Hard X-ray Emission from Sh2-104: A NuSTAR search for Gamma-ray Counterparts
E. V. Gotthelf, K. Mori, E. Aliu, J. M. Paredes, J. A. Tomsick, S. E. Boggs, F. E. Christensen, W. W. Craig, C. J. Hailey, F. A. Harrison, J. S. Hong, F. Rahoui, D. Stern, W. W. Zhang
aa r X i v : . [ a s t r o - ph . H E ] M a y To be Submitted to The Astrophysical Journal
Preprint typeset using L A TEX style emulateapj v. 08/13/06
HARD X-RAY EMISSION FROM Sh E. V. Gotthelf , K. Mori , E. Aliu , J. M. Paredes , J. A. Tomsick , S. E. Boggs , F. E. Christensen , W. W.Craig , C. J. Hailey , F. A. Harrison , J. S. Hong , F. Rahoui , D. Stern , W. W. Zhang To be Submitted to The Astrophysical Journal
ABSTRACTWe present
NuSTAR hard X-ray observations of Sh II region containing sev-eral young massive stellar clusters (YMSCs). We have detected distinct hard X-ray sources coin-cident with localized VERITAS TeV emission recently resolved from the giant gamma-ray complexMGRO J2019+37 in the Cygnus region. Faint, diffuse X-ray emission coincident with the easternYMSC in Sh Sh XMM-Newton and
NuSTAR spectrum of 3XMM J201744.7+365045 is well-fit to an absorbed power-law model with N H = (3 . ± . × cm − and photon index Γ = 2 . ± .
1. Based on possible long-term fluxvariation and the lack of detected pulsations ( ≤
43% modulation), this object is likely a backgroundAGN rather than a Galactic pulsar. The spectrum of the
NuSTAR nebula shows evidence of anemission line at E = 5 . z = 0 . ± . d = 800 Mpc) and L X = 1 . × erg s − . Follow-up Chandra observations of Sh Sh Fermi pulsar PSR J2017+3625 and not the H II region. Subject headings:
ISM: individual (MGRO J2019+37, VER J2019+368, VER J2016+371,3FGL J2021.1+3651, 3FGL J2017.9+3627, 3FGL J2015.6+3709) — pulsars: in-dividual (PSR J2021+3651, PSR J2017+3625) — stars: neutron — supernovaremnants INTRODUCTION
MGRO J2019+37 is the brightest Milagro gamma-raysource in the Cygnus region, with 80% of the Crab Neb-ula flux at 20 TeV (Abdo et al. 2007). The origin andnature of MGRO J2019+37 has long been the subjectof debate as its ∼ ◦ extent overlaps several supernovaremnants (SNRs), H II regions, Wolf-Rayet stars, > Fermi pulsarsand a hard X-ray transient. Recent TeV observationson ∼ ′ scales using the VERITAS telescope clearly re-solve out the giant gamma-ray complex into at leastthree distinct TeV emission regions, each coincident with Columbia Astrophysics Laboratory, Columbia University,550 West 120th Street, New York, NY 10027-6601, USA;[email protected] Departament de F´ısica Qu`antica i Astrof´ısica, Institut deCi`encies del Cosmos, Universitat de Barcelona, IEEC-UB, Mart´ı iFranqu`es 1, 08028, Barcelona, Spain Space Sciences Laboratory, University of California, Berkeley,CA 94720, USA DTU Space-National Space Institute, Technical University ofDenmark, Elektrovej 327, 2800 Lyngby, Denmark Lawrence Livermore National Laboratory, Livermore, CA94550, USA Cahill Center for Astronomy and Astrophysics, California In-stitute of Technology, Pasadena, CA 91125, USA Harvard-Smithsonian Center for Astrophysics, Cambridge, MA02138, USA European Southern Observatory, Karl Schwarzchild-Strasse 2,85748 Garching bei M¨unchen, Germany Jet Propulsion Laboratory, California Institute of Technology,4800 Oak Grove Drive, Pasadena, CA 91109, USA NASA Goddard Space Flight Center, Greenbelt, MD 20771,USA Department of Astronomy, Harvard University, 60 GardenStreet, Cambridge, MA 02138, USA a