Free Access
Issue
A&A
Volume 555, July 2013
Article Number A60
Number of page(s) 21
Section Catalogs and data
DOI https://doi.org/10.1051/0004-6361/201220474
Published online 28 June 2013

© ESO, 2013

1. Introduction

Giant H ii regions (GHRs) are known as the site of massive young clusters which are rich in massive stars. The strong stellar winds and the evolution of these massive stars can disrupt the interstellar medium of these GHRs, resulting in the formation of expanding superbubbles. In the simplest scenario, the creation of superbubbles can be done by scaling the kinetic energy luminosity from the single-star models of Weaver et al. (1977). These models assume that shocked stellar winds dominate the dynamics of these bubbles. However, analysis of several superbubbles created by OB associations in the Large Magellanic Cloud (LMC; Oey 1996) and some extragalactic GHRs (Maíz-Apellániz & Walborn 2001; MacKenty et al. 2000) reveals that such a model can not explain the observations. Recently, Lopez et al. (2011) studied the different feedback processes that are taking place in the 30 Doradus region in the LMC. They suggested that stellar feedback is the main process responsible in shaping the expansion of this GHR. In a different approach, Pellegrini et al. (2011) found that the large-scale structure of 30 Dor is dominated by a system of X-ray bubbles, which are in equilibrium between them. In spite of several studies to investigate the origin of bubbles in GHRs, there is no consensus as to the mechanisms responsible, even in the well-studied case of 30 Dor.

A related, unsolved problem regarding GHRs is the origin of supersonic velocities seen in spatially-integrated nebular profiles. Given that supersonic motions should rapidly be dissipated, an energy source is needed to maintain them. Three possible sources exist: kinetic energy input from stellar winds and supernova explosions (SN), gravity, and photoionization. All of them can supply enough power to maintain the supersonic velocity dispersion but it is not clear which one(s) is/are responsible (Chu & Kennicutt 1994; Tenorio-Tagle et al. 1993, 1996; Melnick et al. 1999). High-resolution spectroscopic data of nearby GHRs is necessary to disentangle these phenomena through the study of the kinematics of the warm gas. In this context, 30 Dor is one of the closest and most appropriate targets for undertaking a more detailed study.

The 30 Dor nebula is the largest H ii region in the Local Group and the most powerful source of Hα emission in the LMC. At its core is the young massive cluster R136, host to a very large concentration of massive hot and luminous stars (see Crowther et al. 2010). Given all the information that can be extracted from 30 Dor, this object has been called the starburst “Rosetta Stone” (Walborn 1991). Due to its nature, 30 Dor has been the target of a range of multiwavelength studies (infrared, optical, ultraviolet and X-ray data) to try to disentangle its structure – both in terms of its stellar populations and the distribution of gas and dust. For instance, Bosch et al. (2009) searched for massive binary stars in the ionizing cluster of 30 Dor, finding a binary candidate rate of 50%. These authors also derived the binary-corrected, virial mass of the cluster, which corresponds to 4.5 × 105   M, suggesting that it could be a candidate for a future globular cluster. More recently, the VLT-FLAMES Tarantula Survey (VTFS, Evans et al. 2011) has opened a new window in the study of massive binary systems in 30 Dor (Sana et al. 2013), via the large number (over 800) of stars for which there are now high-quality spectroscopic data.

Table 1

VLT-FLAMES/MEDUSA observing log for the regular nebular grid.

From imaging with the Hubble Space Telescope, Walborn et al. (2002) found some wind-blown cavities and several filamentary structures. While at X-ray wavelengths, Wang (1999) and Townsley et al. (2006a,b) found several X-ray bubbles and point sources in 30 Dor. In the case of the diffuse X-ray structures, which are spatially associated with high-velocity, optical emission-line clouds (Chu & Kennicutt 1994), Townsley et al. (2006a) found that not every high-velocity feature displays bright X-ray emission. In this sense, optical kinematical analysis of the bright X-ray regions is needed in order to understand this phenomenon.

The kinematics of the warm ionized gas in 30 Dor has also been well studied. For example, Smith & Weedman (1972) used a Fabry-Perot instrument to map the [Oiii] λ5007 emission line, and suggested that the fast internal motions in the region are produced by the winds of WR stars. Chu & Kennicutt (1994) and Melnick et al. (1999) both studied the kinematics of the ionized gas in 30 Dor, using echelle and long-slit spectroscopy, respectively. Chu & Kennicutt (1994) found that 30 Dor displays very complex kinematics, with several fast expanding shells (that are coincident with extended X-ray sources) that can not be explained by stellar-wind models, and they suggested that SN remnants can solve that problem. Melnick et al. (1999) found complex Hα profiles in several regions of 30 Dor. They used their observations to study the mechanisms responsible for the line-broadening observed in the emission lines of H ii regions and found a low-intensity broad component that explains the wings in the integrated profile of 30 Dor. Lastly, Redman et al. (2003) used echelle spectroscopy to study the gas kinematics in the outskirts of 30 Dor. These authors found some high-speed features, that they interpreted as shells formed by stellar winds and SNe.

These studies have each provided useful insights about the kinematics of 30 Dor, but the origin of superbubbles and the supersonic velocity dispersions in GHRs are still open questions. In fact, Melnick et al. (1999) suggested that it would be crucial to study the complete 30 Dor nebula at high spatial resolution. Thus, to address these problems, we have obtained high-resolution spectroscopy of 30 Dor, using the FLAMES-Giraffe spectrograph on the Very Large Telescope (VLT) at the Paranal Observatory. Preliminary results from this programme have been published by Torres-Flores et al. (2011).

