SEQUENTIAL STAR FORMATION IN RCW 34: A SPECTROSCOPIC CENSUS OF THE STELLAR CONTENT OF HIGH-MASS STAR-FORMING REGIONS^*

The H ii region RCW 34 (Rodgers et al. 1960; IRAS 08546−4254 or Gum 29, Gum 1955) is located in the VMR cloud C (Liseau et al. 1992). The region is an optically visible H ii region of about 2' × 2' in extent and slightly offset from the main dense cores defining the VMR cloud C detected by CO observations (Murphy & May 1991; Yamaguchi et al. 1999). Due to a similar velocity (v_lsr = 6 km s⁻¹), Murphy & May (1991) argue that RCW 34 is likely part of the C-cloud. A similar velocity has been measured from the CS line (v_lsr = 5.5 km s⁻¹; Bronfman et al. 1996), confirming the association with cloud C, which would place the region at a distance somewhere between 1 and 2 kpc.

Herbst (1975) and van den Bergh & Herbst (1975) allowed that the exciting star of RCW 34 might be more distant, but favored an association with the Vela-R2 R-association at a distance of 870 pc. Optical spectroscopy by Heydari-Malayeri (1988) implies that the central source vdBH 25a (van den Bergh & Herbst 1975) is of spectral type O8.5V. This suggests a distance of 2.9 kpc, supporting the idea that RCW 34 is more distant than Vela-R2. In Section 4.1, we use near-infrared spectroscopy of the cluster OB stars to derive a distance in rough agreement with Heydari-Malayeri (1988). Heydari-Malayeri also detected a north–south velocity gradient in the ionized gas, which he suggested to be evidence that RCW 34 is an H ii region undergoing a champagne flow. This is supported by Deharveng et al. (2005) who conclude, based on the Midcourse Space Experiment and near-infrared observations, that the region is not the result of the collect and collapse triggered star formation scenario.

Several signs of active star formation are present in RCW 34. Two water masers have been reported (Braz & Epchtein 1983; Caswell et al. 1989) as well as a class II methanol maser (Chan et al. 1996; Walsh et al. 1997). Associated with the water masers, CS emission has been found (Zinchenko et al. 1995) with an estimated mass of 1320 M_☉. Gyulbudaghian & May (2007) reported the detection of 1.2 mm continuum emission toward RCW 34.

The outline of the paper is as follows. Section 2 outlines the data reduction of the SINFONI and Spitzer/IRAC data sets used in this paper. In Section 3, the interaction between the H ii region and the cluster surroundings is studied using the IRAC data as well as the SINFONI emission line maps. Section 4 focuses on the stellar spectra obtained in the SINFONI field; spectral types, luminosity class as well as extinctions are derived; and in Section 5, the stellar content is discussed and an age estimate is obtained for the NIR cluster in RCW 34. In Section 6, we discuss the complete region and combine all the different properties to derive a possible formation scenario for RCW 34; and Section 7 summarizes the main conclusions of the paper.

Near-infrared H- and K-band observations were performed using the Integral Field instrument SINFONI (Eisenhauer et al. 2003; Bonnet et al. 2004) on UT4 (Yepun) of the VLT at Paranal, Chile. The observations of RCW 34 were performed in service mode between 2007 February 4 and 9. The non-AO mode of SINFONI, used in combination with the setting providing the widest field of view (8'' × 8''), delivers a spatial resolution of 025 per slitlet. The typical seeing during the observations was 07 in K. The H+K grating was used, covering these bands with a resolution of R=1500 in a single exposure.

We observed RCW 34 with a detector integration time (DIT) of 30 s per pointing to guarantee a good signal-to-noise ratio (S/N ∼ 70) for the early B-type stars within this cluster. RCW 34 is much larger than the field of view of SINFONI, the near-infrared extent of the H ii region is 104'' × 60'' on the sky. The observations were centered on the central O8.5V star (vdBH 25a) at coordinates: α(2000) = 08^h56^m281, δ(2000) = −43°05'556. We observed this area with SINFONI using a raster pattern, covering every location in the cluster at least twice. The offset in the east–west direction was 4'' (i.e., half the field of view of the detector) and the offset in the north–south direction was 675. A sky frame was taken every 3 minutes using the same DIT as the science observations. The sky positions were chosen based on existing NIR imaging in order to avoid contamination. Immediately after every science observation, a telluric standard of B spectral type was observed, matching as close as possible the air mass of the object. This star is used for the removal of the telluric absorption lines as well as for flux calibration. The stars chosen as telluric standards are HIP 048652, HIP 045117, HIP 052202, and HIP 040629, one for each of the observing blocks.

2.1.1. Data Reduction

Additional data obtained with IRAC (Fazio et al. 2004) on board of the Spitzer satellite have been retrieved from the Spitzer archive (PID: 40791, PI: Majewski). The data were obtained on 2008 January 31 in a mapping mode, covering a large part of the outer galaxy. We selected the frames covering the area around the cluster as well as an adjacent control field. Each frame has an exposure time of 1.2 s and every position is covered twice. The raw frames were processed using the standard pipeline version S17.02 to produce the basic calibration data set. This data set was processed further using MOPEX version 16.2.1 (Makovoz et al. 2006).