Fermi source (Aliu et al. 2014). The bulk of theVERITAS emission from MGRO J2019+37 falls into theelongated (1 . ◦ × . ◦ Fermi pulsarPSR J2021+3651 (Roberts et al. 2002) at the easternedge of VER J2019+368 is not sufficiently energetic topower all the gamma-ray flux in the region, based on thetime required for electrons to diffuse and fill the largeemitting volume relative to their cooling lifetime. In-stead, these authors suggest that massive star-formingactivity associated with the H II region Sharpless 104(herein Sh Sh . ± . ∼ ′ radio diameter (Paredes et al. 2009). The massive COclouds around the star clusters suggests Sh II region (Deharveng et al. 2003). Anassociated H-alpha nebula is clearly resolved in the DSSPOSSII-J image, likely powered by a central O6 V starionizing the region (Lahulla 1985), and possibly a brightnearby IRAS source.A serendipitous XMM-Newton observation of theMGRO J2019+37 field caught the eastern half of the Sh m m m h m m m o ’ o ’ R. A. (J2000) D ec l . ( J2000 ) F G L J2021 . + ( PS R J2021 + ) Sh 2-104CTB 87VER J2016+371Ber 87 3FGL J2017.9+3627(PSR J2017+3625)VER J2019+368 3FGL J2015.6+3709
Fig. 1.—
The TeV gamma-ray map of the MGRO J2019+37region resolved into distinct sources using VERITAS obser-vations. Note that the Milagro source itself fills the field-of-view. Superimposed are VERITAS images in two energybands, 0 . − > Fermi source ( white cir-cles ). The harder, extended TeV emission, VER J2019+368,is associated with the pulsar PSR J2021+3651; Sh level, that suggested several point sources within the ra-dio shell. Most notably, these include ones overlappingthe central star, coincident with a 2 σ ROSAT source,and the eastern YMSC. Just outside the radio shell lies3XMM J201744.7+365045 and a barely detected nebula ∼ ′ in diameter. These results open the possibility ofidentifying a low energy counterpart to the gamma-rayemission, and help identify its origin.As part of the NuSTAR
Galactic Survey programwe have obtained broad band X-ray observations of Sh XMM-Newton source and the nearby diffuse neb-ula. We consider the possibility that these sources arerelated to the star formation regions and/or associatedwith gamma-ray emission. Alternatively, the latter twosources may have an unrelated extragalactic origin. NuSTAR
OBSERVATIONS OF Sh NuSTAR observed the H II region Sh NuSTAR consists of two co-alignedX-ray telescopes, with corresponding focal plane modulesFPMA and FPMB that provide 18 ′′ FWHM imaging res-olution over a 3–79 keV X-ray band, with a characteristicspectral resolution of 400 eV FWHM at 10 keV (Harrisonet al. 2013). The reconstructed
NuSTAR coordinates areaccurate to 7 . ′′ NuSTAR is ∼ < FTOOLS
NUSTARDAS
NuSTAR
Calibration Database (CALDB) files of 2013August 30. The resulting data set provides a total of80.5 ks and 91.6 ks of net good time for the two point-ings, respectively, after removing intervals of high back-ground rates. We also exclude a bright arc of stay lightthat contaminates the eastern edge of the field-of-viewin both detectors during the first observation. The ex-tracted spectra combined data from both FPM detec-tors, grouped into appropriate spectral fitting channels,and modeled using the
XSPEC (v12.8.2) spectral fittingpackage (Arnaud 1996). All spectral fits use the
TBabs absorption model in
XSPEC with the wilm
Solar abun-dances (Wilms et al. 2000) and the vern photoionizationcross-section (Verner et al. 1996).
