A Radio Characterization of Galactic compact Bubbles
Adriano Ingallinera, Corrado Trigilio, Grazia Umana, Paolo Leto, Alberto Noriega-Crespo, Nicolas Flagey, Roberta Paladini, Claudia Agliozzo, Carla Buemi
MMon. Not. R. Astron. Soc. , 1– ?? (2013) Printed 15 October 2018 (MN L A TEX style file v2.2)
A Radio Characterization of Galactic compact Bubbles
A. Ingallinera (cid:63) , C. Trigilio , G. Umana , P. Leto , A. Noriega-Crespo , N. Flagey , R. Paladini , C. Agliozzo , C. Buemi Universit`a di Catania, Dipartimento di Fisica e Astronomia, via Santa Sofia, 64, 95123 Catania, Italy INAF-Osservatorio Astrofisico di Catania, Via S. Sofia 78, 95123 Catania, Italy Spitzer Science Center, California Institute of Technology, Mail Code 314-6, Pasadena, CA 91125, USA Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA NASA Herschel Science Center, California Institute of Technology, Pasadena, CA, USA
ABSTRACT
We report the radio observations of a sub-sample of the 428 galactic compact bubbles discov-ered at 24 µ m with the MIPSGAL survey. Pervasive through the entire Galactic plane, theseobjects are thought to be different kinds of evolved stars. The very large majority of the bub-bles ( ∼ Key words: planetary nebulae: general – radio continuum: general – stars: evolution.
Over 400 compact roundish objects, presenting diffuse emission,were identified at 24 µ m from visual inspection of the MIPSGALLegacy Survey (Carey et al. 2009; Mizuno et al. 2010) mosaic im-ages, obtained with MIPS (Rieke et al. 2004) on board of the Spitzer Space Telescope . These small ( (cid:54) (cid:48) ) rings, disks or shells(hereafter denoted as ‘bubbles”) are pervasive throughout the en-tire Galactic plane in the mid-infrared (IR). Their distribution isapproximately uniform in Galactic latitude and longitude, and theaverage density is found to be around 1.5 bubbles per square de-gree. A further analysis of the GLIMPSE (3 . µ m to 8 . µ m) andMIPSGAL (70 µ m) images indicates that the bubbles are mostlydetected at 24 µ m only. The absence, for most of these objects, of acounterpart at wavelengths shorter than 24 µ m could be interpretedeither as a sign of extreme extinction, which would explain the non-detection of these objects in previous visible or near-IR surveys, oras intrinsic property of the objects. The main hypothesis about thenature of the bubbles is that they are different type of evolved stars(planetary nebulae, supernova remnants, Wolf–Rayet stars, asymp-totic giant branch stars, etc.).Some bubbles present a central source in the middle of thenebula in the MIPSGAL images. Studies by Wachter et al. (2010) (cid:63) E-mail:[email protected] The Multiband Imaging Photometer for Spitzer. The Galactic Legacy IR Mid-Plane Survey Extraordinaire, conductedwith the InfraRed Array Camera (IRAC) on board of the Spitzer Space Tele-scope. show how this central source is usually well detected at shorterwavelengths (down to 2MASS J band or even optical for excep-tional cases). In particular the authors spectroscopically examined62 bright source surrounded by a 24- µ m shell, being able to charac-terize the nature of 45 central sources. They found that 19 of themare compatible with Oe/WN, Wolf–Rayet (WR) and luminous bluevariable (LBV) stars. Furthermore, they also pointed out that it ispossible to explain that many bubbles emit only at 24 µ m assum-ing that this emission is not a continuum from warm dust but itrises from an intense [O IV ] line emission at 25 . µ m, as found byMorris et al. (2006), resulting in an almost pure gas nebula.The presence of very massive stars can also be inferred bythe morphology of the nebula. Gvaramadze, Kniazev, & Fabrika(2010) found that many bubbles, showing central sources, resembleknown nebulae surrounding blue supergiant (BSG), LBV, or WRstars. They confirmed the nature of some bubbles, inferred by amorphological analysis, by means of spectroscopic identificationof their central sources, showing that the mere presence and shapeof the nebula can suggest the possibility of these massive stars.Mid-IR spectroscopic observations with IRS were carried outfor 14 bubbles, 4 in high-resolution mode (Flagey et al. 2011) and10 in low-resolution (Nowak et al., in prep. ). Among the 4 bubblesobserved in high-resolution mode, two show a dust-poor spectrumdominated by highly ionized gas lines of [O IV ], [Ne III ], [Ne V ], [S III ], and [S IV ], typical of planetary nebulae with a very hot centralwhite dwarf ( (cid:38)
200 000K). The other two spectra are dominated The IR Spectrograph on board of the Spitzer Space Telescope.c (cid:13) a r X i v : . [ a s t r o - ph . GA ] N ov A. Ingallinera by a dust continuum and lower-excitation lines. These two bubblesalso show a central source and are, respectively, a nebula surround-ing a WR star (Stringfellow et al. 2012a) and a LBV candidate(Wachter et al. 2010).An extensive search of available catalogues had allowed us toidentify less than 15 per cent of these objects. The majority of thealready known bubbles were found to be planetary nebulae (PNe).Three supernova remnants (SNRs) and one post-asymptotic giantbranch (AGB) star were also identified. Therefore, about 90 percent of the objects within the MIPSGAL bubbles were new discov-eries. Further studies on the bubble catalogue allowed to extend thenumber of classified bubbles to, presently, about 30 per cent.Massive stars play a pivotal role in the evolution of their hostgalaxies. They are among the major contributors to the interstellarultraviolet radiation and, via their strong stellar winds and final ex-plosion, provide enrichment of processed material (gas and dust)and mechanical energy to the interstellar medium, strongly influ-encing subsequent local star formation. Still, the details of post-main sequence (MS) evolution of massive stars are poorly under-stood. On one side, theoretical modelling depends on mass-lossfrom the stars, which in turn is function of poorly constrained pa-rameters such as metallicity and rotation (e.g. Leitherer & Langer1991; Chieffi & Limongi 2013). On the other side, empirical stud-ies had relied on a low number of objects at different stages ofpost-MS evolution (Clark et al. 2005), and only recently IR obser-vations have permitted the discovery of hundreds of new WR andLBV stars (e.g. Shara et al. 2009; Shara et al. 2012; Wachter et al.2010; Wachter et al. 2011; Mauerhan, Van Dyk, & Morris 2011;Stringfellow et al. 2012a; Stringfellow et al. 2012b). Besides beinga powerful ‘game reserve’ for evolved massive stars, the mid-IRbubbles catalogue is the right place where to look for the miss-ing Galactic population of embedded PNe. The small number ofknown PNe (i.e. ∼ ∼ M (cid:62) M (cid:12) ) PNe progenitors.In this paper we present results of a radio continuum study ofa sub-sample of the bubbles aimed at understanding their nature.Spectral information, as derived from multi-frequency radio obser-vations, are an unique tool for a first assessment of the content ofnon-thermal and thermal radio emitters in our sample, sorting outSNRs or, more generally, shocked nebulae (synchrotron emission),from nebulae associated to evolved massive stars (LBV and WR)and PNe (thermal free-free emission). It is eventually showed howradio and IR observations can be combined to establish more ex-haustive classification schemes. From the original sample of the MIPSGAL bubbles only sourceswith δ (cid:62) − ◦ (to be visible with the EVLA ) were selected, re-sulting