Isolated X-ray -- infrared sources in the region of interaction of the supernova remnant IC 443 with a molecular cloud
A. M. Bykov, A. M. Krassilchtchikov, Yu. A. Uvarov, H. Bloemen, F. Bocchino, G. M. Dubner, E. B. Giacani, G. G. Pavlov
aa r X i v : . [ a s t r o - ph ] J a n T HE A STROPHYSICAL J OURNAL , V . 677, A PRIL
IN PRESS
Preprint typeset using L A TEX style emulateapj v. 08/13/06
ISOLATED X-RAY–INFRARED SOURCES IN THE REGION OF INTERACTION OF THE SUPERNOVA REMNANTIC 443 WITH A MOLECULAR CLOUD
A.M. B
YKOV , A.M. K RASSILCHTCHIKOV , Y U .A. U VAROV , H. B LOEMEN , F. B OCCHINO ,G.M. D UBNER , E.B. G IACANI , G.G. P AVLOV The Astrophysical Journal, v. 677, April 2008, in press
ABSTRACTThe nature of the extended hard X-ray source XMMU J061804.3+222732 and its surroundings is investigatedusing XMM-Newton, Chandra, and Spitzer observations. This source is located in an interaction region of theIC 443 supernova remnant with a neighboring molecular cloud. The X-ray emission consists of a number ofbright clumps embedded in an extended structured non-thermal X-ray nebula larger than 30 ′′ in size. Someclumps show evidence for line emission at ∼ ∼ ′′ × ′′ size is prominent in the 24 µ m, 70 µ m, and 2.2 µ m bands, adjacent to a putative SiK-shell X-ray line emission region. The observed IR/X-ray morphology and spectra are consistent with thoseexpected for J/C-type shocks of different velocities driven by fragmented supernova ejecta colliding with thedense medium of a molecular cloud. The IR emission of the source detected by Spitzercan be attributed to bothcontinuum emission from an HII region created by the ejecta fragment and line emission excited by shocks.This source region in IC 443 may be an example of a rather numerous population of hard X-ray/IR sourcescreated by supernova explosions in the dense environment of star-forming regions. Alternative Galactic andextragalactic interpretations of the observed source are also discussed. Subject headings:
ISM: individual (IC 443) — supernova remnants — X-rays: ISM INTRODUCTION
The energy release and the ejection of nucleosynthesisproducts by supernovae (SNe) events are of great impor-tance for our understanding of the physics of the interstellarmedium (ISM). The mixing of the ejected metals with the sur-rounding matter is of special interest when a SN occurs in amolecular cloud, which may cause further star-forming activ-ity.Optical and UV studies of the structure of SN remnants(SNRs) have revealed a complex metal composition of ejectaand the presence of isolated high-velocity ejecta fragmentsinteracting with surrounding media. The most prominentmanifestations of this phenomena are the fast moving knotsobserved in some young “oxygen-rich” SNRs, such as theGalactic SNRs Cas A (e.g., Chevalier & Kirshner 1979; Fe-sen et al. 2006), Puppis A (Winkler & Kirshner 1985),G292.0+1.8 (e.g. Winkler & Long 2006), and also N132Din the LMC and 1E 0102.2–7219 in the SMC (e.g. Blair et al.2000).Ballistically moving ejecta fragments of SNRs can be con-sidered as a class of hard X-ray sources. The prototype wasobserved in the Vela SNR (Aschenbach, Egger, & Trümper1995; Miyata et al. 2001). A massive individual fragmentmoving supersonically through a molecular cloud can have aluminosity L x > ∼ ergs s - in the 1–10 keV band, and isobservable with XMM-Newtonand Chandraat a few kpc dis- A.F. Ioffe Institute for Physics and Technology, St. Petersburg, Russia,194021; [email protected] SRON Netherlands Institute for Space Research, Sorbonnelaan 2, 3584CA Utrecht, The Netherlands INAF – Osservatorio Astronomico “G.S.Vaiana", Piazza del Parlamento1, 90134 Palermo, Italy Instituto de Astronomía y Física del Espacio (IAFE), CC 67, Suc. 28,1428 Buenos Aires, Argentina Pennsylvania State University, 525 Davey Laboratory, University Park,PA 16802 tance (Bykov 2002, 2003). Its X-ray emission is expected toconsist of two components. The first one is thermal X-rayemission from the hot shocked ambient gas behind the frag-ment bow shock, with a spectrum of an optically thin ther-mal plasma of an ISM-cloud abundance. The second emissioncomponent is nonthermal; the interaction of fast electrons ac-celerated at the fragment bow-shock with the fragment bodyproduces a hard continuum as well as line emission (X-rayand IR), including the K-shell lines of Si, S, Ar, Ca, Fe, andother elements ejected by SN. Detection of the X-ray lineemission would help distinguish an ejecta fragment from theother possible source of hard continuum emission associatedwith a SNR, namely, a pulsar wind nebula (PWN).A young SNR of an age of a few thousands years inter-acting with a molecular cloud can produce hundreds of X-ray sources associated with isolated ejecta fragments. Theyshould be particularly numerous in starforming regions likethose in the Galactic center region, where young core-collapsed supernovae in or near molecular clouds are expectedto be present in abundance. The expected observational ap-pearance of isolated ejecta fragments in a molecular clouddiffers from what is seen in the Vela SNR. Ejecta fragmentsinteracting with a dense molecular cloud are slowed down andcrushed, and they are generally more bright. We will arguehere that the X-ray emission spectra of ejecta fragments in amolecular cloud may be dominated by hard non-thermal com-ponents, because a powerful but very soft thermal componentcould be heavily absorbed. On the other hand, the spectraof fast supernova ejecta fragments propagating in a tenuousplasma, as it is the case in the Vela SNR, would be long-lived,less luminous and dominated by thermal emission.The present paper focuses on IC 443. This is a SNR of amedium age, estimated by Chevalier (1999) to be ∼ much smaller than in a young SNR (pos-sibly, only a few). It is, however, the best and most reliable A.M.Bykov et al.laboratory to study this phenomenon since there are only veryfew examples of clearly established SNR-cloud interactions.IC 443 (G189.1+3.0) is an evolved SNR of about 45 ′ size ata distance of 1.5 kpc (e.g. Fesen & Kirshner 1980). Radio ob-servations of IC 443 (e.g. Braun & Strom 1986; Green 1986;Leahy 2004) show two half-shells. This appearance is proba-bly due to interaction of the SNR with a molecular cloud thatseems to separate the two half-shells. The molecular-cloudmaterial has a torus-like structure (Cornett, Chin & Knapp1977; Burton et al. 1988; Troja, Bocchino, & Reale 2006),that can be interpreted as a sheet-like cloud first broken bythe expanding pre-supernova wind and then by the SNR blastwave. Plenty of evidence for shock-excited molecules in thisregion has been found (e.g. DeNoyer 1979; Burton et al.1988; Dickman et al. 1992; Turner et al. 1992; van Dishoeck,Jansen, & Phillips 1993; Tauber et al. 1994; Richter, Graham,& Wright 1995; Cesarsky et al. 1999; Snell et al. 2005). Thecomplex structure of the interaction region, with evidence formultiple dense clumps, is seen in 2MASS images (e.g. Rhoet al. 2001). Three OH (1720 MHz) masers were found inIC 443 (Claussen et al. 1997; Hewitt et al. 2006, and refer-ences therein).Soft X-ray maps of IC 443 based on ROSAT data (Asaoka& Aschenbach 1994) and recent radio observations (Leahy2004) suggest that another SNR, G189.6+3.3, is seen in theIC 443 field (see also the XMM-Newton study by Troja,Bocchino, & Reale 2006). This makes the multiwavelengthobservational picture even more complex to interpret.The field of IC 443 was observed in X-rays with HEAO1 (Petre et al. 1988), Ginga (Wang et al. 1992), ROSAT(Asaoka & Aschenbach 1994), ASCA (Keohane et al. 1997;Kawasaki et al. 2002), BeppoSAX (Preite-Martinez et al.2000; Bocchino & Bykov 2000), Chandra(Olbert et al. 2001;Bykov, Bocchino, & Pavlov 2005; Gaensler et al. 2006;Weisskopf et al. 2007), XMM-Newton (Bocchino & Bykov2001, 2003; Troja, Bocchino, & Reale 2006), and RXTE(Sturner, Keohane, & Reimer 2004).The X-ray emission of IC 443 below 4 keV is dominatedby a number of thermal components (e.g., Petre et al. 1988;Asaoka & Aschenbach 1994; Kawasaki et al. 2002; Troja,Bocchino, & Reale 2006). The thermal-emission morphologyis center-filled, with soft emission filaments visible at ener-gies below 0.5 keV. A gradient of X-ray surface brightnessat the SNR limb was found, as well as strong variations ofabsorbing column density N H , which indicates the complexmolecular-cloud environment of IC 443 in the southern partof the remnant (e.g. Asaoka & Aschenbach 1994).ASCA observations have established that the hard X-rayemission of IC 443 (above 4 keV) is dominated by localizedsources in the southern part of the remnant (Keohane et al.1997). In XMM-Newton observations Bocchino & Bykov(2003; BB03 hereafter) found 12 sources with fluxes over10 - ergs cm - s - in the 2–10 keV band. Six of the de-tected sources are located in a relatively small, of 15 ′ × ′ size, region projected onto the molecular cloud in the South-Eastern part of IC 443. BeppoSAX MECS observations(4–10 keV) showed