SPITZER VIEW OF YOUNG MASSIVE STARS IN THE LARGE MAGELLANIC CLOUD H ii COMPLEX N 44

Our Spitzer observations of N 44 were made with the InfraRed Array Camera (IRAC; Fazio et al. 2004) on 2005 March 27 and the Multiband Imaging Photometer for Spitzer (MIPS; Rieke et al. 2004) on 2005 April 7. The IRAC observations were obtained in the 3.6, 4.5, 5.8, and 8.0 μm bands, using the mapping mode to cover a ∼30' × 30' area in each band. The pixel size was 12 pixel⁻¹, and the common area covered by all four bands was ∼30' × 23'. At each pointing, exposures were made with a five-point cyclic dithering pattern in the 30 s high-dynamic range mode. The total integration time at each pointing was ∼150 s. The mosaicked maps were processed with the Spitzer Science Center (SSC) pipeline (ver. S11.4.0) and provided to us as part of the Post Basic Calibrated Data (PBCD) products.

The MIPS observations were made in the 24, 70, and 160 μm bands using the scan map mode at the medium scan rate. The mapping consists of 16 05 scan legs with a cross-scan step of 148'' to cover a region of 20' × 30' in all three MIPS bands. The MIPS DAT version 3.00 (Gordon et al. 2005) was used for the basic processing and final mosaicking of the individual images. In addition, the 24 μm image has been corrected for a readout offset and divided by a scan-mirror-independent flat field, and the 70 and 160 μm images have been corrected for a pixel-dependent background using a low-order polynomial fit to the source-free regions. The final mosaicked maps have pixel scales of 125, 985, and 160 pixel⁻¹, and exposure times of roughly 160, 80, and 16 s at 24, 70, and 160 μm, respectively.

The common area covered by all IRAC and MIPS bands is 20' × 23'. To achieve a more complete view of the molecular cloud to the north of N 44 superbubble, we have used the Spitzer survey of the LMC (SAGE; Meixner et al. 2006) to extend our IRAC 3.6 and 5.8 μm data in the declination direction. The final IRAC and MIPS field we have analyzed is 20' × 25'.

To carry out aperture photometry for point sources in the IRAC images, we use the IRAF package apphot. First, the sources were identified with the automated source finding routine daofind, using parameters optimized to find the majority of pointlike sources while minimizing the inclusion of peaks of extended dust emission. In regions containing multiple sources superposed on extended emission, daofind does not always identify the same sources at the same locations in the four IRAC bands. Therefore, we identified sources in each of the four IRAC bands and merged the four source lists into a master source list using 12 (= 1 pixel) as the criterion for coincidence. The resulting master list is then used for photometric measurements in all four bands. The photometric measurements were made with a source aperture of 36 (3 pixels) radius and an annular background aperture at radii of 36–84 (3–7 pixels). Finally, an aperture correction was applied, and the fluxes were converted into magnitudes using the correction factors and zero-magnitude fluxes provided in the IRAC Data Handbook and listed in Table  1.

The photometric measurements from long- and short-exposure IRAC observations were averaged with weights proportional to the inverse square of their errors. For sources that were saturated in the long exposures, their measurements from the short exposures were adopted. The results were used to produce a photometric catalog of 17,002 IRAC sources in the 20' × 25' field of N 44. This catalog is presented as an ASCII table and an example is shown in Table  2.

MIPS images have lower angular resolution. Source identification using daofind is feasible only for the 24 μm images. In the 70 and 160 μm images, point sources cannot be easily resolved from one another or from a bright diffuse background; therefore, the few apparent point sources were identified by visual inspection. The photometric measurements were made with parameters appropriate for the point spread functions (PSFs), as recommended in the MIPS Data Handbook and given in Table 1. Note that the aperture corrections adopted for 70 and 160 μm measurements are those for sources of temperatures 15 K and 10 K, respectively. For sources of higher temperatures, such as 1000–3000 K, the adopted aperture corrections will result in fluxes 8% too high in 70 μm and 2% too high in 160 μm, because the source emission peaks at shorter wavelengths; however, these errors do not significantly affect the analysis and conclusions of our study of YSOs. In the 160 μm image, only one point source can be clearly identified; therefore, the 160 μm photometry will be discussed in the text but not included in the photometric catalog. The catalogs of the MIPS 24 and 70 μm bands are merged with the IRAC catalog and included in Table  2.

The Spitzer images of N 44 in the 3.6, 8.0, and 24 μm bands are shown in Figure 2. The 3.6 μm image is dominated by stellar emission, the 8.0 μm image shows the polycyclic aromatic hydrocarbon (PAH) emission, while the 24 μm image is dominated by dust continuum emission (Li & Draine 2001, 2002; Draine & Li 2007). To better illustrate the relative distribution of emission in the different bands, we have produced a color composite with 3.6, 8.0, and 24 μm images mapped in blue, green, and red, respectively. In this color composite, shown in Figure 2(d), dust emission appears red and diffuse, stars appear as blue point sources, red supergiants appear yellow, and dust-shrouded YSOs and AGB stars appear red.

Zoom In Zoom Out Reset image size

We obtained near-IR images in the J and K_s bands with the Infrared Side Port Imager (ISPI) on the Blanco 4 m telescope at Cerro Tololo Inter-American Observatory (CTIO) on 2005 November 14–15. The images were obtained with the 2K × 2K HgCdTe HAWAII-2 array, which had a pixel scale of 03 pixel⁻¹ and a field of view (FOV) of 1025 × 1025. Six fields were observed to map N 44. Each field was observed with ten 30 s exposures in the J band and twenty 30 s exposures in the K_s band (each of the latter was co-added from two 15 s frames to avoid background saturation). The observations were dithered to aid in the removal of transients and chip defects. Owing to the diffuse emission in N 44, we obtained sky frames at ∼20' south of N 44 to aid in sky subtractions. The sky observations were made before and after each set of ten on-source exposures. All images were processed using the IRAF package cirred for dark and sky subtraction and flat fielding. The astrometry of individual processed images was solved with the routine imwcs in the package wcstools. The astrometrically calibrated images are then co-added to produce a total exposure map for each filter. The flux calibration was carried out using 2MASS photometry of isolated sources.

We obtained the SDSS u and Johnson–Cousins BVI broadband images of N 44 with the MOSAIC II CCD Imager on the CTIO Blanco 4 m telescope on 2006 February 2. The MOSAIC Imager consists of eight 2K × 4K SITe CCDs. The CCDs have a pixel scale of 027 pixel⁻¹, yielding a total FOV of 36' × 36'. The entire N 44 was imaged in a single field. All uBVI images have been processed with the standard reduction procedure: bias and dark were subtracted, flat-fielding was applied, and multiple frames in each filter were combined to remove cosmic rays and improve the S/N. The astrometry for the processed images was performed by referencing to stars in the USNO B1.0 catalog. The flux calibration was carried out using photometric measurements of isolated sources in Chen (2007) and Zaritsky et al. (2004).

We have searched the HST archive for Wide Field Planetary Camera 2 (WFPC2) images in the field of N 44. The available observations are listed in Table 3, in which the coordinates, program ID, PI, filter, and exposure time are given. Most of the observations contain multiple exposures for the same pointing and filter, and these images are combined using the IRAF routine crrej to remove cosmic rays and produce a total exposure map. The astrometry was refined for each resultant image by referencing to stars in the USNO B1.0 catalog.