Image analysis
Figure 2 presents the exposure-corrected 3–79 keV
NuSTAR images of the Sh σ = 3 . ′′ >
30 keV) pointsource just north of the Sh . ′ below it, roughly 2 . ′ in diameter.These sources clearly correspond to faint X-ray emissionseen in a short, 20 ks, 2007 XMM-Newton observation,detected serendipitously, at the very edge of the field-of-view (Zabalza et al. 2010).The two bright
NuSTAR sources are embedded in en-hanced diffuse emission that overlaps, at least in part,with the Sh Sh r ∼ . ′
5, correctedfor the PSF, is consistent with the size of the opticalcluster. Further south we find evidence of several otherfaint sources, including a
Swift
X-ray source obtained aspart of the follow-up program of
Fermi sources. Table 1presents the list of detected
NuSTAR sources along withthe significance of detection computed by wavdetect .The source coordinates are accurate to ≈ . ′′
0, registeredusing 3XMM J201744.7+365045, the counterpart to thebright
NuSTAR point source. Finally, we note thatno hard X-ray counterpart is detected for the
ROSAT source at the center of Sh Spectral analysis
To preface our spectral analysis we note that the
NuS-TAR low-energy response (3 keV) is too hard to con-strain the absorbing column for a typical source with N H < cm − . In the following spectral fits using NuSTAR data alone, for definitiveness, we hold the col-umn density fixed to a fiducial value of the Galactic to-tal. Generally, the range of likely column density here isfound to have no significant effect on the resulting spec-tral parameters. We include both a neutral Hydrogen and a molecular Hydrogen component to the column den-sity, to take into account significant local CO emission(see Dame et al. 2001). We compute a total Galactic http://heasarc.gsfc.nasa.gov/cgi-bin/Tools/w3nh/w3nh.pl ard X-ray Emission from Sh <--- Point SourceNebula --->YMSC ---> <--- Point SourceNebula --->YMSC ---> Fig. 2.—
NuSTAR exposure-corrected and smoothed 3-79 keV X-ray images of the field containing Sh Left —
A bright
NuSTAR point source is evident up to 30 keV and significant X-ray nebula is found in the 3 −
20 keV band just south of thepoint source. Both of these sources are of unknown origin. The nebula overlaps the rim of the HII region Sh Right —
The same image with the bright point source removed and scaled tohighlight diffuse emission. Indicated ( circles ) are the detected
NuSTAR sources listed in Table 1. Also shown is the location ofthe nearby
Chandra observation ( outline ) of 3FGL J2017.9+3627 ( contours ). TABLE 1
NuSTAR
Sources in the Sh ±
14 4.42 20 17 36.02 +36 37 58.5 73 ±
16 5.2
Swift source3 20 17 38.91 +36 40 27.1 45 ±
13 3.74 20 17 40.02 +36 36 38.3 24 ± ±
13 4.06 20 17 42.41 +36 43 09.7 48 ±
13 4.17 20 17 44.32 +36 48 12.9 2219 ±
82 27.0 Nebula NuSTAR J201744.3+3648128 20 17 44.42 +36 50 46.4 1406 ±
54 35.0 3XMM J201744.7+3650459 20 17 53.02 +36 31 56.0 52 ±
12 4.910 20 17 55.57 +36 45 33.1 197 ±
34 5.8 Sh II region Note . — Coordinate system is registered to ≈ . ′′ NuSTAR point source. column density of N H = n HI + 2 n H = 2 . × cm − ,a value consistent with the results of the combined XMM-Newton and
NuSTAR fit to the spectrum of 3XMMJ201744.7+365045, presented below. In all cases, thequoted spectral uncertainties are at the 90% confidencelevel for one or two interesting parameter(s) for the oneand two component spectral fits, respectively.The high-energy emission from the eastern YMSC isof great interest, as this source is a natural candidatefor the observed gamma-ray emission. We extracted a
NuSTAR spectrum from YMSC using a r < . ′ aper-ture in the usable 3–10 keV range. This yields a totalof 680 counts of which 71% are from background con-tamination, as estimated from counts extracted from anadjacent aperture ( r < . ′ ) on same chip of each FPM.We consider several appropriate spectral models as thequality of the spectrum is not sufficient to distinguishbetween then. Under the assumption that the X-rayemission is due to colliding winds of component stars in the cluster we fit the raymond thermal plasma model in XSPEC (Raymond & Smith 1977, and updates). The best-fit temperature is kT = 1 . − . χ = 1 . . × − erg cm − s − . We estimate a sourceluminosity of L ∼ erg s − from the plasma coolingcurve (e.g., Maio et al. 2007) and the derived emissionmeasure, computed from the model normalization and adistance of 4 kpc to Sh . . − .
