Table Of Content

PreprinttypesetusingLATEXstyleemulateapjv.2/16/10 CHANDRA OBSERVATION OF THE TEV SOURCE HESS J1834–087 Zdenka Misanovic1, Oleg Kargaltsev2 and George G. Pavlov3,4 1SchoolofPhysics,MonashUniversity,Melbourne,3800VIC,Australia([email protected]) 2Dept. ofAstronomy,UniversityofFlorida,Gainesville,FL32611-2055 ([email protected]fl.edu) 3Dept.ofAstronomyandAstrophysics,ThePennsylvaniaStateUniversity, 525DaveyLab.,UniversityPark,PA16802([email protected]) 4St. PetersburgStatePolytechnical University,Polytechnicheskaya ul. 29,St.Petersburg195251, Russia 1 ABSTRACT 1 Chandra ACIS observedthe field of the extended TeV source HESS J1834–087for 47 ks. A previous XMM-Newton 0 EPICobservationofthesamefieldrevealedapoint-likesource(XMMUJ183435.3–084443)andanoffsetregionoffaint 2 extended emission. In the low-resolution, binned EPIC images the two appear to be connected. However, the high- n resolution Chandra ACIS images do not support the alleged connection. Instead, in these images XMMU J183435.3– a 084443isresolvedintoapointsource,CXOUJ183434.9–084443(L≃2.5×1033ergss−1,foradistanceof4kpc;photon J index Γ ≃ 1.1), and a compact (. 20′′) nebula with an isotropic morphology and a softer spectrum (L ≃ 4.2×1033 7 ergs s−1, Γ ≃ 2.7). The nature of the nebula is uncertain. We discuss a dust scattering halo and a pulsar-wind nebula as possible interpretations. Based on our analysis of the X-ray data, we re-evaluate the previously suggested ] interpretations of HESS J1834–087 and discuss a possible connection to the Fermi LAT source 1FGL J1834.3–0842c. E We also obtained an upper limit of 3× 10−14 ergs cm−2 s−1 on the unabsorbed flux of the SGR J1833–0832 (in H quiescence), which happened to be in the ACIS field of view. h. Subjectheadings: gammarays: individual(HESSJ1834–087,1FGLJ1834.3–0842c)—stars: neutron— p supernova remnants: individual (W41=G23.3–0.3)—X-rays: individual (CXOU - J183434.9–084443,XMMU J183435.3–084443) o r t s 1. INTRODUCTION by energetic electrons injected/accelerated in the vicin- [a Surveys of the Milky Way plane with modern Imag- ityofthepulsar(Lorentzfactor∼108),theVHEPWNe ingAtmosphericCherenkovTelescopes(IACTs)havere- are powered by cooled, less-energetic electrons (Lorentz 1 vealed more than 70 sources of very high energy (VHE) factor ∼107), which have propagatedfurther awayfrom v γ-rays1. While X-ray binaries (XRBs), young stel- thepulsar(e.g.,de Jager & Djannati-Ata¨ı2008). TheIC 2 lar clusters and background AGNs have all been pro- cooling time of these γ-ray-producingelectrons is longer 4 posed as possible sources of emission in the TeV en- than that of the electrons that produce X-rays, and 3 ergy range, the majority of the classified Galactic γ- hence, they trace the cumulative history of the PWN, 1 ray sources are thought to be associated with pul- similar to radio emission (the relic PWN model). If the . 1 sar wind nebulae (PWNe) and supernova remnants pulsar contains a hadronic component, the VHE emis- 0 (SNRs)(Aharonian et al.2005,2006;Gallant et al.2008; sioncouldalsobe producedviaa hadronicprocesswhen 1 Kargaltsev & Pavlov 2010). However, about 30% of the relativistic PW hadrons encounter dense material such 1 cataloged VHE sources lack reliable classifications. Fur- as the host SNR shell. An alternative possibility is that v: thermore, for many classified sources there remains an the SNR shock accelerates protons, which then interact i uncertainty about the actual mechanism responsible for withthedenseISM(e.g.,amolecularcloud)andproduce X theproductionoftheVHEγ-rays. Therefore,multiwave- VHE emission. r length studies and modeling of the VHE source popula- The extended source HESS J1834–087 (18.7± 2.0× a tion are of a great interest at the present time. 10−12 photons cm−2 s−1 at E >200 GeV, photon index The two main mechanisms for the production of high- Γ = 2.5±0.2; Aharonian et al. 2005, 2006) is spatially energy γ-rays are the inverse Compton (IC) scattering coincidentwiththe centerofthe radioshellSNRG23.3– of low-energy photons, and the hadronic collisions, in 0.3 (W41; Kassim 1992), making the remnant a promis- whichtheenergeticprotonscollidingwithambientparti- ingcandidateforgeneratingtheobservedγ-rayemission. clesproduceneutralpions,whichdecayintohigh-energy However,thesmallersizeofthe HESSsource2 (diameter photons (π0 → 2γ; see, e.g., Hinton & Hofmann 2009). ∼ 10′ compared to the SNR diameter of ∼ 30′) and the Rotation-powered pulsars emit highly energetic pulsar projectedlocationwellwithintheSNRshellsuggestthat wind (PW), which is responsible for the synchrotronra- the γ-ray emission does not come from the entire SNR diation observed from radio up to MeV energies (see shell as it has been seen in other SNRs firmly associated Kargaltsev & Pavlov 2008, 2010, for recent reviews). It withHESSsources(e.g.,Acero et al.2009). Albert et al. is believed that the γ-ray PWN can be produced when (2006) detected the TeV source with the Major Atmo- the ambient microwave, optical or infrared photons are spheric Gamma Imaging Cerenkov (MAGIC) telescope, upscattered by relativistic PW electrons up to the GeV confirmeditsextension,andmeasuredthefluxconsistent and TeV energies. While the X-ray PWNe are powered with the HESS measurement. The region was also observed in radio with the Very 1 Seehttp://tevcat.uchicago.edu/ 2 SeeTable3andSection5.1inAharonianetal.(2006). 