As the IRAC data are undersampled, aperture photometry is preferred above psf-fitting photometry. Source detection and photometry were performed using the Daophot package under IRAF. Point sources with a peak level of more than 5 times the rms of the background were detected. Aperture photometry was applied with an aperture of 3 pixel (37) and a sky annulus between 3 (37) and 7 (85) pixel. These small values were chosen due to the variable extended emission. The appropriate aperture corrections and zero points were adopted (Tables 5.1 and 5.7 of the IRAC Data Handbook) to obtain the magnitudes of the objects.

A three-color Spitzer/IRAC image of the region is shown in Figure 1 and, Figure 2 shows a [3.6]–[4.5] versus [5.8]–[8.0] color–color diagram of the stars in the field. Objects not detected in all bands are not included in the analysis. We have not plotted the stars in the control field which only contains stars without IR excess. With the color criteria proposed by Megeath et al. (2004) and Allen et al. (2004), we can identify several classes of objects. Stars without IR excess have colors close to 0.0. The IRAC colors are insensitive to the effective temperature of the photospheres. These objects, plotted as star-shaped symbols, could be fore- or background late-type stars as well as earlier-type main-sequence stars associated with the star-forming region. The objects with [3.6]–[4.5] between 0.0 and 0.8 and [5.8]–[8.0] between 0.4 and 1.1 are classified as class II objects. This region is marked with a box, and the objects showing class II colors are plotted as triangles. Objects with [3.6]–[4.5] > 0.8 and [5.8]–[8.0] > 0.0 or [3.6]–[4.5] > 0.4 and [5.8]–[8.0] > 1.1 are classified as protostars (class 0 and I objects; diamonds), objects with [5.8]–[8.0] > 1.1 and [3.6]–[4.5] between 0.0 and 0.4 are classified as class I/II sources (square symbols). The five numbered objects are located in the SINFONI field and are discussed in more detail in Sections 4.3. and 5.3.

Zoom In Zoom Out Reset image size

The class I/II identification should be considered with caution. Megeath et al. (2004) show that these objects are likely class II sources with the 8 μm flux contaminated by variable extended emission. Figure 1 shows that these objects are located on top of a strong diffuse background. Therefore, we consider them as class II objects. Additional sources of contamination, especially for the detection of protostars, could be planetary nebula, AGB stars, or background galaxies (Whitney et al. 2003). However, the absence of sources with these colors in the control field suggests that the contamination is negligible. We have plotted the different classes on the IRAC 3.6 μm image (Figure 3) to study their potential relation with the extended MIR emission of the bubble. The locations of the stars without infrared excess (white, star shapes) are not related to the extended emission, but evenly distributed over the field outside the bubble. These objects are most likely field stars.

Zoom In Zoom Out Reset image size

The location of the young stellar objects shows a clear correlation with the bubble and the extended emission. Most of the class II sources (yellow triangles) are located inside the bubble. The protostars (red diamonds) are spatially associated with the extended MIR emission. Most of these objects are located near the bright arc in the north, but also some objects are associated with the fainter extended emission in the southern part of the bubble. Although no obvious cluster is visible inside the bubble, the class II objects located inside the bubble suggest a recent star formation event. This might have been responsible for the creation of the bubble (Churchwell et al. 2006). The location of the younger protostars in the outer regions of the bubble indicates the presence of a more recent generation of stars near the edge of the bubble.

The SINFONI observations are centered on the ionizing source, vdBH 25a, and cover both the H ii region as well as the bright arc seen in the IRAC images. Figure 4 shows the three-color composite of three line maps along with contours of the IRAC 3.6 μm image. This image reveals the stellar content of this region and the overlay shows that several near-infrared sources are also detected at IRAC wavelengths. The region is dominated by the two bright stars in the center of the field (one of them is vdBH 25a at the location of the asterisk symbol) and a relatively sparse cluster of about 130 stars down to K = 17 mag, surrounding the central region. For all the point sources visible in this image, a SINFONI H- and K-band spectrum is available and a spectral classification is obtained (Section 4). The H ii region is traced by Brγ (2.16 μm, red) and He i (2.058 μm, blue) emission and the surrounding PDR by the molecular H₂ emission (2.12 μm, green). Just north of the position of the central star, the Brγ emission is strongest. Even further to the north, the Brγ and He i emission drop abruptly and the H₂ emission begins, indicating the interface between the ionized and the molecular gas, coincident with the bright arc seen in the IRAC images.

Zoom In Zoom Out Reset image size

3.2.1. H ii Region

We have extracted spectra at six locations in the H ii region covering most of the H ii region (annotated with numbers I–VI in Figure 4). Table 1 gives the positions and sizes of the six extracted regions.