two sources, 1SAX J0617.1+2221 and1SAX J0618.0+2227, with evidence from the BeppoSAXPDS for the presence of hard emission up to 100 keV forthe former (Bocchino & Bykov 2000). Observations of thissource by Chandra (Olbert et al. 2001; Gaensler et al. 2006;Weisskopf et al. 2007) and XMM-Newton (Bocchino &Bykov 2001) established its plerionic nature. Leahy (2004)argued that the pulsar that powers this plerion is associated with G189.6+3.3 rather than IC 443. The nature of the sec-ond hard source – 1SAX J0618.0+2227 – remained unknown.This source, the brightest in the region (excluding the ple-rion), was resolved with XMM-Newton into two sources –the extended XMMU J061804.3+222732 (of ∼ ′′ size)and the point-like XMMU J061806.4+222832. We will callthem Src 1 and Src 2 respectively (note that the sourceswere listed as Src 11 and Src 12 in BB03). The position ofXMMU J061804.3+222732 in the remnant is illustrated inFigures 1 and 2. A dedicated Chandra observation of Src 1has revealed a complex structure of a few bright clumps em-bedded in extended emission of > ′′ size (Bykov, Bocchino,& Pavlov 2005; BBP05 hereafter). The brightest clumps arethe extended Src 1a and the point-like Src 1b. The appar-ent position of the source in a SNR – molecular cloud inter-action region naturally leads to SNR-related interpretations.The observed X-ray morphology of Src 1 and the spectra of itscomponents are consistent with expectations for a SN ejectafragment interacting with a dense ambient medium. Alterna-tively, Src 1 could be interpreted as a PWN associated witheither IC 443 or G189.6+3.3 (BBP05). However, one cannotexclude the extragalactic origin of the source, that is discussedin some detail in Section 4.7.IC 443 is a candidate counterpart of the EGRET γ -raysource 3EG J0617+2238, with a flux of about 5 × - cm - s - above 100 MeV (Esposito et al. 1996). The spec-trum of Src 1 extrapolated into the EGRET range is con-sistent with that of 3EG J0617+2238 (BBP05). Also theposition of Src 1 is consistent (albeit marginally) with thatof 3EG J0617+2238. Such a γ -ray luminosity can be ex-pected for both the fragment and PWN interpretations. Theforthcoming GLAST mission (e.g. Johnson 2006) willbe able to provide an accurate position and spectrum of3EG J0617+2238, thus helping to solve the issue. Src 1lies far away from the 99% error circle of the TeV-regimesource recently reported by MAGIC (Albert et al. 2007) inthe Western part of IC 443 field. The apparent position of TeVMAGIC source is close to the 1720 MHz OH maser detectedby Claussen et al. (1997).We present here new results of a deep 80 ks observation ofthe region with XMM-Newton, imaging of the region with theSpitzerinfrared observatory, and a new analysis of VLA radioobservations. In § 2 a combined analysis of the new XMM-Newton observations and all the previous high-resolution X-ray data from Chandra and XMM-Newton is presented, in-cluding images, spectra, and time variations in the X-ray do-main. In § 3 archival radio (VLA), IR (2MASS and SpitzerMIPS), and optical (POSS-II) data are used to constrain thenature of Src 1. A discussion of the obtained results and fu-ture prospects are presented in § 4. X-RAY DATA ANALYSIS
XMMU J061804.3+222732 (Src 1) was observed withXMM-Newton in 2000 and 2006, and with Chandra in 2000and 2004 (Table 1). CIAO v.3.0 with CALDB v.3.2.2 wasused for Chandra data processing, SAS v.20060628_1801-7.0.0 for XMM-Newton data processing, and the HEA-SOFT 6.1 suite, including XSPEC v.12.3, for spectral fitting.In the course of XMM-Newtondata reduction, patterns 0–4of EPIC-PN detector events and patterns 0–12 of MOS eventswere used. To filter out the periods of enhanced particle back-ground, we applied a standard method of soft flare detection . http://heasarc.gsfc.nasa.gov/docs/xmm/abc/node7.html -ray–IR sources in IC 443 3 F IG . 1.— Wide-field views of SNR IC 443. Left:
XMM-Newton 2–8 keV image with Spitzer MIPS 24 µ m contours overlaid. Right:
Spitzer MIPS image at24 µ m with VLA 1.4 GHz contours overlaid. The images are produced from the XMM-Newton observations 0114100101–0114100601 and 0301960101, andSpitzer MIPS observations r4616960, r4617216, and r4617472. The white arrow points to the studied region. The filtering did not significantly change the good time for thedata taken in 2000, but reduced it in the dataset of 2006.In the Chandraobservation of the year 2000, Src 1 is offset17 . ′ was used to correct the Chandra event filesfor the systematic position offset.Figure 1 shows a wide-field view of IC 443, both in X-raysand in the medium IR 24 µ m band. Two X-ray images ofthe source region of interest here, obtained with Chandra andXMM-Newton, are presented in Figure 2. The coordinates ofSrc 1a and Src 1b obtained from the Chandra data with thecelldetect algorithm ( α = 06 h m . s , δ + ◦ ′ . ′′
1, and α = 06 h m . s , δ = + ◦ ′ . ′′
8, respectively) do not dif-fer from those reported by BBP05, as well as the position ofSrc 2 ( α = 06 h m . s , δ = + ◦ ′ . ′′ α = 06 h m . s , δ = + ◦ ′ . ′′
9. In addition, the XMM-Newton data show ev-idence for an extended ‘bridge’ of diffuse emission betweenSrc 1 and Src 3 (see Figures 1, 2, and 6). On the timescaleof about 6 years, covered by the XMM-Newton observationsused here (Table 1), it is not possible to detect a proper mo-tion of the sources at a 1.5 kpc distance if they move with atransverse velocity < ∼ - . Source spectra
The line of sight to the source region intersects the super-nova shell(s), the interior of the remnant, and the molecularcloud, each having different physical parameters. Therefore,the X-ray emission detected along the line of sight has mul-tiple thermal and nonthermal components, in addition to theGalactic background. A SNR shell can be characterized bya power-law emission spectrum, the hot low-density interiors http://asc.harvard.edu/cal/ASPECT/fix_offset/fix_offset.cgi — by thermal plasma emission, and a dense molecular cloudmay cause an appreciable absorption.We have performed a series of spectral studies of Srcs 1, 2,and 3, the results being presented in Table 2 (power-law mod-els), Table 3 (thermal models), and in the text of this subsec-tion. Mekal thermal plasma models were used (e.g. Kaastra1992) and the wabs model for absorption was applied (Mor-rison and McCammon 1983). Because of strong gradients inthe soft X-ray surface brightness, one should carefully chooseregions for spectral analysis. For the Chandra analysis, cir-cular source regions of 10 ′′ radius surrounded by 10 ′′ –widebackground annuli were used. For XMM-Newton, whosePSF is broader, source regions of 20 ′′ radius and annuli with10 ′′ width were used. That results in an underestimation ofa source flux caused by subtraction of the source photons re-garded as background photons due to the wide XMM-NewtonPSF wings. The effect is estimated to be about 20% of thesource flux. That may account for the differences in the Chan-draand XMM-Newtonnormalizations of the fitted source fluxreported in Table 2. However, the fitted power-law indices andabsorption values are not affected.When comparing the Chandra and XMM-Newton spectra,such enlarged source extraction regions were also used forthe Chandradata. For the spectral fitting, the count rate spec-tra were grouped with a minimum of 15 counts per bin. Ta-ble 2 summarizes the results of spectral fitting for Src 1, Src 2,and Src 3 with an absorbed power-law (using the wabs modelfor absorption; Morrison and McCammon 1983). Thermalplasma models for Src 1 and Src 3 yield very high and poorlyconstrained values of the temperature. However, the spectrumof Src 2 can be as well described with a mekal model as an ab-sorbed emission of thermal plasma (see Table 3).The high angular resolution of Chandra ACIS is useful forstudying the spectra of the substructure of Src 1 in more de-tail. A 2 ′′ × ′′ elliptical source region around Src 1a, anda 2 ′′ radius source region around Src 1b were selected. Forboth regions, the background counts were extracted from anannulus with inner radius of 10 ′′ and outer radius of 20 ′′ cen- A.M.Bykov et al. TABLE 1X-
RAY OBSERVATIONS OF
XMMU J061804.3+222732.Obs ID Observatory Instrument Date of observation Exposure Good time(YYYY-MM-DD) (ks) (ks)760 Chandra ACIS-S 2000-04-10/11 11.5 9.6PN 23.2 19.50114100301 XMM MOS1 2000-09-27 25.6 25.1MOS2 25.6 25.14675 Chandra ACIS-S 2004-04-12/13 58.4 56.2PN 79.9 52.90301960101 XMM MOS1 2006-03-30/31 81.6 67.6MOS2 81.6 68.2F IG . 2.— X-ray images of the region under consideration. The data are from the Chandraobservation 4675 and from XMM-Newtonobservations 0114100101–0114100601 and 0301960101. Left : Chandrahardness ratio image (a 3.0–7.0 keV countrate map divided by a 0.5–3.0 keV countrate map) together with a contouroutlining the extended near-IR source 2MASS J06180378+2227314 (with the L-shaped Si K-shell X-ray emission line region inside). Broad band Chandramapsof the region have been presented by BPP05.