In this paper we present the data and discuss the general kinematics of 30 Dor, based on analysis of its nebular emission-lines. In future papers we will present analysis of: i) the width of the different emission lines and their contribution to the integrated emission-line profile of 30 Dor; ii) the kinematics of specific regions; iii) the wings detected in the integrated Hα profile of 30 Dor; iv) the connection between the kinematics of the ionized warm gas and the X-ray emission in this GHR. This paper is organized as follows. In Sect. 2 we present the data and the data-reduction process. In Sect. 3 we present the data analysis. In Sect. 4 we show the general results obtained with the current data set. Finally, in Sect. 5 we summarize our findings.

2. Observations

2.1. Data sets and fiber positioning

To study the detailed kinematics of the ionized gas in 30 Dor we used the MEDUSA fiber mode of the VLT FLAMES-Giraffe spectrograph (Pasquini et al. 2002). Given the proximity of 30 Dor, FLAMES allows us to map, at high spectral and spatial resolution and high signal-to-noise, the main nebular emission lines from the gas, i.e., Hα and [Nii] at 6548 and 6584 Å. In this regard, the multiplex of the MEDUSA observing mode enables us to sample a large number of points simultaneously across a relatively wide field-of-view; crucial when studying an extended region like 30 Dor. Given the complex morphology of 30 Dor, three different data sets are used to analyse the its kinematic features: a regular nebular grid, an irregular nebular grid (where the fibers were located in the brightest regions of 30 Dor), and finally a stellar grid. In the following we describe each data set in detail.

2.1.1. The regular grid

The observations of the regular nebular grid were carried out on December 3, 2003, using the HR14A Giraffe setting. This provided spectral coverage ranging from 6300 to 6691 Å, with a notional spectral resolving power of R = 17   740 at the central wavelength. We covered a field-of-view of 10′ × 10′ centered on R136. Given that we were limited by the MEDUSA fiber separations, we used three different fiber configurations in offset positions to adequately sample our field-of-view; a total of 10 fields were observed, as summarised in Table 1. These fields give a regular grid of 32  ×  30 positions with a spatial sampling of 20′′, as shown in the top panel of Fig. 1, where the position of each fiber is indicated by a red circle. The aperture of each MEDUSA fiber corresponds to 1.2′′ on the sky (~0.3 pc).

Our configurations did not cover some positions given that two MEDUSA fibers were broken, producing some voids in the spatial sampling (see Fig. 1). Out of 960 fiber positions, 49 positions in the grid were not observed. We used nine fibers of the irregular grid to cover some of these missing positions (fibers nos. 11, 22, 23, 28, 31, 79, 100, 102 and 112). The positions of these fibers are indicated by blue numbered circles in the top panel of Fig. 1. In the case of fibers 28 and 31, we calculated an average of the two spectra and replaced this value at the corresponding missing position. The remaining 41 missing positions were filled with the average of the eight closest (and spatially linked) spectra. In this sense, we caution the reader that the spectra visible in the positions of the missing fibers do not represent the real emission at these locations (see Fig. 1).

thumbnail Fig. 1

Top panel: GIRAFFE/MEDUSA regular grid overlaid on an optical Digital Sky Survey image of 30 Dor. The spatial separation between each fiber corresponds to 20′′ (~5 pc). Middle panel: MEDUSA irregular grid. Bottom panel: MEDUSA stellar grid (VFTS survey). The field-of-view on the sky shown in the lower two panels is larger than in the top panel to show all the fibers available. The red rectangle in the middle panel indicates the region covered by the regular grid.

Open with DEXTER

The centers of the different configurations, setups, and exposure times used to observe 30 Dor are listed in Table 1. Given the high surface-brightness displayed by 30 Dor, two sets of observations were taken to avoid saturation effects. In the first instance we took three exposures of 60 s and then we took two exposures of 600 s. At the end, we co-added the two observations of 600 s. The fiber positions for the regular grid are listed in Table A.1; when a fiber was taken from the irregular grid (as described above), it is labelled as “IG”.

Thus, from the regular grid we have obtained 911 equally-spaced spectra across an area of 10′ × 10′ centered on R136. With that data in-hand, we sorted all the spectra in right ascension and declination. Given that each fiber gave us spectral information for each equally-spaced point, the regular grid enabled us to produce a high-resolution spectroscopic datacube of 30 Dor, spanning 6300 to 6691 Å. This datacube can provide us spatial and spectral information simultaneously, where the spatial information depends on the fiber positions and the spectral information depends on the wavelengths covered by the spectra. For instance, a slice of this datacube centered at Hα can give us a monochromatic image of 30 Dor, albeit limited in spatial resolution to 20′′ by our fiber separation. Despite this fact, the datacube give us an excellent means by which to investigate the kinematics of the large expanding structures in 30 Dor. To illustrate the spectral coverage and primary nebular emission features in the spectra, the integrated spectrum (summed from all spectra in the datacube) is shown in Fig. 2. Given that the most intense emission lines in Fig. 2 correspond to Hα and [Nii] λ6584, we have derived two sub-datacubes centered in these lines, where these sub-cubes cover a range of 500 km s-1 in velocity.

In this paper we focus on analysis of the regular grid, but we now briefly introduce the other two data sets as they will be employed in the future studies.

