THE TOP 10 SPITZER YOUNG STELLAR OBJECTS IN 30 DORADUS

The Spitzer Space Telescope Legacy Program "Surveying the Agents of a Galaxy's Evolution" (SAGE; Meixner et al. 2006) imaged a 7° × 7° region in the LMC with all the IRAC (3.6, 4.5, 5.8, and 8.0 μm; Fazio et al. 2004) and MIPS (24, 70, and 160 μm; Rieke et al. 2004) bands. The MIPS data are not used here because of their lower angular resolution. We use the IRAC "Single Frame + Mosaic Photometry" Archive.⁶

The "VISTA Magellanic Clouds" survey (VMC; Cioni et al. 2011) is a European Southern Observatory (ESO) public survey, which consists of deep imaging in the three NIR bands Y (1.02 μm), J (1.25 μm), and K_s (2.15 μm), covering both Magellanic Clouds and surrounding areas, with the VIRCAM (VISTA IR Camera; Dalton et al. 2006) at the ESO VISTA telescope. The first release of VMC data includes images of the 30 Doradus Region (a "tile" of about 10 × 13).

We retrieved the 30 Doradus images through the ESO Data Archive Facility. For our analysis, we used the "pawprint" images (six of which form a "tile"). The released VMC data have been processed through the dedicated pipeline of the VISTA Data Flow System (Irwin et al. 2004; Emerson et al. 2004). In addition to the images, the VMC survey contains source catalogs, but the aperture photometry they provide includes only a few of our sources and may be unreliable in regions of bright, variable nebulosity.

Accordingly, we performed point-spread-function (PSF) photometry of our sources in the VMC images, using DAOPHOT (Stetson 1987) software layered in IRAF.⁷ Stars were detected at a minimum 5σ level above the background, with special care to isolate faint companions to the targets.

The VISTA photometric system is tied to the Two Micron All Sky Survey (2MASS; Skrutskie et al. 2006). To derive the photometric cross-calibrations, we have selected 440 2MASS catalog stars with the lowest photometric uncertainties (<10% in all three bands) and located in areas of faint nebulosity, matching them to our VISTA photometric catalog. The transformation equations obtained are

$$\begin{equation*} J_{\rm 2MASS} - J_{\rm VISTA} = 0.06 \times (J_{\rm VISTA} - K_{\rm VISTA}) \end{equation*}$$

and

$$\begin{equation*} K_{\rm 2MASS} - K_{\rm VISTA} = -0.003 \times (J_{\rm VISTA} - K_{\rm VISTA}). \end{equation*}$$

The coefficients of these equations are very similar to those found by Cioni et al. (2011). Our transformations are valid for the color range of the 2MASS sources used to calculate them, i.e., −0.1 < (J − K_s) < 4.2. Table 1 shows that the 30 Dor sources with both J and K_s magnitudes are in the valid range. For sources with only limits on J, the uncertainties in the K_s transformation are also very small, due to the small value of the color-term coefficient. (The Y magnitudes in Table 1 remain in the VISTA photometric system and are not used in our analysis below.)

The Infrared Survey Facility (IRSF) Magellanic Clouds Point Source Survey (Kato et al. 2007) covered a 40 deg² area in the LMC in J (1.25 μm), H (1.63 μm), and K_s (2.14 μm) bands, with the SIRIUS camera at the South African Astronomical Observatory (SAAO). The average seeing throughout the survey was 12 (FWHM).

Figures 1 and 2 display Spitzer and VISTA images of the field investigated at the same scale, with the 10 brightest Spitzer sources and some other relevant objects identified. It is seen that all of them lie within or near the Tarantula filaments to the west and northeast of R136, except for the very luminous and previously little known source S1 to the northwest. (The hemispheric structure of two-stage starbursts, as discussed by Walborn 2002, is well seen.) The photometric data taken from the literature, except for VISTA/VMC as discussed above, for these and other IR sources of interest are listed in Table 1. The alternate designations are from Hyland et al. (1992) or Rubio et al. (1998), while the corresponding SAGE and IRSF identification numbers are given in Table 2. Figure 3 shows that the 30 Doradus sources are among the most luminous young stellar objects (YSO's) in the LMC, as listed in the caption references. The 30 Dor sources are discussed individually in detail in the following section, emphasizing their remarkable structures and diversity.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We were surprised to find one of the most luminous Spitzer sources embedded in extensive diffuse emission at the northwestern extreme of the field, evidently unrelated to R136. S1 appears to be associated with a larger CO feature cataloged as 30Dor-06 by Johansson et al. (1998), and with a H ii region listed as No. 889 by Kastner et al. (2008). In the IR images it appears as a somewhat diffuse double source, with the fainter barely resolved component toward the SE. Fortunately, a recent HST/WFC3 F775W image (GO 12499, PI D.J. Lennon) reveals what lurks within (Figure 5): the object appears as a dusty H ii knot, here named "The Skull Nebula," containing a compact cluster dominated by four I ∼ 19–21 stars in an EW line (the brighter IR "component") and two more toward the SE (the fainter IR "component"; E. Sabbi 2012, private communication). Two more comparable stars within the outer nebulosity to the N and S may be additional members or field objects.