Right : XMM-Newton image in the 2–8 keV band with superimposed contours of 2.122 µ m H emission of Burtonet al. (1987) [adopted from van Dishoeck, Jansen, & Phillips (1993)], indicating the presence of a shocked molecular cloud. The studied X-ray sources areindicated as well as the clump D of van Dishoeck, Jansen, & Phillips (1993). tered between the two sources. The spectrum of Src 1a canbe described as an absorbed ( N H = (1.1 ± × cm - )power-law with photon index Γ = 1.5 + . - . and a thermal plasmacomponent with T = 0 . + . - . keV. The reduced χ of this fit is0.71 at 14 d.o.f. The spectrum of Src 1b can be described asan absorbed ( N H = 1.2 + . - . × cm - ) power-law with pho-ton index Γ = 2.0 + . - . . The reduced χ of this fit is 0.73 at 9d.o.f. The quoted errors are at the 90% confidence level (forone interesting parameter). The confidence contours for thespectral parameters of the whole Src 1 are shown in Figure 3.These results are in a good agreement with those obtained byBBP05.The best-fit value of N H for a power-law model of Src 1a(that dominates the 10 ′′ radius Src 1 in Table 2, line 4) is wellbelow that for the South-Eastern region of IC 443 obtainedwith ROSAT and XMM-Newton (7 × cm - ). If one as-sumes that Src 1a, Src 1b and the extended hard emission areat the same column density N H ∼ × cm - , one has to adda soft thermal component (of T > ∼ N H = 7 × cm - ) can be understood— if Src 1a is a pulsar wind nebula — as a combination ofthermal emission from the surface of a neutron star plus non-thermal emission from the neutron star magnetosphere and/orsurrounding pulsar wind nebula.It should be noted that the high plasma temperatures ob-tained to fit a thermal model to the Chandra spectrum of theregion of a 20 ′′ radius around Src 1 are due to a contribution ofhard spectra of Src 1a, Src 1b and their neighbourhood. If oneconsiders the spectrum of this circle region with a removedinner circle of 11 ′′ radius surrounding Src 1a and Src 1b (theresulting annulus is named Src 1* in Table 3), one obtains afit with N H = 1.1 + . - . × cm - , T = 5.3 + . - . keV with χ =0.93 at 78 d.o.f.A line feature centered at ≈ . F IG . 3.— Confidence contours (68%, 90% and 99% confidence levels) for the spectral fit parameters of Src 1 spectra obtained with Chandra. Upper and central:
Temperature and photon index vs. hydrogen column density for a two-component thermal plasma + power law model.
Lower:
Photon index vs. hydrogen columndensity for an absorbed power law model. The outer contour on the central panel runs outside the shown parameter space.
MOS1 + PN + ACIS spectrum of the L-shaped region canbe modeled as an absorbed N H =(0.3 ± × cm - ] power-law with Γ =1.1 + . - . and a possible Si line at 1.9 ± ′′ radius.The background counts were taken from an annulus with theinner radius of 3 ′′ and the outer radius of 10 ′′ . The spectrum is shown in Figure 4 (bottom row). It contains a feature at3.7 keV that is possibly due to an Ar emission line. There is,however, an alternative possible interpretation of the line as aredshifted Fe K line that assumes that Src 3 is extragalactic.The XMM-Newton spectrum of the faint bridge be-tween Src 1 and Src 3 can be modeled either as absorbed( N H =0.5 + . - . × cm - ) thermal plasma (T = 5.5 + . - . keV)emission with χ ν = 0.79 at 74 d.o.f. or as an absorbed A.M.Bykov et al. F IG . 4.— Upper Left : X-ray spectrum of the L-shaped region (shown in Fig.2) derived from a combination of the MOS1, PN (obs. 2006), and ACIS (obs.2004) data. The model (histograms) corresponds to an absorbed [ N H =(0.3 ± × cm - ] power-law with Γ =1.1 + . - . and a possible Si line at 1.9 ± Upper Right : 68%, 90%, and 99% confidence contours for the parameters of the 1.8 keV Si K-shell line for the L-shaped region, shown inFig.2, obtained from combined data of the MOS1, PN (obs. 2006), and ACIS (obs. 2004) detectors.
Lower Left : X-ray spectrum of Src 3 extracted from Chandraobservations. The spectrum is modeled as an absorbed [ N H =0.6 + . - . × cm - ] power-law with Γ =1.8 + . - . and a possible Ar line at 3.7 ± + . - . keVwidth. Lower Right : 68%, 90%, and 99% confidence contours for the parameters of the 3.7 keV Ar K-shell line for Src 3. All the quoted errors are at the 90%confidence level. ( N H =0.6 + . - . × cm - ) power-law of photon index Γ =2.0 + . - . with χ ν = 0.74 at 74 d.o.f.The 20 ′′ –aperture Chandra spectra of the studied sourcesSrc 1, Src 2 and Src 3 can be simultaneously modeled by anabsorbed ( N H =0.8 + . - . × cm - ) power-law of the photonindex Γ = 1.8 ± χ ν = 1.1 at 315 d.o.f. or by a ther-mal plasma (mekal) model with N H =(0.6 ± × cm - ,T=12 + - keV with χ ν = 1.1 at 315 d.o.f.Using all the data available, we found that Src 1 has notshown a significant time variation, its flux being consistentwith that originally obtained by BB03. Some evidence (at the90% confidence level) is found for time variation in the un-absorbed flux of Srcs 2 and 3, as illustrated in Fig. 5. Theunabsorbed flux of Src 2 in the 0.5–10 keV band decreasedbetween 2000/09 and 2006/03 with 99% confidence. In the 2–10 keV band the flux increased between 2000/04 and 2000/09and later decreased with 90% confidence. The unabsorbedflux of Src 3 increased between 2000/09 and 2006/03 with90% confidence. More observations are needed to firmly con-clude on the issue. RADIO, IR AND OPTICAL DATA ANALYSIS
VLA data analysis
A radio image of IC 443 was obtained from archival VLA data obtained in 1997 at 1465 MHz from observations inthe C and D arrays. The interferometric image has an an-gular resolution of 14 . ′′ × . ′′
2. To recover informationat all spatial frequencies, the synthesis data were combinedwith single dish data from the survey at 1408 MHz carriedout with the 100 m MPIfR telescope (Reich, Reich & Fürst1990). The final high fidelity image has an angular resolu-tion of 37 . ′′ × . ′′
3, PA = 46.5 o , and an average rms noise ∼ . S = (58 ±
2) Jy, is in a good agreement with thetotal integrated flux obtained from single dish observations( S ∼
60 Jy, Mufson et al. 1986), which assures the accuracyof flux density estimates over selected portions of the SNR.Figure 1 (right panel) shows the contours of the λ ∼
20 cmemission of SNR IC 443 superimposed on a 24 µ m Spitzermap. The radio map for the source region of interest madefrom interferometric data only is shown in Figure 6 as con- The VLA of the NRAO is a facility of the NSF, operated under coopera-tive agreement by Associated Universities, Inc. -ray–IR sources in IC 443 7 -14 -13 -12 F l ux ( . - k e V ) e r g s / c m / s Src 2
XMM 2000-09-27 Chandra 2004-04-12/13 XMM 2006-03-30/31Chandra 2000-04-10/11
Src 3 2000 2001 2002 2003 2004 2005 2006 2007Observation time10 -14 -13 -12 F l ux ( - k e V ) e r g s / c m / s Src 2
XMM 2000-09-27 Chandra 2004-04-12/13 XMM 2006-03-30/31Chandra 2000-04-10/11 F IG . 5.— Left:
Fluxes of Src 2 and Src 3 as a function of time in the 0.5–10 keV range. The errors are given at the 68% confidence level.