Seven fields have observations, but not all are useful, for example, the wide U band (F300W) images have very low S/N. The most useful images are those taken with the Hα (F656N) or Strömgren y (F547M) filter. The former shows ionized gas, and the latter shows stars at high resolution.

To construct SEDs for the sources in our Spitzer photometric catalog of N 44, we have expanded the catalog by adding UBV photometric measurements obtained from CCD images taken with the CTIO 0.9 m telescope (Chen 2007). In regions not covered by these 0.9 m UBV observations, we use the UBVI photometry from the Magellanic Cloud Photometric Survey (MCPS, Zaritsky et al. 2004). We have also added near-IR data from the Point Source Catalog of the Two Micron All Sky Survey (2MASS, Skrutskie et al. 2006).

When merging the data sets, we allow a 1'' error margin for matching Spitzer sources with optical or near-IR sources. The final catalog lists each source's right ascension, declination, and magnitudes in the order of increasing wavelengths, i.e., U, B, V, I, J, H, K_s, 3.6, 4.5, 5.8, 8.0, 24, and 70 μm. The entire catalog is presented as an ASCII table, and an example is shown in Table 2. These magnitudes can be converted to flux densities using the corresponding zero-magnitude flux listed in Tables 1 and 4 and then used to construct SEDs for individual sources.

Finally, we have used Hα images of N 44 from the Magellanic Cloud Emission Line Survey (MCELS; Smith & The MCELS Team 1999) to examine the large-scale distribution of dense ionized gas and to compare with images at other wavelengths. As the angular resolution of this survey is ∼2'', we have used additional Hα images taken with the CTIO 0.9 m telescope by M. A. Guerrero (Nazé et al. 2002). These images, with a pixel size of 04 pixel⁻¹, are used to show the immediate environments of YSOs.

YSOs have SEDs that differ from stars because they are shrouded in dust, which absorbs the stellar radiation and irradiates at IR wavelengths; therefore, it is possible to identify YSOs from their IR excesses. It is, however, difficult to decipher the IR excess of massive YSOs, because the distribution of their circumstellar dust is not well known. As their formation mechanism is still uncertain (e.g., Stahler et al. 2000; Zinnecker & Yorke 2007), it is not even known whether massive YSOs ubiquitously possess accretion disks (e.g., Cesaroni et al. 2007).

We will use the commonly accepted physical structure and evolution of low-mass YSOs as a starting point to decipher the SEDs of higher mass YSOs. Low-mass YSOs are believed to be initially surrounded by a small accretion disk and a large infalling envelope with bipolar cavities, and as they evolve, the envelope and disk dissipate. Different evolutionary stages result in different SEDs that have been used to classify low-mass YSOs, i.e., the Class I/II/III system (Lada 1987). In this classification system, a Class I YSO has a compact accretion disk and a large infalling envelope with bipolar cavities; its SED is dominated by emission from the envelope and rises longward of 2 μm. A Class II YSO has dispersed most of its envelope and is surrounded by a flared disk; its SED, dominated by emissions from the central source and disk, is flat or falls longward of 2 μm. A Class III YSO has cleared most of the disk so its SED shows stellar photospheric emission with little or no excess in near-IR.

These geometries of dust disk and envelope applicable to low-mass YSOs have been adopted in radiative transfer models of high-mass YSOs by a number of investigators (e.g., Whitney et al. 2004a; Robitaille et al. 2006). In their models, at an early evolutionary stage, a YSO has a small disk and a large envelope with narrow bipolar conical cavities; its SED is dominated by the envelope emission and shows a generally rising trend from the shortest detectable wavelength to beyond 24 μm. At an intermediate evolutionary stage, the opening angle of the bipolar cavity increases and the star and disk may be exposed; the YSO's SED thus shows double peaks: one below 1 μm (stellar emission) with its intensity dependent on the viewing angle, and the other rising longward of 1 μm but turning flat or falling shortward of 10–20 μm. At a late evolutionary stage when most of the envelope and disk have been dispersed, the SED shows a bright stellar emission with a modest mid-IR excess. These models provide useful links between circumstellar dust structures and SEDs; thus, we will use the general trends of the SEDs to identify YSO candidates in N 44.

Owing to their excess IR emission, YSOs are positioned in redder parts of the color–color and color–magnitude diagrams (CMDs) than normal stars. However, background galaxies and asymptotic giant branch (AGB) stars can also be red sources, and these contaminants exist in non-negligible numbers. To separate YSOs from these contaminants, we have examined several color–color diagrams and CMDs with solely IRAC bands as well as combinations of IRAC with JHK_s or MIPS bands (Gruendl & Chu 2008). We find that for many sources the 2MASS catalog is too shallow to detect their counterparts in JHK_s, and the MIPS observations cannot resolve them from nearby sources or bright diffuse background. To include most of the sources, we decide to concentrate on diagnostic diagrams using only the IRAC bands. The IRAC [3.6]−[4.5] versus [4.5]−[8.0] color–color diagram has been suggested by Simon et al. (2007) and the [8.0] versus [4.5]−[8.0] CMD has been suggested by Harvey et al. (2006) to offer the best separation of YSOs from contaminants. We have adopted the latter approach for the initial selection of massive YSOs, because galaxies and evolved stars have different distributions in brightness and can be effectively excluded using simple criteria.

Figure 3 displays the [8.0] versus ([4.5]−[8.0]) CMD of all sources detected in N 44. The sources in the prominent vertical branch centered at ([4.5] − [8.0])∼0.0 are mostly main-sequence, giant, and supergiant stars. The contaminating background galaxies are concentrated in the lower part of the CMD, and the AGB and evolved stars are distributed mostly in the upper part of the CMD. Below, we discuss the criteria used to exclude these contaminating sources from the high-mass YSO candidates.

Zoom In Zoom Out Reset image size

3.2.1. Excluding Normal and AGB Stars

3.2.2. Avoiding Background Galaxies

3.2.3. Massive YSO Candidates

Background galaxies can be identified from their morphologies if they are resolved. To diagnose unresolved background galaxies, we have to resort to SEDs. Galaxies co-located with YSO candidates in the [8.0] versus ([4.5]−[8.0]) CMD are abundant in gas and dust, such as late-type galaxies or active galactic nuclei (AGN). The SEDs of late-type galaxies are characterized by two broad humps, one from stellar emission over optical and near-IR wavelength range and the other from dust emission over mid- to far-IR range. The observed SED of an AGN depends on the viewing angle with respect to its dust torus; it can be flat from optical to far-IR or obscured in optical, showing only mid- to far-IR emission (Franceschini et al. 2005; Hatziminaoglou et al. 2005; Rowan-Robinson et al. 2005). In the latter case, the SED can be falling or rising at 24 μm. We further use the interstellar environment as a secondary diagnostic for background galaxies, since galaxies are statistically less likely to occupy preferred positions in prominent dust filaments. Based on these considerations, two of our CMD-selected YSO candidates, sources 052042.0−674307.7 and 052106.8−675715.9, are reclassified as background galaxies.