2.1.2. The irregular grid

Additional observations were obtained in the same observing run as the regular grid. This “irregular grid” of MEDUSA positions was selected to observe the brightest nebular regions of 30 Dor, which can be associated with photodissociation regions which lie between the bright nebular and molecular phases. The same high-resolution setup (i.e. HR14A) was used as the regular grid. To reach a high signal-to-noise ratio, nine exposures (each of 600 s) were obtained with this configuration; the positions are shown in the middle panel of Fig. 1 (red circles). The red rectangle indicates the field-of-view covered by the regular nebular grid; as can be seen most of the fibers were located across the filamentary structure of 30 Dor. Each fibre position is listed in Table A.2.

2.1.3. The stellar grid

The observations of the stellar grid were obtained as part of the VFTS (Evans et al. 2011). This multi-epoch spectral survey aims to detect massive binary systems and determine their nature and evolution. The VFTS also aims to study the properties of stellar winds and rotational mixing in O-type stars. To carry out these studies, the survey has observed 800 massive stars, as shown in the bottom panel of Fig. 1. Of relevance here, the red-optical VFTS data were observed with the HR15N Giraffe setup, which has a spectral coverage ranging from 6442 to 6817 Å, at a spectral resolving power of R = 16   000. The HR15N setup provides the additional nebular diagnostic of the [Sii] lines at λλ6716, 6731 Å, which are not covered by the HR14A observations. Given the large spatial coverage of the VFTS survey, these data can provide additional insights in the study of the kinematics of 30 Dor.

2.2. Data reduction

Reduction of the 2003 data was performed with the ESO pipeline GASGANO and EsoRex software. We observed three bias and flats, which were combined to correct our observations. The wavelength calibration was performed by using the ThAr calibration lamp, from which the instrumental resolution was measured to have a FWHM of 0.4 Å, which corresponds to R = 16   400 at Hα. All the spectra were corrected to the heliocentric rest frame.

We have compared our processed data with the data available in the Giraffe archive1. By visual inspection, the two reductions are in good agreement. Cosmic rays were removed using the task crreject in iraf. Once the data were bias/flat-field corrected and wavelength calibrated, we removed the continuum emission present in the spectra by fitting a polynomial (using the iraf continuum task). In a few cases in the regular grid, the fibers lie close to some stars; for these we use a high-order polynomial to remove the continuum emission. To remove the sky emission lines, we fit Gaussians to one of the lowest intensity spectra. This allowed us to create a template of sky lines, which were removed from all the spectra.

3. Analysis

Past efforts to understand the internal kinematics of H ii regions have employed profile widths of their emission lines as a diagnostic (e.g., Chu & Kennicutt 1994). Following a similar approach, we use the Hα line from our observations to characterize the kinematics of the gas, as it presents the highest signal-to-noise ratio and can be easily resolved into different components in our data. To obtain the line widths we fitted Gaussians automatically to each observed Hα profile. In several cases a single Gaussian does not represent the real shape of the multiple profiles that can be seen in this star-forming region (e.g., Melnick et al. 1999). However, this exercise provides some insight about the general behaviour of the Hα emission.

We used the mpfit package in idl (Markwardt 2009) to obtain the center, peak intensity and the observed width (σobs, uncorrected by instrumental and thermal broadening) of each emission line. Thereafter, where relevant, we fitted multiple Gaussians to the observed profiles using the pan package2, which enabled us to select the number of components for each observed profile3. This kind of analysis is necessary given the complexity of the emission lines in 30 Dor. For example, Melnick et al. (1999) showed that more than three Gaussian components are needed to reproduce some of the observed profiles. We note that this kind of exercise is (to some degree) arbitrary, given that in several cases we can not resolve two superposed Hα profiles (which lie at the same radial velocity, for example), which we would fit with just one Gaussian.

thumbnail Fig. 2

Integrated nebular spectrum of 30 Dor obtained from summing all of the spectra from the regular grid. The most intense emission line is Hα, but the intensity axis in the plot is truncated to show the low-intensity emission lines.

Open with DEXTER

thumbnail Fig. 3

ESO/La Silla Hα image of 30 Dor (taken with the EMMI instrument), on which we have over-plotted the Hα profile from each fiber of the regular nebular grid. Each small box is 20′′  ×  20′′ (5 pc  ×  5 pc), with the fibers located at the center of each box. Thick boxes indicate the positions for which we have interpolated the spectra (see Sect. 2.1.1).

Open with DEXTER

The width of the observed Hα profiles take into account the instrumental (σin) and thermal (σth) widths. The former value depends on the resolution of the instrument, while the latter depends on the thermal motions of the Hydrogen. To obtain a corrected value for the width of the Hα emission line, we have subtracted σin and σth from σobs as follows: , where σ represents the true width of the Hα line. As discussed in Sect. 2.2, σin was derived from the calibration lamp exposures and has a value of σin = 7.8 km s-1. For σth we assume hydrogen gas at an electronic temperature of 104 K, from which σth ~ 9.1 km s-1.

thumbnail Fig. 4

Velocity-sliced view of the 30 Dor datacube centered on the Hα line. The velocity with respect to the center of the Hα line is shown in the upper left-hand corner of each panel (negative and positive velocities imply radially approaching and receding velocities, respectively). In this figure, a logarithmic scale is used for the intensity of each map.