in a total ‘northern sample’ of 367 sources. We then checkeda 1 (cid:48) × (cid:48) field centred on each of the MIPSGAL positions in both The Expanded Very Large Array. the NVSS catalogue and in MAGPIS for radio emission, endingup with a total of 55 sources possibly detected at 20cm. Despitethe fact that, for our targeted sources, either NVSS or MAGPIS(or both) data already exist, these cannot be used for the purposeof identifying the 24 µ m MIPSGAL bubbles. In fact, the existingNVSS/MAGPIS data suffer from three main issues: the availabledata were obtained with a typical rms of 0 . − . / beam, onaverage one order of magnitude worst than what is achievable withEVLA; the existing observations were taken at a different time withrespect to ours and time variability effects could potentially affectthe spectral index analysis; the combination of VLA and EVLAdata can be, in principle, very problematic from a technical pointof view.The available NVSS and MAGPIS data yet provided very use-ful indications regarding the size and flux of our selected sample ofsources, and this information was used to guide our observing strat-egy in terms of configuration and time request. Remarkably, 11 ofthe objects selected for EVLA observations were already classified,according to the SIMBAD database. Observations of the bubbles sample were made with the EVLA at6cm (central frequency 4 . C band) in configuration Dduring March 2010 and at 20cm (1 . L band) in config-uration C and CnB during, respectively, March and May 2012.For C -band observations the sample was split in four sub-set, observed in four different days. Each bubble was observed forslightly less than 10 minutes and in two 128-MHz wide spectralwindows (resulting therefore in a total bandwidth of 256MHz) al-lowing us to achieve a theoretical noise level of ∼ µ Jy / beam.We note that calibration errors and required flagging introduce fur-ther sources of noise that eventually dominate over the theoreticalthermal noise.For observations in L band, the previous 6-cm observationswere used to select a sub-sample on which focusing our attention.In particular we selected a sub-sample of 34 bubbles detected orpossibly detected at 6cm, excluding some bubbles that appearedtoo extended at 6cm or whose classification was certain. The largerfield-of-view at 20cm allowed to include other 6 bubbles as fieldsources, resulting in a total sample of 40 bubbles. Though the totalbandwidth was as wide as 1GHz, a lot of radio-frequency interfer-ences (RFI) contaminated our data and the signal-to-noise ratio wasmuch lower than we expected, sometimes one order of magnitudeor more.Since observations were made toward the galactic plane, itwas also necessary to check the confusion limit. At 6cm we ex-pected a value around 7 µ Jy / beam, while at 20cm slightly less than20 µ Jy / beam. Both limits were well below our expected noise lev-els. In Table 1 all observed objects are reported along with theircoordinates and the date and duration of each observation. Besidethe official designation (MGE l ± b ), for each bubble we list a shorteridentification name (second column), derived from shorthands usedduring the identification phase, that will be used in this work as acompact notation. http://third.ucllnl.org/gps/catalogs.html http://simbad.u-strasbg.fr c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles Table 1.
Observations summary. Source dimensions at 24 µ m are from the bubbles catalogue (Mizuno et al. 2010).Designation Bubble RA DEC Obs. day Obs. time Obs. day Obs. time Dimension Classified[MGE] ID (J2000) (J2000) (2010) (min) (2012) (min) at 24 µ m in SIMBAD? C band C band L band L band010.5569+00.0188 3153 18:08:50 . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) a
10 15 (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) a
10 18 (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) PN b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) a (cid:48)(cid:48) SNR b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) a (cid:48)(cid:48) PN b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) c
18 44 (cid:48)(cid:48) . ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) c
15 16 (cid:48)(cid:48) . ◦ (cid:48) (cid:48)(cid:48) c
18 21 (cid:48)(cid:48) . ◦ (cid:48) (cid:48)(cid:48) c
21 21 (cid:48)(cid:48) . ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) PN b [ ] . ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) PN b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) PN b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) SNR b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) a
10 28 (cid:48)(cid:48) PN b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) PN b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) PN b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) a
10 12 (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) PN? b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) a Observed as field source. b [1] Miszalski et al. 2008; [2] Green 2009; [3] Kerber et al. 2003; [4] Chevalier 2005; [5] Parker et al. 2006; [6] Kohoutek 2001. c Observed also on 13–May.c (cid:13) , 1–, 1–
10 12 (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) PN? b [ ] . − ◦ (cid:48) (cid:48)(cid:48) (cid:48)(cid:48) a Observed as field source. b [1] Miszalski et al. 2008; [2] Green 2009; [3] Kerber et al. 2003; [4] Chevalier 2005; [5] Parker et al. 2006; [6] Kohoutek 2001. c Observed also on 13–May.c (cid:13) , 1–, 1– ?? A. Ingallinera
The entire data reduction process was performed using the package
CASA . As a first step, the data were edited and flagged in order toidentify and delete not properly working antennas, bad baselinesand border (and usually noisy) channels. For C -band observationsthe editing process revealed no great corruptions in our data, whilefor L -band observations a large amount of flagging was needed inorder to filter out the conspicuous RFI, leading to less 33 per centof useful data remaining.For all the observations, the bandpass and flux calibrationswere done using 3C286 as calibrator. In order to improve the qual-ity of our gain calibration, depending on the distance from thesource (typically within 10 ◦ ), we used a variety of standard cali-brators spanning a range of flux densities. Data imaging was made using the Clark implementation (Clark1980) of the CLEAN algorithm (H¨ogbom 1974), convolving theresulting ‘clean components’ with a Gaussian PSF.For C band, since all observations were carried out with theEVLA in the same configuration (D), no significant differenceswere found in synthesis beam sizes. Therefore all the images werebuilt using a 4 (cid:48)(cid:48) pixel and a total size of 256 ×