Right:
Fluxes ofSrc 2 in the 2–10 keV range.F IG . 6.— Left:
Contours of radio surface brightness at λ =20 cm superimposed on the XMM-Newton X-ray image in the 2-8 keV energy band. The angularresolution of the radio data is 14 . ′′ × . ′′
2. The radio contours are plotted at 1.2, 2.4, 3.6, 4.9, 6.1, 7.3, 8.0 and 8.6 mJy/beam.
Right:
Contours of 24 µ memission detected by MIPS Spitzer superimposed on the XMM-Newton X-ray image in the 2-8 keV energy band. tours on a 2–8 keV XMM-Newton X-ray image. This fig-ure shows that Src 1 lies at the periphery of IC 443, far fromthe main SNR radio shell (that is situated in the North-Eastof the remnant), but near a localized radio excess. At theangular resolution and sensitivity of the present data, no ra-dio continuum source, either point-like or extended, couldbe associated with any of the X-ray sources. The local ra-dio flux density, obtained by integrating the radio emissionover the region containing 97% of the XMM-Newton countsof XMMU J061804.3+222732, is (60 ±
3) mJy.
Spitzer MIPS imaging and photometry
The field of IC 443 was the target of Spitzer MIPS scanobservations r4616960, r4616960, and r4616960 performedon 2005 November 9 (PI: G. Rieke). The Multiband Imag-ing Photometer for Spitzer (MIPS) aboard the Spitzer SpaceTelescope (Werner et al. 2004) is capable of imaging andphotometry in broad medium-IR spectral bands centered at24 µ m, 70 µ m, and 160 µ m, and low-resolution spectroscopybetween 55 µ m and 95 µ m (Rieke et al. 2004). The 24 µ mband covers the range of 21.3 µ m – 26.1 µ m, the 70 µ m bandcovers the range of 61.5 µ m – 80.5 µ m. Srcs 1, 2, and 3 were outside the field of view of the Spitzer IRAC near-IR camera.We used the standard MOPEX 030106 software (Makovozet al. 2006) to construct mosaic images and extract pointsources from the archival BCD-level data (pre-processed bythe S13 pipeline) according to the recipes in the Spitzer cook-books and the MIPS data handbook . The first framesof each sequence were ignored. A total of 5940 individualframes were mosaiced for each of the MIPS bands. The netexposure of the mosaic maps is equal to 65–92 frames (2.62s each) for the 24 µ m band and 12–15 frames (3.15 s each)for the 70 µ m band (different parts of the map were obtainedwith different effective exposures). Outlier detection was per-formed to exclude moving and solar system objects. A wide-field Spitzer MIPS image of the whole remnant is shown inFigure 1.Using the APEX software suite (Makovoz et al. 2006),we detected two point-like sources, at α = 06 h m . s , δ = + ◦ ′ ′′ and at α = 06 h m . s , δ = + ◦ ′ ′′ . Thesources were detected only in the 24 µ m band with fluxes1.77 ± ± http://ssc.spitzer.caltech.edu/documents/datademos/ http://ssc.spitzer.caltech.edu/mips/dh/mipsdatahandbook3.2.pdf A.M.Bykov et al. F IG . 7.— Src 1 environment as seen by Spitzer MIPS. The sources detected by Chandra are shown as crosses (3–8 keV) and a dashed rectangle (1.8–2.0 keV).The black contour denotes the extended near-IR source 2MASS J06180378+2227314. The dashed circles denote the beamsize of MIPS. respectively. The latter source coincides with the absorbednear-IR source 2MASS J06180406+2227345 detected only inthe K s band ( J > . H > . K s = 15 . ± . s (2.02 – 2.30 µ m) flux of the source ranges from0.64 to 1.21 mJy (considering the uncertainty of the extinc-tion value), that is (1.2–2.2) × - ergs cm - s - .According to the extinction maps of Schlegel, Finkbeiner &Davis (1998), the total Galactic absorption towards Src 1 cor-responds to A V ≈ . ± .
7. This sets an upper limit becausethe source is only ∼ A V ∼ B = 16 . ± . V = 14 . ± . J = 12 . ± . H = 11 . ± . K s = 11 . ± . s flux of the source ranges from 20.1 to 27.6mJy, that is (3.6–5.0) × - ergs cm - s - . Most likely,2MASS J06180359+2227227 is a foreground star.An extended excess of IR emission is seen in the 24 µ mmosaic map (left panel of Figure 7), coinciding with the ex-tended emission region detected with Chandra ACIS in thewestern part of Src 1 and with 2MASS J06180378+2227314,an extended (14 ′′ × ′′ ) source of near-IR emissionlisted in the 2MASS XSC catalog with observed isophotal K s = 12.86 ± µ m band(right panel of Figure 7), its apparent size being comparablewith the beam size. The aperture photometry estimates ofthe excess (with aperture corrections applied) are 11 . + . - . mJy(90% err.) ± σ error of the pipeline) for the 24 µ mband and 840 + - mJy (90% err.) ±
170 mJy (3 σ error ofthe pipeline) for the 70 µ m band. The apertures of a 6 ′′ and8 ′′ radius were used for the 24 µ m and 70 µ m bands, respec-tively. These values correspond to (2.8 + . - . ± . × - ergs cm - s - for the 24 µ m band and (9.7 + . - . ± . × - ergs cm - s - for the 70 µ m band. Notice, that the latter valueis actually an upper limit. Depending on the extinction value, the dereddened flux of 2MASS J06180378+2227314 is 6.4–10.7 mJy, that is (1.2–1.9) × - ergs cm - s - in the K s band.The extended source is seen neither in the J and Hbands of the 2MASS survey (Skrutskie et al. 2006) nor inthe archival blue, red, and infrared images of the POSS-II survey (Reid et al. 1991). With the daophot pack-age (Stetson 1987) integrated into the NOAO IRAF soft-ware suite, the following upper limits were obtained for2MASS J06180378+2227314: 6.6 × - ergs cm - s - inthe 2MASS J band, 2.9 × - ergs cm - s - in the 2MASSH band, 8.0 × - ergs cm - s - in the POSS-II blue band(3750–5500 Å), 3.0 × - ergs cm - s - in the POSS-II redband (5900–7100 Å), and 3.0 × - ergs cm - s - in thePOSS-II IR band (7350–8750 Å). The limits assume the esti-mated extinction for the source ( A V = 6).The point-like X-ray source Src 3 is also seen as a weakIR source in all the bands of the 2MASS survey, with J =17 . ± . H = 16 . ± . K s = 16 . ± . DISCUSSION
Below we discuss the multiwavelength data obtained for the ∼ . ′ ∼ Γ ∼ F IG . 8.— Upper panels:
Src 1 environment in the 2MASS near-IR bands. The source regions and positions are marked in the same manner as in Figure 7. Thewhite X-mark denotes the position of 2MASS J06180359+2227227.