7) with a similar χ and flux as found forthe thermal model.For 3XMM J201744.7+365045, we extracted a highquality NuSTAR spectrum using a r = 45 ′′ source aper-ture and a r = 1 . ′ background region offset from thesource. The source spectrum is found to dominate thebackground up to 20 keV, but emission is evident to at Gotthelf et al. Fig. 3.—
NuSTAR and
XMM-Newton spectra of 3XMMJ201744.7+365045. The spectra are fitted simultaneously to anabsorbed power-law model with their normalizations left free. Theupper panel present
NuSTAR (black) and
XMM-Newton
EPIC pn(red) and EPIC MOS (green) spectral histograms along with thebest-fit model (solid lines) given in Table 2. The lower panel showsthe residuals from the best-fit model in units of sigma. least 30 keV. The spectrum is well-fitted in the 3–20 keVenergy band to an absorbed power-law model with thecolumn density held fixed to the Galactic total. Thebest fit spectral index is Γ = 2 . ± . χ = 0 .
82 for28 DoF. The unabsorbed flux in the 2–10 keV band is(1 . ± . × − erg cm − s − . A blackbody model isexcluded by the fit, as is, for lack of line features, ther-mal plasma models. The spectral results are presentedTable 2.To better estimate the source column density for3XMM J201744.7+365045 we extracted and fit the XMM-Newton spectrum simultaneous with the
NuSTAR data, allowing the flux normalization to be independent.This resulted in N H = (3 . ± . × cm − and spec-tral index Γ = 2 . ± . XMM-Newton flux mea-surements obtained 7 years earlier, the
NuSTAR value islower, formally by a factor of ∼
3. However, the formerflux is not well established due to poor statistics, thehigh background, and the far off-axis source detectionon the edge of the
XMM-Newton
EPIC instruments, andthe relative flux calibration between instruments.To examine the X-ray nebula NuSTARJ201744.3+364812, we extracted spectra using a r = 1 . ′ radius aperture and the background regiondefined above. This aperture encompassed nearly all thenebula extent to the background level. Of the ∼ ∼ . − . χ ν = 1 . . − . × − erg cm − s − (90% confidence level)and the unabsorbed flux is 8 . × − erg cm − s − .The poor χ statistic for this fit is mainly due to a line-like feature around 5 . NuSTAR spectrumof the point source 3XMM J201744.7+365045. Introduc-ing a Gaussian line to the fit, to better characterize thecontinuum, yield a line measured at 5 . ± . χ ν = 0 .
94 for 54 DoF.The F-test value of 13 .
09 corresponds to a false positivesignificance of the added spectral line of ℘ = 2 . × − .However, this significance should be interpreted with care(Protassov et al. 2002). Formally, an analysis of the F-test probability using the XSPEC script simftest con-firms a highly significant detection of an emission linefeature associated with the source.As no emission line is known at this energy we con-sider the possibility of a redshifted Fe line from a galaxycluster hidden behind the Galactic plane. A spectral fitusing the Raymond-Smith model for a thermal plasmaproduced an excellent fit again, with χ ν = 0 .
86 for 55DoF, with the column density fixed at the Galactic to-tal. The best-fit parameters give a kT = 5 . +1 . − . keV,redshift z = 0 . ± .