256 pixels, in sucha way that each map covers approximately a 17 (cid:48) × (cid:48) area (theprimary beam is about 9 (cid:48) FWHM). In some maps we were able toclean down to a rms ∼ µ Jy / beam, with an average beam sizearound 25 (cid:48)(cid:48) × (cid:48)(cid:48) . The typical noise was one order of magnitudegreater than the confusion limit.For L band instead, since multiple configurations were used,for each image a best choice between a pixel size of 3 (cid:48)(cid:48) or 4 (cid:48)(cid:48) wasadopted. Also, the size of images was allowed to vary to best ac-commodate for field sources cleaning. The typical rms was about0 . − / beam with an average beam size of 18 (cid:48)(cid:48) × (cid:48)(cid:48) . The typ-ical noise was two orders of magnitude greater than the confusionlimit.In C band we expected that only sources with dimensions sig-nificantly less than 2 (cid:48) (EVLA largest angular scale) could be rea-sonably well imaged, and this would also permit a total flux den-sity recovery. The more a source is extended the less reliable isits flux density measurement. Therefore, at the end of the imagingprocess, we cautiously excluded eight bubbles (namely 3259, 3282,3310, 3328, 3558, 3910, 4485 and 4595) from the remainder of thiswork since they were suspected to be resolved out by the EVLA. Asingle-dish analysis for these bubbles is in progress.Radio maps and 24- µ m images of some Bubbles are presentedin appendix A (online only). The majority of the bubbles observed were detected in both bands.In particular for C band we detected 44 bubbles out of 55, with3 uncertain detections and 8 non-detections. For L band we de-tected 23 bubbles out of 40, with 3 uncertain detections and 14non-detections.Since one of the main goals of this work was to characterisethe radio emission of the bubbles as an important aid to their classi-fication, a very accurate flux density determination was needed. To avoid introducing methodological errors or biases, a unique proce-dure in this calculation was adopted. First of all the sources weredivided into two classes depending on whether they were resolvedor not.For point sources (not resolved) the flux density was deter-mined using the CASA task imfit , which fits an elliptical Gaus-sian component to an image. Given that the maps units are jan-sky/beam, the total flux density for a point source is equal to thepeak value of the fitted Gaussian, i.e. S = S p . The error was com-puted as the quadratic sum of the error derived from the fit, the maprms and the calibration error (this one, negligible in both bands): ∆ S = (cid:113) σ + σ + σ . (1)The flux density calculation for extended sources proved muchmore difficult. For extended sources detected or resolved in oneband only, the strategy was to localise the source boundary as thelowest brightness level at which we were confident to encompassonly our object. Theoretically, one should go down to σ rms , be-low which the source becomes indistinct with respect to the back-ground. However, the artefacts in interferometric images usually donot allow to look so deep and, for many bubbles, we were forced tostop at higher levels. Selected then an appropriate region for eachobject, the flux density was calculated by means of an integrationover this area, performed directly with the CASA viewer . The to-tal error was estimated as the map rms multiplied by the square rootof the integration area expressed in beams.For sources resolved in both bands we proceeded as follows.First the map with the higher angular resolution was degraded byconvolving the clean components with the lower resolution beamand adding back the residual map. Then, for each bubble, we se-lected a region large enough to cover the source in both bands, andused this to estimate the flux and corresponding error as in the pre-vious case.Furthermore an approximate size for resolved bubbles wascalculated as follows: the observed size of the source, Ω o , is ex-pressed as Ω o = Ω s + Ω b (2)where Ω s is the ‘real’ angular size of the source and Ω b is the beamsolid angle. The quantity Ω o can also been expressed in term ofnumber of beams, N b , a quantity already computed for the determi-nation of flux densities Ω o = N b Ω b , (3)hence Ω s = ( N b − ) Ω b (4)and we calculated the corresponding mean size as (cid:104) θ s (cid:105) = (cid:113) b maj b min ( N b − ) , (5)where b maj and b min are, respectively, the beam major and minoraxis. The results obtained are listed, along with some useful char-acteristics of each map, in Table 2 for C band and in Table 3 for L band.As mentioned in the previous section, the determination of aspectral index for as many bubbles as possible was critical for thiswork. Once the flux densities were estimated as described above,the spectral index α is defined as S ν ∝ ν α , (6)with an associated error given by c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles Table 2.
Flux densities at 6cm. Among the 44 bubbles detected at this frequency 8 are, likely, resolved out (see section 2.4) and for one (Bubble 3173) the fluxdensity measurement is not reliable. Therefore only 35 bubbles are listed.Bubble Map rms Beam PA Flux density Resolved? (cid:104) θ s (cid:105) Notes(mJy/beam) (mJy)3188 0.24 23 . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) Self-calibrated . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . Self-calibrated ∆ α (cid:39) (cid:115)(cid:18) ∆ S L S L (cid:19) + (cid:18) ∆ S C S C (cid:19) ln ν C ν L , (7)where subscripts C and L refer, respectively, to 6cm and 20cm ob-servations. The error on frequencies was neglected. The analysis of the spectral indices, obtained as described above,suggests that many bubbles are free-free emitters, with the majorityoptically thick at 20cm (see Table 4 and Figure 1). Only Bubbles3367 and 4486 may have spectral index values compatible withnon-thermal emission.Among all the observed objects, two bubbles, 3654 and 3706,were already classified as PNe (Kerber et al. 2003). These twosources appear resolved in both bands in our images.As mentioned in Section 3.1, not all the bubbles were detected,especially in L band. It is very likely that the bubbles detected in C band but not in L band are characterised by positive spectral indicesand also, due to the higher rms in L band, are simply below the Table 4.
Spectral index for sources detected in both bands.Bubble Flux density Flux density α Resolved?at 20cm (mJy) at 6cm (mJy)3222 21 . ± . . ± . . ± .
13 no3333 4 . ± . . ± . . ± .
26 yes3354 12 . ± . . ± . . ± .
08 yes3367 6 . ± . . ± . − . ± .
19 no3438 10 . ± . . ± . . ± .
07 yes3448 12 . ± . . ± . − . ± .
08 yes3654 64 . ± . . ± . − . ± .
01 yes3706 10 . ± . . ± . . ± .
28 yes3866 9 . ± . . ± . . ± .
30 no4436 6 . ± . . ± . − . ± .
05 no4465 1 . ± . . ± . . ± .
48 no4473 34 . ± . . ± . . ± .
09 no4486 20 . ± . . ± . − . ± .
06 yes4497 15 . ± . . ± . . ± .
06 no4552 15 . ± . . ± . − . ± .
07 yes4589 8 . ± . . ± . . ± .
06 no4602 14 . ± . . ± . . ± .
10 no4607 5 . ± . . ± . . ± .
09 noc (cid:13) , 1–, 1–
09 noc (cid:13) , 1–, 1– ?? A. Ingallinera
Table 3.
Flux densities at 20cm. The flux densities (or their upper limits) for 7 bubbles were not reliable and are not listed.Bubble Map rms Beam PA Flux density Resolved? (cid:104) θ s (cid:105) Notes(mJy/beam) (mJy)3188 1.46 25 . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) Resolved-out at . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . Peak intensity . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . (cid:48)(cid:48) Self-calibrated . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ < . Upper limit only . (cid:48)(cid:48) × . (cid:48)(cid:48) − ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ . ± . C o un t s Figure 1.
Spectral index statistical distribution. detection limit. It is possible to estimate a minimum spectral indexfor each non-detected bubble assuming an upper limit for their fluxdensity as follows: (1) for point sources in C band the flux densityupper limit at 20cm is simply assumed as three times the rms ofthe respective L -band map, (2) for extended sources the size of thesource as imaged in C band is reported in number of beams of the L -band map and the square-root of this number is multiplied by three Table 5.