Lower panels:
Surroundings of Src 3 in the 2MASS near-IR bands. The 10 ′′ radius circledenotes the region used for spectral analysis of the X-ray data. The contour inside the circle denotes the position of Src 3 as seen by Chandra in the 0.3 – 10 keVband at the 2.5 × - cps/pix level. TABLE 2P ARAMETERS OF ABSORBED POWER - LAW MODELS OF S RCS
1, 2,
AND N H a Γ Norm b χ ν /dof*Src 1 (20 ′′ )/bkg2 XMM 2000 A 1064/505/507 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 XMM 2006 A 4574/1899/2150 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 Chandra 2004 2973 0 . + . - . . + . - . . + . - . - ′′ )/bkg1 Chandra 2004 1314 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 XMM 2000 A 955/418/470 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 XMM 2006 B –/944/1003 1 . + . - . . + . - . . + . - . - ′′ )/bkg2 Chandra 2004 1829 1 . + . - . . + . - . . + . - . - ′′ )/bkg1 Chandra 2004 1270 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 XMM 2000 B –/235/225 0 . + . - . . + . - . . + - . - ′′ )/bkg2 XMM 2006 A 2730/940/1080 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 Chandra 2004 1251 0 . + . - . . + . - . . + . - . - ′′ )/bkg1 Chandra 2004 620 0 . + . - . . + . - . . + . - . - OTE . — ‘A’ means PN, MOS1, MOS2 data combined, while ‘B’ means MOS1 and MOS2 data combined; ‘bkg1’ denotes a background annuluswith inner radius of 10 ′′ and outer radius of 20 ′′ , ‘bkg2’ denotes a background annulus with inner radius of 20 ′′ and outer radius of 30 ′′ . The sourcecounts in the third column are given as PN/MOS1/MOS2 for XMM and ACIS for Chandra. All errors quoted in the table are at the 90% confidence levelfor one interesting parameter. a N H is in units of 10 cm - . b Normalization parameter is the spectral flux in photons cm - s - keV - at 1 keV. keV with a flux of a few times 10 - ph cm - s - at the 90%confidence level, which might be attributed to a Si K-shellline. A feature at 3.7 keV was found in the X-ray spectrum ofSrc 3 at the 99% confidence level, which might be attributedto an Ar K-shell line, unless the line is a redshifted Fe K lineof an extragalactic source. A firm detection of the lines fromthe localized clumps with XMM-Newton is hampered by thepresence of strong hard non-thermal continuum emission.An extended source of IR emission was found at the North-Western edge of Src 1 with a dereddened flux in the 24 µ mband of Spitzer MIPS of about 3 × - ergs cm - s - and anupper limit for the 70 µ m band of about 10 - ergs cm - s - .The near-IR flux of the source is about 2 × - ergs cm - s - in the 2MASS K s band. The source is not seen in the Jand H bands of the 2MASS survey, nor in the POSS-II op-tical bands. The upper limits are 3 × - ergs cm - s - forthe H band, 7 × - ergs cm - s - for the J band, 3 × - ergs cm - s - for the POSS-II infrared band (0.7–0.9 µ m),3 × - ergs cm - s - for the POSS-II red band (0.6–0.7 µ m), and 8 × - ergs cm - s - for the POSS-II blue band(0.4–0.6 µ m). These upper limits will be used below to con-strain the model we propose for J0618.Earlier studies of molecular emission from the extendedregion of apparent interaction of the IC 443 SNR with theneighboring molecular cloud have indicated the presence ofemission from both fast and slow shocks (e.g. Burton et al.1988; Dickman et al. 1992; van Dishoeck et al. 1993, Snell etal. 2005). Molecular clouds are known to have highly inho-mogeneous internal structure (e.g. Blitz 1993). A molecularcloud consists of numerous dense clumps with a rather smallvolume filling factor, embedded in an interclump matter of amodest density of 5–20 cm - . The presence of a wide rangeof dense molecular emission clumps down to about 1 ′′ scalehas been established in the cloud around Src 1 by Richter,0 A.M.Bykov et al. TABLE 3P
ARAMETERS OF ABSORBED PLASMA ( MEKAL ) MODELS OF S RCS
1, 2,
AND N H a T , keV Norm b χ ν /dof*Src 1 (20 ′′ )/bkg2 XMM 2000 A 1064/505/507 0 . + . - . . + - . . + . - . - ′′ )/bkg2 XMM 2006 A 4574/1899/2150 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 Chandra 2004 2973 0 . + . - . . + . - . . + . - . - ′′ )/bkg1 Chandra 2004 1314 0 . + . - . + . - . . + . - . - ′′ )/bkg2 Chandra 2004 1781 1 . + . - . . + . - . . + . - . - ′′ )/bkg2 XMM 2000 A 955/418/470 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 XMM 2006 B –/944/1003 1 . + . - . . + . - . . + . - . - ′′ )/bkg2 Chandra 2004 1829 0 . + . - . . + . - . . + . - . - ′′ )/bkg1 Chandra 2004 1270 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 XMM 2000 B –/235/225 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 XMM 2006 A 2730/940/1080 0 . + . - . . + . - . . + . - . - ′′ )/bkg2 Chandra 2004 1251 0 . + . - . . + . - . . + . - . - ′′ )/bkg1 Chandra 2004 620 0 . + . - . . + . - . . + . - . - OTE . — ‘A’ means PN, MOS1, MOS2 data combined, while ‘B’ means MOS1 and MOS2 data combined; ‘bkg1’ denotes a background annuluswith inner radius of 10 ′′ and outer radius of 20 ′′ , ‘bkg2’ denotes a background annulus with inner radius of 20 ′′ and outer radius of 30 ′′ . The sourcecounts in the third column are given as PN/MOS1/MOS2 for XMM and ACIS for Chandra. All errors quoted in the table are at the 90% confidence levelfor one interesting parameter. Src 1* is the annulus region of Src 1(20 ′′ ) with the removed 11 ′′ radius circle containing Srcs 1a and 1b. a N H is in units of 10 cm - . b Mekal model normalization parameter: K = - π D (1 + z ) R n e n H dV , where D is the angular size distance to the source in cm, n e and n H is the electronand hydrogen density in cm - , integration is done over source volume. Graham, and Wright (1995).The apparent coincidence of the extended structured non-thermal X-ray continuum source J0618 with local excesses ofIR emission can be understood in the framework of a modelof interaction of a molecular cloud with a fast ballisticallymoving object. If the object was ejected by a SNR, it couldbe either a massive fast moving ejecta knot or a pulsar windnebula. In both cases shock waves will be driven into thecold matter of the cloud. Such a scenario and the expectedproperties of the emitted IR and X-ray radiation will be dis-cussed below. A large power [above 10 erg s - ] released bya fragment of velocity about 300 km s - being decelerated bya dense cloud is emitted as UV photons. The UV emissionwill create an HII region surrounding the fragment. The de-tected Spitzer emission and all of the IR/optical upper limitsare consistent with the continuum emission of an HII regionof a temperature about 10 K and a number density above 100cm - . However, to explain the detected near-IR flux of thesource in the 2MASS K s band one should consider an addi-tional emission component most likely due to line emission ofshocked molecular hydrogen. The radio, IR, and optical emis-sion of the HII region surrounding J0618 will be addressed inSections 4.2 and 4.3. X-ray emission from isolated fast ejecta fragments
An important distinctive feature of J0618 is the presence ofextended non-thermal hard X-ray emission. This can be ex-plained in the framework of a model of interaction of a mas-sive isolated ejecta fragment moving with a velocity above300 km s - through a molecular cloud. In the model by Bykov(2002, 2003), X-ray emission from an ejecta fragment of 0.2– 0.3 pc size ( ∼ ′′ - ′′ at 1.5 kpc, the angular size of sucha fragment in IC 443), interacting with a molecular cloud wasestimated. The fragment mass was assumed to be > ∼ - M ⊙ ,containing ∼ - M ⊙ of Si. The “knot” traveled through aninter-clump medium of a number density ∼
100 cm - with avelocity of ∼
500 km s - (the corresponding postshock tem-perature is ∼ . Γ < ∼ . ∼ - ph cm - s - . The K-shell linesare excited by both non-thermal and thermal electrons.In the case of J0618, a slightly more massive fragmenthighly structured due to interaction with the dense molecu-lar clump would be more realistic. A range of sub-fragmentvelocities around the mean value ∼ - is consis-tent with the age of IC 443 of about 30,000 year advocated byChevalier (1999). Such a fragment would provide line fluxesof a few times 10 - ph cm - s - , consistent with the esti-mated fluxes of the putative Si and Ar lines (Figure 4) fromJ0618. The effects of clumping of the metal-rich ejecta wouldresult in an intermittent spatial structure of X-ray line emis-sion from the source. It is worth to mention that the intrinsicabsorbing column ∆ N H of a fragment could be substantial ( > cm - ), especially if the fragment contains metal-richejecta material.The relatively low velocity and the high absorbing columnof an ejecta fragment in a dense molecular cloud makes its ob-servational appearance to be very different from that observedin Vela shrapnel A. The observed emission of Vela shrapnelA is strongly dominated by an optically thin thermal compo-nent (of T about 0.5 keV) of a shock heated plasma (Miyataet al. 2001). Contrary to the Vela case, a thermal emission ofa thin plasma of temperature ∼ Energetics of an X-ray emitting SN ejecta fragment