02, and abundance 0 . Z ⊙ . The2–10 keV absorbed flux is 6 . +0 . − . × − erg cm − s − (90% confidence level) and the unabsorbed flux is 7 . × − erg cm − s − . We note that the measured rangeof kT is essentially independent of the column density,from zero to 3 . × cm − , the point source value.This N H range yields a measured temperature range of kT = 5 . − . kT obtained using the Galactic column.For a galaxy cluster at the implied redshift distanceof 800 Mpc, the total X-ray luminosity is L x = 1 . × erg s − . This is within an order of magnitude ofthe value derived from the luminosity - temperature re-lation for clusters (Novicki et al. 2002). Moreover, basedon the inferred temperature, the observed nebula size iswell predicted for a putative cluster (Mohr et al. 2000).Given the large uncertainties in these relations we takethe results as reasonably evidence that the X-ray nebulais due to a background galaxy cluster unrelated to thegamma-ray emission. Timing Analysis
The high time resolution of
NuSTAR allows a searchfor pulsations from 3XMM J201744.7+365045 down toperiods of P ∼ XMM-Newton coordinates. The
NuSTAR light curve is found to be stable during theobservation on all timescales. A Fast Fourier Transform(FFT) finds no evidence of red noise, indicative of accret-ing systems in the power spectrum. We also searched fora coherent signal using both the FFT method and the Z n test statistic for n = 1 , , ,
5, and the H-test, to be sensi-tive to both broad and narrow pulse profiles. We initiallyrestricted the timing search to photon energies in the 3–25 keV range and used an aperture of 30 ′′ to optimize thesignal-to-noise ratio. We repeated our search for an ad-ditional combination of energy ranges 3 < E <
10 keV,10 < E <
25 keV and aperture sizes r > ′′ . None ofthese resulted in a significant detection. After taking intoaccount the estimated background emission, we place anupper limit on the pulse fraction f p ≤
43% for a sinu-soidal signal in the 3–25 keV band for the 20 ′′ aperture. DISCUSSION
TeV gamma-ray emission from MGRO J2019+37is well-separated into three regions by high-resolutionVERITAS observations, each associated with a
Fermi ard X-ray Emission from Sh Fig. 4.—
NuSTAR spectra of the new X-ray nebula NuSTAR J201744.3+364812.
Left —
The spectrum fitted with an absorbed power-law(solid lines). The lower panel shows the residuals from the best-fit model given in the text, in units of sigma.
Right —
The same spectrumfitted with the addition of an emission line.
TABLE 2
NuSTAR and
XMM-Newton
Spectra of 3XMM J201744.7 + Parameter
NuSTAR only
XMM-Newton only
NuSTAR + XMM-Newton a N H (cm − ) 2 . × (fixed) 3 . . − . × (3 . ± . × Γ 2 . ± . . − . . ± . F abs (2–10 keV) 1 . ± . × − . × − . ± . × − F unabs (2–10 keV) 1 . × − . × − . × − χ ν (dof) 0.82 (29) 0.68 (36) 0.70 (62) Note . — Power-law model fits are obtained in the 0.5–10 keV and 3–20 keV energy bandsfor the
XMM-Newton and
NuSTAR spectra, respectively. For the joint
NuSTAR and
XMM-Newton spectral fits, the indexes and column densities are linked. The uncertainties are 90%confidence limits for two interesting parameters, except for the
NuSTAR only data, which isfor one interesting parameter. The given fluxes are in units of erg cm − s − . a The listed flux values are for the
NuSTAR spectra, while the
XMM-Newton spectra arejointly fit with a relative normalization factor to account for the flux variation. source (Aliu et al. 2014). To the north, VER J2016+371is likely associated with
Fermi emission from the blazarB2013+370 (Kara et al. 2012) and/or the filled-centersupernova remnant CTB 87 (Aliu et al. 2014). To thesouth, localized VERITAS emission, significant at the 3 σ confidence level, is coincident with the recently discov-ered Fermi pulsar, PSR J2017+3627, discussed below.To the east, the large elliptical morphology of the spec-trally harder VER J2019+368 suggests a blend of twooverlapping sources. If the arguments of Paredes et al.(2009) are correct, the
Fermi pulsar PSR J2021+3651easily accounts for the eastern most TeV emission fromVER J2019+368, while Sh NuSTAR data reveals faint emission from the east-ern YMSC of Sh L ∼ erg s − . In other YMSCs, hard > NuSTAR spectrum, then the plasmatemperature of ∼ < Sh . +1 . − . , which is justbarely compatible with NGC 3576 N.The coincidence of gamma-ray emission near star-forming regions suggests a physical connection betweenthe two – for example, W49A (Brun et al. 2010), West-erlund1 (Luna et al. 2010), and Carina Nebula, althoughthe gamma-rays from the latter is likely dominated bythe CWB η Carinae (Tavani et al. 2009; Farnier et al.2011). However, this connection remains far from clear,at least on an individual basis. For the case of the YMSCin Sh Sh . ′′ . ± .