2 Misanovic, Kargaltsev & Pavlov Large Array (VLA), as part of the Galactic Plane sur- 2.1. Spatial analysis vey in continuum and in CO molecular line emission, 2.1.1. Previous multiwavelength observations and also in X-rays with XMM-Newton. Based on these Figure1showstheVLA20-cmimageofW41(Whiteet observations,Tian et al. (2007) estimated the remnant’s age (104 −105 yrs) and the distance (4 kpc), and sug- al. 2005), revealing fine details of the highly structured partial shell with a diameter of ∼30′, and also some ex- gested that the observed γ-ray emission is produced by tendedradioemissionnearthecenter. However,nopoint a hadronic process when the particles accelerated by radio sources are visible, and no detection of radio pul- the SNR shock interact with a giant molecular cloud sations from within the SNR shell has been reported so (GMC) projected near the center of W41 (see also far. The extended VHE source HESS J1834–087 is lo- Yamazaki et al.2006). Anotherexplanationwasputfor- cated near the SNR center. With an angular extent of ward by Bartko & Bednarek (2008), who proposed the ′ ′ 5.4 in radius (see Table 3 in Aharonian et al. 2006) the nearby pulsar J1833–0827 (=B1830–08; 24 away from TeV source appears to be significantly smaller than the the center of the HESS 1834–078) as the source of rel- SNR shell (Figure 1, left). ativistic electrons. These electrons were ejected at the The middle panel of Figure 1 shows the innermost pulsar’s birth place, presumably near the SNR center, region of the extended TeV source as it is seen in a i.e.,withintheextentoftheHESSsource(arelicPWN). 17 ks exposure by XMM-Newton EPIC (for the anal- Gaensler & Johnston (1995) argued that the association ysis of XMM-Newton data see Tian et al. 2007, M+09). of the PSR J1833–0827 and W41 is plausible, given the M+09 detected a total of 16 point-like sources in the pulsar’scharacteristicageof148kyrandthecorrespond- ing projectedvelocity of ∼250km s−1, requiredto move FoV of the MOS detectors, with one of the brightest sources well within the extent of the TeV emis- the pulsar from its birthplace to the current location. sion. Furthermore, in the EPIC images this bright However,ina recentanalysisofthe XMM-Newton data, point-likesource(XMMU J183435.3–084443)appearsto Schmitt et al. (2010, to be submitted to ApJ) rules out be accompanied by extended X-ray emission, also lo- theassociationofB1830–08andW41. Schmittetal. de- cated within the extended TeV source. The spectral tectedapossiblePWNsurroundingthepulsar,butsince analysis of the XMM-Newton data indicates that both the morphology of the nebular emission does not resem- the point-like source and diffuse emission (designated bleabow-shock,theyconcludedthatthepulsardoesnot XMMU J183435.3–084443 and G23.234–0.317, respec- movefastenoughto travelfromthe center ofW41 to its ′ tively; M+09) are highly absorbed and most probably current position, approximately 26 away. non-thermal, although the spectral parameters for the Subsequent analysis of the XMM-Newton observa- diffuseemissionarenotwellconstrainedduetothelarge tions (Mukherjee et al. 2009, hereafter M+09) revealed background contribution. GLIMPSE3 8µm image (Fig- a bright point-like source XMMU J183435.3–084443 at ure 1, bottom) shows no correlation between the diffuse the center of W41 apparently connected to a region of X-ray and IR emission. faint extended emission also located within the extent of the HESS source. M+09 have argued that the new 2.1.2. Images pulsar/PWN candidate is the most likely counterpartof OurChandraobservationresolvedXMMU J183435.3– HESS J1834–087. 084443 into a point source, which we designate To discriminate between these alternative scenarios, CXOU 183434.9–084443 (centered at R.A.=18h 34m we have observed the region of the TeV source with the 34s.944, Decl.=–08o 44′ 43′.′09), and a compact diffuse Chandra ACIS detector. In Section 2 we present the emission extending up to ≃ 20′′ from the point source multiwavelengthdataonHESSJ1834–087,which,inad- (Figure2,top),withmoreorlessisotropicsurfacebright- dition to the Chandra observation,also include archived ness distribution, and an average surface brightness of radio, infrared (IR) and γ-ray data. We further discuss 0.8 cts arcsec−2 in an annulus with radii 1′.′5 and 10′′ ourresultsinSection3andsummarizetheminSection4. (S/N≈14). To separate the X-ray diffuse emission from 2. OBSERVATIONSANDRESULTS the PSF wings of the pulsar candidate image, we have simulated a PSF, using MARX4 and assuming the mea- We observed the region around HESS J1834–087 on sured spectral properties of CXOU J183434.9–084443. 2009 June 7 with Chandra ACIS (ObsID 10126) for The comparisonof the data and point source simulation 47.5 ks (46.5 useful scientific exposure) in timed expo- (Figure 3) suggests a good agreement between the data sure (TE) mode using very faint (VF) telemetry format. and simulation in a small aperture (approximately up The point source was imaged on the S3 chip approxi- ′′ ′′ to 1 radius) while the extended emission is easily seen mately 15 away from the nominal aimpoint, placing a at larger radii. The 2MASS image of the same region large part of the extended emission detected by XMM- (Figure2,bottom)showsnonear-infrared(NIR)sources Newton on the same chip. In addition, S1, S2, S4, I1 ′′ within 6 from CXOU J183434.9–084443position. and I2 chips were also activated during the observation. WesearchedforotherpointsourcesintheACISimage We used CIAO 4.2 and CALDB 4.2.2 for our analysis. in particular within the extent of the HESS source. We Wehaveappliedadditionalparticlebackgroundcleaning for the VF mode, using the pulse heights in the outer 3 The Galactic Legacy Infrared Mid-Plane Survey; see 16 pixels of the 5 × 5 event island to help distinguish http://www.astro.wisc.edu/sirtf between “good” X-ray events and “bad” events that are 4 For the MARX simulation we used the same off-axis and roll most likely associated with cosmic rays. We also made angles,andthedetectorparametersasintherealobservation,but extensiveuseofthearchivalmultiwavelengthdataofthis increased the exposure time by a factor of 100 (rescaling the re- sultingimage)toreducethestatisticalerrors. Wefoundthatusing regionobtainedwithXMM-Newton,VLAandSpitzerto theditherblurparameterof0′.