Table 1. Properties of the H ii Gas

Region	R.A. (2000)	Decl. (2000)	Size	vlsr	He i/Brγ	AV
	(h m s)	(° ' '')	('')	(km s−1)		(mag)
I	08 56 29.8	−43 05 52.4	2.4 × 4.6	18 ± 13	<0.13	5.3 ± 0.2
II	08 56 28.1	−43 05 50.8	13.7 × 4.9	16 ± 8	0.038 ± 0.002	5.1 ± 0.1
III	08 56 27.3	−43 05 39.2	2.3 × 2.7	16 ± 9	<0.17	15.9 ± 0.6
IV	08 56 25.7	−43 05 55.7	5.0 × 3.0	10 ± 12	0.032 ± 0.006	5.6 ± 0.2
V	08 56 27.7	−43 06 05.0	7.7 × 4.0	13 ± 9	0.047 ± 0.004	4.4 ± 0.1
VI	08 56 27.7	−43 06 11.9	7.7 × 3.7	11 ± 10	0.06 ± 0.01	1.4 ± 0.2

Download table as:  ASCIITypeset image

As an example, the upper panels of Figure 5 show the spectra extracted from region II, just north of the ionizing star.

Zoom In Zoom Out Reset image size

To check the hypothesis of a champagne flow, we estimate the column density of the molecular cloud in which the H ii region is embedded. In an H ii region, the relative line intensities of the hydrogen recombination lines are mostly determined by the Einstein coefficients, with a very weak dependence on density and electron temperature (Storey & Hummer 1995). For example, the ratio of the Br10 (1.73 μm) and the Brγ (2.16 μm) lines is predicted to be 0.33. Deviations from this value are caused by foreground extinction. We have converted the measured Br10/Brγ ratio to A_V using the Cardelli et al. (1989) extinction law. The derived A_V values for the six extracted regions are listed in the Column 7 of Table 1. The bulk of the H ii region has an A_V of around 5–6 mag (regions I, II, and IV). In region III, on the border with the molecular cloud, we measure an A_V of 15.9 mag, while toward the south the extinction declines to 4.4 mag and 1.5 mag for regions V and VI, respectively. This extinction gradient can be explained as a density gradient in the surrounding molecular cloud and is consistent with the presence of a champagne flow.

The (extinction corrected) ratio of the 1.70 μm He i line over Brγ is dependent on the temperature of the ionizing star and can be used to constrain its spectral type (Lumsden et al. 2003). The extinction correction line ratios for the six regions are given in the Column 6 of Table 1. The maximum ratio is measured just south of the location of the ionizing stars, and has a value of 0.047 ± 0.004 in region V and 0.06 ± 0.01 in region VI. Comparing this value to the predictions of Lumsden et al. (2003), an effective temperature for the central star around 35,000 K is expected, which corresponds to a spectral type between O8 V and O9 V (Martins et al. 2002), consistent with the optical spectral type determination of Heydari-Malayeri (1988).

3.2.2. PDR and Outflows

The IRAC detections of class 0 and I sources shows that stars are still forming close to the bright arc. An additional argument for active star formation is the presence of outflows, detectable by H₂ line emission (bottom panels of Figure 5).

The H₂ emission in the near-IR can be either thermally or non-thermally excited. The thermal emission originates from shocked neutral gas being heated up to a few 1000 K (outflows), while the non-thermal emission is due to fluorescence by non-ionizing UV photons between 912 and 1108 Å. These two mechanisms can be distinguished as they preferentially populate different energy levels giving rise to different line ratios. The fluorescence mechanism preferably excites the higher vibrational levels, while for thermal excitation the lower levels are preferred (Davis et al. 2003).

Figure 4 shows the location of the 2.12 μm H₂ emission (green) in relation to the emission from ionized gas (Brγ and He i). The H₂ emission has a very filamentary structure and appears just north of the ionized gas, co-located with the bright arc in the IRAC images. Additionally, two compact sources are visible in the northeast. To identify the excitation mechanism of the arc of filamentary emission (region (a)) as well as that of the two compact regions ((b) and (c), top left panel of Figure 6), we construct excitation diagrams for these regions.

Zoom In Zoom Out Reset image size

First, the line flux ratios are extinction corrected. The ratio of the 1–0 S(1) (2.12 μm) over the 1–0 Q(3) (2.423 μm) is insensitive to the ionization mechanism and has an expected value of 0.7 (Turner et al. 1977). Therefore, the observed ratios are used for an extinction estimate in the same manner as discussed in Section 3.2.1 (Table 2). Region (a) covers a large area, 262 arcsec² and shows a large-extinction variation. The derived properties are dominated by the strongest emission regions, located close to the H ii region where the extinction is the lowest.

Table 2. Physical Properties of the H₂ Gas

Region	Trot	N(H2)	AV	vlsr
	(K)	(cm−2)	(mag)	(km s−1)
a	3720 ± 90	3.6 ± 0.6 × 1017	9	+1 ±  14
b	2217 ± 90	1.8 ± 0.4 × 1018	13	−14 ± 14
c	1980 ± 115	4.6 ± 2.0 × 1018	20	−26 ± 16

Download table as:  ASCIITypeset image

Figure 6 also show the excitation diagrams of the three regions. In these diagrams, the measured column densities of the lines are plotted against the energy of the upper level (see Martin-Hernández et al. 2008, for a detailed description of this diagram). The total column density was calculated using the description of Zhang et al. (1998). Different vibrational levels are indicated with different symbols. Region (a) has the most lines detected as it covers the largest area (the complete arc). For the spectra of regions (b) and (c), the weaker lines seen in the spectrum of region (a) are not detected and 3σ upper limits are given instead. The solid line in the diagrams is a single temperature fit to all the data points, while the dashed line in the diagram of region (a) represents a linear fit to only the 1–0 S lines. For region (a), it is clear that the column densities are not represented by a single temperature gas, but that the line fluxes of the 2–1 and 3–2 vibrational levels are higher than expected from the 1–0 line fluxes. A likely excitation mechanism of the H₂ gas in the arc is fluorescence by non-ionizing UV photons. We have searched for spatial variations of the temperature and column density in region (a) by calculating the excitation diagrams for several sub-regions, but could not detect any significant deviations from the values of the total area.