The known AGB stars in N 44 have SEDs peaking between 1 and 8 μm, as shown in Figure 4. These SEDs are similar to those of Galactic AGB stars, corresponding to dust temperatures of ∼400K–1000 K (Rowan-Robinson et al. 1986). These known AGB stars in N 44 are bluer than our selection criteria for YSOs; however, the more obscured AGBs may have lower dust temperatures, exhibit redder colors, and occupy the same regions as the YSO candidates in the CMD (e.g., Buchanan et al. 2006). We identify such obscured AGB or evolved stars based on their SEDs, whose shapes are similar to those shown in Figure 4 but peaking at longer wavelengths. We have also used the interstellar environment as a secondary criterion to identify AGB and evolved stars, since evolved stars are not expected to be located at preferred positions in diffuse dust emission. From these considerations, two of our YSO candidates, sources 052221.0 − 680515.3 and 052351.1 − 675326.6, are reclassified as AGB/evolved stars.

Zoom In Zoom Out Reset image size

Warm interstellar dust may show SEDs similar to those of circumstellar dust in YSOs. As the angular resolution of IRAC images is ∼2'', corresponding to 0.5 pc for a LMC distance of 50 kpc, some small dust clumps may be identified as point sources and included in the YSO candidate list. Owing to our conservative compilation of master source list for IRAC photometry (Section 2.1), a star projected near a dust clump may also make its way into our YSO candidate list. To identify these two types of YSO imposters, we use the ISPI JK_s images that have higher angular resolution. In the ISPI K_s images, stars or YSOs appear unresolved, while dust clumps may appear as extended emission. The J − K_s color further differentiates between stars and YSOs, as stars are brighter in J and YSOs brighter in K_s. Aided by the ISPI images, we find that 23 of our 99 YSO candidates are interstellar dust clumps, and another 12 are stars projected near dust clumps.

The results of our examination of 99 YSO candidates are given in Table 5, which lists source name, ranking of the brightness at 8 μm, magnitudes in the U, B, V, I, J, H, K_s, 3.6, 4.5, 5.8, 8.0, 24, and 70 μm bands, source classification, and remarks. Note that some of the YSOs appear as single sources in IRAC images but are resolved into multiple sources in ISPI and MOSAIC images; for these sources, the photometric measurements in Table 5 were made for the dominant YSO sources and these entries are remarked in Column "Flag."

After excluding background galaxies, evolved/AGB stars, and dust clump imposters, 60 YSO candidates remain. As these are most likely bona fide YSOs, we will simply call them YSOs in the rest of the paper.

YSOs in N 44 have been identified using other methodologies by, e.g., Gruendl & Chu (2008) and Whitney et al. (2008). Gruendl & Chu (2008) used the same Spitzer data of N 44 and similar selection criteria, but identified only 49 YSOs. This is understandable because their automated search for point sources was unable to find faint sources near a bright neighbor or over a bright background. Such faint sources can be found and measured more easily with human intervention, as we did in this paper for the small area of N 44 only.

The comparison with the YSO sample of Whitney et al. (2008) is not as straightforward because they used a very complex set of criteria. Within the area of N 44, they found only 19 YSO candidates, of which one was identified as a planetary nebula (PN). In Figure 5(a), we mark the locations of these YSO candidates and our YSOs in the 8 μm image of N 44, and in Figure 5(b), we compare our sources with Whitney et al. (2008) YSO candidates in a [8.0] versus ([4.5]−[8.0]) CMD. The differences can be summarized as the follows.

• 1.  
  Within the YSO wedge bounded by [4.5]−[8.0] ⩾ 2.0 and [8.0] ⩾14.0−([4.5]−[8.0]), we find 60 YSOs, while Whitney et al. (2008) find only 13 YSO candidates, a subset of our sample. To investigate why the majority of the YSOs in N 44, including bright sources with [8.0] < 7.0, are missed in Whitney et al. (2008), we further compare our sources with their original catalog, i.e., the SAGE point source catalog (kindly provided by Remy Indebetouw and Marta Sewilo). We find that the missed YSOs do not result from different selection criteria, but from defining point sources and applying signal-to-noise (S/N) thresholds when making final lists in these two studies. For a survey of the entire LMC, the SAGE catalog used stringent parameters for the automated search for point sources and Whitney et al. (2008) applied a high-S/N threshold. The former discards any irregular point sources compared to IRAC PSFs. The latter tends to exclude sources near bright neighbors or over bright background, as shown in Figure 5(a), since varying backgrounds make photometry more difficult, and the SAGE pipeline increases the photometric uncertainties in such areas, reporting conservative S/N ratios.
• 2.  
  Six YSO candidates of the Whitney et al. (2008) sample were rejected by our stringent color–magnitude criteria for YSO selection. The SEDs of these six sources are shown in Figure 5(c). With the addition of optical fluxes in the SEDs, it can be more clearly seen that at least YSO candidate W546 is likely an evolved star with circumstellar dust and that W530 and W574 may be background galaxies.

Zoom In Zoom Out Reset image size

We conclude that our sample of massive YSOs in N 44 is the most complete among the three compared.

The observed SED of a YSO can be compared with model SEDs, and the best-fit models selected by the χ² statistics can be used to infer the probable ranges of physical parameters for the YSO. This can be carried out with the fitting tool Online SED Model Fitter⁶ (Robitaille et al. 2007). The database of this fitting tool includes model SEDs pre-calculated for 20,000 YSO models each with 10 different viewing angles. A typical YSO model assumes a stellar core, a flared accretion disk, and a rotationally flattened infalling envelope with bipolar conical cavities. To fully describe this complex structure, each model is defined by 14 parameters (Robitaille et al. 2006). The stellar parameters include the star's mass (M_⋆), radius (R_⋆), and temperature (T_⋆); these are sets of parameters at different ages (t_⋆) along the pre-main-sequence stellar evolutionary tracks modeled by Bernasconi & Maeder (1996) and Siess et al. (2000). For a given set of M_⋆, R_⋆, and T_⋆, the stellar radiation determined from stellar atmosphere models of Kurucz (1993) or Brott & Hauschildt (2005) is adopted and fed to the radiative transfer code to produce the YSO's model SED. The disk parameters include the disk's accretion rate ($\dot{M}_{\rm disk}$), mass (M_disk), inner radius (R^min_disk), outer radius (R^max_disk), scale height factor (z_factor), and flaring angle (β). The envelope parameters include envelope accretion rate ($\dot{M}_{\rm env}$), envelope outer radius (R^max_env), cavity density (ρ_cavity), cavity opening angle (θ_cavity), and the ambient density (ρ_ambient) at which the density of the envelope reaches the lowest value as the density of the ambient ISM. Outside of the envelope is the foreground extinction (A_V). The 20,000 YSO models are generated to sample each of the disk and envelope parameters and the ambient density for a variety of pre-main-sequence stellar model (Robitaille et al. 2006).