Bubbles detected only in C band. For flux densities in L band apossible range is provided as described in the text.Bubble S ( L ) (mJy) S ( C ) α Resolved?min max (mJy) (mJy)3188 0.1 4.5 1 . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) + . . ± . (cid:38) − . . ± . (cid:38) + . . ± . (cid:38) + . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . times the map rms. Assuming pure black-body emission ( α = L -band fluxdensity values. In Table 5 we provide the results of this estimate. c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles The determination of the radio spectral index in the previous sec-tion has allowed us to make preliminary hypotheses of the natureof the bubbles. However a multi-wavelength approach is necessaryto fully characterise these objects.In addition to MIPSGAL and GLIMPSE observations, manybubbles were detected in other IR bands, from 1 . µ m to 160 µ m. In particular we took into account data from on-line cataloguesof: the 2-Micron All Sky Survey (2MASS) at 1 . µ m ( J band),1 . µ m ( H band) and 2 . µ m ( K s band) (Cutri et al. 2003); theWide-field IR Survey Explorer (WISE) at 3 . µ m, 4 . µ m, 12 µ mand 22 µ m (Cutri et al. 2012); the Midcourse Space Experiment(MSX) at 8 . µ m, 12 µ m, 15 µ m and 21 µ m (Egan, Price, & Krae-mer 2003); the IR Astronomical Satellite (IRAS) at 12 µ m, 25 µ mand 60 µ m; the Japanese satellite AKARI at 9 µ m, 18 µ m, 60 µ m,90 µ m, 140 µ m and 160 µ m.In the Table 6 for each bubble listed in the Table 4, except 3654and 3706, a brief summary of all the available IR observations willbe presented. In the last comment we report a possible classificationfor each bubble as reported in literature or derived in this work.Beside the IR archive search, we also looked for possible de-tections in H α using the SuperCOSMOS H-alpha Survey (SHS;Parker et al. 2005). The survey detects all known PNe in Table 1(except 3558 and 3654, not covered by the survey), but also bub-bles 3193, 4436, 4602 and 4607. Our radio spectral index analysishas shown that these four bubbles are thermal emitters (see Tables4, 5 and 6). If we assume that the H α emission is a good tracer ofthe radio free-free continuum, the detection of these four bubblesin SHS corroborates our classification. However only the Bubble4602 is clearly detected in H α , while the other three nebulae ap-pear very faint and barely visible (we cannot even exclude a fakedetection). We therefore cautiously avoid a quantitative analysis inthis moment.In the following subsections, we will make use of this infor-mation to attempt a classification of the bubbles whose nature isstill uncertain. Herschel observations detected bubbles also at longer wavelengths, butthey will not be discussed in this work.c (cid:13) , 1–, 1–
Bubbles detected only in C band. For flux densities in L band apossible range is provided as described in the text.Bubble S ( L ) (mJy) S ( C ) α Resolved?min max (mJy) (mJy)3188 0.1 4.5 1 . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) + . . ± . (cid:38) − . . ± . (cid:38) + . . ± . (cid:38) + . . ± . (cid:38) − . . ± . (cid:38) − . . ± . (cid:38) − . times the map rms. Assuming pure black-body emission ( α = L -band fluxdensity values. In Table 5 we provide the results of this estimate. c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles The determination of the radio spectral index in the previous sec-tion has allowed us to make preliminary hypotheses of the natureof the bubbles. However a multi-wavelength approach is necessaryto fully characterise these objects.In addition to MIPSGAL and GLIMPSE observations, manybubbles were detected in other IR bands, from 1 . µ m to 160 µ m. In particular we took into account data from on-line cataloguesof: the 2-Micron All Sky Survey (2MASS) at 1 . µ m ( J band),1 . µ m ( H band) and 2 . µ m ( K s band) (Cutri et al. 2003); theWide-field IR Survey Explorer (WISE) at 3 . µ m, 4 . µ m, 12 µ mand 22 µ m (Cutri et al. 2012); the Midcourse Space Experiment(MSX) at 8 . µ m, 12 µ m, 15 µ m and 21 µ m (Egan, Price, & Krae-mer 2003); the IR Astronomical Satellite (IRAS) at 12 µ m, 25 µ mand 60 µ m; the Japanese satellite AKARI at 9 µ m, 18 µ m, 60 µ m,90 µ m, 140 µ m and 160 µ m.In the Table 6 for each bubble listed in the Table 4, except 3654and 3706, a brief summary of all the available IR observations willbe presented. In the last comment we report a possible classificationfor each bubble as reported in literature or derived in this work.Beside the IR archive search, we also looked for possible de-tections in H α using the SuperCOSMOS H-alpha Survey (SHS;Parker et al. 2005). The survey detects all known PNe in Table 1(except 3558 and 3654, not covered by the survey), but also bub-bles 3193, 4436, 4602 and 4607. Our radio spectral index analysishas shown that these four bubbles are thermal emitters (see Tables4, 5 and 6). If we assume that the H α emission is a good tracer ofthe radio free-free continuum, the detection of these four bubblesin SHS corroborates our classification. However only the Bubble4602 is clearly detected in H α , while the other three nebulae ap-pear very faint and barely visible (we cannot even exclude a fakedetection). We therefore cautiously avoid a quantitative analysis inthis moment.In the following subsections, we will make use of this infor-mation to attempt a classification of the bubbles whose nature isstill uncertain. Herschel observations detected bubbles also at longer wavelengths, butthey will not be discussed in this work.c (cid:13) , 1–, 1– ?? A . I nga lli n e r a Table 6.
Synoptic table of IR observations. Legend: ‘C’ only central source, ‘N’ only diffuse emission, ‘B’ both central source and diffuse emission, ‘P’ point source due to low resolution, ‘–’ no source detected. Inthe last column the ‘?’ indicates a candidate while ‘
RadTh ’ that we can only state that we are observing a radio thermal emitter.Bubble 2MASS WISE IRAC MSX IRAS AKARI Comments J / H / K s [3.4]/[4.6]/[12]/[22] [3.6]/[4.5]/[5.8]/[8] [8.3]/[12]/[15]/[21] [12]/[25]/[60] [9]/[18]/[65]/[90]/[140]/[160]3222 –/C/C C/C/N/N C/C/B/B P/P/P/P –/P/– –/–/–/–/– PN? (Urquhart et al. 2009)3333 –/–/– –/–/–/N –/–/–/– –/–/–/– –/–/– –/–/–/–/– RadTh (This work) II region? (Anderson et al. 2011)3367 –/C/C C/C/N/N C/C/C/N –/–/–/– –/–/– –/P/–/–/–/– PN? (This work) RadTh (This work) (This work)
RadTh (This work) (This work)
RadTh (This work)
RadTh (This work)
RadTh (This work)
RadTh (This work) c (cid:13) R A S , M N R A S , ?? Radio Characterization of Galactic compact Bubbles In Section 3 we discussed the derivation of the radio spectral indexbetween 20 cm and 6cm for all those bubbles whose flux density iswell determined. We found that most of the bubbles have a posi-tive or slightly negative spectral index, indicating that we are verylikely observing thermal free-free emission typically in opticallythick regime, with a large amount of sources presenting a spectralindex of 0. This behaviour was somehow expected, since the ma-jority of the already classified bubbles are PNe (see Table 1). Fur-thermore also other kinds of evolved stars (such as LBV or WR) arecharacterized by a radio free-free emission, with only SNR show-ing clear non-thermal features.For 5 bubbles, a potential classification is available from theliterature, according to which 4 are PNe candidate (denoted assquares in Figure 2) and 1 is a H II region candidate (denoted astriangles in Figure 2). For these sources, the spectral index derivedfrom our analysis is consistent with the existing classification.Two sources, i.e. Bubbles 3367 and 4486, are characterized byrather negative spectral index values. Their spectral indices wereestimated as − .
30 and − .
25 respectively, values too low to beascribed to pure free-free emission. However, the errors associatedwith these measurements are significant, so the thermal emissionhypothesis cannot be entirely ruled out.
