike 1974) wascalculated for each case. This is defined as AIC = k − L ),where k is the number of free parameters in the model and L themaximum likelihood obtained in the fit. The definition of thisstatistical criterion is such that the best available model is theone that minimizes the AIC. The models used were two pointsources, a Gaussian and a disk, for which the extensions andlocations were optimized to maximize the likelihood. The loca-tions of the two point sources were obtained by re-optimizingthe positions of the catalogued sources 4FGL J1925.4 + + + ff ers a better descriptionof the data compared to the other models and it is thus chosenfor the rest of the analysis. We also obtained significance mapsof the residual gamma-ray emission resulting from each of themorphological models. As we saw in these maps, the Gaussianmodel more properly accounts for the emission in the region.In order to estimate the significance of the source extension, wefollow Lande et al. (2012) and calculate TS ext = L ext / L ps ),where L ext and L ps are the likelihoods resulting from fitting theextended source and a point source. The resulting value, TS ext = .
3, indicates that the extension is significant beyond a pointsource for the LAT.The best-fit coordinates of the Gaussian centroid and 68%-containment radius (with their 1 σ statistical uncertainties) areRA = . ± . ◦ , Dec = . ± . ◦ (J2000) and 0 . + . − . ◦ . See https: // fermi.gsfc.nasa.gov / ssc / data / analysis / documentation / Pass8_edisp_usage.htmlArticle number, page 2 of 10. Araya ID : GeV emission in the region of the supernova remnant G51.26 + RA J2000 (deg) D ec J ( d e g ) G051.010+00.060LAT 68%G51.26+0.09
Fig. 1: LAT TS map for a point source hypothesis with events having energies above 5 GeV showing the excess emission abovethe background. The location and size of the SNR G51.26 + + + ∆ AICTwo point sources 13.8Disk 6.6Gaussian 0
Notes. ∆ AIC is defined as the di ff erence between the AIC value for agiven model and that of the model with the lowest AIC value. The 68%-containment region of the best-fit Gaussian template isalso shown in Fig. 1.
Once an appropriate model for the morphology of the GeV emis-sion is obtained, we apply this model to events above an energyof 200 MeV in the ROI and obtain the spectrum of the source.The Gaussian template found in this section was added to themodel, and a search for new point sources in the ROI was car-ried out to improve the background description. Once the newpoint sources are added, we let the normalizations of the sourceslocated within 10 ◦ of the ROI center and the other spectral pa-rameters of the sources located within 5 ◦ of the ROI center freeto vary.Two phenomenological spectral energy functions were usedin separate fits to the emission seen in the region of G51.26 + Article number, page 3 of 10 & A proofs: manuscript no. g51 to quantify the significance of curvature (i.e., deviation from apower law) in the spectrum. It was found that the addition ofa parameter using the log parabola produces a negligible change( ∼ .
2) in the likelihood function, indicating that the spectrum ofthe source is not significantly curved in the LAT energy range.This justifies the use of a power law for the spectrum in the mor-phological analysis using events with energies above 5 GeV. Ifthe spectral function is written as dNdE = N (cid:32) EE (cid:33) − Γ , where E = N = (1 . ± . stat ± . sys ) × − MeV − cm − s − and Γ = . ± . stat ± . sys . The overall test statisticvalue of the source is TS =
443 above 200 MeV, which translatesto a detection significance of 21 σ . The residual map obtainedwhen adding the source shows no significant emission leftoverin the region, meaning that the model found for the source issatisfactory. A spectral energy distribution (SED) was obtainedby measuring the flux of the source in ten energy bins. In eachbin the normalization of the source of interest was fit togetherwith the normalizations of the sources located within 2 ◦ of thecenter of the ROI, as well as those of the di ff use and isotropicbackgrounds. The spectral index of the source was kept fixedto 2, but the results were not significantly a ff ected by using adi ff erent value for the spectral index. If the TS of the source in abin was below 4, a 95% confidence level upper limit on the fluxwas estimated.Two factors were considered for estimating the systematicerrors in the spectral parameters. The e ff ect of the uncertainty inthe Galactic di ff use emission model was estimated with the useof the eight alternative models developed by Acero et al. (2016)in making the LAT catalog of SNRs. The files were scaled appro-priately to account for di ff erences in energy dispersion betweenPass 7 and Pass 8 reprocessed data . The uncertainties were cal-culated as in Acero et al. (2016) for the source parameters in theglobal fit using the entire energy range as well as in the individ-ual energy bins used for obtaining the SED. The uncertainties inthe e ff ective area of the LAT were also propagated onto the spec-tral parameters of the global fit, as well as to the normalizationsof the SED fluxes, using a set of bracketing response functionsas recommended by Ackermann et al. (2012). In these alternativefits the pivot energy was used as the value of the scale parame-ter E , which was estimated with the covariance error matrix ofthe global fit. For the individual SED fluxes the statistical andsystematic errors were combined in quadrature.