Open with DEXTER

4. Results

4.1. General view of the 30 Dor spectroscopic datacube

To visualize the complex morphology of the ionized emission in 30 Dor, in Fig. 3 we have plotted the Hα profile of each fiber from the regular grid (normalized by the peak of the intensity in each case) over an Hα image obtained with the EMMI instrument at the La Silla Observatory4.

In Fig. 3 it is possible to correlate the high-resolution spectral information with features seen in the image. The Hα profiles in many instances are extremely complex. For example, in the cavity located 1′ to the east of R136, the Hα profiles clearly display multiple components. In fact, the fiber located at the position α = 05h38m53.21sδ = −69°05′59.89′′ (J2000) displays at least five components (see the last panel of Fig. 7). In general, most of the multiple Hα profiles are located in the regions where expanding shells have been previously identified (see Chu & Kennicutt 1994). On the other hand, Fig. 3 shows that the most intense Hα emitting regions in 30 Dor (based on its EMMI image) display simple and narrow profiles. In several cases, these bright regions are associated with photodissociation regions (PDRs), which can halt some expanding structures, producing simple profiles. In addition, Fig. 3 shows that some low-intensity regions display simple profiles (for example, in the north-east region), which suggests that no expanding structures lie at that location. By inspecting Fig. 3 it is also possible to identify low-/high-velocity components. For instance, at α = 05h37m49.66sδ = −69°02′39.91′′ we find a low-intensity component having an approaching velocity larger than 200 km s-1 compared to the systemic value.

thumbnail Fig. 5

Velocity-sliced view of the 30 Dor datacube centered on the [Nii] λ6584 line. The velocity with respect to the center of the [Nii] λ6584 line is shown in the upper left-hand corner of each panel (negative and positive velocities imply radially approaching and receding velocities, respectively). In this figure, a logarithmic scale is used for the intensity of each map.

Open with DEXTER

Given the kinematic richness of 30 Dor, we have derived a velocity-sliced view of the datacube, centered on the Hα and [Nii] λ6584 lines. In Fig. 4 we show 20 frames of the datacube. The systemic (zero) velocity was defined by the Gaussian fit to the integrated Hα line (i.e. from Fig. 2), in which the line-center was 6568.65 Å, equivalent to vr = 267.4 km s-1 (where the rest wavelength of the Hα emission line was taken from Hirata & Horaguchi 1995) and the velocity interval between each frame was 25.1 km s-1. In this case, negative and positive velocities imply radially approaching and receding velocities, respectively.

By inspecting the first four frames of Fig. 4, we can see several high-velocity components with respect to the mean velocity of 30 Dor, especially in the neighborhood of the Hodge 301 cluster (the high velocity components in the first and second frames). These components reach approaching velocities larger than 230 km s-1. By inspecting the datacube we found several low-intensity components that present velocities even larger than 250 km s-1. For instance, Redman et al. (2003) found several discrete high-speeds knots in 30 Dor, which have velocities of  ± 200 km s-1. In the last frames of Fig. 4 we also note several high-velocity components – some of these can be associated with the expanding structure #5 from Chu & Kennicutt (1994, their Fig. 1c).

Another interesting feature that appears in Fig. 4 is the detection of shell-like structures which are visible across all the observed field of 30 Dor (see Sect. 4.4). These structures appear as small voids that increase in size as we move from the blue to the red side of the Hα line. It is interesting to note that bluewards of the systemic Hα emission the shells are clearly defined, while they are not seen in the frames redwards of the central velocity. Internal extinction produced by dust can produce this signature. Also, molecular clouds located at these positions could hide the red side of the Hα emission.

In Fig. 5, we show a similar velocity-sliced view, but now centered on the [Nii] λ6584 emission line. As in the previous case, the systemic (zero) velocity was derived from the Gaussian fit to the integrated [Nii] λ6584 line, which gives a radial velocity of vr = 266.1 km s-1 (6589.29 Å). As to be expected from the reduced intensity of the [Nii] line compared to Hα, most of the emission is detected at lower expanding velocities than seen in the Hα map. While there is a clear correlation between the morphology of the shell-like structures detected in both lines, inspection of Figs. 4 and 5 shows that several Hα and [Nii] λ6584 high-velocity components do not correlate spatially, which suggests that different physical processes are exciting the gas at these locations. For example, in Fig. 6 we show a high-velocity component in the [Nii] λ6584 line (black solid line) which presents a strong intensity with respect to their Hα counterpart (α = 05h38m41.97sδ = −69°04′40.04′′). In the same figure, we see the spectrum of the next observed fiber (red dashed line), which present a high-velocity component in Hα but with no [Nii] λ6584 emission (α = 05h38m41.96sδ = −69°05′00.03′′). We note that in this comparison we are using two adjacent observed fibers. These points all help to illustrate the capabilities of these data in disentangling the complex structures of 30 Dor.

4.2. Complexity of the Hα profiles in 30 Doradus

4.2.1. From single to multiple profiles

The wide variety of Hα profiles in the data are illustrated by the examples in Fig. 7, which show the four narrowest and broadest profiles (upper and lower panels, respectively). All these profiles correspond to those observed and not to the averaged profiles described in Sect. 2.1.1. A single Gaussian was fitted to each observed profile (as described in Sect. 3), with the coordinates of the fiber position and σobs derived from the fitting process shown in the upper right of each panel. In this analysis, the center of the Gaussian fit gives the overall systemic radial velocity at the position of the fiber, while the width of this fit gives the velocity dispersion of the ionized gas, which must be corrected by the instrumental and thermal widths.