The excitation diagrams of regions (b) and (c) are well represented by a single temperature. We note, however, that the 3–2 lines are not detected in either region, and the 2–1 lines are mostly upper limits. We conclude that the excitation diagrams of both regions are consistent with shock excited emission in an outflow. Their elongated shapes, especially for region (b), are also suggestive of outflow. The linear fits of the excitation diagrams also allow us to determine the column density and temperature of the emitting gas (Table 2). An additional constraint on the nature of the H₂ emission regions is the velocity of the different regions.

The obtained velocities are listed in the last column of Table 2. The two outflows have a clear blueshifted velocity compared to that of the extended PDR, with region (c) having the largest value of −26 ± 16 km s⁻¹. These relatively low velocities as well as the absence of [Fe ii] at 1.644 μm suggest that both shocks are C-type shocks (Stahler & Palla 2005). The more destructive shocks (J-type) would have strong [Fe ii] emission as well as larger shock velocities (Hollenbach & McKee 1989). The velocity measured in the arc (region (a)) of 1 ± 14 km s⁻¹ is more blueshifted than the velocity of the H ii region, but consistent with that of the Vela molecular cloud: 6 km s⁻¹.

For outflow (b), a driving source could not be identified in the IRAC images, most likely the source is deeply embedded and not visible at 8 μm. For source (c), however, we could identify a possible driving source. About 3'' southeast of the H₂ emission, a deeply embedded protostar is identified on the IRAC images with a very bright 4.5 μm flux (source No. 28 in Figure 2). The location of the source is shown in Figures 4 and 6 where the 3.6 μm contours of source No. 28 are overlaid on the SINFONI images. The excess 4.5 μm emission is most likely coming from H₂ emission closer to the source, too deeply embedded to detect at near-infrared wavelengths.

Summarizing, RCW 34 consists of three distinct regions. A large bubble is detected in the IRAC images, with a cluster of class II stars inside. At the northern edge of this bubble the H ii region is located, ionized by vdBH 25a (O8.5V). The detected north–south extinction gradient as well as the Hα velocity gradient found by Heydari-Malayeri (1988) suggest that the H ii region is a champagne flow.

In between the molecular cloud and the H ii region, a PDR is located traced by H₂ emission and 8 μm emission. The star formation inside the molecular cloud seems to be more recent as masers, class 0/I sources and outflows are present in that area.

The SINFONI H+K spectra of the three brightest cluster members (sources No. 1, No. 2, and No. 3) show Brγ absorption in the K band and Br10–14 absorption in the H band (Figure 8). Weaker lines of He i are also visible in H (1.70 μm) and K (2.058 and 2.11 μm). To obtain a spectral type estimate, we apply the K-band classification scheme for OB stars from Hanson et al. (1996). Their classification links a K-band spectral type to an optical spectral type. For late O and early B stars, the classification is based on He i (2.11 μm) and Brγ. For the H band, we used the classification of Hanson et al. (1998) and Blum et al. (1997) where the important lines are He i (1.700 μm) and Br11 (1.68 μm).

Zoom In Zoom Out Reset image size

Table 3 shows the measured equivalent widths (EW) of the relevant lines in the H- and K-band spectra of the three OB stars. In Column 6, the K-band spectral type is given, with the corresponding optical spectral type listed in Column 7. The brightest star, No. 1, has a spectral type between O8V and B1V based on the presence of He i (2.11 μm) and moderately strong Brγ. Star No. 2 has a much stronger Brγ absorption and is therefore classified as B1V–B2.5V. The fainter OB star (No. 3) does not have He i (2.11 μm) in its spectrum and is therefore classified as B3V–B7V. The EWs of the H-band lines are in agreement with the K-band spectral type. The only deviating value is the very strong Br11 absorption of No. 3, which is about equally strong as the Brγ line. The strength of Br11 is compatible with that of a supergiant.

The spectral type of star, No. 1, (O8V–B1V) is compatible with the optical classification as a O8.5V star by Heydari-Malayeri (1988). As the optical classification is more reliable than our near-IR classification, we will adopt a spectral type of O8.5V for No. 1 as the main ionizing star for RCW 34. This spectral type is in agreement with the Lyman continuum flux derived from radio continuum data (Caswell & Haynes 1987), and is also compatible with the He i and Brγ emission of the H ii region (Section 3.2.1). This corresponds to a mass of 20 M_☉ for star No. 1 (Martins et al. 2005).