Zoom In Zoom Out Reset image size

With reference to Figures 1 and 2, we propose an origin of star formation at this location triggered by adjacent SN events. To the NE of S1 are the three luminous, "isolated" WN stars R144, R146, and R147 (Feast et al. 1960), apparently associated with the large kinematical (Chu & Kennicutt 1994) and X-ray (Townsley et al. 2006) shell at the north of 30 Dor. This object could have been produced by the explosion of another former WR star. To the SW of S1 lies the 20 Myr old cluster Hodge 301 that has produced 40 SN (Grebel & Chu 2000) and is surrounded by another giant shell. Numerous dark globules can also be seen along the interface between the two shells in optical images, suggesting that further star formation is likely to occur in these compressed filaments–a novel feature of 30 Doradus.

This relatively fainter source is IRSW-11 of Rubio et al. (1998). (It is also NIC07a of Brandner et al. 2001, in which the cross-identification with Rubio et al. was missed.) It is located at the northern end of a diffuse structure that is bright in the Spitzer image (Figure 1) but dark in JHK (Figure 2). As noted by Walborn et al. (1999), with appropriate stretch and scale the object appears multiple in the NICMOS data, and analysis of its PSF indicates that it is extended, with a Gaussian sigma of 1.29 pix versus 1.04 pix for a point source in K. Here we analyze the S2 photometry as if from a single source, with the caveat that it may actually be a tight multiple system.

This luminous source apparently within a Tarantula filament is at the apex of an inverted triangular asterism with the "isolated" WN star R135 (Feast et al. 1960) and another source brighter at shorter wavelengths that we designate S3K. The reversal of their relative magnitudes between Figures 1 and 2 is striking; see also Figure 4. We suggest that the massive wind of the WN star may have triggered the formation of S3 in the adjacent nebular filament. As further analyzed below, S3K may be a star rather than a YSO.

S4 is one of the most luminous sources in 30 Dor at all NIR wavelengths, also IRSW-30 of Rubio et al. (1998), NIC03a of Brandner et al. (2001), and P3 of Hyland et al. (1992). As extensively discussed based on HST optical and IR images by Walborn et al. (1999, 2002), it is located in the head of a large, bright-rimmed dust pillar oriented toward R136. They measured magnitude 20.3 in V and 19.3 in I; here we report 11.3 in K and 5.8 at 8 μm (Table 1). This is a prime object for astrophysical investigation, including spectroscopy from visual through IR. IRSW-26 is a companion to the SW, but it is more than 3 mag fainter in K and evidently not significant in the Spitzer bands (Figure 4), so we analyze S4 as a single object.

S5 or Rubio et al. (1998) IRSW-127 is located within a prominent boomerang-(or stapler-)shaped dust cloud silhouetted against the bright stellar and nebular field just north of R136 in optical images. Again, this cloud is bright in the Spitzer image. It is likely located in the immediate foreground of R136. The source IRSW-133 immediately to the south is brighter at J, in which S5 is invisible (Figure 4), but the latter is brighter at K and likely dominates in the Spitzer images. These sources were also measured at H and K_s by Campbell et al. (2010, their Section 5.4), with results similar to ours, although the three available K magnitudes are in relatively poor agreement, perhaps due to source confusion or variability. C. Evans (2012, private communication) has pointed out that IRSW-133, which Campbell et al. suggested to be a massive star, is VFTS 476 in the current VLT-FLAMES Tarantula Survey (Evans et al. 2011); indeed, it has been classified as O((n)) from the presence of He ii absorption lines in the VFTS data (N. R. Walborn et al., in preparation), with heavy nebular emission-line contamination preventing a more precise description. Seale et al. (2009) report Spitzer spectroscopy of S5 consistent with a YSO nature.

S6 is the very luminous P2 of Hyland et al. (1992) and NIC16a of Brandner et al. (2001). It is a point source in HST optical and IR images, as shown by Walborn et al. (2002) who measured an I mag of 20.9 and placed a V limit of 23. Beware of potential confusion with a Long Period Variable (LPV) about 2'' NW, which oscillates around 17th mag in I except during dust-formation episodes when it becomes fainter than S6 (OGLE-LMC-LPV-76683; Soszynski et al. 2009). The LPV is always much fainter than S6 at longer wavelengths (Figure 4). To within the HST resolution, S6 appears to be an instance of monolithic, "isolated" massive star formation.