We have used the above SED fitting tool to analyze 36 YSOs in our sample that appear single or are clearly the dominant source within the IRAC PSF. The input parameters of the SED model fitter include the fluxes of a YSO and uncertainties of the fluxes. The uncertainty of a flux has two origins: the measurement itself and the absolute flux calibration. The fluxes and their measurement errors are given in Table 5. The calibration errors are 5% in U, 3% in B, V, 10% in I, J, K_s, 3.6, 4.5, 5.8, 8.0, and 24 μm, and 20% in 70 μm (Chen 2007; Gruendl & Chu 2008; IRAC Data Handbook; MIPS Data Handbook). The total uncertainty of a flux is thus the quadratic sum of the measurement error and the calibration error. The resulting best-fit and acceptable SED models are shown in Figure 7, in which the YSOs are arranged in the order of Types I, I/II, II, II/III, and III (from our empirical classification in Section 5), and within each type in order of increasing [8.0] magnitude. In each panel, data points are plotted in filled circles and upper limits are plotted in filled triangles; error bars are plotted but they are usually smaller than the plot symbols. The best-fit model, with minimum χ², is shown in solid black line, and the radiation from the stellar core reddened by the best-fit A_V is shown in dashed black line. For most YSOs, their SEDs can be fitted similarly well by a range of models, and these "acceptable" models are plotted in gray lines (see Robitaille et  al. 2007 for definitions of χ² and acceptable fits).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The results of the SED model fits are given in Table 7, where the YSOs are listed in the same order as in Figure 7. The source name, [8.0] magnitude, and type from our empirical classification are listed on the left of the table, and physical parameters of the best-fit models are listed to the right. Among the 14 parameters of each model, we have only listed the stellar parameters (M_⋆, T_⋆, R_⋆, t_⋆), accretion rates ($\dot{M}_{\rm env}$, $\dot{M}_{\rm disk}$), disk mass (M_disk), and foreground extinction (A_V). We have also listed the viewing angle (i, the angle between the sightline and the polar axis), and the total luminosity (L_tot). In addition to the parameters of the best-fit model, we have also used the acceptable models to show a possible range of stellar mass (M_⋆ Range), and used $\dot{M}_{\rm env}/M_\star$ and M_disk/M_⋆ from the acceptable models to estimate a possible range of its evolutionary stage, the Stage Range, as defined by Robitaille et al. (2006). The possible ranges of stellar mass and evolutionary stage are also given in Table 7.

Table 7. Inferred Physical Parameters from SED Fits to YSOs

					Physical Parameters of the Best-Fit Model
Source Name	[8.0] (mag)	Type	M* Range	Stage Range	M* (M☉)	T* (K)	R* (R☉)	Age (yr)	$\dot{M}_{\rm env}$ (M☉/yr)	$\dot{M}_{\rm disk}$ (M☉/yr)	Mdisk (M☉)	i (°)	Ltot (L☉)	AV (mag)	Remark
052212.6−675832.2	5.10	I	20-50	I	26	16000	37.7	8E+03	7.6E−03	0.0E+00	0.0E+00	18	92000	6.1	DC, UCHIIa
052350.1−675719.7	7.63	I	17-25	I	21	33000	7.5	3E+04	1.2E−03	0.0E+00	0.0E+00	56	62000	1.1	DC
052216.9−680403.6	8.52	I	8-15	I,II	8	16000	8.8	2E+05	1.0E−05	1.8E−07	6.0E−02	41	4300	7.2
052211.9−675818.1	8.97	I	8-28	I,II	28	4200	300.0	1E+03	5.9E−04	0.0E+00	0.0E+00	56	26000	2.1	DC
052343.6−680034.2	6.75	I/II	17-37	I,II	20	18000	23.3	2E+04	5.9E−03	0.0E+00	0.0E+00	18	53000	27.5	DC, UCHIIa
052219.8−680436.8	6.87	I/II	17-24	I	21	8000	93.7	5E+03	3.9E−04	0.0E+00	0.0E+00	49	33000	1.9	DC, UCHIIa
052257.6−680414.1	8.70	I/II	10-16	I,II	16	8400	56.6	1E+04	6.5E−03	0.0E+00	0.0E+00	18	15000	29.7	CO peak
052335.6−675235.4	7.59	II	14-17	II,III	17	33000	5.3	9E+05	1.7E−08	6.7E−11	1.9E−06	18	30000	0.9
052255.2−680409.5	7.63	II	13-18	I,II	13	30000	4.6	3E+05	4.0E−06	7.8E−07	3.6E−01	81	15000	0.0	UCHIIa
052202.2−675753.6	7.67	II	17-36	I,II	18	6400	125.0	6E+03	1.4E−04	8.8E−06	9.4E−02	87	23000	0.2	DC?
052256.8−680406.9	7.76	II	11-12	I	12	29000	4.3	2E+05	1.7E−05	1.8E−08	2.3E−02	87	11000	0.7
052202.0−675758.2	7.77	II	9	II	9	24000	3.7	2E+06	0.0E+00	6.1E−07	6.8E−02	56	4100	9.9	DC
052202.8−674701.9	7.77	II	8-15	I,II,III	15	8900	47.3	1E+04	2.0E−04	3.0E−06	6.9E−02	49	13000	0.1	DC tip
052046.6−675255.1	7.81	II	17	II	17	33000	5.3	1E+06	0.0E+00	8.7E−07	1.5E−01	87	32000	0.0
052203.1−674703.5	7.81	II	8-14	I,III	8	16000	8.8	2E+05	1.0E−05	1.8E−07	6.0E−02	41	4300	0.4
052207.3−675826.8	8.38	II	7-25	I,II	11	4400	76.8	5E+03	1.8E−04	1.8E−04	4.0E−01	18	2700	7.8	DC
052231.8−680319.2	8.40	II	8-15	I,II	8	12000	13.9	1E+05	8.9E−05	5.2E−10	8.0E−04	18	3200	2.9	DC?
052206.4−675659.2	8.41	II	8-15	I,II	13	5900	78.2	1E+04	4.2E−03	4.0E−06	2.2E−01	18	6600	1.2
052331.5−680107.9	8.42	II	8	I	8	16000	8.8	2E+05	1.0E−05	1.8E−07	6.0E−02	69	4300	0.9
052203.9−675743.7	8.98	II	8-15	I,II	8	22000	3.4	8E+05	5.0E−07	5.5E−06	1.2E−01	31	2700	2.0	DC
052138.0−674630.3	9.52	II	9-10	I	9	7000	31.4	4E+04	4.5E−04	1.2E−05	2.4E−02	18	2200	3.5
052212.0−674713.9	10.10	II	2-9	I	9	5100	48.7	3E+04	1.3E−03	5.1E−07	8.7E−02	18	1400	1.9
052242.0−675500.5	10.24	II	5-10	I	9	6200	31.4	5E+04	6.3E−05	4.8E−07	6.1E−02	41	1300	3.6
052259.0−680346.3	10.46	II	4-11	I,II	6	18000	3.6	5E+05	1.0E−05	2.9E−08	1.7E−03	18	1200	1.9
052204.8−675744.6	8.76	II/III	8-33	I,II,III	9	6000	43.2	3E+04	6.1E−05	3.9E−09	2.4E−03	31	2200	0.7	DC edge
052158.2−675554.2	8.81	II/III	5-14	I,III	12	7100	51.8	2E+04	3.8E−03	1.7E−07	2.5E−03	18	6100	1.0
052155.3−675634.9	8.89	II/III	8-9	I,II	8	22000	3.4	4E+05	6.2E−05	8.2E−08	1.6E−02	63	2300	0.5
052212.3−675813.4	8.97	II/III	5-28	I,II,III	12	7100	51.8	2E+04	3.8E−03	1.7E−07	2.5E−03	18	6100	1.0	superposed on DC
052136.0−675443.4	9.32	II/III	6-11	II,III	6	13000	5.8	5E+05	2.3E−07	1.9E−09	3.2E−03	41	890	0.1	dust pillar tip
052129.7−675106.9	4.12	III	34	II	34	42000	7.8	5E+05	1.2E−06	1.1E−10	4.6E−04	41	160000	0.0	∼O9I, this study
052147.1−675656.7	7.84	III	17	III	17	33000	5.3	1E+06	0.0E+00	8.7E−13	1.0E−06	81	32000	0.1
052207.3−675819.9	8.01	III	9-15	II	9	25000	3.7	3E+05	3.0E−06	1.8E−08	1.1E−03	41	4600	0.4
052315.1−680017.0	9.10	III	20-21	III	21	36000	5.9	1E+06	0.0E+00	4.7E−12	2.4E−06	75	50000	0.0	B0-O5Vb
052157.0−675700.1	9.96	III	9	I	9	9200	20.1	6E+04	9.6E−05	9.9E−06	5.5E−01	18	2800	0.0	O8.5Vc
052159.6−675721.7	10.09	III	16	III	16	32000	5.1	1E+06	0.0E+00	1.5E−11	4.7E−07	31	25000	0.0	O7.5Vc
052340.6−680528.5	10.95	III	12	III	12	28000	4.4	2E+06	0.0E+00	9.3E−14	7.4E−07	87	11000	0.0	B[e]c