′35givesthebestmatchbetweenthe compare them with the Chandra images. dataandsimulation. Chandra observation of J1834-087 3 found a faint (∼30 cts in 0.5–8 keV) point-like source N at R.A.=18h 34m 42s.6, Decl.=–08o 45′ 01′.′95, approx- ′ imately 2 east of the pulsar candidate (Figure 4, top). The source coincides with the region of the enhanced HESS J1834-087 E “tail” emission detected by M+09, although no point sourcewasdetectedatthatpositionintheXMM-Newton data. This could be due to the lower angular resolution and higher EPIC background or due to the tran- sient nature of the source. The Chandra spectrum of W41 this source is soft, with an absorption column almost an order of magnitude lower than that of the candidate 20 cm pulsar/PWN. Furthermore, the X-ray source coincides with a 2MASS source, suggesting that it is a foreground star. Most of the point-like X-ray sources detected in the XMM-Newton data (Table 1 of M+09) are either outside the FoV of the activated ACIS chips, far off- axis,ornearthe chipedges,butwediddetectthe bright sources11and15onI2andI3chips,respectively(follow- ing the source designation by M+09). The X-ray spec- MOS1+MOS2 tra of these sources are relatively soft, with absorption 2-10 keV columnsN =(1−9)×1021 cm−2,suggestingthatthey H are relatively nearby. Source 15 coincides with a faint 2MASSsource,alreadylistedinTable1ofM+09. Using our more accurate Chandra position, we also find two bright2MASSsourceswithintheChandraerrorcircleof the source 11. We also found that the position of the recently detected SGR J1833–0832(Gelbord et al. 2010; Gög˘u¨¸s et al.2010)iswithintheACISFoV,neartheedge ofthe S1chip(Figure 4,bottom),butwedidnotdetect 2’ any source at this position. After correcting for the vi- gnetting,andassumingthespectralparametersreported byGelbord & Vetere(2010),we obtainedanupper limit of3×10−14 ergscm−2 s−1 ontheunabsorbedfluxofthe SGR in quiesence. While looking for high-energy sources within the ACIS FoV, we found Fermi LAT source 1FGL J1834.3– GLIMPSE 0842c (0.11 Crab in 1–100 GeV, photon index Γ = 2.24±0.04; Abdo et al. 2010) located ∼5′ northwest of 8 micron CXOU 183434.9–084443. The 95% error ellipse of the 1FGL J1834.3–0842c position also includes the north- western part of the SNR shell (Figure 4, bottom). The only X-ray source detected within the error ellipse of the Fermi LAT source is a relatively faint source, CXOU 183430.3–084142.8, with a bright optical coun- terpart, located at the edge of the GeV emission. This source is apparently a foreground star, most likely not associated with the GeV emission. Figure4(top)showsthelarge-scaleACISimage,which covers the region of the extended HESS source. The heavilysmoothedimageshowssomefaintextendedemis- sion,whichmustbeapartofthelarge-scaleemissionseen in the XMM-Newton images. Indeed, a closer comparison of the two images (see Figure 1) shows that the ex- Fig.1.— Top: VLA 20-cm imageof the SNR G23.3–0.3 (W41) tendedemissionvisibleintheACISimagecoincideswith showing its highly structured partial shell with a diameter of ∼30′ and some enhanced emission near the center (White et al. the brightest regions seen in the XMM-Newton data. 2005). The circle (radius 5.′4) shows the position and extent of However,the originofthis emissionremains elusive. We HESSJ1834–087asmeasuredbyAharonianetal.(2006). Middle: seenoclearconnectionbetweenCXOU183434.9–084443 Broad-band(0.2–10keV)XMM-Newton17ksimageoftheinterior andtheextendedemission3′−4′ east-southeastofit. Its of W41 combining MOS1 and MOS2 data (see also M+09). The image with the pixel size of 8′′ is smoothed with a Gaussian of associationwith the SNR is also questionable, as it does FWHM 40′′. The cross marks the position of CXOU 183434.9– notseemtocoincidespatiallywiththeradioshellorwith 084443 (see text). Higher-resolution images of the vicinity of the diffuse radio emission in the SNR interior (Figure 4, CXOU183434.9–084443areshowninFigure2. Thetail-likeemis- middle and bottom). To investigate the nature of this sion detected in the XMM-Newton observation is marked in all panels. Bottom: GLIMPSE (8 micron) image of the same region extended emission, its spectrum must be measured in a oftheskyasinthemiddlepanel. 4 Misanovic, Kargaltsev & Pavlov long observation. parameter when fitting the spectra simultaneously, with XMMU J183435.3–084443 has also been detected by other parameters untied). the Swift (Landi et al. 2006). M+09 included the three After obtaining the best-fit value for N , we fixed H archived Swift observations in their analysis, but could this parameter before fitting the ACIS spectra of the not make any firm conclusions about intrinsic flux vari- point source and the nebula individually and calculat- abilityofXMMUJ183435.3–084443(thecountratemea- ing the errors for other model parameters and for the suredin the third observationshoweda 3σ increase over fluxes (Table 1). The unabsorbed luminosities are cal- the first two, or 2σ increase over the mean rate). To culated for the 0.5–8keV band assuming a distance of search for variability in the ACIS observations, we pro- 4 kpc (Tian et al. 2007). We note that the large ab- duced light-curveswith a bin size rangingfrom ∼102 to sorption measured from the X-ray spectra is more than 103 seconds, but could not see any significantflux varia- an order of magnitude higher than the estimated N HI tions on the time-scales relevant for the ACIS data (i.e., value for the Galaxy in that direction (∼2×1022 cm−2; considering the length and time resolution of the ACIS Dickey & Lockman 1990), implying a large amount of observation). OurChandradataallowedustosearchfor molecular material in the line of sight (see Tian et al. pulsations in the 6–100 s period range, but we did not 2007, and references therein). find any periodic signal. 