This intense "double source" is Rubio et al. (1998) IRSN-122/126, the latter dominating Hyland et al. (1992) P1; it is also Brandner et al. (2001) NIC12b/d. As extensively discussed by Walborn et al. (1999, 2002), they are embedded within a massive dust pillar oriented toward R136 and parallel to the adjacent Knot 1 pillar from which a young stellar system has just emerged. Like S4 and S6, these luminous sources can be measured at I (20.4 for S7A and 21.2 for S7B), with V limits of 22 for both; hence they are also outstanding candidates for future spectroscopic investigation over a wide wavelength range.

In striking contrast to S6, this bright source (P4 of Hyland et al. 1992; IRSN-137 of Rubio et al. 1998; NIC15b of Brandner et al. 2001, but mislabeled there as IRSN-134) is surrounded by a rich cluster of much fainter ones as shown by Walborn et al. (2002), indicating qualitatively different modes of massive star formation for these two sources. This circumstance creates difficulties with the photometry of S8, because the VISTA and IRSF images from which the YJHK magnitudes listed in Table 1 are derived resolve it from the cluster, whereas the Spitzer images do not. Rubio et al. distinguish a relatively bright "tail" of the blended cluster toward the SW as the separate source IRSN-134, but only the NICMOS images reveal the actual structure of the cluster. We have noted that the cluster is surrounded by a thick annulus of emission in the Spitzer/MIPS 24 μm image, which warrants further study.

This remarkable source is a constituent of the intricate YSO/ZAMS phenomena associated with the Knot 2 complex of Walborn & Blades (1997). With reference to Figure 3 and Section 3.2 of Walborn et al. (1999): S9 is the brighter of two extended sources aligned across and approximately equidistant from the central, young optical stellar system (Walborn et al. 2002). The major axis of S9 has an extent of 1'', or 0.25 pc ∼50,000 AU. Both extended sources display (opposite) color gradients perpendicular to their radius vectors from the central system, suggesting (counterclockwise) rotation. No evidence of the presumed jets impacting the surrounding dust clouds to produce them has been found to date. The spectral type of the brightest stellar component in Knot 2, VFTS 621 or Parker (1993) 1429, has been found to be O2 V((f*))z from the VFTS data (N. R. Walborn et al., in preparation), thus a very massive young object.

It is surprising that S9 would be one of the brightest Spitzer sources in 30 Dor, but then, its nature is entirely unknown. Seale et al. (2009) report Spitzer spectroscopy of this source consistent with a YSO, although the slit width is at least ∼4'' and S9 is not dominant at 8 μm (Figure 4). Again, the NICMOS images show that it is nonstellar, so that a very large circumstellar envelope would be implied in that case, but the point-symmetric counterpart on the opposite side of the central stellar system suggests a more exotic interpretation. Moreover, we have found that S9 and Knot 2 are apparently associated with the brightest 30 Dor source in all of the Herschel 100–500 μm bands, from the HERschel Inventory of The Agents of Galaxy Evolution (HERITAGE; Meixner et al. 2010). Finally, there are also two water masers associated with this complex (Ellingsen et al. 2010), one of which is very near S9, while the other is roughly opposite across the central stellar system, approximately mimicking the alignment of the IR "mystery spots" at a lower P.A. (Figure 4; Walborn et al. 1999). Clearly further detailed, combined analysis of the multiple rich datasets now available for this region will prove rewarding.

At first glance S10 appears to be one of the brightest sources in the Spitzer images, but on closer inspection it is somewhat extended, and indeed, the higher resolution VISTA images reveal it to be three much fainter sources that we designate A, B, C. Moreover, they may comprise part of a larger structure associated with the luminous WN star VFTS 682, newly discovered by that survey (Bestenlehner et al. 2011). The WN star appears to have excavated a cavity, around the periphery of which there is an array of IR sources, including the S10 system. Another bright K source that we designate S10K forms a triangular asterism with VFTS 682, and S10 at the apex (Figure 4). Rubio et al. (1998) IRSN-169 and its fainter NE companion IRSN-170 lie near the opposite side of the cavity (midway between VFTS 682 and S9 in Figures 1 and 2); the blended Spitzer data for the two sources are ascribed to "S11" in Tables 1 and 2 and Figure 3. There is also a water maser exactly coincident with IRSN-169 (Ellingsen et al. 2010). Thus, along with S3/R135, this structure may represent another case of a massive WN wind triggering further star formation.