Notes. ^aIndebetouw et al. (2004). ^bChen (2007). ^cOey & Massey (1995).

Download table as:  ASCIITypeset image

The SED fits of the 36 YSOs, as displayed in Figure 7, show different degrees of goodness of fit among our empirically defined YSO types. The seven Type I and I/II YSOs have model SEDs agree well with the observed SEDs, although the best and acceptable models span a large mass range. The 21 Type  II and II/III YSOs show good agreement between model and observed SEDs, except at 4.5 μm, where the observed fluxes are systematically lower than the modeled. This discrepancy will be discussed below in Section 6.2.1. Among the eight Type III YSOs analyzed, only three of them show good agreement in the SED fits, while the other five exhibit significant discrepancies between model and observed SEDs. These discrepancies and possible causes will be discussed in Section 6.2.2.

6.2.1. PAH Emission in Massive YSOs

6.2.2. Discrepancies in Type III YSOs

Among the eight Type III YSOs, five show significant discrepancies between the best-fit models and the observed SEDs: 052129.7 − 675106.9, 052315.1 − 680017.0, 052157.0 − 675700.1, 052159.6 − 675721.7, and 052340.6 − 680528.5 (Figure 7). The brightest object, 052129.7 − 675106.9, is peculiar and will be discussed later in Section 7.5. The other four YSOs show deep V- or U-shape SEDs with the flex point in the IRAC bands, where the best-fit model fluxes are much higher than the observed fluxes, exhibiting the most prominent discrepancies.

To understand these discrepancies, we can divide the SED of a YSO into three segments, optical/near-IR (UBVIJK_s), near/mid-IR (IRAC bands), and mid/far-IR (MIPS bands and beyond), and look into the dominant origin of emission for each segment. At the shortest wavelengths, the optical/near-IR segment, the radiation is dominated by stellar photospheric emission; for example, Figure 8 shows the optical/near-IR segment of the SEDs of four Type III YSOs can be well fitted by stellar atmosphere models (Kurucz 1993). At longer wavelengths, the stellar emission diminishes and dust emission rises. The near/mid-IR segment consists of emission from PAHs and warm dust, while the mid/far-IR segment consists of emission mostly from colder dust. Qualitatively, warm dust that emits at near/mid-IR wavelengths needs to be near the star in order to reach the required emitting temperatures; therefore, the near/mid-IR segment of a YSO's SED is dominated by disk emission. Colder dust that emits at mid/far-IR wavelengths is farther away from the star, and thus the mid/far-IR segment of a YSO's SED is dominated by envelope emission.

Zoom In Zoom Out Reset image size

The aforementioned discrepancies between observations and best-fit models of SEDs of Type III YSOs in the IRAC bands suggest that the YSOs do not have as much disk emission as in the models. Below we analyze the SEDs quantitatively to understand their physical implications. The mid- to far-IR parts of the V- or U-shaped SEDs of these Type III YSOs (Figure 7) suggest that the dust continuum peaks at ⩾24 μm. The corresponding blackbody temperature of the dust is thus ⩽120 K. The distance of such dust temperature to the central star can be estimated if the stellar spectral type and the implied stellar effective temperature are known. Among these four YSOs, 052157.0 − 675700.1, 052159.6 − 675721.7, and 052340.6 − 680528.5 have been spectroscopically classified as O8.5V+neb, O7.5V+neb, and B[e] stars, respectively (Oey & Massey 1995), and 052315.1 − 680017.0 has (U − B) = −0.79, (B − V) = 0.09, and V = 14.10 (Table 5) that are consistent with the colors and magnitudes of a B0-O5 V star (Schmidt-Kaler 1982) reddened by A_V∼1.2. For a dust temperature of 120 K and an albedo of 0.5, the closest distance of such dust to a B0 V star with a stellar effective temperature of 30,000 K is 860 AU; this distance is larger for stars with higher effective temperatures. This result implies a low dust content, or a hole, within the central ∼1000 AU.

While our assessment of the observed SEDs suggests R^min_disk > 860 AU, no models with R^min_disk > 100 AU are available in the database of pre-calculated SEDs for model fitting (Robitaille et al. 2007). To illustrate how R^min_disk affects the SEDs, we have used the Whitney et al. (2003) radiative transfer code⁷ to calculate SEDs for a YSO model with stellar parameters appropriate for the observed spectral types but with customized disk parameters. The YSO model #3019532 is chosen because its T_⋆ = 34,718 K and R_⋆ = 5.628 R_☉ are representative of late-type O stars. In this Stage III YSO model, the envelope has dissipated completely. We have calculated two SEDs with disk radii of 50–1000 AU and 1000–2000 AU, respectively; these model SEDs are overplotted on the observed SED of 052157.0 − 675700.1 in Figure 9. The model SED with R^min_disk = 50 AU show prominent excess emission in the near/mid-IR bands, while the model SED with R^min_disk = 1000 AU adequately reproduce the V-shaped SED without excess near/mid-IR emission. This comparison suggests that dust has been cleared out to nearly 1000 AU. This clearing of dust is consistent with the expectation from YSOs at late evolutionary stages. We thus conclude that the large discrepancies between the best-fit models and the observed SEDs are caused by the absence of appropriate model SEDs in the database of the Online SED Model Fitter.

Zoom In Zoom Out Reset image size

Similar V-shaped SEDs have been observed in Herbig Ae/Be stars RCW34 and TY Cra, and the SED modeling suggested an analogous evolutionary stage (Hillenbrand et al. 1992). Similar near-IR dips have also been observed in SEDs of the low-mass YSOs DM Tau and GM Aur; these dips have been interpreted as the clearing of dust from the inner regions of their disks (Rice et al. 2003; Calvet et al. 2005).