2.2. Spectral analysis To minimize the contribution from the nebula, we extracted the X-ray spectrum of CXOU J183434.9–084443 ′′ fromacircularregionwitharadiusof1 centeredonthe point source (see Figure 3), while the background photons were extracted from an annulus with the inner and ′′ ′′ outer radii of 1.5 and 5.5, respectively. A total of 220 ′′ counts were detected in the r = 1 aperture in the 0.5– 8 keV band. The comparison with the simulated point sourcesuggeststhatapproximately30ofthesecountsare from the nebula, while the contribution from the background is negligible (1–2 counts), implying an aperture- corrected5 count rate of 4.80±0.04 cts ks−1. For spectral fitting these counts were grouped to a minimum of 15 counts per bin. To measure the nebular spectrum, we extracted a to- ′′ ′′ talof208photonsfromthe1.5–10 annuluscenteredon CXOU J183434.9–084443, while the background contri- ′′ bution was measured from an annulus with 10 < r < 20′′. We estimate that there are ∼50 counts from the true backgroundand ∼10 counts from the PSF wings in the same aperture, implying a count rate of 3.2±1.1 cts ks−1 for the background- and PSF-subtracted emission. The nebula spectrum wasalso groupedto a minimum of 15 counts per bin. The small number of counts and the lack of soft emission (the spectra are virtually cut off below 3 keV; see Figure 5) did not allow us to constrain well the spec- tralmodelparameters. Wefitthespectrawithabsorbed power-law (PL) models and show the results in Table 1 and Figures 5 and 6. We first fit each spectrum inde- pendently and noticed that the absorption columns for the point sourceand for the diffuse emissionregionwere similar (2.8+2.8 and 2.8+2.9 ×1023 cm−2, respectively), −1.3 −1.5 suggesting a common distance. To determine the value of the absorption column more accurately, we then per- formed a simultaneous fit with the MOS1 and MOS2 spectra6 of XMMU J183435.3–084443, assuming again a common absorption column density (i.e., linking this 5 From the radial profile of the simulated PSF and the data (Figure 3) we estimated the point source contribution of 85% in thisaperture. 6 The MOS spectra were only used to measure the absorption column, because they contain not only photons from the pulsar candidate but also a large fraction of the compact nebula that couldnotberesolved. Chandra observation of J1834-087 5 TABLE 1 Absorbed PLmodel fitsto the spectra of CXOUJ183434.9–084443andsurroundingextended emission. NH Γ PLNorm. χ2ν/d.o.f AbsorbedFlux Luminosity 1023 cm−2 10−4cm−2s−1keV−1 10−13ergscm−2s−1 1033ergss−1 Pointsource 2.72(frozen) 1.09+−00..4402 0.95+−00..9427 0.81/13 2.56+−00..2222 2.3+−00..64 Nebula 2.72(frozen) 2.73+−00..5543 6.8+−93..19 0.64/13 0.98+−00..1132 4.1+−41..89 Note. — Theabsorptioncolumndensitywasfrozenatitsbest-fitvaluedeterminedinthejointfitwithXMM-Newtondata(MOS1andMOS2 spectraof the point-like sourceXMMU J183435.3–084443). The observed flux and unabsorbedluminosity, LX = 4πd2FXunabs, are calculatedfor the3–8keVand0.5–8keVenergyband,respectively,forthedistanced=4kpc. Thelisteduncertaintiesareata1σconfidencelevel. Theobserved fluxandunabsorbedluminosityofCXOUJ183434.9–084443areaperture-corrected. 30’’ Fig.3.— Theradialprofiles(in3–8keV)ofthesimulationand datashowingthecontributionoftheextendedemissioninthevicin- ityof the point source. The dashed linewith errorbarsshows the PSFprofileofthepointsourcesimulatedusingMARX(seetext). Thehistogramshowstheradialdistributionofthesurfacebright- nessmeasuredinannularregionscenteredonthepointsource. The backgroundlevelof0.15ctsarcsec−2hasbeensubtractedfromthe data. The dashed linewithout errorbarsshows possiblecontribu- tionofadusthalo(seeSection3.1.1andAppendix,equations A6 andA7). TeV emission could arise either as a result of interaction between the SNR shock and the GMC near W41 (Tian et al. 2007), or alternatively, it could be due to Fig.2.— Top: Chandra ACIS 90′′×90′′ two-color (red: 0.5–2 a PWN powered by CXOU J183434.9–084443 (M+09). keV; blue: 2–8 keV) image of the region around the pulsar and Below we discuss the constraints that our observation compact PWN candidate coinciding with HESS 1834–078. The places on the nature of CXOU J183434.9–084443 and imageisbinnedtoapixelsizeof1′′andsmoothedwithaGaussian HESS J1834–087. We also explore a possible link be- ofFWHM∼4′′. Bottom: 2MASSimageofthe sameregion. The tween these sources and the recently discovered Fermi positionofthepulsarcandidateismarkedbyacirclewitharadius of5′′. LAT source 1FGL J1834.3–0842c which is also located within SNR W41. 3. DISCUSSION 3.1. CXOU J183434.9–084443 and its extended Chandra resolved the compact extended emission sur- emission rounding CXOU J183434.9–084443, which is by far the brightest X-ray source located within the extent of The point source appears to have a strongly absorbed HESS J1834–087. It was previously suggested that the spectrum that fits hard PL. The flux, PL slope, and 6 Misanovic, Kargaltsev & Pavlov 2’ nts s keV−1 −1 102×10−3−3 u o C 0−4 1 × 5 2 χ 0 2 − 5 Energy, keV 2 - 8 keV V−1 10−3 e k s −1 ounts 10−4 C 2 1 χ 0 1 − 2 − 2 5 Energy, keV Fig.5.