We fitted YSO models to the spectral energy distributions (SEDs) of selected, unresolved sources. We used the Robitaille et al. (2006) grid of 20,000 2D YSO radiative transfer models for masses from 0.1 through 50 M_☉. For each model in the grid, SEDs are calculated at 10 viewing angles, resulting in 200,000 pre-computed SEDs that can be compared to the observations using the Robitaille et al. (2007) fitting tool. Since these models are computed for single objects, we selected those sources from our sample that appear single in the near-IR and mid-IR images: S2, S3, S3K, S4, S6, S7A, S7B, and S10K. For the fitting, we used VISTA J and K_s, IRSF H, and IRAC 3.6–8.0 μm fluxes as available (Table 1), as well as HST/WFPC2 F814W (∼I) and F555W (∼V) fluxes or upper limits for S4, S6, S7A, and S7B (these magnitudes are cited in the individual source discussions above). The fits for these sources are shown in Figure 6. All of the YSO fits indicate Class I objects.

Zoom In Zoom Out Reset image size

The formal fit for S3K as a star is superior to that as a YSO, so both fits are shown for this source. This result is related to the fact that the Spitzer fluxes are smaller than those at shorter wavelengths, unlike all the other cases. The stellar fit yields T_eff = 4750 K and A_V = 5.85. While the apparent close association of S3K with S3, R135, and nebular structure shown in the figures may be a chance alignment, there is also the interesting possibility that it could be a post-YSO, PMS object.

The physical parameters estimated from the fitting are listed in Table 3: best and average stellar luminosity, effective temperature, and mass; envelope and disk masses; and interstellar extinction. Since the Robitaille et al. (2006) models assume a Galactic gas-to-dust ratio, we scaled the results for the envelope and disk masses to the 30 Dor ratio by multiplying them by a factor of 2.1 (Dobashi et al. 2008). The averages are over all YSO model fits with χ² per data point between $\chi ^{2}_{{\rm best}}$ and $\chi ^{2}_{{\rm best}}$ + 2, as also listed in the table. The uncertainties of the averages quoted (in parentheses) in Table 3 represent standard deviations of the means. For sources with very high χ²/pt and only one fitted model (S6 and S7B), there are of course no averages and the derived parameters are highly uncertain. For other fits with χ²/pt > 5, the physical parameters may not reflect well the actual properties of the sources. Source S3K is also listed in Table 3 as a YSO and only its luminosity is provided; however, as already noted this source is more likely to be a star with the parameters given above. The $\chi ^{2}_{\min }$/pt for the stellar photosphere fit is ∼6, far lower than that for the YSO fit, ∼109 as listed in the table.

Table 3. The SED Fitting Results: Physical Parameters

Source	L⋆ (104 L☉)			T⋆ (104 K)			M⋆ (M☉)			Menv (102 M☉)			Mdisk (M☉)			AV			$\chi _{\rm min}^{2}$/ptb	nfitsc
Name	Best	Ave		Best	Ave		Best	Ave		Best	Ave		Best	Ave		Best	Ave
S2	3.4	7.7(0.7)a		1.2	2.1(0.1)		20.1	23.0(0.7)		2.2	6.0(0.7)		0	0.15(0.05)		6.5	5.4(2.7)		0.4	109
S3	8.2	10.5(2.5)		3.8	3.4(0.6)		25.2	26.7(3.2)		13.3	14.4(7.7)		0	0		10.0	9.7(0.4)		14.7	3
S3K	1.7	...		...	...		...	...		...	...		...	...		...	...		109.3	1
S4	10.7	8.2(2.5)		3.9	3.8(0.1)		27.4	24.5(2.9)		10.9	6.4(4.5)		0.19	0.10(0.09)		1.8	1.2(0.6)		8.7	2
S6	3.7	...		3.4	...		18.4	...		12.0	...		0.13	...		3.0	...		37.9	1
S7A	3.0	2.7(0.4)		2.0	2.4(0.4)		15.2	16.4(0.6)		1.6	3.1(1.2)		0.04	0.13(0.09)		3.2	2.8(1.7)		1.4	9
S7B	7.3	...		1.7	...		23.4	...		71.5	...		0	...		0.1	...		41.1	1
S10K	0.7	2.4(1.8)		2.5	2.0(0.1)		9.1	10.9(1.5)		0.5	2.8(1.9)		0.28	0.11(0.02)		4.4	4.5(2.4)		0.5	24

Notes. ^aThe uncertainties represent standard deviations of the means. ^b$\chi ^{2}_{\rm min}$/pt is a χ² per data point for the best-fit model. ^cn_fits is the number of fits with χ²/pt between $\chi ^{2}_{\rm min}$/pt and $\chi ^{2}_{\rm min}$/pt+2.

Download table as:  ASCIITypeset image

The models assume single sources; at the distance of the LMC, even the sources that appear unresolved in our data could in fact be multiple or even protoclusters (see also Carlson et al. 2011). Our derived parameters are valid only on the assumption that a single source dominates the luminosity at all wavelengths.