The models of an isolated X-ray source involving the in-teraction of a supernova blast wave with dense ambient mat-ter suggest that the keV emission is due to bremsstrahlungof shock-accelerated electrons (Bykov et al. 2000; Bykov2002). Synchrotron X-ray emission would require electronswith TeV energies accelerated by shocks of speed well above1,000 km s - , which is hard to expect for a middle-aged SNRin a dense ambient medium, unless the moving object is a pul-sar wind nebula.The radiative efficiency of nonthermal bremsstrahlung is-ray–IR sources in IC 443 11known to be low at keV energies. This means that toproduce a hard X-ray continuum at a rate ˙ ǫ r , an electronof energy E e dissipates energy at a rate ˙ ǫ e via Coulomblosses in a medium of an average charge Z , and ˙ ǫ r / ˙ ǫ e ≈ . × - Z ( E e / m e c ) — see e.g. Akhiezer & Berestet-sky (1957). Therefore, to produce the X-ray emission atthe observed level of about 6 × d . ergs s - by elec-trons accelerated to 50–100 keV, the dissipated power mustbe about ˙ ǫ e ≈ d . Z - ( m e c / E e ) ergs s - . The total dissi-pated power ˙ ǫ kin = 10 ˙ ǫ e η - - . Here η - ≡ η/ . < ∼ ˙ ǫ kin ∼ d . Z - η - - ergs s - .The electron energy E e = 50 keV wasassumed in the estimation. Since Z < ∼
10 in an oxygen/silicon-rich gas an enhanced metallicity of an ejecta fragment (depen-dent on mixing of the ejecta with the ambient matter) couldsomewhat compensate the bremsstrahlung inefficiency pro-viding ˙ ǫ kin ∼ d . ergs s - . Note that the power requiredto produce radio-emitting relativistic electrons in clump D isalso just above 10 d . ergs s - .The upstream ram pressure power dissipated at the for-ward shock of a ballistically moving fragment, ˙ ǫ sh ≈ × n a3 v r d . ergs s - , is the source of gas heating and par-ticle acceleration. Here r is the shock radius of J0618 (mea-sured in 20 ′′ ). The mechanical power ˙ ǫ kin in the shock modelmust not exceed ˙ ǫ sh , resulting in the condition n a3 v r Z η - ≥
12 to be fulfilled. Therefore, ejecta fragments of velocity v ∼ n a3 ∼ . ˙ ǫ sh ≈ ˙ ǫ kin , if Z > ∼
2. Thus, the observed patchy structure of theX-ray emission that is apparent in the Chandra ACIS imagesof J0618 could be attributed to structured metal clumps of theejecta of IC 443.How can one directly detect or constrain the power dissi-pated by the shock in the model of IC 443 ejecta fragment bal-listically moving through a molecular cloud? The gas temper-ature behind the standard single-fluid strong MHD shock canbe estimated as T ≈ . · v ( K ). For a shock with ener-getic particle acceleration efficiency > ∼
10% that we considerhere the postshock temperature should be reduced by a factorof about 1.2 to account for the effects of the energy flux car-ried away with energetic particles (see e.g. Bykov 2002). Thepostshock gas cooling distance estimated by Hartigan, Ray-mond, and Hartmann (1987) is about 1.8 × v . n - cm.The shock of a velocity 300 km s - in the molecular cloud isradiative with the cooling layer angular size of 0.15 ′′ n - . Ifthe extended structure of Src 1a seen in Figure 2 of BBP05 isindeed due to the thermal emission of hot postshock gas then n a3 ∼ ′′ andthe luminosity of soft thermal component below 10 ergs s - discussed in BBP05. Most of the shock power, however, is notin the soft X-ray emission of shock heated gas, but rather inUV–optical emission of the radiative shock dominated by UVlines of OVI at 1035 Å, Ly α and He II at 304 Å(e.g. Hartigan,Raymond, and Hartmann 1987).At the same time the UV photons produced by the radia-tive bow shock of ejecta fragment with a luminosity L UV =10 erg s - L , will be absorbed and reprocessed, mostly toIR emission, in an expanding HII region surrounding the bowshock of J0618. Note here that the shape of the HII regioncould be different from that of the non-thermal X-ray nebu-lae, but it should overrun the bow shock of J0618. From theapparent position of J0618 in the molecular cloud and the es-timated fragment velocity one may conclude that it enteredthe molecular cloud about 1,000 years ago. Assuming that all of the UV photons produced by the radiative bow shockare absorbed in a spherical layer of an homogeneous ambientmatter, one can obtain the estimation of the HII region radiusof about 6 × L / n - / cm (see e.g. Spitzer 1978). In fact,the geometry of J0618 is more complex than a spherical HIIregion because of the motion of the extended emitting frag-ment and because of the strong inhomogeneity of the ambientmolecular cloud down to the arcsecond scale as was observedby Richter, Graham, and Wright (1995). In the interclumpmatter of density below 100 cm - the scale size of the HII re-gion would exceed the apparent X-ray size of J0618. On theother hand, the number density of the clump D located in aclose vicinity of J0618 was estimated by van Dischoeck etal. (1993) to be above 10 cm - and the size of the HII re-gion is consistent with the size of the 24 µ m emission excessaround clump D [of about (2–3) × - ergs cm - s - ] appar-ent in Spitzer image presented in the right panel of Figure 6.Although very simplified, such a model allows us to estimatethe radio, IR, and optical emission of the shock-produced HIIregion. Radio–IR–optical continuum and line emission of theHII region
The emission of the HII region consists both of continuumemission of a thin thermal plasma of a kinetic temperatureabout 10,000 K and of a rich emission line spectrum (seee.g. Spitzer 1978). The specific appearance and the chemistryof the mostly neutral photodissociation region in molecularclouds were reviewed by Hollenbach and Tielens (1999). Inthe standard case the interstellar HII regions are powered bymassive luminous stars. In the case of J0618 the source ofthe ionizing radiation is an extended fast moving bow shockand the ionized region is likely unsteady. That makes anaccurate modelling of the system rather complicated, so weshall present here some approximate estimations of the ex-pected IR/optical fluxes from the HII region. The observed24 µ m Spitzer MIPS, 2MASS K s and 1.4 GHz radio emis-sion, as well as the upper limits provided by the other Spitzer,2MASS and POSS-II observations of J0618 were modelled ascontinuum emission of a hot ionized plasma of a temperatureranging from 8,000 to 20,000 K. The modelled fluxes werecorrected for interstellar extinction for a wide range of A V values below 8, since A V ∼ N H ∼ × cm - – the maximal value allowed by the fits tothe Chandra data presented in Fig. 3. Thus all the opticaland IR measurements of J0618 were used to constrain boththe kinetic temperature of the HII region and the parameter Y = n × r , where n e3 is the electron number density in theHII region measured in units of 10 cm - , and r is defined inSection 4.2. It was found that the measured 24 µ m flux of thesource can be explained by a thermal continuum of a 10,000K temperature HII region and Y ∼ V > ∼
6. Sincethe continuum emissivity is ∝ n the apparent patchy imageof 24 µ m emission could be due to the clumpy structure of theejecta fragment.The continuum emission from the HII region is consis-tent with the upper limits obtained from Spitzer, 2MASS andPOSS-II observations of J0618. The radio flux density fromsuch an HII region is about 30 mJy. The flux density ob-tained by integrating the 1.4 GHz radio emission over a larger(an arcminute scale size) region with VLA was about 60 mJy.Therefore the 1.4 GHz radio flux is not in a conflict with theHII region model for the parameter Y ∼ α line from theHII region is below the upper limit of 3 × - ergs cm - s - obtained from the POSS-II red band (0.6–0.7 µ m) observationdiscussed above.The estimated parameter Y ∼ n e ∼
160 cm - . The bow shock radius estimated from X-rayimage is r ∼ n a < n e / < ∼
40 cm - , generally con-sistent with that expected in the interclump matter of a molec-ular cloud. A substantial part of the powerful UV emissionfrom the radiative shock of J0618 will irradiate the nearbydense molecular clump D providing another HII region of asurface brightness and high luminosity well in excess of thatfrom J0618 that is clearly seen in Fig. 6. In the frame of theejecta fragment model of J0618 we estimated that at least asubstantial amount of 24 µ m flux from clump D [that is about(2–3) × - ergs cm - s - ] can be attributed to the HII regionexcited by the UV emission from the nearby source J0618.That model provides, however, only about one third of theobserved 2MASS K s band flux of J0618, and that can notbe simply relaxed with an appropriate HII region parameterschoice. Thus, an extra contribution in the 2MASS K s bandat the level of ∼ . × - ergs cm - s - is required. Thatcontribution may come, most likely, from the emission linesof the shocked molecular hydrogen in J0618 and would re-quire the presence of C-type molecular shocks of velocity ∼