3. Properties of the ISM CO (J = → ) data After inspecting the GRS data cubes we have found several ra-dial velocity intervals where CO (J = →
0) emission wasdetected anywhere near the field of view around the GeV emis-sion, and the results are shown in Fig. 2. The line velocities usedin this work are all measured with respect to the local standardof rest. See https: // fermi.gsfc.nasa.gov / ssc / data / access / lat / Model_details / Pass8_rescaled_model.html
Inspection of the data cubes from the HI4PI survey in the regionof G51.26 +
4. Discussion
In this section we present the physical models that account forthe spectrum of the GeV emission and discuss possible sourcesresponsible for it. We made use of the properties of the ISMshown in Section 3. We fit the GeV data using one-zone lep-tonic and hadronic scenarios to derive the particle distributionsrequired using the naima package (Zabalza 2015), and used theonly radio fluxes available, which were measured in the largerregion associated to the original THOR candidate, G51.21 + We adopt a particle distribution in energy ( (cid:15) ) which is a powerlaw with an exponential cuto ff given by dnd (cid:15) = n (cid:15) − s e − (cid:15)/(cid:15) c , where n and s are the normalization and spectral index of thedistribution, respectively, and (cid:15) c is the cuto ff energy. As seedphoton fields for the calculation of the inverse Compton scat-tering fluxes (IC) we adopt the cosmic microwave background(CMB), a far-infrared (FIR) field and stellar optical and near-infrared (NIR) photons. For the latter two, the densities and tem-peratures adopted are the same as those estimated at a Galacto-centric distance of 8 kpc (FIR: 26 K, 0.35 eV cm − , NIR: 2000 K,0.7 eV cm − , Shibata et al. 2011), but the results do not changesubstantially if these parameters are modified.We perform three independent fits to the gamma-ray SEDpoints obtained in this work under the assumption that one of thethree main mechanisms for high energy emission known to oper-ate in Galactic sources dominates in each case: pion decay emis-sion from hadronic processes, IC from leptons interacting withambient photons and bremsstrahlung emission also from highenergy leptons interacting with ambient material. We note thatthe gamma-ray data could also be explained by combinations ofthese processes but given how little is known about the sourceand its environment our aim is to probe the properties of the par-ticles in the more simple, individual scenarios described. Fig. 4shows the SED and the resulting best-fit models in each case. Acalculation of the AIC using the likelihood function obtained inthe fits yields AIC = s = . + . − . and (cid:15) c = + − , while the Article number, page 4 of 10. Araya ID : GeV emission in the region of the supernova remnant G51.26 + RA J2000 (deg) D ec J ( d e g ) -0.75 — 9.2 km/s K km/s
RA J2000 (deg) D ec J ( d e g ) K km/s
RA J2000 (deg) D ec J ( d e g ) K km/s
RA J2000 (deg) D ec J ( d e g ) K km/s
RA J2000 (deg) D ec J ( d e g ) K km/s
Fig. 2: CO (J = →
0) integrated intensity in the relevant velocity ranges covered by the GRS (Jackson et al. 2006). The greencontours represent the TS values seen in Fig. 1 in steps of 10 from 10 to 80.