The color of early-type stars is a good measure of the extinction. For the intrinsic colors, we used the values given by Koornneef (1983). We used the extinction law of Cardelli et al. (1989); the derived extinction toward the three stars is listed in Table 4 and ranges from A_V = 3.9 to 4.2 mag (A_K = 0.42 to 0.45 mag). The error in the A_V values reflects both the photometric errors and the uncertainties in (J−K)₀ due to uncertainties in their spectral classification. These values are in good agreement with those derived from the Br10/Brγ ratio of the H ii spectra (Section 3.2.1).

The spectral types and extinctions of the OB stars, coupled with their absolute magnitudes, allow a distance determination. For the O8.5V star No. 1, we obtain the absolute magnitude from Martins & Plez (2006) and Aller et al. (1982). This results in a distance of 2500 ± 200 pc for the O8.5V star. The uncertainty reflects the difference in the absolute magnitudes between Martins & Plez (2006) and Aller et al. (1982). Hanson et al. (1997) and Blum et al. (2000) provide absolute magnitudes for zero-age main sequence (ZAMS) stars. Their results are about 1 mag fainter than corresponding values for the main sequence of the Hertzsprung–Russell diagram (HRD). If we adopt the ZAMS magnitudes, we obtain a distance of 1800 ± 200 pc. The cluster age is probably a few megayears, hence we prefer the main-sequence values. Thus, we adopt 2500 ± 200 pc as the distance to RCW 34. This distance is slightly closer than the 2.9 kpc derived by Heydari-Malayeri (1988) using the same spectral type and optical magnitudes. The difference results from the different absolute magnitudes and extinction we apply.

Using the 2.5 kpc distance, we can refine the spectral type classification of the two B stars in the cluster. Using the distance modulus, the A_V, and the observed brightness, we derive an absolute K-band magnitude of M(K) = −2.1 mag for No. 2 and M(K) = −0.8 mag for No. 3. With the calibration of Aller et al. (1982) this results in a B0.5–B1 V spectral type for No. 2 and a B2–B3V spectral type for No. 3 (Table 4), still compatible with the original near-infrared spectroscopic classification. These spectral types correspond to a mass of 11 and 7 M_☉, respectively (Aller et al. 1982).

Besides the 3 OB stars, we found 21 stars showing absorption lines typical of later spectral type (Figure 9). The most prominent lines are the CO first overtone absorption bands between 2.29 and 2.45 μm, and absorption lines of Mg i and other atomic species in the H band as well as Ca i and Na i in the K band. To determine if these stars are cluster members, we need to obtain their spectral type and luminosity class.

Zoom In Zoom Out Reset image size

For the classification of the late-type stars, we use the reference spectra of Cushing et al. (2005) and Rayner et al. (2009). These spectra are taken with the IRTF telescope¹⁵ and cover the spectral types between F and M in all luminosity classes with a resolution between R = 2000 and 5000. These reference spectra were convolved and rebinned to the resolution of SINFONI. We corrected for the differences in spectral slope due to the extinction of our SINFONI targets using the extinction law of Cardelli et al. (1989). The atomic lines, like Mg i and Na i, are used for the temperature determination, while the CO and OH lines serve as a luminosity indicator.

Comparison of the SINFONI spectra with reference spectra indicates that the CO (2.29 μm) absorption is usually deeper than in dwarf reference spectra, but not as deep as in the giant spectra. In the few cases where a luminosity class IV reference spectrum was available, the spectrum provided a better match to the observed spectrum. This suggests that our late-type stars are low- and intermediate-mass pre-main-sequence (PMS) stars; indeed, Luhman (1999) and Winston et al. (2009) find that PMS spectra have a surface gravity intermediate between giant and dwarf spectra. If the stars were giants, dust veiling could make the K-band CO lines weaker. However, the H-band CO and OH lines are also much weaker, while the atomic lines have the expected EWs. Therefore, this seems to be a surface gravity effect, suggesting a PMS nature of our late-type stars.

We used the relation from Kenyon & Hartmann (1995) to convert the spectral type into effective temperature. This relation applies only for main-sequence stars; for PMS stars, a different relation may hold (Hillenbrand 1997; Winston et al. 2009). Cohen & Kuhi (1979) show that the temperature of PMS stars might be overestimated by values between 500 K (G stars) and 200 K (mid-K). We took this source of error into account when calculating the errors in the effective temperature.

Table 5 shows the result of the classification for all the late-type stars brighter than K = 14 mag. The spectral types range from early G to mid-K; about half of the stars have a surface gravity indicative of a PMS nature, their luminosity class is marked as "PMS." The stars classified as dwarfs are candidate foreground stars, but this does not exclude them as PMS cluster members, as Winston et al. (2009) also find PMS stars with surface gravity very close to the dwarf values.

With the knowledge of the spectral type, the observed color can be converted into an extinction estimate. The intrinsic J−K color for dwarfs and giants can differ as much as 0.4 mag for late K stars (Koornneef 1983). As the late-type stars are PMS stars, we used the mean of the color for dwarfs and giants as the intrinsic J−K color. The difference between the mean value and the dwarf and giant values is used as the error in the intrinsic color. The derived extinctions are listed in Table 6. Source No. 17 does not have a J-band magnitude as the photometry is too contaminated by source No. 1; it is excluded from further analysis.