The narrowest Hα profile has a σobs = 14.3 km s-1 (upper left-hand panel in Fig. 7), which corresponds to σ = 7.8 km s-1 once corrected for σth and σin. This profile is located in the south-west of the nebula, which is ionized by the LH99 stellar association (Lucke & Hodge 1970) and is also the location of the SN remnant 30 Dor B (Danziger et al. 1981). Two low-intensity components can also be seen in the profile, and these could be associated with an expanding structure. Nonetheless, the main structure of this profile is symmetric and can be well-fitted by a single Gaussian (red dashed line in Fig. 7). The other three narrow Hα profiles shown in Fig. 7 display a few low-intensity components, but the main part of the profiles are also well-fitted by a single Gaussian.

In contrast, the broad profiles shown in Fig. 7 clearly display multiple strong components. As expected, these profiles can not be fit by a single Gaussian and their measured widths are the result of a coarse fit that takes into account all the components. All of these broad, multiple profiles lie in shell-like structures. It is interesting to note that the third broad profile shown in Fig. 7 lies just 20′′ (equivalent to  ~5 pc) from the narrow profile located at α = 05h38m56.945sδ = −69°06′19.940′′. This shows the extremely complex structure of 30 Dor even at small spatial scales.

4.2.2. Multiple Gaussian fits: a few examples

To determine the line-broadening mechanism in 30 Dor, Melnick et al. (1999) fitted multiple Gaussian components to the observed Hα profiles of different regions. They found that a broad, low-intensity component was necessary to reproduce the wings of all the observed Hα profiles. As noted earlier, we have performed multiple Gaussian fits to a subset of profiles using the pan package. In particular, we have examined the two narrowest and broadest Hα profiles from single-component fits on the FLAMES data (see Fig. 7), with results of these fits shown in Fig. 8. In each panel of the figure the red (dashed) lines indicate the different Gaussian components used in the fits; the coordinates of the fiber position and the σobs of each fitted component are given in the upper right of the panels. In Fig. 8 we also plot the residuals of the fits, i.e. the difference between the fitted and the observed profile, in which the intensity axis was limited to  ± 5% of the peak of the observed profile.

For the narrowest Hα profile in the data (upper left in Fig. 8), we have fitted three Gaussian components. The two low-intensity components detected at this position present widths smaller than σobs ~ 26 km s-1. The strongest component at this position has a corrected width of σ = 6.1 km s-1, i.e., lower than the value obtained by fitting just one Gaussian (σ = 7.8 km s-1, from Fig. 7). By inspecting the residual of these fits, no low-intensity broad component is necessary to explain the observed profile (the residuals are negligible compared with the observed emission). This is of wider interest as this Hα profile is located in 30 Dor B, associated with a SN remnant (where we can expect a complex kinematic). In the case of the second-narrowest profile, we have fitted three Gaussian components. The central, most intense feature is fit with a narrow component with a corrected width of 7.8 km s-1, together with two broader, low-intensity components with corrected widths of 19.2 km s-1 and 31.8 km s-1.

The scenario is quite different when multiple Gaussians are fitted to the broadest and more complex Hα profiles. In the lower left-hand panel of Fig. 8, we show that at least six components are necessary to reproduce the observed emission line at that position. The broadest component has a corrected width of σ = 28.5 km s-1. As in the previous cases, the residual is negligible compared with the intensity of the profiles. Finally, in the lower right-hand panel of Fig. 8 we show another spectrum which requires six components to fit the observations. In this case, the broadest component has a corrected width of σ = 48.8 km s-1. It is interesting to note that four of the six broad components used in the Gaussian fits shown in the lower panel of Fig. 8 display supersonic widths (once corrected by σth and σins). Despite other Gaussian components can be fitted in these wide profiles, our results suggest the presence of a highly turbulent gas at these locations.

thumbnail Fig. 6

Example of a high-velocity component seen in the [Niiλ6584 line. The black solid and red dashed lines correspond to the spectra of two contiguous fibers on the sky. The black solid spectrum shows a [Nii] high velocity component which is intense in comparison with its Hα counterpart. This effect is not detected in the red dashed spectrum.

Open with DEXTER

We note that Melnick et al. (1999) found that most of the Hα profiles in their analysis required the presence of a low-intensity broad component with σ = 45 km s-1, Moreover, this broad component was required to fit their integrated Hα profile of 30 Dor, centered at the radial velocity of this GHR. This component was not identified in the two narrowest profiles observed by FLAMES, nor in the second broadest profile (shown in Fig. 8). In fact, the broad components that we have identified are not at the systemic radial velocity of the 30 Dor gas; the origin of the broad component in 30 Dor, if real, is still a mystery.

thumbnail Fig. 7

Examples of the different Hα morphologies detected in 30 Dor. In the first four panels we show the narrowest Hα profiles found in this study. In contrast, the last four panels show the broadest velocities arising from single-profile fits (which are clearly a consequence of multiple components in the observed lines). In each panel we note the coordinates of the observation and σobs of the single-profile Gaussian fit (indicated by the overplotted red dashed lines).

Open with DEXTER

thumbnail Fig. 8

Multiple Gaussian fits to the Hα profiles in 30 Dor with the narrowest and broadest single-component velocity dispersions (upper and lower panels, respectively). The black (solid) and red (dashed) lines represent the observed and fitted profiles, respectively; each component is numbered. Residuals of the fits are shown below each fit, which are relatively minor when compared with the intensity of the profiles. In each panel we list the position of the spectra and the σobs of the different fitted components.