Article number, page 5 of 10 & A proofs: manuscript no. g51
RA J2000 (deg) D ec J ( d e g ) -43.7 km/s RA J2000 (deg) D ec J ( d e g ) -26.9 km/s RA J2000 (deg) D ec J ( d e g ) -14.0 km/s RA J2000 (deg) D ec J ( d e g ) -6.3 km/s RA J2000 (deg) D ec J ( d e g ) RA J2000 (deg) D ec J ( d e g ) RA J2000 (deg) D ec J ( d e g ) RA J2000 (deg) D ec J ( d e g ) RA J2000 (deg) D ec J ( d e g ) Fig. 3: HI emission maps in the region of G51.26 + . × (cid:18) d kpc (cid:19) erg,where d is the source distance. The value has been normalizedto a sample distance to the source of 2 kpc but it could be easilyadjusted for other values. For the calculation of the synchrotronfluxes shown in Fig. 4 (a) we set the magnetic field value to 3 µ G,which is a typical value in the ISM. However, a di ff erent valuefor the magnetic field could be used instead if the total energycontent of the particles is adjusted appropriately. Given the lackof information and data from this region, we can only really con-strain the parameters of the spectral distribution shape ( s and (cid:15) c ).It is interesting to note that the particle spectral index obtainedhere results in a slope for the synchrotron flux ( α = − .
84) whichis compatible with the value measured by Driessen et al. (2018)of α = − . ± .
21 for G51.21 + + α = − . ± .
05. It is then expected for the radio spectrum in theregion of G51.26 + α = − . s = . + . − . and (cid:15) c = + − TeV. Given the relatively hard spectral energy distribution required for thehigh-energy leptons ( s ∼ n (cid:38)
120 cm − and for the total energy inthe leptons to be less than ∼ . × (cid:18) d kpc (cid:19) erg, in order tosuppress the IC contribution. Fig. 4 (b) shows this scenario for n =
120 cm − . A magnetic field of 10 µ G was used for the calcu-lation of the synchrotron fluxes. As can be seen, this model failsto account for the synchrotron fluxes which should not be toodi ff erent than the shown upper limits. The predicted synchrotronfluxes could be increased for a higher magnetic field but the en-ergy output quickly becomes too large in the X-ray band. How-ever, no source of X-rays is known in the region of the GeVemission. Future X-ray observations of this source are encour-aged.Finally, in the hadronic scenario, the particle spectral indexand cuto ff energy required are s = . + . − . and 84 + − TeV,respectively. As in the previous fits, the particle cuto ff energycannot be well constrained using Fermi -LAT data alone since thegamma-ray SED itself is best described by a simple power lawfunction. The total energy required in the cosmic rays is 8 . × Article number, page 6 of 10. Araya ID : GeV emission in the region of the supernova remnant G51.26 + − − − − − E (MeV) − − − E d N d E ( M e V / c m s ) IC (CMB)IC (FIR)IC (NIR)SynchrotronTotal gamma-ray flux (a) − − − − − E (MeV) − − − E d N d E ( M e V / c m s ) IC (CMB)IC (FIR)IC (NIR)SynchrotronBremsstrahlungTotal gamma-ray flux (b) E (MeV) − − − E d N d E ( M e V / c m s ) Pion decay (c)
Fig. 4: SED of the gamma-ray source shown in Fig. 1 with the models obtained in each scenario where the high-energy emissionresults predominantly from IC emission from leptons (a), bremsstrahlung emission from leptons (b) and neutral pion decay fromhadronic interactions by high-energy cosmic rays (c). The gamma-ray fluxes are from this work and the radio fluxes correspond tothe original THOR candidate SNR (G51.21 + (cid:16) cm − n (cid:17) (cid:18) d kpc (cid:19) erg, where n is now the particle numberdensity of the target material for hadronic interactions. Fig. 4 (c)shows the resulting fit to the data in this scenario.Based only on the spectral fits to the SED shape and given theradio flux upper limits, we conclude that if the GeV emission isproduced mainly by a single particle population and one physicalmechanism is dominant, then the IC and hadronic scenarios aremore likely than the bremsstrahlung-dominated scenario to bethe source of the gamma rays. Several young star clusters are known to accelerate cosmic raysmost likely at the shocks of the collective winds from massivestars, resulting in the emission of high-energy photons from thematter that surrounds the clusters (see Aharonian et al. 2019,and references therein). The gamma-ray emission associated tothese objects is extended, with hard GeV spectral indices ( Γ ∼ . − .