— Top: The X-ray spectrum of the pulsar candidate extracted from the Chandra observation, and the corresponding best-fit absorbed PL model. The absorption column is frozen to its best fit-value (see Table 1) determined from the fit combining 20 cm theChandraandXMM-Newtondata (seetext). Bottom: TheX- rayspectrum of the PWN candidate extracted from the Chandra dataandthecorrepondingabsorbedPLmodel,assumingthesame absorptioncolumnasforthepulsarcandidate (Table1). SGR 1833-0832 11 SGR J1833-0832 15 1FGL J1834.3-0842c HESS J1834-087 Fig.6.—90%and99%confidencecontoursandtheunabsorbed FWFHigM.∼4.—10′′,ToCph:andLraargAe-CscISaleim, asgmeo(o2t–h8edkeVw)ithencaomGpaaussssiniagnthoef aflnudxePsW(0N.5–c8ankdeiVd)atien.units of10−12 ergscm−2 s−1 forthe pulsar region of the extended tail-like emission detected with XMM- Newton (see Figure 1). The arrow shows the point source with abright2MASScounterpart(seetext). Middle: VLAimageofthe the lack or any counterparts at lower frequencies make sameregion. Theenhanced extended radioemissioninthe center CXOU J183434.9–084443 a plausible candidate for a ofW41doesnotseemtocoincidewiththediffuseX-rayemission. young remote pulsar. However, other options, such as Bottom: Atwo-colorimage(red: 20cm;blue: 0.5-8keV)showing anAGNseenthroughthe Galacticplane,oraquiescent, thewholeW41andalsotheprojectedACISchipsactivatedinthe Chandraobservation. Theextent ofthe TeV source ismarkedby obscured XRB (similar to those discovered with INTE- thebigcircle(radius5.′4)whiletheGeVsourceisshownbyits95% GRAL), remain viable. Although the latter interpreta- positional error ellipse (6.′0×7.′2). Several X-ray sources detected tioncouldexplainthe observedlargeN bytheintrinsic in the FoV are marked, including the star candidate at the edge H oftheGeVemission(smallcircle),andsources11and15asnum- absorption, we do not find any evidence of variability bered in Table 1 by M+09. The position of the SGR J1833–0832 isalsomarkedwithasmallcircle. Chandra observation of J1834-087 7 or an iron line which are commonly seen in obscured contribution of a dust halo. Although the dust halo in- HMXBs (Walter et al. 2006). It seems rather plausible terpretation of the extended emission does not rule out that a large amount of molecular material is generally theyoungpulsaroptionforthepointsource,itallowsfor presentalongthisline ofsight;forinstance,Gög˘u¨¸s et al. additional possibilities, such as a magnetar, an ANG, or (2010)foundN ≃1023 cm−2 whilefittingthespectrum an XRB. H of the nearby SGR 1833–0832in the active state. An important piece of evidence, which we will con- 3.1.2. A pulsar-wind nebula? siderindetail,isacompact,r .20′′,extendedemission, which has been clearly resolved from the point source Let us now assume that CXOU J183434.9–084443 is (Figures 2 and 3). This emission appears to be approx- a pulsar, and a substantial fraction of the observed ex- imately symmetric and exhibits a much softer spectrum tended emission around it is a PWN. The symmetric compared to that of the point source (Γ = 2.7±0.5 vs. morphology of the candidate compact PWN would then 1.1±0.4). Two most likely interpretations are a PWN suggestthattheputativepulsardoesnotmoveveryfast. arounda young pulsar in W41, or a dust scattering halo Even if the pulsar was born at the geometrical center of around the strongly absorbed point source. A combina- the SNR (i.e., ∼ 2′ from its current position) 100 kyrs tion of the two is also a possibility. In fact, a hint of a ago, its transverse speed would only be 22d4 km s−1, plateau(ornon-monotonicbehavior)seeninFigure3be- which is in agreement with the nearly isotropic PWN tween 3′′–10′′ may indicate at two possible components shape. Thus, the large-scale emission east of the pulsar with different dependence on r. isnotexpectedtobeakintolongcollimatedtailsformed behind supersonically moving pulsars, such as observed 3.1.1. A dust halo around a point source? by Kargaltsev et al. (2008). Indeed, the high-resolution The very large absorbing column measured in X-rays X-rayimages ofthose tails show that the surface bright- suggestsasignificantdustcolumnthatshouldleadtoan ness is usually the highest near the pulsar,and it gradu- extended dust scattering halo (e.g., Predehl & Schmitt allydecreaseswith the distancefromthe pulsar,i.e., the 1995). The halo brightness grows with the increas- opposite of what we see in this case. It could be, how- ing intervening dust column, which is usually as- ever, that the large-scale X-ray emission is akin to that sumed to be proportional to the hydrogen absorp- in the Vela X PWN, where the X-ray emission is offset tion column, N , measured from X-ray spectra. from the pulsar and thought to come form the relic pul- H More specifically, the scattering optical depth τ ≃ sar wind crushed and pushed aside by the reverse SNR scat S(N /1022 cm−2)(E/1 keV)−2 for S ≃ 0.5 found by shock (e.g., LaMassa et al. 2008). H Predehl & Schmitt (1995) gives τ < 1 at E > 3.7 The slope of of the extended emission spectrum, fit- scat keV for CXOU J183434.9–084443, whose spectrum vir- ted with the absorbed PL model, is rather steep, Γ = tually cuts off below 3 keV (see Figure 5). Although 2.7±0.8,albeituncertain. ForaPWN,suchasteepslope taking into account multiple scatterings would be more is unusual because the X-ray spectra of most7 PWNe accurate,the largeuncertaintiesof the other parameters haveΓ=1−2(e.g.,seeFigure6inKargaltsev & Pavlov (such as the dust grainproperties and dust distribution; 2008). The steep slope may, however, be indicative of see Appendix) do not warrant this extra complication. strong synchrotron cooling. The ratio of the extended Therefore, we have used the single scattering approxi- and point source luminosities, LPWN/LPSR ∼ 1.8, is mation to calculate the halo profile by convolving the typical of PWNe (see Figure 5 in Kargaltsev & Pavlov spectralintensity of a halo with the detector response in 