Open with DEXTER

thumbnail Fig. 9

Single and two-component Gaussian fits to the integrated Hα profile (upper and lower panels, respectively). The fitted components are shown by the red (dashed) lines and their widths are indicated in the upper right of each panel.

Open with DEXTER

4.2.3. The integrated Hα profile of 30 Doradus

Fits to the integrated Hα profile (from Fig. 2) are shown in Fig. 9. For a single-component fit we found a corrected width of σ = 26.5 km s-1 (σobs = 29.1 km s-1). Note from the figure that the blue and red wings of the integrated profile can not be fitted by one component – a second, broader component is required, as shown in the lower panel of Fig. 9. By using narrow, high-intensity and broad, low-intensity components, we obtain a much better fit to the observed Hα profile of 30 Dor. In this case, the narrow and the broad components have corrected widths of σ = 21.2 km s-1 (σobs = 24.4 km s-1) and σ = 47.6 km s-1 (σobs = 49.1 km s-1), respectively. Our result for the narrow component is in good agreement with the corrected width from Melnick et al. (1999, σ = 22 km s-1) from the data of Chu & Kennicutt (1994); our broader component is marginally wider than the σ = 44 km s-1 from Melnick et al. By comparing the bottom panel of Fig. 9 with the Fig. 4 from Melnick et al., we note that our broad component is less intense (with respect to the height of the profile) than their broad fit, perhaps due to the different spatial extents that the two studies cover.

thumbnail Fig. 10

Radial line-of-sight systemic velocity map (upper panel) and velocity dispersion map (lower panel) of 30 Doradus from analysis of the Hα emission in the FLAMES spectra. North at the top and east to the left. The radial velocity map was corrected by the systemic velocity of 30 Dor.

Open with DEXTER

4.3. The velocity map of 30 Doradus

As discussed in Sect. 3, we have fitted a single-component Gaussian to each Hα profile. We have used the center and σ (corrected by σin and σth) of each fit to derive the radial line-of-sight systemic velocity field and the velocity dispersion map of 30 Dor, as shown in Fig. 10. In the case of the radial velocity field, it was corrected by the systemic velocity of 30 Dor. Although a single Gaussian fit does not represent the exact radial velocity of regions that display multiple profiles, these fits still give us important information regarding the complexity of the observed profiles. i.e., regions that present multiple profiles will have large values of σ (as shown in Fig. 7). In this sense, both maps help us to understand the overall dynamics of 30 Dor.

In the top panel of Fig. 10 we show the velocity field, centered on its integrated Hα line (see Fig. 9). Approaching and receding regions are represented by blue and red colors, respectively. Green regions are at the systemic velocity of the gas in 30 Dor. In order to show the whole kinematic behaviour of 30 Doradus, we chose a dynamic range of 70 km s-1 in the velocity map. As shown in the velocity map, the north-eastern and south-western regions remain at roughly the systemic velocity (267.4 km s-1, see Sect. 4.1). In the case of the south-western region, the velocity dispersion map (lower panel of Fig. 10) shows several narrow Hα profiles, especially in the neighbourhood of 30 Dor B (in which we find the narrowest Hα profile detected in the FLAMES spectra).

Another remarkable feature to note from Fig. 10 is the presence of large expanding structures. The north-eastern edge of shell #2 (see Chu & Kennicutt 1994) appears to have negative velocities (with respect to the systemic velocity) while the cavity of this structure appears to have positive velocities, giving a 3D view of this shell. Several profiles within this cavity display large values for σobs, which are the result of fitting a single Gaussian over at least two Hα components. In general, Fig. 10 shows that the outer regions of 30 Dor are at the same velocity of the integrated profile of all the spectra, and also that the outer regions can typically be fitted by a single Gaussian component.

thumbnail Fig. 11

Top panel: large expanding structures previously catalogued in 30 Dor (nos. 1, 2, 3 and 5) with the five new expanding structures (nos. 6, 7, 8, 9 and 10) identified here. Bottom panel: χ2 map for the Hα cube of 30 Dor, derived from a single Gaussian fit on each observed profile.

Open with DEXTER

4.4. Large expanding structures in 30 Doradus

We have used the Hα datacube to search for large expanding structures. These structures are characterized by double components in their emission-line profiles, signature of expanding ionized gas at these positions. To identify these structures, we have applied two methods. In a first instance, we have integrated the Hα profiles shown in Fig. 3 over regions of 1′ × 1′, given the poor spatial sampling of our data. Then, these integrated profiles were visually inspected in order to search for double components. In Fig. 11, we superimpose these integrated Hα profiles over the Hα image. At least ten large expanding structures can be identified and are overlaid on the image in Fig. 11. Some of these – shells 1, 2, 3, and 5 – were catalogued previously (Cox & Deharveng 1983; Wang & Helfand 1991), so we adopt the same numbers. Structures 6, 7, 8, 9 and 10 are identified for the first time by this work. These expanding regions present double components in their Hα emission, as can be noted from Fig. 11.

