The circumstellar disk, envelope, and bi-polar outflow of the Massive Young Stellar Object W33A
The Young Stellar Object (YSO) W33A is one of the best known examples of a massive star still in the process of forming. Here we present Gemini North ALTAIR/NIFS laser-guide star adaptive-optics assisted -band integral-field spectroscopy of W33A and its inner reflection nebula. In our data we make the first detections of a rotationally-flattened outer envelope and fast bi-polar jet of a massive YSO at near-infrared wavelengths. The predominant spectral features observed are Br , H, and a combination of emission and absorption from CO gas. We perform a 3-D spectro-astrometric analysis of the line emission, the first study of its kind. We find that the object’s Br emission reveals evidence for a fast bi-polar jet on sub-milliarcsecond scales, which is aligned with the larger-scale outflow. The hybrid CO features can be explained as a combination of hot CO emission arising in a disk close to the central star, while cold CO absorption originates in the cooler outer envelope. Kinematic analysis of these features reveals that both structures are rotating, and consistent with being aligned perpendicularly to both the ionised jet and the large-scale outflow. Assuming Keplerian rotation, we find that the circumstellar disk orbits a central mass of 10, while the outer envelope encloses a mass of 15. Our results suggest a scenario of a central star accreting material from a circumstellar disk at the centre of a cool extended rotating torus, while driving a fast bi-polar wind. These results therefore provide strong supporting evidence for the hypothesis that the formation mechanism for high-mass stars is qualitatively similar to that of low-mass stars.
keywords:ISM: individual: W33a – stars: pre-main-sequence – ISM: Hii regions
It is still unclear how massive stars ( 8) are formed. This is due in part to the rapid formation timescales of such stars (yrs), which mean that the star can reach the main-sequence (MS) while it is still heavily embedded in its natal molecular cloud. Also, it is still unknown how a massive protostar can continue to accrete despite the emmense outward radiation pressure it exerts on its surroundings (for reviews of massive star formation see e.g. Beuther et al., 2007; Hoare et al., 2007).
It has been suggested that massive stars may form through a mechanism similar to that of lower mass stars, whereby matter is accreted from a circumstellar disk, with the outward radiative force escaping via the poles and driving a fast bipolar wind. Several numerical studies have succeeded in creating stars with masses 30this way (e.g. Yorke & Sonnhalter, 2002; Krumholz et al., 2009), while it has been shown that the fast bipolar winds should strongly influence the infalling material from the envelope (Parkin et al., 2009). The observational evidence to support this scenario however is limited and circumstantial. The CO bandhead emission at 2.3m has been modelled as arising in a circumstellar disk (e.g. Chandler et al., 1993, 1995; Bik & Thi, 2004a; Blum et al., 2004); large-scale bi-polar outflows have been observed to originate from heavily embedded objects (e.g. Beuther et al., 2002); and some authors have found evidence of accretion disks in individual massive protostars (Patel et al., 2005; Beltrán et al., 2006; Torrelles et al., 2007).
Massive Young Stellar Objects (MYSOs) are thought to represent the phase in a massive star’s formation immediately before the star reaches the main-sequence (MS) but whilst it is still accreting, and as such are important objects with which to study the formation of massive stars. W33A is a well-known MYSO. It lies at a distance of 3.8kpc and has a source luminosity of 10(Faúndez et al., 2004). It is deeply embedded (e.g. Gibb et al., 2000), but is visible in the band despite this. The morphology of the object’s large-scale nebula suggests a bipolar outflow, but as yet no convincing evidence exists for an accretion disk or a bipolar wind close to the central star. The object’s weak radio emission has only been marginally resolved at radio wavelengths (Rengarajan & Ho, 1996; van der Tak & Menten, 2005). The close environment of W33A has been probed using mid-IR interferometry (de Wit et al., 2007, de Wit et al, 2009, submitted), with the visibilities well fitted by a model in which the emission arises in the warm dusty envelope close to the surface of the cavity created by a bipolar outflow.
To provide a detailed study of W33A’s inner nebula, and ultimately to investigate the physical processes occuring in the formation of the central star, in this paper we present high spatial resolution integral field spectroscopy of W33A. These data allow us to study the spatial variations in emission from the ionised outflow, the dense, hot molecular material close to the star, and the cooler molecular material in the object’s inner envelope and outflow.
2 Observations & data reduction
In Fig. 1 we show the wide-field -band image of the W33A region taken from the UKIDSS Galactic Plane Survey (Lucas et al., 2008). The image clearly shows what appears to be a large-scale outflow extending 0.8 towards the south-east at a position-angle (PA) measured to be (1355)°. Diffuse emission is also seen to the north-west of the flux peak, which may be the opposite lobe of the outflow. A velocity resolved CO map of this region conclusively shows that the material is outflowing from the central source (see Fig. 2 of de Wit et al., submitted).