The emission at 24 µ m and 6cm have a different origins. In fact, theemission at 24 µ m originates both from warm thermal dust emis-sion, and from gas forbidden lines, such as [O IV ] at 25 . µ m(Flagey et al. 2011). The radio emission at 6cm, instead, origi-nates from either thermal free-free emission or synchrotron emis-sion. However it was shown by several authors that a strong corre-lation between mid-/far-IR and radio emission exists (de Jong et al.1985; Helou, Soifer, & Rowan-Robinson 1985; Pinheiro Gonc¸alveset al. 2011).In Figure 3 the flux density at 24 µ m from MIPSGAL plottedagainst the flux density at 6cm from our observations (Table 2),for all the bubbles with measured 6-cm flux density with the onlyexception of the Bubble 3313 (see below). The figure evidencesa clear correlation between the emission in the two bands. If wedefined for each bubble the quantity q = log S IR S ra (8)we find that q = . ± .
4, where the error is computed as the stan-dard deviation of the distribution. A linear fit to the ensemble of thelog S IR vs. log S ra values retrieveslog S IR = . S ra + . q is a good representation of log ( S IR / S ra ) .Bubble 3313 has a much higher S IR / S ra value ( ∼ µ m this source appearsvery extended (about 80 (cid:48)(cid:48) ) and might be interacting with Bubble3312 (Gvaramadze, Kniazev, & Fabrika 2010; Wachter et al. 2010).Spectroscopic near-IR studies of the central sources of these twobubbles reveal that both can be classified as WR stars of the samespectral type WN9h (Burgemeister et al. 2013). Our radio observa-tions at 6cm show a very faint irregular nebula around the centralstar of Bubble 3313, less extended than the 24- µ m nebula, with no emission around the other bubble or in any other region where the24- µ m emission is present (see Figure A17 in appendix A). Despitethe fact that this bubble is detected in the MAGPIS 20-cm tile, noemission is visible from our maps at 20cm. Indeed it is possiblethat the extended emission is below our detection limit (especiallyat 20cm) and/or that it was resolved out (especially at 6cm). Forthese reasons the flux density computation is not considered reli-able enough and the bubble was not included in this part of theanalysis.Although the emission at 24 µ m is well correlated with theemission at 6cm, we cannot use this effect to classify our sources.For example, if we compute q for 8 known PNe, we find a value of1 . ± . Combining our radio observations with IRAS archive data it is pos-sible to discriminate whether a source is a PN candidate or not. Al-though IRAS poor resolution did not allow us to resolve individualPNe, its sensitivity was enough to detect these objects at least at thedistance of the galactic center (Pottasch et al. 1988).Unfortunately, only few bubbles studied here have archivalIRAS fluxes, and none has a flux density determination in morethan two bands. Using the IRAS Point Source Catalogue andarchival VLA 6-cm data (Becker et al. 1994) for a sample of knownPNe and H II regions, we were able to generate color plots usefulfor our classification purposes.As a first step, it is important to notice that, following the dis-cussion in Section 4.2, the IRAS flux densities at 25 µ m are well-correlated with the radio flux densities at 6cm (Figure 4). This plotis quite similar to Figure 3. It is however interesting to notice howthe two plots span a different range of values in flux parameterspace, with the MIPSGAL and EVLA observations extending thecoverage towards lower flux densities. We also notice that, thoughPNe and H II regions partly overlap in Figure 4, H II regions be-come dominant at very high flux densities. All our 6 bubbles, forwhich both flux density values are available, are located in thelower-left region of the plot, so they are all compatible both withPNe and H II regions.A more interesting result can be obtained by plotting the IRASflux density values at 60 µ m against the radio flux densities at 6cm(Figure 5). In this plot it is still evident how IR and radio flux den-sity values correlate but it is also possible to notice how PNe repre-sent a population clearly separated from other H II regions (despitesome exceptions). From this plot, we might be tempted to clas-sify Bubble 4436 as a PN candidate. However, this hypothesis isnot supported by the distribution of IRAS 60 micron vs. 25 micronfluxes (Figure 6). In this case, PNe still occupy a well-defined andseparate region of space with respect to H II regions, but bubblesand PNe do not share the same region in the plot, with bubbles hav-ing a much lower flux density than both PNe and H II regions. In-deed, their low surface brightness is likely the reason these sourceswere not detected by the IRAS survey. Therefore, it is difficult tosay which classification is more appropriate for Bubble 4436, givenits outlier behaviour when compared to already classified objects.Using all IRAS bands combined with 6-cm data, we also gen-erated color-color diagrams. However, none of them was useful forour classification attempt, since no particular trend was observed. c (cid:13) , 1–, 1–
4, where the error is computed as the stan-dard deviation of the distribution. A linear fit to the ensemble of thelog S IR vs. log S ra values retrieveslog S IR = . S ra + . q is a good representation of log ( S IR / S ra ) .Bubble 3313 has a much higher S IR / S ra value ( ∼ µ m this source appearsvery extended (about 80 (cid:48)(cid:48) ) and might be interacting with Bubble3312 (Gvaramadze, Kniazev, & Fabrika 2010; Wachter et al. 2010).Spectroscopic near-IR studies of the central sources of these twobubbles reveal that both can be classified as WR stars of the samespectral type WN9h (Burgemeister et al. 2013). Our radio observa-tions at 6cm show a very faint irregular nebula around the centralstar of Bubble 3313, less extended than the 24- µ m nebula, with no emission around the other bubble or in any other region where the24- µ m emission is present (see Figure A17 in appendix A). Despitethe fact that this bubble is detected in the MAGPIS 20-cm tile, noemission is visible from our maps at 20cm. Indeed it is possiblethat the extended emission is below our detection limit (especiallyat 20cm) and/or that it was resolved out (especially at 6cm). Forthese reasons the flux density computation is not considered reli-able enough and the bubble was not included in this part of theanalysis.Although the emission at 24 µ m is well correlated with theemission at 6cm, we cannot use this effect to classify our sources.For example, if we compute q for 8 known PNe, we find a value of1 . ± . Combining our radio observations with IRAS archive data it is pos-sible to discriminate whether a source is a PN candidate or not. Al-though IRAS poor resolution did not allow us to resolve individualPNe, its sensitivity was enough to detect these objects at least at thedistance of the galactic center (Pottasch et al. 1988).Unfortunately, only few bubbles studied here have archivalIRAS fluxes, and none has a flux density determination in morethan two bands. Using the IRAS Point Source Catalogue andarchival VLA 6-cm data (Becker et al. 1994) for a sample of knownPNe and H II regions, we were able to generate color plots usefulfor our classification purposes.As a first step, it is important to notice that, following the dis-cussion in Section 4.2, the IRAS flux densities at 25 µ m are well-correlated with the radio flux densities at 6cm (Figure 4). This plotis quite similar to Figure 3. It is however interesting to notice howthe two plots span a different range of values in flux parameterspace, with the MIPSGAL and EVLA observations extending thecoverage towards lower flux densities. We also notice that, thoughPNe and H II regions partly overlap in Figure 4, H II regions be-come dominant at very high flux densities. All our 6 bubbles, forwhich both flux density values are available, are located in thelower-left region of the plot, so they are all compatible both withPNe and H II regions.A more interesting result can be obtained by plotting the IRASflux density values at 60 µ m against the radio flux densities at 6cm(Figure 5). In this plot it is still evident how IR and radio flux den-sity values correlate but it is also possible to notice how PNe repre-sent a population clearly separated from other H II regions (despitesome exceptions). From this plot, we might be tempted to clas-sify Bubble 4436 as a PN candidate. However, this hypothesis isnot supported by the distribution of IRAS 60 micron vs. 25 micronfluxes (Figure 6). In this case, PNe still occupy a well-defined andseparate region of space with respect to H II regions, but bubblesand PNe do not share the same region in the plot, with bubbles hav-ing a much lower flux density than both PNe and H II regions. In-deed, their low surface brightness is likely the reason these sourceswere not detected by the IRAS survey. Therefore, it is difficult tosay which classification is more appropriate for Bubble 4436, givenits outlier behaviour when compared to already classified objects.Using all IRAS bands combined with 6-cm data, we also gen-erated color-color diagrams. However, none of them was useful forour classification attempt, since no particular trend was observed. c (cid:13) , 1–, 1– ?? A. Ingallinera S(6 cm) (mJy)10 S ( c m ) ( m J y ) PN?HII reg.??