6.3.1. Type versus Stage

6.3.2. Evolutionary Stages of YSOs in Ultracompact H ii Regions

Some young massive stars are known to be associated with small (diameter ⩽10¹⁷ cm), dense (⩾10⁴ cm⁻³) regions of ionized gas, called ultracompact H ii regions (UCHIIs). Five UCHIIs have been identified in N 44: B0522 − 5800, B0523 − 6806(NE), B0523 − 6806(SE), B0523 − 6806(SW), and B0523 − 6806 (Indebetouw et al. 2004). All five UCHIIs are coincident with Spitzer sources within 1''. The Spitzer counterpart of B0523 − 6806(SE), 052324.8 − 680641.6, is faint and falls below the cutoff line of [8.0] ⩾14.0−([4.5]−[8.0]) in the [8.0] versus [4.5]−[8.0] CMD, where background galaxies and YSOs with masses ⩽4 M_☉ are located. As low-mass stars cannot produce UCHIIs, this source is most likely a background star-forming galaxy. We have examined the MOSAIC I, ISPI JK_s, and Spitzer IRAC and MIPS images of this source. The I and J images show a brighter source and a faint source separated by less than 1''. The faint source becomes the brighter of the two in the K_s band, and is the dominant source in the Spitzer bands. This source is not in a giant molecular cloud or surrounded by bright diffuse PAH emission. These results support the identification of B0523 − 6806(SE), or 052324.8 − 680641.6, as a background star-forming galaxy.

The Spitzer counterparts of the other four UCHIIs are bright and have been identified as Type I-II YSOs (see Table 8). As these sources are among the top 10 most luminous YSOs in N 44 at 8.0 μm, they have good 24 μm flux measurements and consequently models of their SEDs are well constrained. The stellar masses determined from the best-fit models to their SEDs can be translated into spectral types, assuming the relationship for main sequence stars. These spectral types can be compared with those implied by the ionizing fluxes determined from radio continuum observations (Indebetouw et al. 2004). As seen in Table 8, excellent agreement exists between these two independently determined spectral types.

Table 8. Properties of YSOs with UCHIIs

UCHII	YSO ID	Type	Stage Range	M⋆ (M☉)	Spec. Type	Spec.a Type	$\dot{M}_{\rm env}$ (M☉/yr)	$\dot{M}_{\rm crit}$b (M☉/yr)
B0522 − 5800	052212.6 − 675832.2	I	I	26	O7-8 V	O7 V	7.6E−03	4.1E−05
B0523 − 6806 (NE)	052343.6 − 680034.2	I/II	I,II	20	O8-9 V	O9 V	5.9E−03	1.8E−05
B0523 − 6806 (SW)	052219.8 − 680436.8	I/II	I	21	O8-9 V	O8.5 V	3.9E−04	2.2E−05
B0523 − 6806	052255.2 − 680409.5	II	I,II	13	B1-2 V	B0 V	4.0E−06	1.0E−05

Notes. ^aThe spectral type is determined from radio observations (Indebetouw et al. 2004). ^bThe critical infalling rate is adopted from Churchwell (2002).

Download table as:  ASCIITypeset image

The development of a UCHII depends on not only the ionizing flux provided by the central star, but also the opacity of the circumstellar medium. For infalling rates higher than some critical value, $\dot{M}_{\rm crit}$, the circumstellar medium will have such high opacities that the ionized region will be too small and too optically thick to be detectable (Churchwell 2002). For the respective spectral types of the central stars, the $\dot{M}_{\rm crit}$ of the four UCHIIs in N 44 are computed to be ∼(1–4) × 10⁻⁵ M_☉ yr⁻¹ (Table 8). Compared to the $\dot{M}_{\rm env}$ determined from the best-fit models for these four YSOs (Table 8), it is seen that only the Type II YSO 052255.2 − 680409.5 has $\dot{M}_{\rm env} < \dot{M}_{\rm crit}$, and the other three YSOs have $\dot{M}_{\rm env} \gg \dot{M}_{\rm crit}$. We have further examined whether any acceptable models of the latter three YSOs yield smaller envelope accretion rates. We find that only 052343.6 − 680034.2 has a few models with $\dot{M}_{\rm env} \le \dot{M}_{\rm crit}$, but these models have T_⋆ = 38,000–40,000 K, higher that that for an O9 V star (T_⋆∼33,000 K) estimated from the ionization requirement of the UCHII. All the other acceptable models have $\dot{M}_{\rm env} \gg \dot{M}_{\rm crit}$.

Our limited sample shows that the Type II YSO in UCHII has $\dot{M}_{\rm env} < \dot{M}_{\rm crit}$, and the Type I and I/II YSOs in UCHIIs have $\dot{M}_{\rm env} \gg \dot{M}_{\rm crit}$. As Types I and I/II YSOs are still dominated by envelopes, it is possible that most of the infalling envelope material is used in forming an accretion disk, instead of the stellar core, as modeled by Yorke & Sonnhalter (2002). Therefore, $\dot{M}_{\rm env}$ should not be interpreted as the accretion rate of mass onto the stellar core.

In our sample of 60 YSOs in N 44, 24 do not have mass estimates from SED model fits as their IRAC fluxes are contaminated by neighboring stars; nevertheless, it can be judged from their lower brightnesses that they are probably on the low-mass end in the sample. The other 36 YSOs have reliable SEDs that can be modeled to determine their masses; their mass estimates from the best and acceptable fits are listed in Table 7. Although many of these YSOs have large uncertainties in their mass estimates, i.e., large M_⋆ Range, 30 of them have M_⋆ Range all greater than 8 M_☉ and are thus most likely bona fide massive YSOs. The remaining six YSOs have M_⋆ Range extending from intermediate to high mass; these may also be massive YSOs, particularly for the four with best-fit masses ⩾8 M_☉.

It has been suggested that the criterion [8.0] ⩽8.0 may be used to select massive YSOs in the LMC (Gruendl & Chu 2008). We find that indeed the M_⋆ Ranges of the brightest YSOs, with [8.0] ⩽8.0, are all ⩾8 M_☉. This criterion may be too conservative, as almost all (25 out of 27) YSOs with [8.0] ⩽9.0 still have masses ⩾8 M_☉

At the high-mass end, nine YSOs have M_⋆ Ranges all ⩾17 M_☉; these are most likely O-type YSOs. Five YSOs that show M_⋆ Range with an upper mass limit ⩾17 M_☉ and a lower mass limit ≪17 M_☉; these may or may not be O-type stars. To improve the census of O-type YSOs, we have checked the optical spectral classifications of the Type III YSOs to search for O stars that are missed by our estimates of masses from SED model fits. We find that two Type III YSOs with optical spectral types of O7.5 V and O8.5 V (Oey & Massey 1995) have masses determined from the SED model fits to correspond to B0–3 V spectral types (052157.0 − 675700.1 and 052159.6 − 675721.7). Therefore, there exist at least 11 O-type YSOs in N 44. These most massive YSOs will be discussed further in Section 7.1.