2008). The spin-downpowerofthe putative pulsarpow- the 3–8 keV energy range. We find that for the parame- ering the PWN can be estimated as E˙ = 4πr2cp , ters Θ = 360′′ and S ≃ 1, and for the dust distribution where r is the terminationshock radius and p s isamthbe s amb function f(x) defined in the Appendix (equation A8), ambient pressure. In the Sedov expansion phase, the the dust halo model generally describes the observedra- pressure inside the SNR could be estimated from its ra- dialprofile (see Figure 3). However,some deviations are dius by using a simple formula p = 3E/4πR3, where amb noticeable, hinting at a possibility of a second emission itisusuallyassumedthatthetotalSNRexplosionenergy component. The dust distribution function f(x), used E =1051 ergs(e.g.,see O’C. Drury et al.2009). The es- in the calculation of the halo model shown in Figure 3, timated pressure pamb,−9 = pamb/10−9 dyne cm−2 and suggeststhatatleastsomedustmustbelocatednearthe the termination shock radius scaled to a plausible value source,inagreementwiththeextremeabsorption,which r = r /1017 cm (corresponding to 2.7′′ at the W41 s,17 s ckonuolwdnbteoaettxriisbtuitnedthtiostrheegilooncaolfmskoylec(uAllabrecrltoeutdsalt.h2a0t0a6r)e. distanceofd4 =4kpc)giveE˙ =4×1036rs2,17pamb,−9 erg s−1 – a value typical for a young Vela-like pulsar. This We should, however, point out that, in addition to estimate is sensitiveto the value of r , whichcould be the freedom of choosing the functional dependence of s,17 dustdistributionwiththedistance,theapproximatedust a factor of a few smaller then the scaling chosen8. The scattering model we use has two other free parameters luminosityofthe compactPWN, ≃4.1×1033d24 ergs−1, (S and Θ), which can attain values within rather broad could also be used to estimate E˙ from the L –E˙ corre- X ranges,dependingontheunknownpropertiesofthedust grains along this particular line of sight (see Appendix). 7 A recently discovered PWN candidate in HESS J1632–408 Therefore, the mere fact that the model qualitatively (Balboetal.2010)alsoshowsasteepspectrumwithΓ=3.4+−00..68. describes the observed radial profile does not guarantee However, the authors did not consider a possible contribution of a dust halo, which may be quite substantial given the very large thattheextendedemissionisindeedadusthalo. Onthe NH≃1.3×1023 cm−2 theymeasuredinX-rays. other hand, the symmetric shape and the softer spec- 8 Toriradiiinbright,well-resolvedPWNeexceedrs byafactor trum of the extended emission support at least partial of2onaverage(Bambaetal.2010). 8 Misanovic, Kargaltsev & Pavlov lation(seeKargaltsev & Pavlov2008,forrecentresults). highly speculative. Although this correlationshows a very largescatter, the On the other hand, a failure to find pulsations from plausible range for E˙ ∼1036−1037 erg s−1 is in a good CXOUJ183434.9–084443,downto a restrictive limit for agreement with the above estimated value. a young pulsar, would strongly suggest that it is an un- related source. In this case, one would be forced to con- 3.2. The nature of HESS J1834–087, 1FGL clude that there is no plausible candidate for a young J1834.3–0842c, and CXOU J183434.9–084443 pulsar within W41. This would imply that the TeV emission is most likely produced by the interaction of HESS J1834–087 belongs to the growing group of protonsacceleratedintheSNRshockwithaparticularly Galactic TeV sources that lack firm classifications or densematerialinthenearbymolecularcloud(Tian et al. identificationswithlower-energycounterparts. Thereare 2007; Albert et al. 2006). Obtaining a firm evidence of currently about 25 TeV sources in this group, of which this would be of a greatinterest because so far there are only five can be considered truly “dark” sources (i.e., veryfew TeV sources (HESS J1745–303A,HESS J1800– those without any detected counterparts) while the rest 240AB,HESSJ1848–018)wheretheinteractionwiththe of the sources are coincident with one or more lower- nearbymolecularcloudisconsideredasapossible(albeit energy sources of a known nature (SNRs, massive open notfirmly established) mechanismofthe VHE emission. stellar clusters, GMCs, X-ray binaries), which could, in Since HESS J1834–087 is located well within the W41 principle, power the TeV emission (although there are shell and has a significantly smaller diameter, it could no clear-cut cases). Spatial coincidence with W41 puts only be associated with a part of the shell projected HESS J1834–087 in the latter category. As such a good within the W41 interior on the sky. The offset between positional match between a TeV source and a bright ra- the TeV and 1FGL sources would be even more difficult dio SNR is very unlikely to happen by chance, HESS to explain than in the PWN interpretation. Also, one J1834–087 must have physical relation to W41. This would have to conclude that the large-scale X-ray emis- naturally leaves only two options for the origin of HESS sion(seeSection2.1.2)mustbeassociatedwiththeSNR J1834–087, a relic PWN or an SNR shock interacting interior, which would require rather high temperatures with the GMC, both of which have been previously dis- to explain the measured hard spectrum (M+09). cussed (M+09, Tian et al. 2007). Keepinginmindtheuncertaintyintheinterpretations, Our detection of extended emission around weopttoplotthemultiwavelengthPWNspectralenergy CXOU J183434.9–084443 does not provide a con- distribution (Figure 7) but refrain from any modeling, vincing argument in favor of either of these two possible which would be premature due to the large number of interpretations. Proving that CXOU J183434.9–084443 optionsaffordedbythe existingdata. Wenote,however, is a pulsar (e.g., by detecting pulsations) would