Figure 2.
Flux densities comparison at 20cm and at 6cm. The red (lower) area delimits the range of expected values for free-free emission, with the redline (bottom) representing a pure black-body emission ( α =
2) and the green (top) a pure optically thin free-free emission ( α = − . − . −
1, typical of an optically thin synchrotron emission. Noticeably, the majority of the points lie close to the greenline. S(6 cm) (mJy)10 S ( µ m ) ( m J y ) PNPN?HII reg.??
Figure 3.
Correlation between MIPSGAL flux densities at 24 µ m and radio data at 6cm from our EVLA observations. The grey dotted lines represent fluxdensity ratios of 10, 100 and 1000. As we discussed in the introduction, one of the characteristicof bubbles is that they are mostly detected only at 24 µ m. TheGLIMPSE survey, in fact, failed to detect extended emission forthe majority of the bubbles, despite the great sensitivity of IRAC.However, in seven cases, a faint nebular emission appears in theGLIMPSE data and for five of these we performed aperture pho-tometry using the Aperture Photometry Tool . For Bubbles 3222 and 4607 it was impossible to derive a reliable flux density: in factthe first nebula is very small and dominated by its central sourcewhile the second is faint and immersed in a confused fore- andbackground. To this end, we subtracted foreground point-sources,performed an interpolation of the empty pixels using the informa-tion from the surrounding background, and then estimated the skybackground as the median value of a sufficiently large region inproximity of the source. In addition to aperture photometry, whenthe central source is visible within the bubble, we extracted point-source photometry from the online GLIMPSE catalogue.Information on the nature of a source detected in all IRACbands come directly from the 3-color image obtained by superpo- c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles S(6 cm) (mJy)10 S ( µ m ) ( m J y ) PN??
Figure 4.
Correlation between IRAS flux densities at 25 µ m and radio data at 6cm. Small crosses are archive PNe and small points archive H II regions; largermarkers represent our bubbles with IRAS archive values and our radio data. It is possible to notice that the two flux densities are well-correlated and that thePNe are usually characterized by lower flux density values. S(6 cm) (mJy)10 S ( µ m ) ( m J y ) PN?HII reg.??
Figure 5.
Correlation between IRAS flux densities at 60 µ m and radio data at 6cm. Small crosses are archive PNe and small points archive H II regions; largermarkers represent our bubbles with IRAS archive values and our radio data. It is possible to notice that the PN population is characterized by a lower value ofthe two flux density ratio and is well-separated from the H II regions. sition of the monochromatic maps at 8 µ m, 5 . µ m and one amongthe other two bands. As discussed in Murphy et al. (2010), PNe usu-ally appear red, while H II regions appear either yellow or white.This is due, for H II regions, to PAH emission (yellow) or broad-band thermal emission by dust (white) (Cohen et al. 2011). An in-spection of the GLIMPSE 3-color images for Bubbles 3367, 3448,4473 and 4602, reveals a red color for all of them. Of these, two,namely 3448 and 4602, are classified in the literature as PN can-didates (Kohoutek 2001; Gvaramadze, Kniazev, & Fabrika 2010),while nothing is found about the nature of the other two. From whatemerges from this discussion and follows in the next section, it can be concluded that Bubbles 3367 and 4473 could also be consideredPN candidates. It is remarkable, in particular, how Bubble 4473morphologically resembles Bubble 4602 in the GLIMPSE images.On the other hand, Bubble 3354, classified as H II by Anderson etal. (2011), shows the expected yellow appearance.All the 5 bubbles considered show a nebular emission at 8 µ m,while for only one (Bubble 4602) this nebular emission is detectedin all four bands. It was shown that the ratio between the flux den-sity at 8 µ m and at 20cm ranges in a well-determined interval andthat different kinds of PNe are characterized by different values ofthis ratio (Cohen et al. 2011). In Figure 8 we plot the GLIMPSE c (cid:13) , 1–, 1–
Correlation between IRAS flux densities at 60 µ m and radio data at 6cm. Small crosses are archive PNe and small points archive H II regions; largermarkers represent our bubbles with IRAS archive values and our radio data. It is possible to notice that the PN population is characterized by a lower value ofthe two flux density ratio and is well-separated from the H II regions. sition of the monochromatic maps at 8 µ m, 5 . µ m and one amongthe other two bands. As discussed in Murphy et al. (2010), PNe usu-ally appear red, while H II regions appear either yellow or white.This is due, for H II regions, to PAH emission (yellow) or broad-band thermal emission by dust (white) (Cohen et al. 2011). An in-spection of the GLIMPSE 3-color images for Bubbles 3367, 3448,4473 and 4602, reveals a red color for all of them. Of these, two,namely 3448 and 4602, are classified in the literature as PN can-didates (Kohoutek 2001; Gvaramadze, Kniazev, & Fabrika 2010),while nothing is found about the nature of the other two. From whatemerges from this discussion and follows in the next section, it can be concluded that Bubbles 3367 and 4473 could also be consideredPN candidates. It is remarkable, in particular, how Bubble 4473morphologically resembles Bubble 4602 in the GLIMPSE images.On the other hand, Bubble 3354, classified as H II by Anderson etal. (2011), shows the expected yellow appearance.All the 5 bubbles considered show a nebular emission at 8 µ m,while for only one (Bubble 4602) this nebular emission is detectedin all four bands. It was shown that the ratio between the flux den-sity at 8 µ m and at 20cm ranges in a well-determined interval andthat different kinds of PNe are characterized by different values ofthis ratio (Cohen et al. 2011). In Figure 8 we plot the GLIMPSE c (cid:13) , 1–, 1– ?? A. Ingallinera S(25 µm) (mJy)10 S ( µ m ) ( m J y ) PN??
Figure 6.