Our data were taken using the Gemini Near-infrared Integral Field Spectrograph (NIFS, McGregor et al., 2003) on the nights of 16 April and 25 May 2008. The observations were pointed such that the bright flux centre was at the northern edge of the instrument field-of-view, in order to obtain data on the base of the outflow as well as the central source. The instrument applies the image-slicing technique to separate the 33 field-of-view into 29 slices, which are then aligned on the detector and passed through a diffraction grating. We employed the K grating in combination with the HK filter and central wavelength set at 2.2m. This setup achieved a spectral resolution of over the wavelength range 2.0-2.4m across the full NIFS field-of-view.
Our observations made use of the Gemini ALTAIR laser-guide star (LGS) adaptive optics system. The system uses a natural guide star for tip-tilt correction, and the laser guide star for wavefront sensing. For tip-tilt correction we used the star 2UCAC-25155527, which has and is located 6.9 from the centre of our field-of-view (see Fig. 1).
The observational strategy involved observing the target and neighbouring sky in an ABBA pattern, the location of the sky field is shown in Fig. 1. The integration time per individual observation was 180s, with each integration read-out in the faint_mode setting of 16 non-destructive reads. A total of 7 separate integrations were used, giving a total integration time of 20mins. In addition to the science target, we observed the A0 V double HR6798 to characterize the telluric absorption. The GCAL calibration unit was used to take continuum lamp observations plus associated dark frames for flat-fielding purposes. Argon-Xenon arc exposures were obtained for wavelength calibration, as well as ronchi flats which are used to characterize the spatial distortion of NIFS data.
2.2 Data reduction
The preliminary stages of data-reduction were done using the Gemini instrument-specific pipeline which runs on the iraf platform. Science observations had their associated sky frames subtracted to remove the dark current and sky emission lines. The data were then divided through by the normalized flat-field to correct for pixel-to-pixel variations in sensitivity, and any bad pixels flagged in the earlier reduction steps were interpolated over. The arc and ronchi flat frames were used to derive the transformation matrix of the 2-dimensional ‘sliced’ image onto a 3-dimensional () datacube. The application of this transformation rectifies the data both spatially and spectrally, and it is at this step that the wavelength calibration is applied.
Removal of telluric absorption and combination of the individual datacubes was done in IDL, using custom-written routines and the GSFC IDL library111http://idlastro.gsfc.nasa.gov. Correction for the Br absorption in the telluric standard was done by dividing through by a synthetic AO V template spectrum, interactively altering the centroid and absorption strength of the Br feature in the template until a satisfactory match to the standard star was found. At each spatial pixel of the science datacube, the spectrum was extracted and divided by the telluric standard. If the spectrum had sufficient signal-to-noise (i.e. over 100 counts in the continuum) the two spectra were first cross-correlated to achieve optimal telluric correction, particularly important in the region of the OH absorption lines at 2.0-2.1m where sub-pixel misalignments can introduce artifacts into the spectrum.
The individual science datacubes of repeat observations were spatially aligned by first taking the median image of each cube across all spectral pixels, then cross-correlating these median images with one another. We then took the sum of the spectra at each spatial pixel, using the median of these spectra as a template from which to reject any further bad pixels and cosmic-ray hits.
2.3 Adaptive optics performance
In order to illustrate the spatial resolution of the data and the PSF correction that was achieved through the ALTAIR adaptive optics system, in Fig. 2 we show the wavelength-integrated image of the telluric standard stars HR6798A+B. To better demonstrate the shape and size of the corrected PSF, the image was first split into two, with each sub-image containing one star. The sub-images were normalized such that the corrected flux from each star was the same, then the sub-images were recombined. On the image we have drawn contours indicating 67% and 95% encircled energy. While the star to the north appears slightly elliptical, the southern star is circular to within the errors. For each star we find 67% of the PSF flux is contained within a radius of 0.15 of the flux centre. The full-width half-maximum (FWHM) of each star was measured to be 0.13.
2.4 Variations in spectral resolution and wavelength calibration
The accuracy of the wavelength calibration and the consistency of spectral resolution was determined from analysis of datacubes constructed from the arc lamp observations. For each arc line that was used to determine the wavelength solution we fitted gaussian line profiles, and measured the line-width and velocity offset. We found that the absolute wavelength calibration had a root-mean-square standard deviation of 4 km s across the field-of-view, with no discernable trends with spatial position. The spectral resolution was similarly found to be km s across the field, again having no trend with spatial position, other than that it was systematically poorer by 6 km s in the south-east 0.50.5region. As very little data was used from this region, we did not correct for this effect.
3.1 Nebular morphology
In Fig. 3 we show the total two-dimensional image when the datacube is summed across all wavelength channels. The central source is not a point-source, and instead is noticeably extended in the north-south direction (see Fig. 4). In particular, we see a bright central peak (labelled ‘A’ in Fig. 3) and a ‘finger’ pointing southward (labelled as ’B’). The ‘L’-shaped inner nebula, which can be marginally resolved in the UKIDSS -band image and which forms the base of the outflow, is resolved in our data into three bright knots of emission, which we label ‘C’, ‘D’ and ‘E’. As we show later, the emission from these knots is predominantly scattered light.