provide that the GeV and TeV spectra generallyseem to match, a very strong support to the relic PWN hypothesis. which hints at their common origin and a possible inac- Despite our highly significant detection of extended curacy in the determination of the 1FGL J1834.3–0842c emission around CXOU J183434.9–084443,it is difficult position. Alternatively, the observed GeV emission co- to unequivocally establish its origin because of the very inciding with the part of the shell could be produced by high absorption and a high likelihood of a significant protons interacting with hadrons in the SNR shell and dust scattering halo. It appears that a better way to producingneutralpionsthatdecayemittingtheobserved go about establishing the nature of CXOU J183434.9– γ-rays. Therelativisticprotonsmaybeacceleratedeither 084443 would be a sensitive timing search in X-rays in the PW (e.g., Arons 2009; Bednarek 2007) or by the and GeV. However, even if CXOU J183434.9–084443 SNR shock interacting with the molecular clouds, which turns out to be a pulsar, there will remain a question has been also previously proposed as the source of the about the origin of the nearby Fermi LAT source 1FGL TeV emission (Tian et al. 2007; Albert et al. 2006). J1834.3–0842c (see Figure 4), which is located within W41 but offset from both the center of the TeV source 4. CONCLUSIONANDSUMMARY and from CXOU J183434.9–084443, and even more OurChandra observationresolvedtheextendedX-ray offset from the patch of the large-scale X-ray emission, emission in the immediate vicinity of CXOU J183434.9– which could be a relic PWN akin to the Vela X PWN 084443, by far the brightest X-ray source within the (see above). Unless the 1FGL source is not real9 or its HESS J1834–087 extent. The source and the accompa- position is inaccurate, a separate interpretation would nyingextendedemissionarestronglyabsorbedinX-rays be required. While awaiting for better Fermi image, we leadingtosignificantuncertaintiesintheirspectra,which can speculate that 1FGL J1834.3–0842c might be able fit PL models with quite different slopes. The spectrum to provide a long-soughtevidence of hadrons in a pulsar of the extended emission is markedly softer suggesting wind (e.g., Arons 2009; Bednarek 2007). Indeed, the thattheemissionmightbeadustscatteringhalo. Alter- positional coincidence with the SNR suggests that the natively, the difference in the slopes could be due to the GeV emission can be produced by pulsar-wind protons strong synchrotron cooling, if the extended emission is interacting with the dense material of the part of the a PWN. Although our approximate dust scattering halo SNRshell,whichintriguinglyturnsouttobethenearest calculations allow us to obtain a reasonable fit to the to the putative pulsar. We must admit, however, that observed radial profile of the extended emission, large until the pulsar nature of CXOU J183434.9–084443 uncertainties in the dust scattering model preclude us is firmly established the above reasoning will remain from definitive judgment on the nature of the extended emission,leaving a PWN optionas a still viable alterna- 9 Itismarkedas“confused”inthe1FGLcatalog, whichmeans tive. The previously reported faint, large-scaleextended that it might be the result of an imperfect model of the diffuse Galacticbackground(seeAbdoetal.2010,fordetails). emissionappearstobedisjoinedfromCXOUJ183434.9– Chandra observation of J1834-087 9 complicates the interpretation of the GeV emission suggesting that either the 1FGL J1834.3-0842c’sposition is inaccurate or it could be an intriguing case of a pulsar wind interacting with the SNR shell via hadronic mechanism. The X-ray timing and deeper imaging observations are required to understand the nature of CXOUJ183434.9–084443andofthe offsetlarge-scaleX- ray emission. Together with improved GeV data, this will make it possible to identify HESS J1834–087 and 1FGL J1834.3-0842c, and determine the nature of their relationship with W41. We also serendipitously observed the recently discov- eredSGRJ1833–0832,which,however,wasnotdetected. We placealimit ofof3×10−14 ergss−1 cm−2 onits unabsorbed flux, which is a factor of 40 dimmer than was measured by Gög˘u¨¸s et al. (2010). Support for this work was provided by the National Fig.7.— Broad-bandspectrum of the PWN candidate showing theX-rayspectrum(ACIS,black)andthecataloguedGeV(Fermi, Aeronautics and Space Administration through Chan- red) and TeV (HESS, blue) emission. The TeV nebula spatially dra Award Number GO9-0076X issued by the Chan- overlaps withthe X-rayPWN candidate, whilethe GeV emission dra X-ray Observatory Center, which is operated by islocated between theTeV nebulaandpartoftheSNRshell(see the Smithsonian Astrophysical Observatory for and on Figure4). behalf of the National Aeronautics Space Administra- 084443inabetterresolutionChandraimage,whichrules tion under contract NAS8-03060. The work was also outapulsartailhypothesisandsuggestthatit mightei- partially supported by NASA grants NNX09AC84G therbe apartofadisplacedrelic PWNora surprisingly and NNX09AC81G, and NSF grants No. 0908733 and hot plasma in the SNR interior. 