Correlation between IRAS flux densities at 60 µ m and at 25 µ m. Small crosses are archive PNe and small points archive H II regions; larger markersrepresent our bubbles with IRAS archive values. Also in this plot it is possible to notice that the PN population is characterized by a lower value of the twoflux density ratio and is well-separated from the H II regions. -40-30-20-10010203040 GLON offset (arcsec)-40-30-20-10010203040 G L A T o ff s e t ( a r c s e c ) -40-30-20-10010203040 GLON offset (arcsec)-40-30-20-10010203040 G L A T o ff s e t ( a r c s e c ) -40-30-20-10010203040 GLON offset (arcsec)-40-30-20-10010203040 G L A T o ff s e t ( a r c s e c ) -40-30-20-10010203040 GLON offset (arcsec)-40-30-20-10010203040 G L A T o ff s e t ( a r c s e c ) -40-30-20-10010203040 GLON offset (arcsec)-40-30-20-10010203040 G L A T o ff s e t ( a r c s e c ) Figure 7.
Three-color superposition of GLIMPSE tile cut-outs at 3 . µ m (blue), 5 . µ m (green) and 8 µ m (red) for Bubbles 3354, 3367, 3448, 4473 and 4602. Itis remarkable how Bubble 3354 appears different in shape and color with respect to the others and how 4473 and 4602 are morphologically and chromaticallysimilar. flux densities against the radio values from our data. It is possibleto notice how Bubble 3354 clearly does not satisfy the selectioncriterion, in agreement with a classification as an H II region andnot a PN. The other 4 bubbles all lie inside the area where PNeshould be found. In particular the unclassified Bubble 4473 is veryclose to the median ratio value of 4.7, with a calculated ratio of 4.5. These 4 bubbles can be divided in two groups, according to theirratio value: the first group comprises Bubbles 4473 and 4602 andthe second group Bubbles 3367 and 3448. We have already talkedabout the morphological similarities of Bubbles 4473 and 4602: theresult found here may suggest that these two objects could sharemany of their physical characteristics. The other two bubbles ap- c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles S(20 cm) (mJy)10 S ( µ m ) ( m J y ) PN?HII reg.??
Figure 8.
GLIMPSE flux densities at 8 µ m against radio flux densities at 20cm derived from our observations. The coloured area represents the ratio intervalwhere PNe are usually located according to Cohen et al. (2011), with a confidence level of 1 σ (darker area) and 3 σ (lighter area). pear different from the first two. Indeed, for Bubble 3367 no mor-phological consideration can be done, while Bubble 3448 seems tohave bipolar structure. If all these bubbles will be confirmed to bePNe their morphological and physical differences may be due tointrinsic properties or their evolutionary stage. The classification of bubbles is very complicated and a definitiveanswer on this topic is far from being given here. However, fromthis analysis it has clearly emerged that the multi-wavelength ap-proach that we presented is a powerful tool for achieving a sensibleclassification.For at least 21 bubbles, previously unclassified, the spectralindex analysis suggests that they are thermal free-free emitters.Important results have been obtained when our radio data havebeen combined with archival data from IR observation with Spitzerand IRAS. We have shown that correlation and color-color plotscan help to discriminate among different types of objects.A word of caution is necessary concerning the IR-radio cor-relation. Although we have demonstrated that such a correlation –which is known to characterize various classes of astronomical ob-jects – holds true also for Galactic bubbles, yet it cannot be usedalone for classification purposes.We have discussed the morphology of the bubbles at differentwavelengths, considering a peculiar shape as indicative of somekind of circumstellar envelope. These considerations are applicableonly to few sources. Indeed, many bubbles are barely resolved andtheir lack of significant feature may be both an intrinsic property oran instrumental limit.
ACKNOWLEDGMENTS
This work is based on observations made with the Very Large Ar-ray of the National Radio Astronomy Observatory, a facility of the National Science Foundation operated under cooperative agree-ment by Associated Universities Inc., and on data products fromthe
Spitzer Space Telescope , which is operated by the Jet Propul-sion Laboratory, California Institute of Technology under a contractwith NASA. Archive search made use of the SIMBAD databaseand the VizieR catalogue access tool, operated by the Centre deDonn´ees astronomique de Strasbourg.
REFERENCES
Anderson L. D., Bania T. M., Balser D. S., Rood R. T., 2011, yCat,219, 40032Becker R. H., White R. L., Helfand D. J., Zoonematkermani S.,1994, ApJS, 91, 347Burgemeister S., Gvaramadze V. V., Stringfellow G. S., KniazevA. Y., Todt H., Hamann W.-R., 2013, MNRAS, 429, 3305Carey S. J., et al., 2009, PASP, 121, 76Chevalier R. A., 2005, ASPC, 342, 422Chieffi A., Limongi M., 2013, ApJ, 764, 21Clark B.G., 1980, A&A, 89, 377Clark J. S., Negueruela I., Crowther P. A., Goodwin S. P., 2005,A&A, 434, 949Cohen M., Parker Q. A., Green A. J., Miszalski B., Frew D., Mur-phy T., 2011, MNRAS, 413, 514Cutri R. M., et al., 2003, tmc..book,Cutri R. M., et al., 2012, yCat, 2311, 0de Jong T., Klein U., Wielebinski R., Wunderlich E., 1985, A&A,147, L6Egan M. P., Price S. D., Kraemer K. E., 2003, AAS, 35, 1301Flagey N., Noriega-Crespo A., Billot N., Carey S. J., 2011, ApJ,741, 4Green D. A., 2009, BASI, 37, 45Gvaramadze V. V., Kniazev A. Y., Fabrika S., 2010, MNRAS, 405,1047Helfand D. J., Becker R. H., White R. L., Fallon A., Tuttle S.,2006, AJ, 131, 2525Helou G., Soifer B. T., Rowan-Robinson M., 1985, ApJ, 298, L7 c (cid:13) , 1–, 1–