In a second instance, the large expanding structures were searched by using the Gaussian fit on each individual profile. Given that simple Hα profiles can be fitted with a single Gaussian, the χ2 value for these fits will be lower than the value presented by Hα profiles that display double or multiple components. This fact can indicate the location of expanding structures, as profiles that are not well fitted by a single Gaussian. In this sense, we have derived a map with the χ2 values obtained from a single Gaussian fit to the Hα datacube of 30 Dor, which is shown in the bottom panel of Fig. 11. In this case, the Gaussian fit was performed on each observed Hα profile, which was normalized by the total intensity of each profile. Inspecting the bottom panel of Fig. 11, we detect several structures that can not be fitted with a single Gaussian and that could be associated with expanding regions. Interestingly, most of these structures lie on the same position of the expanding regions detected by visual inspection. This correlation is clear for regions #6 and #8. Between regions #1, #2 and #8, the χ2 map shows another peak in the χ2 values. Inspection of the profiles at that location (top panel of Fig. 11) suggest the presence of a small expanding structure. Despite the analysis of the χ2 map is clearly more quantitative than the visual inspection of the Hα profiles, it results necessary to combine both information in order to determine the size and the real nature of the expanding structures.

Of particular interest is structure #6. At this position we do not find any star that could be producing an expansion of the ionized gas. Studies of the diffuse X-ray emission seen by the Chandra X-ray Observatory appear to indicate a single-temperature thermal plasma in this region, but with spectral signatures of both collisional ionization equilibrium and non-equilibrium ionization (Townsley et al., in prep.). This may indicate a plasma that has undergone a recent shock and is transitioning back to ionization equilibrium, possibly consistent with a SN remnant. This scenario is compatible with the expanding structure found in this work; we will analyse the main properties of these newly-detected expanding bubbles in a future study.

Another method to search for expanding structures in GHRs is via analysis of long-slit observations, e.g. the analysis in 30 Dor by Chu & Kennicutt (1994). To mimic their long-slit observations, we have produced 2D cuts along selected rows of the FLAMES datacube (i.e. at different declinations). In Fig. 12 we show these 2D cuts of rows 10, 11, 12, 13 and 14 of the datacube, respectively (see Table A.1 for the RA and Dec of these positions). Row #10 passes across the southern region of the expanding structure #2 (see Fig. 11), while rows #11, #12, #13 and #14 go northwards across the structure #2. In these spectra, we can clearly see the Hα and [Nii] λλ6548, 6584 lines, with Hα the brightest feature. By inspecting the Hα emission in rows 11–14 from Fig. 12, we can note the presence of an expanding structure (as indicated by the black arrows). This structure, which can be identified with shell #2 in Fig. 11, appears as a semi-arc of emission close to the center of the Hα line (see the red-dashed ellipse, in the case of row #14). The expansion of this structure can be clearly identified on the blueward side of the Hα line (see black arrows), but it is not easily seen on the redward side. This structure may be off-center with respect to the systemic Hα emission of 30 Dor.

thumbnail Fig. 12

2D cut obtained from the spectroscopic datacube of 30 Doradus. These spectra correspond to rows 10, 11, 12, 13 and 14 from bottom to top (the numbers are indicated in the top left corner of each panel). The brightest lines correspond (from left-to-right) to [Niiλ6548, Hα and [Niiλ6584. The black arrows indicate the position of an expanding structure. As an example, the red-dashed ellipse indicates the location of this expanding structure in row #14.

Open with DEXTER

5. Summary and conclusions

We have presented new VLT-FLAMES spectroscopy obtained to study the kinematics of the ionized gas in 30 Doradus. These data consist of regular and irregular grids of nebular observations, combined with a stellar grid. The regular grid was combined into a datacube, allowing us to analyse the primary kinematic features of 30 Dor, with the main points now briefly summarised.

  • 1.

    The kinematics of the ionized gas in 30 Dor arecomplex, with a diverse range of single and multiple Hα profiles. In the brightest regions the Hα profiles are found to be simple and narrow. However, the narrowest Hα profile observed in the data lies close to 30 Dor B, where a past SN explosion occurred. This is surprising in the sense that 30 Dor B might be expected to be dominated by the kinematics of the SN remnant (i.e., multiple emission profiles and high-velocity components).

  • 2.

    We have applied multi-Gaussian fits to the Hα profiles of the two narrowest and broadest single-component fits detected in 30 Dor. We do not detect the presence of a broad, low-intensity component, as reported by Melnick et al. (1999) for all of their observed Hα profiles.

  • 3.

    However, the integrated Hα profile of 30 Dor does display broad wings, and required Gaussian fits that include both narrow and broad components (as per Melnick et al.).

  • 4.

    We have derived the velocity field and a velocity dispersion map of 30 Dor. By inspecting these maps, we found that the outer parts of our observed regular grid (i.e. a field-of-view of 10′ × 10′ centered on R136) are at the same velocity of the integrated Hα profile of 30 Dor.

  • 5.

    Using the spectroscopic datacube of 30 Dor, we have identified at least five previously unclassified expanding structures. In one case (structure #6), we did not find any star associated with the expanding gas at that location.

Areas for future work include a detailed analysis of the supersonic velocity in the integrated profile of 30 Dor (besides the origin of the wings in this profile) and a more in-depth study of the individual kinematic structures identified and discussed here.


3

For consistency, we have checked the results derived from mpfit and pan in the case when just one Gaussian is fitted, finding the same results.

4

This image was taken on December 2002, under the program 70.C-0435(A), using the filter Ha#596, with a total exposure time of 1340 s.

Acknowledgments

We would like to thank the referee for the useful comments that improved this paper. We also acknowledge the invaluable help of Leisa Townsley in the interpretation of some of the expanding structures in 30 Dor. S.T.-F. acknowledges the financial support of FONDECYT (Chile) through a post-doctoral position, under contract 3110087. M.R. wishes to acknowledge support from FONDECYT (Chile) grant No. 1080335. M.R. was supported by the Chilean Center for Astrophysics FONDAP No. 15010003.