3) and likely hadronic. In fact, star forming regions withmassive stars may very well be the long sought sources of PeVcosmic rays in the Galaxy (Aharonian et al. 2019). The most prominent known HII region that partially overlapsthe gamma rays seen in Fig. 1 is G051.010 + . ± . CO (J = →
0) emission in the velocity range from43.4 to 52.2 km / s in comparison to the GeV emission from Fig.1. The molecular gas associated to the HII region can be seento be displaced from the GeV emission (although not shown,some molecular material also exists around a velocity of 41 km / snear the southern-most edge of the HII region at a declination of ∼ . ◦ ). Indeed the location of the peak of the GeV emis-sion occurs in a region with no substantial CO emission in thisvelocity interval. The GeV emission detected from star-formingregions is correlated with the gas around them. Based on thedisplacement of the gamma-ray emission from the star formingregion G051.010 + SNRs are known cosmic ray accelerators which can producegamma rays, and thus this scenario is considered now. As canbe seen in Fig. 1 the gamma-ray emission has a peak near thecenter of the SNR G51.26 + Article number, page 7 of 10 & A proofs: manuscript no. g51
RA J2000 (deg) D ec J ( d e g ) Fig. 5: Integrated CO (J = →
0) intensity in the velocity range 43.4 to 52.2 km / s from Fig. 2 showing the molecular gasassociated to the HII region G051.010 + + VGPS data (see also the maps shown by Andersonet al. 2017; Wang et al. 2018). The much more compact SNRG51.04 + + ∼ ◦ . Future observations could confirm the pres-ence of non thermal emission from one or more SNRs in theregion (or perhaps a single SNR of which G51.26 + Article number, page 8 of 10. Araya ID : GeV emission in the region of the supernova remnant G51.26 + gas that could explain this morphology. There is a “wall” of gasseen in the velocity interval -0.75–9.2 km / s that could explain thewestern-most part of the shell, but no other molecular clouds areseen in correlation with the overall morphology. Adopting a stan-dard galactic rotation model (Brand & Blitz 1993) we calculatedthe kinematic distances associated to these velocities which are0.5 kpc and 10–11 kpc. These values are likely either too smallor too large for a typical SNR, but even if this molecular gas waslocated at the same distance as the source, it would still fail toexplain most of the morphology of the SNR.On the other hand, Fig. 3 shows the neutral hydrogen in theISM where several cavity-like features are seen. The winds ofmassive stars that are the progenitors of SNRs are known to cre-ate bubbles where the SNR later expands. We could speculatethat the intensity gradients of the emission from the HI gas seenin Fig. 3 at the line velocities of -43.7, -14.0, 18.1, 28.4, 52.9and 69.7 km / s could be caused by such process, in which casethe distance to the SNR cannot be constrained. We note also thataccording to the Galactic rotation model, the maximum allowedvelocity at the location of the source is ∼
51 km / s and thereforethe features above this value could be caused by high-velocitygas with undetermined distance. The same could be said of the CO emission seen in the velocity interval 52.4–58.8 km / s inFig. 2. This is the only velocity interval in the data from theGRS where molecular gas is seen at the location of the peak ofthe GeV emission, a correlation which would be expected in ahadronic scenario for the GeV emission.In terms of the energetics, an origin of the gamma rays inthe SNR scenario is not problematic. From the photon spectrumdetermined in Section 2, we estimated the luminosity of the GeVemission in the 1–500 GeV energy interval, for a distance d tothe source, which is 1 . × (cid:18) d kpc (cid:19) erg s − . This value istypical of GeV-emitting SNRs (Acero et al. 2016).SNRs also show a variety of GeV spectral indices andshapes, thought to be caused by the di ff erences in environmentsand evolutionary stages. With a spectral index of ∼ .