0908611. G. G. P was partly supported by the Ministry Adjacent to the TeV source, and overlapping with a ofEducationandScienceoftheRussianFederation(con- part of the SNR shell, there is extended GeV emission, tract 11.634.31.0001). the Fermi LAT source 1FGL J1834.3-0842c. The off- setfromHESS J1834–087andCXOUJ183434.9–084443 APPENDIX DUSTHALO MODEL Dust halos, often seen around bright point-like X-ray objects, are formed by scattering of source X-ray photons on dust grains. Here we will only discuss the case of dust optically thin with respect to the photon scattering, τ .1, scat andconsideronlyazimuthallysymmetrichalos(whichimplies thatthedustdistributionacrosstheline ofsight(LOS) is uniform within the interval of angles θ at which we see the halo). In this case the spectral halo intensity (phcm−2 s−1 keV−1 arcmin−2) is given by the equation 1 f(x) dσ (E,θ ) s s I (θ,E)=F(E)N dx , (A1) halo H x2 dΩ Z0 s where F(E) is the point source spectral flux (photonscm−2 s−1 keV−1), x=(D−d)/D is the dimensionless distance from the X-ray source to the scatterer (D and d are the distances from the observer to the source and the scatterer, respectively), θ ≃ θ/x (for small angles) is the scattering angle, f(x) is the dimensionless dust density distribution s 1 alongthe LOS( f(x)dx=1),anddσ (E,θ )/dΩ isthe differentialscatteringcrosssectionper onehydrogenatom, 0 s s s averagedover the dust grain distribution over sizes and other grain properties (see, e.g., Mathis & Lee 1991). R To understand the halo properties from simple analytical expressions, we will use the Rayleigh-Gans (RG) approximation, in which the total scattering cross section ∝ E−2; this approximation works better for higher energies, E &0.5–2keV,depending onthe dust model. For some dustmodels, the averageddifferential crosssectionin the RG can be approximated as (Draine 2003) dσ (E,θ ) σ (E) 1 s s ≈ s , (A2) dΩ πθ2 (1+θ2/θ2 )2 s s,50 s s,50 where Θ S θ ≈ and σ (E)≈ 10−22cm2 (A3) s,50 E s E2 arethemedianscatteringangleandthetotalcrosssection,respectively;E istheenergyinkeV.TheconstantΘinfirst ′′ eq. (A3) depends on the dust model; Draine (2003) derived Θ = 360 from the dust model of Weingartner & Draine ′ (2001), while Bocchino et al. (2010) found Θ=7.4 for the model of Smith & Dwek (1998). 10 Misanovic, Kargaltsev & Pavlov It follows from the second eq. (A3) that the scattering optical depth is τ (E)≈SN E−2. (A4) scat H,22 The factor S in the second eq. (A3) is a constant of the order of 1; e.g., S ≈ 1.3 from Figure 6 of Draine (2003), while Predehl & Schmitt (1995) found a mean value S ≈ 0.49 for a number of halos observed with ROSAT (but the scatter was very large), while Mathis & Lee (1991) discuss models with S =0.903, 1.09, and 0.47 (see their Table 1). Costantini (2004; PhD thesis) estimated τ (1keV) for a number of halo sources observed with Chandra; the values sca of S derived from her results show a very strong scatter, S from 0.018 to 2.26. The scatter itself may be natural, as the dust properties may be different for different sources. It should be noted that the correlationof τ (1keV) with visual extinction A : sca V τ (1 keV)=(0.056±0.01)A (A5) scat V (Predehl & Schmitt 1995) is better than that with N , but A is rarely known for the objects of interest. H V Substituting (A2) and (A3) in (A1), we obtain the spectral intensity profile S 1 f(x) θE 2 −2 I (θ,E)=F(E)N 1+ dx. (A6) halo H,22πΘ2 x2 xΘ Z0 " (cid:18) (cid:19) # For comparison with the point source + halo profile observed in the energy range E < E < E , the sum of the 1 2 spectralintensities shouldbe convolvedwiththe detector response,withallowanceforthe imagespreadcausedby the telescope and the detector. We have checked that the energy redistribution in the detector only slightly affects the broadband radial profile for a smooth incident spectrum. Therefore, assuming the observable halo size to be much larger than the PSF width, we obtain E2 S 1 f(x) θE 2 −2 I (θ)= dEA (E)F(E) ψ(θ,E)+N dx 1+ , (A7) obs eff  H,22πΘ2 x2 xΘ  ZE1  Z0 " (cid:18) (cid:19) #  whereA (E)isthedetector’seffectivearea,andψ(θ,E)isthenormalizedPSF,whichcanbetakenfromasimulation eff   (e.g., with MARX). The first term in Equation(A7) corresponds to the point source, while the second term describes the halo. The above equation can be integrated for a given set of halo parameters, and compared directly with the data. In ′′ particular, for the dust halo model shown in Figure 3 we picked Θ=360 , S =1, and the dust distribution function x−1, x≤x f(x)= o o (A8) 0, x>x o (cid:26) with x =0.25. According to (A6), this distribution function corresponds to o F(E)N Sa a I (θ,E)= H,22 arctana− halo 2π(x Θ)2 1+a2 o (cid:18) (cid:19) ≈ F(E)NH,22Sa 1 a≫1 (A9) 4x2Θ2 (3π)−1a3 a≪1 o (cid:26) where a=x Θ/θE. o REFERENCES Abdo,A.A.,etal.2010, ApJS,188,405 Gaensler,B.M.,&Johnston, S.1995,MNRAS,275,L73 Acero,F.,Ballet,J.,Decourchelle, A.,Lemoine-Goumard,M., Gallant,Y.A.,etal.2008, inAmericanInstituteofPhysics Ortega,M.,Giacani,E.,Dubner,G.,&Cassam-Chena¨ı,G. ConferenceSeries,Vol.983,40YearsofPulsars:Millisecond 2009, A&A,505,157 Pulsars,MagnetarsandMore,ed.C.Bassa,Z.Wang, Aharonian,F.,etal.2005, Science,307,1938 A.Cumming,&V.M.Kaspi,195 Aharonian,F.,etal.2006, ApJ,636,777 Gelbord,J.M.,etal.2010, GRBCoordinates Network,Circular Albert,J.,etal.2006,ApJ,643,L53 Service,10526,1,526,1 Arons,J.2009,inAstrophysicsandSpaceScienceLibrary,Vol. Gelbord,J.M.,&Vetere,L.2010,GRBCoordinatesNetwork, 357,AstrophysicsandSpaceScienceLibrary,ed.W.Becker, CircularService,10531, 1,531,1 373 Gö˘gu¨¸s, E.,etal.2010,ApJ,718,331 Balbo,M.,Saouter,P.,Walter,R.,Pavan, L.,Tramacere,A., Hinton,J.A.,&Hofmann,W.2009,ARA&A,47,523 Pohl,M.,&Zurita-Heras,J.2010, A&A,520,A111 Kargaltsev,O.,Misanovic,Z.,Pavlov,G.G.,Wong,J.A.,& Bamba,A.,Mori,K.,&Shibata, S.2010,ApJ,709,507 Garmire,G.P.2008, ApJ,684,542 Bartko,H.,&Bednarek,W.2008,MNRAS,385,1105 Kargaltsev,O.,&Pavlov,G.G.2008, inAmericanInstitute of Bednarek,W.2007,Ap&SS,309,179 PhysicsConferenceSeries,Vol.983,40YearsofPulsars: Bocchino,F.,Bandiera,R.,&Gelfand,J.2010,A&A,520,A71 MillisecondPulsars,Magnetars andMore,ed.C.Bassa, deJager,O.C.,&Djannati-Ata¨ı,A.2008,ArXiv0803.0116 Z.Wang, A.Cumming,&V.M.Kaspi,171 Dickey,J.M.,&Lockman,F.J.1990,ARA&A,28,215 Draine,B.T.2003,ApJ,598,1026