All the stars with A_V> 12 mag (No. 6, No. 8, No. 16, and No. 24) are located north of the Brγ emission in the region where the H₂ and IRAC dust emission is present. Their extinction is consistent with being embedded in the molecular cloud north of the H ii region. The objects situated amidst the Brγ emission show a visual extinction between 4 and 10 mag, consistent with that measured from the nebular line spectrum (Section 3.2.1).

To confirm the PMS nature of the objects, we looked for Brγ emission from the source. We found two point sources in the continuum subtracted Brγ line map. One of the sources corresponds to No. 23, a G8 PMS star. Following Muzerolle et al. (1998), this Brγ emission could arise from accreting matter. The other Brγ emitting point source is No. 27 (Table 6). This star is rather faint (K =  15.5 ± 0.1 mag) and the continuum has too low S/N to detect any absorption features. The emission line spectrum also shows He i (2.058 μm) and the Brackett lines in the H band, identical to the surrounding H ii region (Figure 5). The Brγ emission from this object could arise from a disk being photo-evaporated by the main ionizing star. These "proplyds" have been found in Orion (O'Dell 2001) and other high-mass star-forming regions like M8 (Stecklum et al. 1998) and RCW 38 (DeRose et al. 2009). The spatial resolution of our data (∼06, 1500 AU at 2.5 kpc) did not allow us to resolve the proplyd, which is consistent with the sizes (up to 500 AU) measured for the Orion Proplyds (O'Dell 1998). The observed K-band magnitude and lower limit of source No. 27 does seem consistent with near-infrared observations of the Orion Proplyds (Rost et al. 2008).

In the spectra of, e.g., source No. 9 and No. 17, emission and absorption of Brγ is detected as well. These lines are artifacts from the Brγ removal using the median filtering. In the Brγ line image, no extra emission or absorption is present.

Two objects brighter than K = 14 mag are unclassified. These objects (No. 21 and No. 22) have very red spectra and a rather low S/N. We could not identify any features in their spectra. As the spectra are extremely red, they are likely dominated by dust emission and therefore the spectrum of the underlying star might not be visible. Like the late-type stars with high extinction, these featureless spectra are located in the high-extinction area, in the north of the SINFONI field and therefore are most likely embedded stars.

The remaining sources are too faint to obtain a proper classification. In the spectra of some of these objects, CO absorption at 2.29 μm can still be detected, but the spectra are of too low S/N to detect the atomic lines needed for an effective temperature classification. A handful of the objects have a J−K color around 1.0 mag (see Figure 7), indicating that these might be late-type foreground stars with very little extinction. Most of the faint objects are concentrated around J−K = 2.0 mag. This color is similar to that of the brighter PMS stars around A_V = 5–10 mag. The redder and more deeply embedded objects are not observable as they are too faint to detect in the J band. The objects with K>14 mag, clustered around J−K = 2 mag, are most likely PMS stars with masses below 1.5–2 M_☉.

In total, 17 sources (out of 26) have been detected in one or more Spitzer/IRAC bands with only three sources detected in all four bands. Most of the sources are only detected at 3.6 and 4.5 μm. At longer wavelengths, the high and variable background emission makes the detection of point sources very difficult. The objects detected in all four bands are labeled in Figure 2. One of the PMS stars (No. 11) is identified as a class II object. The other two objects (No. 14 and No. 21) are classified as class 0/I and will be discussed in Section 5.3.

To determine if the PMS stars have an infrared excess, we construct the spectral energy distribution (SED) and compare it with low-resolution Kurucz spectra (Kurucz 1993) for the 1–8 μm range. This allows us to determine the IR excess even for objects detected in only 1 or 2 IRAC bands. The Column 5 of Table 5 shows that nine PMS stars show a detectable IR excess (marked as a "+"), while for five sources no IR excess is detected (marked as a "–"). The remaining sources are marked with an "n": no IRAC counterpart could be detected due to blending with other sources or spatial coincidence with the bright, extended arc emission. The infrared excess is already present at 3.6 μm with an excess of more than a magnitude in some cases. The detected IR excess of at least nine of the PMS star candidates reinforces the hypothesis that these stars are young and still surrounded by a circumstellar disk.

The SINFONI spectra, however, suggest that the near-infrared excess of most PMS stars is low, as all these stars have absorption lines in their K-band spectra. Should a strong IR excess be present these lines would not have been detected. The K-band spectrum of No. 11 suggests that some veiling might be present; the CO was still stronger than that of the best-fitting dwarf reference spectrum, but the Na i and Ca i lines in the K band were very faint. In the spectra of the two sources where no absorption features have been detected (No. 21 and No. 22), a NIR excess is most likely present. These sources also have very red colors and a strong IRAC excess.

To discuss the stellar content in detail, we first need to separate the cluster members from the fore- and background stars. The O8.5V star (source No. 1) ionizes the H ii region and is also responsible for the zone of interaction with the molecular cloud to the north. Analysis of the brightness and spectral type of the B stars (sources No. 2 and No. 3) shows that they are located at the same distance (2.5 kpc) and have a similar extinction as the main ionizing source and therefore are considered to be cluster members.

It is less obvious that all the late-type stars are cluster members as well. As discussed in Section 4.2, most of the late-type stars have a surface gravity typical for PMS stars. Additionally, about half of the stars have an IR excess, supporting the result from the spectroscopy that those stars are PMS stars surrounded by an accretion disk. The extinction derived for these stars is consistent with the extinction derived using the H ii region and the H₂ emission.