We examine the star formation properties of the molecular clouds in N 44 as these clouds contain the bulk material to form stars. The NANTEN CO survey of the LMC (Fukui et al. 2001) shows three large concentrations of molecular material in N 44, i.e., the central, southern, and northern peaks (Figure 10). These three concentrations exhibit different numbers of massive stars formed in the last few Myr, as evidenced by their different amounts of ionized gas. The central molecular peak is associated with prominent Hα emission from a supershell and bright H ii regions along the shell rim. Star formation has been occurring at this site for an extended period of time, with the supershell encompassing 10 Myr old massive stars and the bright H ii regions containing ∼5 Myr old massive stars (Oey & Massey 1995). The southern molecular peak shows one bright H ii region and several smaller, disjoint H ii regions. The massive stars of these H ii regions have not been studied spectroscopically, but the absence of shell structures produced by fast stellar winds and supernovae implies that massive stars are most likely also ∼5 Myr old or that star formation has started only in the last few Myr. The northern molecular peak has a couple small H ii regions, indicating that only modest massive star formation has taken place.

Zoom In Zoom Out Reset image size

Figure 10 shows that almost all of the YSOs in N 44 are found in molecular clouds and about 75% of the YSOs are congregated toward the three molecular peaks. The YSOs of the three molecular peaks show different characteristics in their spatial distributions and interstellar environments. The central molecular peak has the highest concentration of YSOs, as 21 of them aggregate in the prominent H ii regions along the southwest rim of the supershell. The southern molecular peak has loosely distributed YSOs, and most of these 11 YSOs are associated with the disjoint H ii regions. The northern molecular peak has 12 YSOs that are also loosely distributed, but the majority of the YSOs are not associated with any ionized gas.

The YSOs of the three molecular peaks in N 44 also show differences in their mass distributions. In Figure 10, we have marked the YSOs with circles in three sizes that represent O-type stars with M_⋆⩾ 17 M_☉, B-type stars with M_⋆⩾ 8 M_☉, and intermediate-mass stars with M_⋆< 8 M_☉, respectively. It is striking that ∼80% of the O-type and B-type YSOs are in or adjacent to H ii regions. The YSOs with intermediate masses or without mass estimates do not show such strong bias in spatial distribution, although this may be partly due to an observational bias, as these YSOs are fainter and it is more difficult to detect faint YSOs over the bright background dust emission in H ii regions. The central molecular peak, having the most prominent H ii regions, possesses the highest concentration of O- and B-type YSOs. The southern molecular peak has a respectable number of O- and B-type YSOs, while the northern molecular peak has no O-type YSOs at all.

The characteristics of the current star formation in the three molecular peaks of N 44 appear to be dependent on the massive star formation that occurred in the recent past. It is possible that the pattern of star formation is controlled by properties of the molecular clouds. The central, southern, and northern peaks correspond to the giant molecular clouds (GMCs) LMC/M5221 − 6802, LMC/M5239 − 6802, and LMC/M5221 − 6750 cataloged by Mizuno et al. (2001). Their FWHM line-widths (ΔV) at the peak position are 7.2, 15.8, and 3.8 km s⁻¹, respectively. It is interesting to note that the ΔV of the southern molecular peak of N 44 is the highest and that of the northern molecular peak is nearly the lowest among all GMCs in the LMC. While the small ΔV of the northern molecular peak reflects the low level of stellar energy feedback in the last few Myr, the larger ΔV of the other two peaks do not seem to scale with star formation activity. Detailed mapping of these three molecular clouds is needed to search for fundamental differences among these three clouds that might be responsible for their different star formation characteristics.

To examine the immediate surroundings of the YSOs in N 44, we have searched the HST archive and found useful WFPC2 images of three fields that contain YSOs. These three fields encompass the H ii regions N 44C, N 44F, and N 44H, respectively, and their locations in N 44 are shown in Figure 1(a). These H ii regions and their associated YSOs and ionizing stars are individually discussed below.

N 44C is a bright H ii region located at the southwest rim of the N 44 superbubble. The HST Hα image of N 44C shows an overall shell morphology: the northern part is bright and centered on the ionizing O7 V star (Oey & Massey 1995), while the southern part is faint and consists of multiple circular filaments superposed by radial filaments streaming to the southwest (Figure 11(a)). It is not clear whether these southern radial and circular filaments are physically associated with each other. N 44C abuts against a dark cloud to the north, which coincides with the core of the GMC M5221 − 6802 revealed by high-resolution ESO-SEST observations (Chin et  al. 1997). The observed density variations suggest that N 44C is a blister H ii region on the surface of a molecular cloud.

Zoom In Zoom Out Reset image size

Seven YSOs are within the FOV of the HST Hα image of N 44C. One is projected at the base of a bright-rimmed dust pillar in the northwest part of N 44C. Four are in the dark cloud 2–3 pc exterior to the northwest edge of N 44C; all four are superposed on diffuse 8 μm PAH emission (Figure 11(b)), indicating that they may be associated with PDRs. One is located further northwest and projected in the N 44 supershell rim; it is also superposed on large-scale diffuse 8 μm PAH emission and may be associated with PDRs. The last one is located in the southwest outskirts of N 44C; it does not have prominent PDRs around it. It is remarkable that the majority of the YSOs associated with N 44C are in a molecular core and superposed on prominent PDRs. Apparently they were formed in an interstellar environment that has been affected by energy feedback from stars that were formed a few Myr earlier.

N 44F is a bright, ring-shaped H ii region located at the northwest outskirts of the superbubble. The H ii region is ionized by an O8III star (Will et al. 1997). Two prominent bright-rimmed dust pillars can be readily recognized in the WFPC2 Hα image (Figure 12(a)). One of these dust pillars has a YSO, 052136.0 − 675443.4, emerging at its tip, reminiscent of those seen in the Eagle Nebula (M 16; Hester et al. 1996). The mass estimated from SED fits for this YSO is ∼6–11 M_☉, more massive than those with ∼3–4 M_☉ seen in the Eagle Nebula (Thompson et al. 2002).

Zoom In Zoom Out Reset image size

N 44H is located to the southeast of the N 44 superbubble. This H ii region contains the luminous blue variable (LBV) HD 269445 (Stahl et al. 1984; Humphreys & Davidson 1994) and four blue stars whose photometric measurements suggest spectral types of early-B (Chen 2007); thus, the LBV is most likely responsible for photoionizing the H ii region. The YSO 052249.2 − 680129.0 is projected within N 44H toward its bright northwest rim, ∼60'', or 15 pc, from the LBV. The WFPC2 Hα image of this YSO shows a close pair of sources and another source at ∼3'' south (Figure 13(a)). The 8 μm image (Figure 13(b)) shows a bright source and a faint extension to the south that are coincident with the optical pair and southern source, respectively. It is possible that all three sources are YSOs.

Zoom In Zoom Out Reset image size

We use the YSOs and their interstellar environment to investigate whether some of the current star formation in N 44 is triggered. Figure 10 shows that on a large scale the central supershell of N 44 exhibits the most prominent association between stellar energy feedback and star formation. The alignment of massive YSOs in the southwest rim of the supershell suggests that the expansion of the supershell into the molecular cloud has triggered the star formation. This triggering mechanism may have been going on for ∼5 Myr and caused the formation of the massive stars in the bright H ii regions along the shell rim.