08 mJy for GMRT J201744.8+365045. Com-paring this to the 20 cm flux of 11.15 mJy (White et al.2005) yields a spectral index of α = 1 .
2, where F ν = ν − α .The combination of the radio and X-ray point sourceand diffuse emission suggest a pulsar and its wind nebula,perhaps born in the star formation region, which providesa natural source of seed photons for generating upscat-tered gamma-rays (cf., HESS J1837–069/PSR J1838–0655; Gotthelf & Halpern 2008). Although the offsetbetween the point source and the nebula is somewhat un-usual, PWN systems often show complex X-ray morphol-ogy, as revealed by Chandra (e.g., Crab, MSH 15 −
52; seeKargaltsev & Pavlov 2008).The X-ray spectrum of 3XMM J201744.7+365045 is,however, somewhat steep for a pulsar, more consistentwith that of a hidden optical AGN behind the Galacticplane. The radio spectrum also prefers an AGN inter-pretation over a pulsar. The likely coincidence with abright point-like radio source, the lack of detected pul-sations ( > ∼ Sh Fermi source 3FGL J2017.9+3627, 0 . ◦ II region. A search for pulsations from 3FGL J2017.9+3627by the Einstein@Home distributed computing pulsarproject (Anderson et al. 2006; Allen et al. 2013) detecteda 167 ms signal, consistent with a 2 Myr old rotation-powered pulsar (Clark, C., in prep.). The inferred spin-down power of PSR J2017+3625, ˙ E = 1 . × erg s − ,suggests that it likely lies at a distance of 450 pc, givenits gamma-ray flux of 4 . × − erg s − cm − (Aceroet al. 2015) and a gamma-ray efficiency of L GeV / ˙ E = 0 . Fermi pulsar. On the other hand, the lackof significant X-ray detection of a candidate NS in theunpublished 10 ks
Chandra observation (ObsID 14699)suggests that the pulsar is further away. For this ob-servation, we estimate a flux limit of F x (2 −
10 keV) . × − erg s − cm − for a typical pulsar power-lawspectrum ( N H = 1 . × cm − ; Γ = 1 . & L (1 −
10 TeV) ∼ × − erg s − , represents an effi-ciency of L TeV / ˙ E ∼ .
03, plausible for a > yr pulsar(e.g., see Kargaltsev et al. 2013).In conclusion, it is possible that PSR J2017+3625accounts for most, if not all, of the coincident VERI-TAS TeV excess and that the harder TeV photons near Sh NuSTAR sources, the Sh Sh NuSTAR sources presentedin this study. This will require high resolution
Chandra observations to allow a comparison between these sourcesand several overlapping optical/IR stars and an unclassi-fied radio source, that may or may not be related to theX-ray and/or gamma-ray emission.This work was supported under NASA Contract No.NNG08FD60C, and made use of data from the
NuSTAR mission, a project led by the California Institute of Tech-nology, managed by the Jet Propulsion Laboratory, andfunded by the National Aeronautics and Space Admin-istration. We thank the
NuSTAR
Operations, Softwareand Calibration teams for support with the executionand analysis of these observations. This research hasmade use of the
NuSTAR
Data Analysis Software (NuS-TARDAS) jointly developed by the ASI Science DataCenter (ASDC, Italy) and the California Institute ofTechnology (USA). E.V.G. acknowledges partial support https://einstein.phys.uwm.edu/gammaraypulsar/FGRP1 discoveries.html ard X-ray Emission from Sh XMM-Newton