Sources No. 13 and No. 15 show no evidence for a PMS nature or for an IR excess, but have higher visual extinctions (5.6 and 7 mag, respectively) than observed from the H ii emission in that area. Their early spectral types makes it difficult to derive a luminosity class because of the CO bands are not present. The high extinction suggests that they are background stars, but then the stars should be (super) giants as dwarfs are too faint to be seen as background stars.

Two more stars have A_V values inconsistent with their location: sources No. 14 and No. 26 show very low extinction (0.25 and 1.8 mag, respectively) while they are located toward the high-extinction area north of the H ii region. Star No. 14 has virtually no extinction and a surface gravity typical for main-sequence stars, however, it coincides with an IRAC source having colors of a class 0/I object. Also, on the Brγ line map some faint Brγ emission is seen surrounding the star. The fact that this star has no extinction is inconsistent with its IRAC colors and a protostar nature. If the star were a protostar, it would be more deeply embedded and have a much larger A_V. It is most likely a chance alignment of a foreground late G star (as seen in the NIR) with a deeply embedded protostar only seen at mid-IR wavelengths. Additionally, star No. 26 (A_V = 1.8 mag), also located toward the high-extinction area, does not show any PMS characteristic nor an IR excess and also is most likely a foreground star. The locations of the stars in the SINFONI field are shown in Figure 10. The different classes of objects do not cover a preferred area of the SINFONI field, but rather are spread across the entire field.

Zoom In Zoom Out Reset image size

Summarizing, after analysis of their location, reddening and spectral features, we conclude that most of the stars are probably cluster members, with two exceptions where the stars are probably foreground stars (No. 14 and No. 26). The nature of sources No. 13 and No. 15 is unclear, but their high extinction suggests that they are related to the cluster.

To constrain the mass of the PMS stars and the age of RCW 34, we construct an HRD to compare the observed parameters with PMS evolutionary tracks. The derived extinctions allow us to de-redden the K-band magnitude, convert to absolute magnitude with the derived temperature, and plot the points in a K versus log(T_eff) diagram (Figure 11, left panel). We exclude the foreground stars No. 14 and No. 26 to enable a comparison with the isochrones. The symbols reflect the different spectral types. Filled symbols indicate that a Spitzer/IRAC excess emission was detected.

Zoom In Zoom Out Reset image size

Overplotted as a dotted line is the 2 Myr main-sequence isochrone taken from Lejeune & Schaerer (2001). The 2 Myr isochrone is chosen to illustrate the main sequence; 1 Myr or 5 Myr isochrones would have served equally well. A 20 M_☉ star does not evolve significantly off the main sequence for the first 8 Myr of its life (Lejeune & Schaerer 2001), therefore we adopt 8 Myr as an upper limit for the age of RCW 34.

The solid lines in the left panel represent the evolutionary tracks taken from Da Rio et al. (2009) using the models of Siess et al. (2000) for four different masses; 4, 3, 2, and 1.5 M_☉. Comparison of the location of the PMS stars with these evolutionary tracks yields an approximate mass varying from ∼3 M_☉ for the brightest stars to 1.5 M_☉.

Overplotted in the right panel of Figure 11 are the PMS isochrones (solid lines) between 0.1 and 10 Myr. The location of the G- and K stars is consistent with a range of ages. The brightest object (No. 6) lies between the 0.3 and 0.5 Myr isochrones; however, the majority of the objects span of the 1–3 Myr isochrones, with the more massive objects closer to the 1 Myr isochrone. For the less massive objects, the spread in age is larger, most likely due to higher uncertainties in the spectral type determination. The location of the PMS stars suggests an age of 2 ± 1 Myr.

Comparison of the location of the PMS stars in the HRD with those of the Herbig AeBe stars (van den Ancker et al. 1998) shows that the PMS stars are younger than the Herbig AeBe stars and will evolve from their present G- and K-spectral type to late B, A, or early F spectral type when they become main-sequence stars. This also explains the fact that no A or F stars are spectroscopically detected, as they are still in the PMS phase at cooler temperatures.

Even though the PMS stars are younger than the Herbig AeBe stars, they have, in contrast to the Herbig stars (e.g., Eiroa et al. 2001), no K-band excess. This could be the result of a different disk structure, but also caused by the fact that the PMS stars are emitting most of their energy in the near-infrared wavelength regime, where the SEDs of Herbig AeBe stars peak in the optical, thereby outshining the contribution of the circumstellar disk.

The extinction measurements show that the sources north of the H ii region are more deeply embedded in the molecular cloud. A younger age for this region is suggested by the presence of several protostars as well as outflows and a dense molecular cloud, which would suggest a younger age for the PMS object in this area as well. The PMS stars north of the H ii region (No. 6, No. 8, No. 16, and No. 24) all have the brightest absolute magnitude in their temperature bin with source No. 6 as an extreme example. Their position in the HRD could suggest that they are the youngest of the stars of given spectral type. However, their high extinction leads to a selection effect (only the brightest sources are detected). Fainter sources would fall below the limit where SINFONI spectra could be classified. So, if there were an age difference, the current data would not be able to quantify this. Deeper spectroscopic observations to find more of the high-extinction PMS stars north of the H ii region would be required to make any definitive statements.