Examined closely, the majority of massive YSOs are located at the edges of H ii regions or in PDRs of dark clouds. These are suggestive examples of star formation triggered by external thermal pressure raised by photoionization or photodissociation. YSOs are also found in bright-rimmed dust pillars. Among them, the YSO in N 44F, as aforementioned in Section 7.2, is in a simple interstellar structure and its immediate surroundings can be examined with WFPC2 images (Figure 12(a)), providing an opportunity to investigate whether the star formation is triggered. For this YSO to be formed from triggering instead of merely being exposed by the ionization front of the H ii region, the formation timescale of the YSO should not be longer than the time for the ionization front to traverse from the tip to the bottom of the pillar. For the pillar's projected length of ∼1.25 pc and a shell expansion velocity of 12 km s⁻¹ (Nazé et al. 2002), the traverse time is ∼0.1 Myr. This timescale is comparable to the formation timescale for a 10M_☉ YSO, ∼0.1 Myr (Beech & Mitalas 1994), making it a plausible case of triggered star formation. To further determine statistically the importance of triggering for star formation, a systematic survey of the immediate surroundings of YSOs with high-resolution images, such as HST images, is needed.

The supershell of N 44 shows a breakout at the south rim, and hot gas has been escaping and deflected to the east. The juxtaposition of this hot gas flow and the star forming regions associated with LH49 in the southern GMC is intriguing and may suggest triggered star formation. However, there is no direct physical evidence that the hot gas outflow compresses the molecular cloud to form stars. In fact, the thermal energy of the hot gas outflow, ∼1 × 10⁵⁰ ergs (Magnier et al. 1996), is much smaller than the kinetic energy of the southern GMC, ∼3 × 10⁵¹ ergs estimated from a mass of 2.1×10⁶ M_☉ and a velocity dispersion of 12.6 km s⁻¹ (Mizuno et al. 2001). Furthermore, the star formation in the southern GMC is mostly concentrated in regions not in contact with the hot gas flow. We conclude that the breakout of N 44's supershell is not responsible for triggering the star formation in the southern molecular concentration.

It is interesting to note that a small nebula is detected around YSO 052207.3−675819.9 projected in the dark cloud northwest of N 44C. A closeup of this nebula in Figure 14 shows that it is detected in the Hα emission-line image but not in the y continuum image, indicating that it is an H ii region, instead of a reflection nebula. This small H ii region has a size of 22 × 15, or 0.53 pc× 0.36 pc, and an average Hα surface brightness⁸ of 1.9 × 10⁻¹⁴ ergs cm⁻² s⁻¹ arcsec⁻², corresponding to an emission measure of 9.4 × 10³ cm⁻⁶ pc. Adopting an average diameter of 0.45 pc as the path length of Hα emission, the rms electron density of the H ii region is then ∼145 cm⁻³. This H ii region has a size comparable to those of compact H ii regions, i.e., ≲0.5 pc, but its emission measure and rms electron density are much lower than the typical values for compact H ii regions, ≳10⁷ cm⁻⁶ pc and ≳5 × 10³ cm⁻³ (Franco et al. 2000). The lower densities of this small H ii region suggest that it has expanded and is more evolved than a compact H ii region.

Zoom In Zoom Out Reset image size

This small H ii region provides an opportunity for us to use its ionization requirement to assess the spectral type of the ionizing star and compare its implied mass with that determined from the YSO SED fitting. The UBV photometry of YSO 052207.3 − 675819.9, given in Table 5, suggests that its central star is a ∼B2 V star reddened by A_V ∼ 1.3. Applying this extinction correction to the observed Hα flux, we obtained an Hα luminosity of L_Hα = 3.7×10³⁴ ergs s⁻¹, which requires an ionizing power of Q(H⁰) = 7.4 × 10¹¹L_Hα photons s⁻¹ = 2.7 × 10⁴⁶ photons s⁻¹. This ionizing power can be provided by a main-sequence star of effective temperature of 26,200 K (Panagia 1973), corresponding to a spectral type of B1 V (Schmidt-Kaler 1982).⁹ The mass estimated from SED model fits for this YSO is 9–15 M_☉, corresponding to spectral type B1-2 V. It is satisfactory that the three independent methods have produced consistent mass estimates.

This YSO in a small H ii region illustrates a caveat in analyzing objects that are not fully resolved by Spitzer images. The SED model fits of YSO 052207.3 − 675819.9 infers that it is at evolutionary Stage II. The presence of a visible H ii region and the visibility of the YSO in UBV bands indicate that the star has little circumstellar dust and that the YSO is likely at Stage III. This discrepancy in evolutionary stage is most likely caused by the inclusion of dust emission from the H ii region in the flux measurements, since it is not resolved in the ISPI and Spitzer images. It is thus important to use high-resolution images to examine the immediate surroundings of massive YSOs in order not to be misled by the SEDs.

The source 052129.7−675106.9 is coincident with a small compact H ii region, identified as an Hα knot and cataloged as N 44A by Henize (1956). Previously, this IR source was identified as IRAS 05216 − 6753 and classified as an H ii region (Roche et al. 1987) or an obscured supergiant or AGB (Wood et al. 1992; Loup et al. 1997). Figure 15 shows that the central star of this small H ii region is bright in optical wavelengths. Chen (2007) measured V = 14.17 ± 0.02, (B−V) = 0.79 ± 0.03 and (U − B) = −0.64 ± 0.06. These magnitudes and colors cannot be reproduced by any combination of normal spectral type and luminosity class. The observed (U−B) indicates a blue star, so the intrinsic color (B−V)₀ must be in the range of 0 to −0.3; therefore, E(B − V) = 0.8–1.1, or A_v = 2.5–3.4, and M_V = −6.8 to −8, suggesting that this star is a blue supergiant of luminosity class I. It cannot be an AGB star.

Zoom In Zoom Out Reset image size

We can make another independent estimate of the spectral type based on the Hα luminosity of N 44A. Using the flux-calibrated MCELS images of N 44, we subtracted the red continuum from the Hα image and measured a continuum-subtracted Hα flux of 4.7×10⁻¹² ergs s⁻¹ cm⁻² from N 44A. To achieve feasible ionizing power, the star has to be blue with (B − V)₀ ∼ −0.3, thus we adopt an extinction of A_v ∼ 3.4 and find an Hα luminosity of 1.4 × 10³⁷ ergs s⁻¹. This Hα luminosity requires the ionizing power of an O9 I star.

We thus suggest that the central star of N 44A is an O9 I star. The strong mid-IR excess of this star indicates the existence of abundant dust around the star. The anomalous combination of (U − B) and (B − V) colors may be caused by the inclusion of extra scattered light in the U band. If the star is a young star, the dust and H ii region would both be interstellar. If the star is an evolved star, the dust and H ii region would both be ejected stellar material and show enhanced elemental abundances. The current data do not allow us to distinguish whether the central star of N 44A is a YSO or an evolved massive star, such as an LBV. High-resolution optical spectra of the Hα and [N ii] lines are needed to search for N-enriched ionized gas. If the ionized gas is enriched, the central star of N 44A is an evolved star. If the ionized gas is not enriched, the central star of N 44A must be young, otherwise its stellar wind would have blown away the small H ii region.