2, the high-energy spectrum of the source shows a similar shape to that ofCas A’s in the energy range 1–500 GeV (Abeysekara et al. 2020;Yuan et al. 2013) and that of Tycho’s SNR in the GeV range(Archambault et al. 2017). Both of these are young SNRs withbright X-ray and radio emission, which is not seen in the regionof G51.26 + + . ◦ for the GeV emissionalong the northwest direction, the physical size of the sourcewould be 8 . (cid:18) d kpc (cid:19) pc, where d is the source distance. The di-ameters of SNRs are typically of the order of tens of parsecs,and therefore the observed source extension is consistent withan SNR for reasonable distances of the order of kpc.The SED models seen in Fig. 4 are also generally consistentwith an SNR scenario for the gamma rays. The typical energy inthe shocks of SNRs available for the acceleration of particles isof the order of 10 erg. The maximum total energy required inthe leptonic scenario amounts to 8 . × (cid:18) d kpc (cid:19) erg, while forthe hadronic scenario corresponds to 8 . × (cid:18) d kpc (cid:19) erg fora target density of 1 cm − and, of course, lower for higher den-sities. Although the distance to the source cannot be constrainedthe order of magnitudee of the total energy in the particles is con- sistent with measured values for SNRs (see, e.g., Araya & Cui2010; Fraija & Araya 2016; Xing et al. 2019). Pulsars can also produce extended gamma-ray radiation throughthe interaction of the relativistic electrons and positrons, accel-erated by the pulsar wind nebula (PWN), with ambient photons.A scenario where an SNR produces the observed radio emis-sion and contains within it a PWN that shines in the gamma-rayrange seems plausible. This would also explain the Gaussian-like morphology of the GeV emission. X-ray observations ofG51.26 + in the re-gion of the gamma-ray emission yields three pulsars, PSRJ1924 + + + + + . × ergs − ), three orders of magnitude lower than the observed gamma-ray luminosity for the same distance. Even if the spin-downpower of the pulsar was of the order of 10 erg s − (compara-ble to those of the pulsars associated to the brightest gamma-rayPWN, see e.g. H. E. S. S. Collaboration et al. 2018a), an e ffi -ciency of energy transfer of the order of 10% would be neededfor a time of the order of 10 Myr to a supply the necessary en-ergy in the leptons ( ∼ × erg for a distance of 10 kpc).At a distance of ∼
10 kpc for both PSR J1924 + + ∼
100 pc (taking the size of the radio source). Given the cor-relation between the radio and GeV emissions, it would be rea-sonable to assume that the radio comes from the SNR associatedto the pulsar. However, a 100-pc SNR would be in an advancestage of evolution, likely dissipating in the ISM. This situationseems unlikely from the radio morphology observed. This leavesPSR J1926 +
5. Conclusions
We have found an extended gamma-ray source with a hard (pho-ton index ∼ .
18) spectrum in the GeV range using data from the
Fermi satellite at the location of the SNR G51.26 + https: // / research / pulsar / psrcat / Article number, page 9 of 10 & A proofs: manuscript no. g51 the hard spectral index, a bremsstrahlung-dominated scenario forthe gamma-rays is ruled out.We rule out the star forming region G051.010 + + Acknowledgements.
Thanks to L. Anderson, X. Sun and J. Stil for their com-ments regarding the radio data. This work was possible due to funding by Uni-versidad de Costa Rica and its Escuela de Física under grant number B8267. Thisresearch is based on observations made with NASA’s Fermi Gamma-Ray SpaceTelescope, developed in collaboration with the U.S. Department of Energy, alongwith important contributions from academic institutions and partners in France,Germany, Italy, Japan, Sweden and the U.S. It also makes use of molecular linedata from the Boston University-FCRAO Galactic Ring Survey (GRS). The GRSis a joint project of Boston University and Five College Radio Astronomy Obser-vatory, funded by the National Science Foundation under grants AST-9800334,AST-0098562, AST-0100793, AST-0228993 and AST-0507657.
References