30 km s - in the close vicinity of J0618. Moreover, theatomic fine-structure lines could contribute to the 24 µ m emis-sion detected by SpitzerMIPS if supernova ejecta drive J-typeshocks of about 100 km s - (and faster) into dense molecularclumps. We discussed above only the emission produced bythe forward bow shock of the fragment, however slower re-verse shock will also be present and since the ejecta fragmentbody is likely very inhomogeneous the real structure is likelyeven more complex. Relevant line emission models will bebriefly reviewed in the next subsection. IR-line emission in radiative shock models
Atomic fine structure lines of [OI] (63 µ m) and [FeII] (26 µ m) could dominate the emission in the Spitzer MIPS 70 µ mand 24 µ m bands. The lines are known to trace fast radia-tive shocks in molecular cloud material. A comprehensivestudy of radiative shocks in interstellar clouds has been doneby Hollenbach and McKee (1989; HM89 hereafter). Theyhave shown that the intensity ratio of [OI](63 µ m)/[FeII](26 µ m) is about 10 for a shock wave in a cloud of a density n cl =10 cm - . They estimated the [OI] (63 µ m) IR line inten-sity as I ≈ . × - ergs s - sr - for a shock of a velocity v sh =150 km s - and demonstrated that it scales roughly lin-early with n cl v sh . The combination of optical line emissionof OI at 6300 Å, CII (2326 Å), and OII (3726 Å) providesthe dominating gas coolant in transparent systems of n cl ≤ cm - . The line intensities were obtained by HM89 underthe assumption of standard solar composition with accountfor the interstellar gas depletion to dust grains. The IR-opticalline intensities from shocks driven by (and into) metal-richejecta could be produced with lower pre-shock number densi-ties n cl than those modeled by HM89, and should be re-scaledrespectively. This can reduce the pre-shock density requiredto match the observed fluxes.Applying the HM89 model intensities to an IR source of an angular area A (measured in units of 100 arcsec ) asdiscussed in section 3.2, one obtains a [FeII] (26 µ m) lineflux F ≈ . × - A n v ergs cm - s - . Here n isthe pre-shock number density in units of 10 cm - and v is the shock velocity in 10 km s - . To reach the 24 µ mflux of about 3 × - ergs cm - s - as estimated from theSpitzer MIPS data, one would need n v A ∼
1. As theestimated emission area is A ∼
2, one needs n v ∼ . µ m)line could provide a sizeable part of the 24 µ m Spitzer MIPSflux. The associated model flux of the [OI] (63 µ m) line, F ≈ . × - A · n v ergs cm - s - is consistent withthe upper limit derived above. It is apparent from Figure 2that Src 1 is located near the shocked molecular clump Dpresented in a map by van Dishoeck, Jansen, & Phillips(1993). Typical densities of molecular clumps, studied byvan Dishoeck et al. (1993) are > ∼ cm - . The condition n v ∼ . v ∼ . v ∼ µ m and 1.7 µ m havea rather flat shock velocity dependence above 100 km s - . Theintensity of the brighter FeII line at 1.3 µ m scales roughly lin-early with n cl , and it is about 0 . I . This estimate is closeto, but still consistent with the upper limit of the flux in the2MASS J band discussed in Section 3.2. A near-IR NI (1.04 µ m) line flux was predicted by HM89 to be at the level com-parable with that of [OI] (63 µ m), and a SI (1.1 µ m) line fluxsimilar to that of FeII at 1.3 µ m.From the apparent lack of extended emission in the POSS-IIred band (0.6–0.7 µ m) the flux upper limit for OI (6300 Å) canbe estimated as 3 × - ergs cm - s - (assuming the reason-able extinction to Src 1 as A V ≈ µ m)ratio for a transparent system with shock velocity v > ∼ v <
1, and, for example, a fast J-type (see Draine 1980 for adefinition of J- and C-type shocks) radiative shock of v ∼ . n ∼
1, could explain both the observed [FeII]26 µ m line flux and the derived upper limit for the OI (6300Å) line flux. Note that optical line fluxes from a radiativeshock propagating into oxygen-rich ejecta material could bereduced by heat-conduction effects (e.g. Borkowski & Shull1990).An important diagnostic IR-line ratio sensitive to the shockvelocity is [NeII](12.8 µ m)/[FeII](26 µ m). The ESO VLTSpectrometer and Imager for the Mid-Infrared (VISIR) is anoptimal instrument for observations in the two mid-infraredatmospheric windows: the 8–13 µ m N band with a 19 . ′′ × . ′′ µ m Q band with a 32 . ′′ × . ′′ µ m) predicted by the radiative shock model ofHM89 and the continuum emission from Src 1 HII region,thus discriminating between the two models.The model of a single fast radiative J-type shock de-veloped by HM89 underpredicts the near-IR flux (above10 - ergs cm - s - ) observed in the K s band from the ex-tended source 2MASS J06180378+2227314, closely associ-ated with the Spitzer MIPS 24 µ m excess (Figure 7). Thus, acombination of fast and slow shocks is needed to explain theIR emission from J0618. The same conclusion has been madein most of the studies of molecular emission of the south-western region of IC 443. Burton et al. (1990) argued that-ray–IR sources in IC 443 13both fast dissociative J-type and slow C-type shocks must bepresent in the extended southern cloud of IC 443 to explainthe observed line fluxes [see also Snell et al. (2005) for arecent discussion]. Analyzing different shock models, includ-ing a time-dependent one, Snell et al. (2005) concluded thatno single-shock model can explain the existing observations,and a range of shock velocities is required. Shocks in a molecular cloud driven by ejecta fragments
A wide range of shock velocities in a molecular cloud is re-quired to explain the IR and X-ray observations of Src 1. Slowmolecular C-type shocks are required to explain the observed2MASS K s band emission.Molecular emission occurs when a fast ejecta fragment col-lides with molecular clumps of different densities, ρ cl , pro-ducing multiple shocks of velocities v s ∝ ρ - / . This can ex-plain the shock velocity range of 10 km s - ≤ v s ≤
100 km s - ,required to explain both the fluxes detected by Spitzer MIPSand 2MASS K s and the upper limits on the optical lines.In the considered case the presence of a fast shock of veloc-ity > ∼
300 km s - seems to be required to explain the X-raydata discussed above. Moreover, it is apparent from Figure 1that while the 1.4 GHz radio emission is dim in the south-eastern part of the shocked molecular cloud in comparisonwith that at the north-east, there is a localized excess in thevicinity of molecular clump D adjacent to J0618. If the radioemission is the synchrotron emission produced by relativisticelectrons in the GeV regime in a likely enhanced magneticfield of the dense clump D, then the model requires the pres-ence of a fast shock of velocity well above 100 km s - toaccelerate the relativistic particles in the vicinity of Src 1.The shock ram pressure estimations for multiple shockmodels were discussed by many authors (e.g. Burton et al.1990; Richter, Graham, and Wright 1995; Cesarsky et al.1999; Snell et al. 2005). A fast shock of velocity v s ∼ - incident on a clump of a moderate number density10 –10 cm - would have a high ram pressure nv ∼ –10 cm - [ km s - ] . It seems to be uneasy to attribute theestimated range of ram pressures to a uniform SN blast wavecolliding with a cloud. A SN blast-wave ram pressure, as es-timated from the properties of shocked X-ray emitting gasinside the remnant, is smaller. On the other hand, strongram-pressure inhomogeneities can be produced if a molecu-lar cloud is hit by SN ejecta driving the blast wave. The ejectaare likely to be rather fragmented. Each can be consideredas an ensemble of ejecta fragments with a range of veloci-ties and densities produced by early ejecta instabilities. Thefragments could provide a wide range of ram pressures in themolecular cloud (up to 10 cm - [ km s - ] ) driving shocks ofdifferent velocities, which are required to explain the observa-tional data. Isolated fast moving dense ejecta fragments couldpenetrate deeper into the dense parts of the cloud driving thehigh-pressure shocks on relatively small scales. Time variability and ejecta fragments statistics