References

Appendix A: Position of the fibers

In Table A.1 we list the fiber positioning of the regular nebular grid (which does not include the positions of the broken fibers). In the cases where we used a fiber from the irregular nebular grid to complete our regular grid, we label that fiber as “IG” (irregular grid). In Table A.2 we list the fiber positions of the irregular nebular grid.

Table A.1

Fiber positions of the regular nebular grid.

Table A.2

Fiber positions of the irregular nebular grid.

All Tables

Table 1

VLT-FLAMES/MEDUSA observing log for the regular nebular grid.

Table A.1

Fiber positions of the regular nebular grid.

Table A.2

Fiber positions of the irregular nebular grid.

All Figures

thumbnail Fig. 1

Top panel: GIRAFFE/MEDUSA regular grid overlaid on an optical Digital Sky Survey image of 30 Dor. The spatial separation between each fiber corresponds to 20′′ (~5 pc). Middle panel: MEDUSA irregular grid. Bottom panel: MEDUSA stellar grid (VFTS survey). The field-of-view on the sky shown in the lower two panels is larger than in the top panel to show all the fibers available. The red rectangle in the middle panel indicates the region covered by the regular grid.

Open with DEXTER
In the text
thumbnail Fig. 2

Integrated nebular spectrum of 30 Dor obtained from summing all of the spectra from the regular grid. The most intense emission line is Hα, but the intensity axis in the plot is truncated to show the low-intensity emission lines.

Open with DEXTER
In the text
thumbnail Fig. 3

ESO/La Silla Hα image of 30 Dor (taken with the EMMI instrument), on which we have over-plotted the Hα profile from each fiber of the regular nebular grid. Each small box is 20′′  ×  20′′ (5 pc  ×  5 pc), with the fibers located at the center of each box. Thick boxes indicate the positions for which we have interpolated the spectra (see Sect. 2.1.1).

Open with DEXTER
In the text
thumbnail Fig. 4

Velocity-sliced view of the 30 Dor datacube centered on the Hα line. The velocity with respect to the center of the Hα line is shown in the upper left-hand corner of each panel (negative and positive velocities imply radially approaching and receding velocities, respectively). In this figure, a logarithmic scale is used for the intensity of each map.

Open with DEXTER
In the text
thumbnail Fig. 5

Velocity-sliced view of the 30 Dor datacube centered on the [Nii] λ6584 line. The velocity with respect to the center of the [Nii] λ6584 line is shown in the upper left-hand corner of each panel (negative and positive velocities imply radially approaching and receding velocities, respectively). In this figure, a logarithmic scale is used for the intensity of each map.

Open with DEXTER
In the text
thumbnail Fig. 6

Example of a high-velocity component seen in the [Niiλ6584 line. The black solid and red dashed lines correspond to the spectra of two contiguous fibers on the sky. The black solid spectrum shows a [Nii] high velocity component which is intense in comparison with its Hα counterpart. This effect is not detected in the red dashed spectrum.

Open with DEXTER
In the text
thumbnail Fig. 7

Examples of the different Hα morphologies detected in 30 Dor. In the first four panels we show the narrowest Hα profiles found in this study. In contrast, the last four panels show the broadest velocities arising from single-profile fits (which are clearly a consequence of multiple components in the observed lines). In each panel we note the coordinates of the observation and σobs of the single-profile Gaussian fit (indicated by the overplotted red dashed lines).

Open with DEXTER
In the text
thumbnail Fig. 8

Multiple Gaussian fits to the Hα profiles in 30 Dor with the narrowest and broadest single-component velocity dispersions (upper and lower panels, respectively). The black (solid) and red (dashed) lines represent the observed and fitted profiles, respectively; each component is numbered. Residuals of the fits are shown below each fit, which are relatively minor when compared with the intensity of the profiles. In each panel we list the position of the spectra and the σobs of the different fitted components.

Open with DEXTER
In the text
thumbnail Fig. 9

Single and two-component Gaussian fits to the integrated Hα profile (upper and lower panels, respectively). The fitted components are shown by the red (dashed) lines and their widths are indicated in the upper right of each panel.

Open with DEXTER
In the text
thumbnail Fig. 10

Radial line-of-sight systemic velocity map (upper panel) and velocity dispersion map (lower panel) of 30 Doradus from analysis of the Hα emission in the FLAMES spectra. North at the top and east to the left. The radial velocity map was corrected by the systemic velocity of 30 Dor.

Open with DEXTER
In the text
thumbnail Fig. 11

Top panel: large expanding structures previously catalogued in 30 Dor (nos. 1, 2, 3 and 5) with the five new expanding structures (nos. 6, 7, 8, 9 and 10) identified here. Bottom panel: χ2 map for the Hα cube of 30 Dor, derived from a single Gaussian fit on each observed profile.

Open with DEXTER
In the text
thumbnail Fig. 12

2D cut obtained from the spectroscopic datacube of 30 Doradus. These spectra correspond to rows 10, 11, 12, 13 and 14 from bottom to top (the numbers are indicated in the top left corner of each panel). The brightest lines correspond (from left-to-right) to [Niiλ6548, Hα and [Niiλ6584. The black arrows indicate the position of an expanding structure. As an example, the red-dashed ellipse indicates the location of this expanding structure in row #14.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.