Anderson L. D., Bania T. M., Balser D. S., Rood R. T., 2011, yCat,219, 40032Becker R. H., White R. L., Helfand D. J., Zoonematkermani S.,1994, ApJS, 91, 347Burgemeister S., Gvaramadze V. V., Stringfellow G. S., KniazevA. Y., Todt H., Hamann W.-R., 2013, MNRAS, 429, 3305Carey S. J., et al., 2009, PASP, 121, 76Chevalier R. A., 2005, ASPC, 342, 422Chieffi A., Limongi M., 2013, ApJ, 764, 21Clark B.G., 1980, A&A, 89, 377Clark J. S., Negueruela I., Crowther P. A., Goodwin S. P., 2005,A&A, 434, 949Cohen M., Parker Q. A., Green A. J., Miszalski B., Frew D., Mur-phy T., 2011, MNRAS, 413, 514Cutri R. M., et al., 2003, tmc..book,Cutri R. M., et al., 2012, yCat, 2311, 0de Jong T., Klein U., Wielebinski R., Wunderlich E., 1985, A&A,147, L6Egan M. P., Price S. D., Kraemer K. E., 2003, AAS, 35, 1301Flagey N., Noriega-Crespo A., Billot N., Carey S. J., 2011, ApJ,741, 4Green D. A., 2009, BASI, 37, 45Gvaramadze V. V., Kniazev A. Y., Fabrika S., 2010, MNRAS, 405,1047Helfand D. J., Becker R. H., White R. L., Fallon A., Tuttle S.,2006, AJ, 131, 2525Helou G., Soifer B. T., Rowan-Robinson M., 1985, ApJ, 298, L7 c (cid:13) , 1–, 1– ?? A. Ingallinera
H¨ogbom J.A., 1974, A&AS, 15, 417Kerber F., Mignani R. P., Guglielmetti F., Wicenec A., 2003,A&A, 408, 1029Kohoutek L., 2001, A&A, 378, 843Leitherer C., Langer N., 1991, IAUS, 148, 480Mauerhan J. C., Van Dyk S. D., Morris P. W., 2011, AJ, 142, 40Miszalski B., Parker Q. A., Acker A., Birkby J. L., Frew D. J.,Kovacevic A., 2008, MNRAS, 384, 52Mizuno D. R., et al., 2010, AJ, 139, 1542Morris P. W., Stolovy S., Wachter S., Noriega-Crespo A., PannutiT. G., Hoard D. W., 2006, ApJ, 640, L179Murphy T., Cohen M., Ekers R. D., Green A. J., Wark R. M., MossV., 2010, MNRAS, 405, 1560Parker Q. A., et al., 2005, MNRAS, 362, 689Parker Q. A., et al., 2006, MNRAS, 373, 79Pinheiro Gonc¸alves D., Noriega-Crespo A., Paladini R., MartinP. G., Carey S. J., 2011, AJ, 142, 47Pottasch S. R., Olling R., Bignell C., Zijlstra A. A., 1988, A&A,205, 248Rieke G. H., et al., 2004, ApJS, 154, 25Shara M. M., et al., 2009, AJ, 138, 402Shara M. M., Faherty J. K., Zurek D., Moffat A. F. J., Gerke J.,Doyon R., Artigau E., Drissen L., 2012, AJ, 143, 149Stringfellow G. S., Gvaramadze V. V., Beletsky Y., Kniazev A. Y.,2012, IAUS, 282, 267Stringfellow G. S., Gvaramadze V. V., Beletsky Y., Kniazev A. Y.,2012, ASPC, 465, 514Urquhart J. S., et al., 2009, A&A, 501, 539Wachter S., Mauerhan J. C., Van Dyk S. D., Hoard D. W., KafkaS., Morris P. W., 2010, AJ, 139, 2330Wachter S., Mauerhan J., van Dyk S., Hoard D. W., Morris P.,2011, BSRSL, 80, 291Zijlstra A. A., Pottasch S. R., 1991, A&A, 243, 478
APPENDIX A: IMAGES
In this appendix we show radio contour plots from our data at 6cmand 20cm for all the bubbles listed in Table 6. For eight of them wealso present a superposition of MIPSGAL 24 µ m and radio contourat 6cm. The poor resolution of radio observations, along with avery elongated beam in some cases, only allows us to state that theradio emission is usually co-spatial with the IR, with the importantexception of Bubble 3354. c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles Figure A1.
Radio contours are 10, 15, 20 and 25mJy / beam (left) and 5, 10, 15, and 20mJy / beam (right). Figure A2.
Radio contours are 0.35, 0.70, 1.05 and 1 . / beam (left) and 0.5, 1.0, 1.5, 2.0 and 2 . / beam (right). Figure A3.
Radio contours are 1, 2, 3, 4 and 5mJy / beam (left) and 1, 2, 3, and 4mJy / beam (right).c (cid:13) , 1–, 1–
Radio contours are 1, 2, 3, 4 and 5mJy / beam (left) and 1, 2, 3, and 4mJy / beam (right).c (cid:13) , 1–, 1– ?? A. Ingallinera
Figure A4.
Radio contours are 2.5, 3.5, 4.5 and 5 . / beam (left) and 3, 5, 7 and 9mJy / beam (right). Figure A5.
Radio contours are 2, 4, 6, 8 and 10mJy / beam (left) and 1.5, 3, 4.5, and 6mJy / beam (right). Figure A6.
Radio contours are 3, 6, 9 and 12mJy / beam (left) and 2, 4, 6 and 8mJy / beam (right).c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles Figure A7.
Radio contours are 2, 4, 6, 8 and 10mJy / beam (left) and 3, 6, 9, and 12mJy / beam (right). Figure A8.
Radio contours are 1, 2, 3, 4 and 5mJy / beam (left) and 2, 4 and 6mJy / beam (right). Figure A9.
Radio contours are 0.3, 0.6, 0.9, 1.2 and 1 . / beam (left) and 0.7, 0.8, 0.9, and 1 . / beam (right).c (cid:13) , 1–, 1–
Radio contours are 0.3, 0.6, 0.9, 1.2 and 1 . / beam (left) and 0.7, 0.8, 0.9, and 1 . / beam (right).c (cid:13) , 1–, 1– ?? A. Ingallinera
Figure A10.
Radio contours are 7, 14, 21, 28 and 35mJy / beam (left) and 6, 12, 18, 24 and 30mJy / beam (right). Figure A11.
Radio contours are 4, 8, 12 and 16mJy / beam (left) and 3, 6, 9, and 12mJy / beam (right). Figure A12.
Radio contours are 3, 6, 9, 12 and 15mJy / beam (left) and 3, 6, 9 and 12mJy / beam (right).c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles Figure A13.
Radio contours are 3, 6, 9 and 12mJy / beam (left) and 2.5, 5, 7.5, and 10mJy / beam (right). Figure A14.
Radio contours are 2, 4, 6 and 8mJy / beam (left) and 2, 4, 6 and 8mJy / beam (right). Figure A15.
Radio contours are 4, 8, 12 and 16mJy / beam (left) and 4, 8, 12, and 16mJy / beam (right).c (cid:13) , 1–, 1–
Radio contours are 4, 8, 12 and 16mJy / beam (left) and 4, 8, 12, and 16mJy / beam (right).c (cid:13) , 1–, 1– ?? A. Ingallinera
Figure A16.
Radio contours are 4, 6, 8 and 10mJy / beam (left) and 2, 4 and 6mJy / beam (right).c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e c (cid:13) , 1–, 1–
Radio contours are 4, 6, 8 and 10mJy / beam (left) and 2, 4 and 6mJy / beam (right).c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e c (cid:13) , 1–, 1– ?? A. Ingallinera G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e c (cid:13) , 1–, 1–
Radio contours are 4, 6, 8 and 10mJy / beam (left) and 2, 4 and 6mJy / beam (right).c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e c (cid:13) , 1–, 1– ?? A. Ingallinera G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e c (cid:13) , 1–, 1– ?? A. Ingallinera G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e c (cid:13) , 1– ?? Radio Characterization of Galactic compact Bubbles G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e Figure A17.
Superposition of radio contours at 6cm for Bubble 3313 on MAGPIS 20-cm map (left, inverted colours) and MIPSGAL 24- µ m image (right).Radio contour levels are (in both cases) 0 .
26, 0 .
28, 0 .
30, 0 .
40, 0 .
60, 0 .
80, 1 .
00, 1 .
20 and 1 . / beam. It is possible to notice how at 20cm part of thecircular shell observed at 24 µ m is clearly detected. However, the 6-cm emission seems to come from the interior zone of the nebula, with a possible arcstructure (top-right in the images) that traces the 24- µ m emission.c (cid:13) , 1–, 1–