As the SINFONI observations show, large-extinction variations are present in RCW 34. The high extinction toward the north means that we miss sources that are too deeply embedded to be visible in K band. In our IRAC catalog, we found two sources with no near-infrared counterpart (No. 28 and No. 29 in Table 6). Both sources have IRAC colors of deeply embedded protostars (class 0/I).

Source No. 28 is located in the northeast corner of the SINFONI field, only 3'' away from one of the outflows detected in the H₂ line emission (outflow (c), Section 3.2.2). The other embedded source (No. 29) is located very close to source No. 1, but is not related. Thomas et al. (1973) and Williams et al. (1977) detected No. 29 as a mid-infrared excess toward vdBH 25a. The IRAC images show, however, that it is a separate source. The SED of No. 29 is difficult to determine; at 3.6 μm, the contribution of the O8.5V star is still present, and at 5.8 and 8.0 μm, this object is the brightest source at MIR wavelengths. To determine a better SED, high-resolution and deeper mid-infrared observations are needed to discriminate between the emission from the protostar and the arc in the north. Additionally, as discussed above, the IRAC counterpart of source No. 14 could be another embedded protostar not seen at NIR wavelengths.

Outside the SINFONI field, several other protostars are located. Some of these objects seem to have a K-band counterpart in the SOFI image of Bik (2004), suggesting that the objects in our SINFONI field are even more embedded than those located a bit further away from the arc.

IRAC imaging reveals a large bubble surrounded by extended 8 μm emission. Inside this bubble is a cluster of intermediate- and low-mass class II objects. Typical ages for class II objects are of the order of a few megayears (Kenyon & Hartmann 1995; White et al. 2007), not distinguishable from the age of the stars in the H ii region.

The class 0/I stars, identified mainly close to the bright arc at the northern edge of the H ii region, might be significantly younger. Whitney et al. (2003) suggest an age of ∼10⁵ yr for class I sources; however, White et al. (2007) present evidence that the low-mass class I objects in Taurus gave a similar age as the class II sources. The location of the class 0/I sources coincides with methanol masers, CS emission, high extinction and outflows, and therefore are most likely younger than the objects inside the bubble and the H ii region.

Comparing the 2MASS K-band magnitudes of the class II stars in the bubble with the K-band magnitudes of the PMS stars in the H ii region shows that 13 out of 17 class II objects are about a magnitude brighter. The other four are of similar magnitude. Additionally, in a (J−H) versus (H−K) color–color plot, most of the class II sources have a near-infrared excess and cover the same region in the color–color diagram as the Herbig Ae stars (Eiroa et al. 2001).

The difference in K-band magnitude could be explained with the presence of a K-band excess, suggesting that the class II sources are more similar to the Herbig Ae stars and are therefore more evolved than the PMS stars. An optical spectroscopic classification is needed to confirm this.

The location of the central star of the H ii region, near the northern edge of the bubble suggests that this source might not have created the bubble. Also the Hα (Parker & Phillipps 1998), as well as the Brγ emission, peaks at the location of the O star. In the rest of the bubble only faint Hα emission is present, indicating the absence of massive stars inside the bubble. Apart from the class II sources detected in the bubble, one source in the center (Figure 3) is detected without infrared excess. The magnitude at 3.6 μm (10.0 ± 0.02 mag) is compatible with a B0V star at the distance of RCW 34. However, the optical photometry of the source taken from the USNO catalog (Monet 1998) is not compatible with a (reddened) early-type star. Therefore, we conclude that this star is an unrelated, foreground object. This raises the question whether other mechanisms could be responsible for the bubble. The scenario that the bubble would be created by a supernova seems unlikely. As the stars in the bubble are only a few megayears old, the supernova needs to come from an early-type O star. Assuming a normal IMF, the other O stars still would have to be visible.

Massive stars affect their environment by ionizing photons and stellar winds. Estimating the mechanical luminosity (E_mech = 1/2 Mv²) needed to blow a bubble with a radius of 2.5 pc inside a medium with a typical molecular cloud density (10³ cm⁻³), expanding at 10 km s⁻¹ would be on the order of 10⁴² erg s⁻¹. As a comparison, the wind luminosity of a mid-B star is around 10³⁰ erg s⁻¹ (Tan 2004).

Assuming ∼10⁴⁴ ionizing photons s⁻¹, typical for a mid-B star and a density of 10³ cm⁻³, the Strömgren radius is 0.014 pc. This is a factor 100 lower than the observed radius of the bubble (2.5 pc). Assuming that the 20 observed class II stars all emit the same number of ionizing photons and assuming a density of 100 cm⁻³, the Strömgren radius would be comparable to that of the measured bubble size.

Another possibility is that the OB star, which created the bubble, has been expelled from the center of the cluster due to a binary interaction, similar to what is suggested for the origin of BN in Orion (Rodríguez et al. 2005). The central star of the H ii region could be one of the stars from the disrupted binary, with the other star moving in the opposite direction. With the current data we cannot exclude this scenario, detailed velocity measurements of the star in the H ii region could give more insights into this scenario.