The lifetime of an isolated ejecta fragment in a densemedium is an important factor with regard to time variabil-ity of the X-ray emission. A fast moving knot is deceler-ating due to the interaction with the ambient gas. The dragdeceleration time of a fragment of velocity v , mass M and ra-dius R can be estimated as τ d ≈ M - ( n a3 v R - ) - years.Here R - = R / (0 . M - = M / - M ⊙ . The num-ber density n a3 of the ambient matter is in units of 10 cm - , and the fragment velocity v in units of 10 km s - . Hy-drodynamical crushing of a fast knot occurs on a timescale τ c ∼ χ / R v - (e.g. Chevalier 1975; Sutherland & Dopita1995; Wang & Chevalier 2001, and references therein). Thedensity contrast χ = ρ k /ρ a ( ρ k is the knot density) is of theorder of 1 for a large enough fragment, and τ d < τ c for largeenough clumps. To reach its apparent position in the molecu-lar cloud of IC 443, an isolated ejecta fragment must be mas-sive enough, M - > ∼
1, to overcome strong drag decelerationin the dense matter, as discussed below.Particle acceleration occurs on a few-years timescale if thefast particle diffusion coefficient does not exceed 10 cm s - (see Bykov 2002 for a discussion). Thus, given the fragmentlifetime estimated above, some time variability of the X-rayemission, both in hard continuum and lines, can be expectedon a few-years timescale (and longer) for a fragment of ve-locity v > ∼ n a3 > ∼
3, typicalfor molecular clumps. Variable X-ray emission has been re-cently observed with Chandra in a few molecular clouds inthe Galactic Center region by Muno et al. (2007). Using theabove estimates, one can argue that variable X-ray emissioncould be produced due to an interaction of metal-rich super-nova ejecta with dense clumps of molecular clouds.According to the log N – log S distribution of X-ray emittingknots simulated by the method of Bykov (2003), the proba-bility of another similarly bright ejecta fragment getting intothe field is rather low because of the short lifetime of the frag-ment in a dense molecular clump. However, some smaller andless massive fragments propagating in the interclump mediumcould be seen in the cloud as weak point-like X-ray sources ofluminosities L x < ∼ ergs s - . In order to firmly identify thesources as such fragments, a deeper observation is required.The hydrodynamic simulations available (Klein, McKee, &Colella 1994; Wang & Chevalier 2001) predict a complex ir-regular structure of the fragment body due to hydrodynamicinstabilities. In this case the X-ray image would show an ir-regular patchy structure, instead of a smooth regular head-tailstructure. The patches are due to emission of dense piecesof the fragmented knot illuminated by the shock-acceleratedenergetic particles. The observed morphology of X-ray emis-sion with bright clumps of a few arcsec size in Src 1, such asSrc 1a and Src 1b, as well as Src 3, can be explained in thatmodel. It is worth to note that ballistically moving ejecta frag-ments could have rather large non-radial velocities (with re-spect to the apparent center of SN explosion). The non-radialvelocity components will be substantial if the fragments origi-nate from ejecta instabilities at relatively late evolution stages.Thus, one can conclude that the morphology, spectra, andX-ray luminosity of the J0618 complex are generally consis-tent with those expected for a ballistically moving SN ejectafragment interacting with a molecular cloud, though it isworth to discuss also some alternatives. Alternative interpretations of J0618
A possible interpretation of the extended hard X-ray sourceJ0618 associating it with an interaction of the ejecta of SNRIC 443 with the nearby molecular cloud has been discussedabove in some detail. Another SNR-related possible interpre-tation is a low luminosity pulsar wind nebula originating ei-ther from IC 443 or from another SNR G189.6+3.3, advancedby BBP05.Extragalactic sources could also contribute to the observedappearance of J0618. Spectra the of point-like sources Src 2and Src 3 could be interpreted as that of AGNs. Within such4 A.M.Bykov et al.an interpretation the feature at about 3.7 keV in the spectrumof Src 3 could correspond to a redshifted (with 0.6 < z < z cluster of galaxies.The angular size of Src 1 ( < ′ at photon energies > z > z clusters). The thermal fit of thediffuse Src 1* presented in Table 3 with the temperature about5 keV is consistent with the cluster interpretation.Optical and IR identifications of galaxies (of a few arcsec-ond scale size at z ∼
1) in the hypothetical cluster are the mostobvious test to check the interpretation and measure the red-shift of the putative cluster. With the optical data availableand given the substantial absorption (3 < A V < V = 17–19, whilethe brightest galaxies in most of the known X-ray clusters at z < ∼ V > ∼ z galaxy cluster inter-pretation. The observed extension of the 2MASS K s sourcecorrelated with observed diffuse X-ray emission is not easy tounderstand in the cluster of galaxies interpretation; it rathersupports a Galactic SNR-related origin of J0618. CONCLUSIONS
The multi-wavelength observations presented here indicatea possible physical connection of the X-ray source J0618 withthe neighboring IR Spitzerand 2MASS sources. That connec-tion, if real, can be understood in a scenario where J0618 orig-inates in an interaction of the IC 443 SNR with the adjacentmolecular cloud. The correlation would require the presenceof both fast and slow shocks in the clumpy molecular-cloudmaterial. The X-ray line features apparent in the spectra ofthe clumps favor a scenario in which the shocks are producedby a fast ballistically moving SN ejecta fragment penetrat-ing into a structured molecular cloud. The model provides aphysical picture coherent with the current observational data,although alternative scenario cannot be rejected yet.Alternatively, Src 1 can be interpreted as a massive X-ray cluster of galaxies at a redshift z > REFERENCES
Akhiezer, A.I., & Berestetsky, V.B. 1957, Quantum Electrodynamics, transl.by Consultants Bureau Inc. (Oak Ridge: Tech. Information Service Ext.)Albert, J., et al., 2007, ApJ, 664, L87Arce, H.G., Goodman, A.A., 1999, ApJ, 512, L135Asaoka, I., & Aschenbach, B. 1994, A&A, 284, 573Aschenbach, B., Egger, R. & Trümper, J. 1995, Nature, 373, 585Blair, W.P., et al. 2000, ApJ, 537, 667Blitz, L. 1993, in Protostars and Planets III, ed. E.H. Levy & J.I. Lunine,(Tucson: Univ. of Arizona), 125Bocchino, F., & Bykov, A.M. 2000, A&A, 362, L29Bocchino, F., & Bykov, A.M. 2001, A&A, 376, 248Bocchino, F., & Bykov, A.M. 2003, A&A, 400, 203 (BB03)Borkowski, K.J. & Shull, J.M. 1990, ApJ, 348, 169Braun, R., Strom, R.J. 1986, A&A, 164, 193Bykov, A.M. 2002, A&A, 390, 327Bykov, A.M. 2003, A&A, 410, L5Bykov, A.M., Chevalier, R.A., Ellison, D.C. & Uvarov, Yu.A., 2000, ApJ,538, 203Bykov, A.M., Bocchino, F., & Pavlov, G.G. 2005, ApJ, 624, L41 (BBP05)Burton, M.G. 1987, Quarterly Journal of the RAS, 28, 269Burton, M. G., Geballe, T. R., Brand, P. W. J. L., & Webster, A. S. 1988,MNRAS, 231, 617Burton, M.G., Hollenbach, D.J., Haas, M.R., & Erickson, E.F. 1990, ApJ,355, 197Cesarsky, D., Cox, P., Pineau des Forets, G., van Dishoeck, E.F., Boulanger,F., & Wright, C.M. 1999, A&A, 348, 945Chevalier, R.A. 1975, ApJ, 200, 698Chevalier, R.A. 1999, ApJ, 511, 798Claussen, M.J., Frail, D.A., Goss, W.M., & Gaume, R.A. 1997, ApJ, 489, 143 Cornett, R. H., Chin, G., & Knapp, G. R. 1977, A&A, 54, 889DeNoyer, L.K. 1979, ApJ, 232, L165Dickman, R. L., Snell, R. L., Ziurys, L. M., & Huang, Y.-L. 1992, ApJ, 400,203Draine, B.T. 1980, ApJ, 241, 1021Esposito, J. A., Hunter, S. D., Kanbach, G., & Sreekumar, P. 1996, ApJ, 461,820Fesen, R.A., & Kirshner, R.P. 1980, ApJ, 242, 1023Johnson, R.P. 2006, AIP Conf. Proc., 842, 1010Gaensler, B.M., Chatterjee, S., Slane, P.O., van der Swaluw, E., Camilo, F.,& Hughes, J.P. 2006, ApJ, 648, 1037Green, D.A. 1986, MNRAS, 221, 473Hartigan, P., Raymond, J., & Hartmann, L. 1987, ApJ, 316, 323Hewitt, J.W., Yusef-Zadeh, F., Wardle, M., Roberts, D.A.,& Kassim, N.E.,2006, ApJ, 652, 1288Hollenbach, D., & McKee, C.F. 1989, ApJ, 342, 306Hollenbach, D.J., & Tielens, A.G.G.M. 1999, Rev. Mod. Phys., 71, 173Kaastra, J.S. 1992, An X-Ray Spectral Code for Optically Thin Plasmas,(Internal SRON-Leiden Report, updated version 2.0)Kawasaki, M.T., Ozaki, M., Nagase, F., Masai, K., Ishida, M., & Petre, R.2002, ApJ, 572, 897Keohane, J.W., Petre, R., Gotthelf, E.V., Ozaki, M., & Koyama, K. 1997, ApJ,484, 350Klein, R.I., McKee, C.F., & Colella, P. 1994, ApJ, 420, 213Lagage, P.-O., Pantin, E., Durand, G., Smette, A., Doucet, C., Belorgey, J., &Pel, J.-W. 2006, Proc. SPIE, 6269, 626913Leahy, D.A. 2004, AJ, 127, 2277Makovoz, D., Roby, T., Khan, I. & Booth, H., 2006, Proc. SPIE, 6274, 10Mewe, R., Gronenschild, E.H.B.M., & van den Oord, G.H.J. 1985, A&AS,62, 197 -ray–IR sources in IC 443 15-ray–IR sources in IC 443 15