3.2 Spatial variations of spectral properties
The unique feature of IFU data is that it is possible to chart the variations of the observed spectral properties across the field of view. We begin in this section with a description of the principle spectral features, followed by an analysis of how these features vary with spatial position.
In Fig. 5 we plot the spectrum of the bright central source extracted from region A (see Fig. 3). The steeply-rising continuum indicates that the central source is heavily reddened. In order to highlight the spectral features, in the bottom panel of the figure we show the spectrum when normalized by a 4th-degree polynomial fit to the continuum (labelled as ‘region A’ in the Figure). In this normalised spectrum we see emission features of Br , H, Na i and the CO 2-0, 3-1, 4-2 and 5-3 bandheads. Also seen are narrow absorption lines in the region of the CO bandheads, attributable to low J transitions of the CO v=2-0 state, as well as a broad absorption feature around 2.28m which we attribute to CHOH ice (also observed in this object’s spectrum by Taban et al., 2003). We do not confirm the presence of He i emission detected by Taban et al. (2003), while there is a suggestion of emission from the Mg ii doublet at 2.14m. Table 1 shows the measured equivalent widths of the atomic emission lines as measured from the region A aperture.
3.2.1 Spectrum of the central source versus surrounding nebula
To analyse spectral differences between region A and the rest of the field, we have created an average ‘background’ spectrum by integrating over all spatial pixels in the observed field except those in region A. In the top panel of Fig. 5 we have also plotted the ratio of the region A spectrum to this background spectrum, labelled ‘region A/background’. The continuum of this ratio spectrum is shallower than that of the central source, which is to be expected if the emission in the southern nebula is predominantly scattered light.
In the bottom panel of Fig. 5 we plot the ratio of the spectra when normalized by a continuum fit. Here the emission due to Br , Na i, Mg ii and CO disappears, suggesting that the emission morphologies of these features follow that of the continuum, i.e. they are also scattered. The H lines go into ‘absorption’ indicating that this emission originates mainly in the surrounding nebula rather than on the bright source at region A. The CHOH and CO absorption are still seen in the residual spectrum, implying that there is an increased level of opacity due to these features along the line-of-sight to region A compared with the surrounding nebula.
3.2.2 Variations in continuum emission
In the previous section we showed that the slope of the -band continuum emission is not uniform across the field. To further illustrate the spatial variations in continuum emission we created narrow band continuum images at three featureless regions of the spectrum: 2.083m, 2.185m, and 2.273m, each with a bandwidth of 0.003m. These images were then combined into an RGB image, with the shortest wavelength image assigned to the blue channel and the longest to the red. The resulting image is shown in Fig. 6.
In the image, the southern nebula is seen predominantly in ‘blue’ light (i.e. the shortest wavelength), while the central source appears red. This behaviour can be understood as a combination of scattering and extinction. The central source at region A is heavily absorbed, making the direct light from region A appear red. Meanwhile, the emission to the south is mostly light from region A which has been scattered, and so appears blue.
3.2.3 The Br and CO emission
In this section we map the spatial morphology of the emission due to Br and CO (the CO absorption is studied in Sect. 3.2.6). First we subtract the continuum at each spatial pixel of the datacube by fitting a 4th-degree polynomial to featureless regions of the continuum. We then integrate the emission across the spectral feature in question. For the CO emission, we took only the bright blue edge of the first bandhead (2.293-2.300m) – this feature contains the majority of the flux, and is uncontaminated by the low-J absorption lines.
The continuum-subtracted emission images of Br and CO are shown in Fig. 7. Both images display morphology similar to that of the continuum image (traced by the contours), with each of the five emission knots clearly visible. It is unlikely that the extended Br and CO emission is formed ‘in-situ’ in the southern parts of the nebula, as the two spectral features require different temperatures and densities. It is far more likely that in the southern nebula, as with the continuum emission, we are seeing scattered light which originates within a few AU of the central source.
In Fig. 8 we further explore the scattered light explanation for the Br and CO emission by taking the ratio of the two continuum-subtracted images. If the emission in the southern nebula attributable to these transitions is due to scattering by material in the outflow, we should in principle see very little morphological difference between the emission morphologies. Figure 8 shows that no features corresponding to the southern nebula are seen in the ratio image. Instead, we see a slight increase in the ratio of Br to CO in the southern regions compared that around region A. This can be understood as a combination of the wavelength dependence of scattering, which boosts the Br /CO ratio in the south, and extinction by dense material around the central source, which attenuates Br with respect to CO around region A.
3.2.4 Variations in line-profile with viewing-angle
As we have shown in previous sections, the line-emission observed in the southern nebula is most likely scattered light which orginates at the flux centre (i.e. region A). We can then use this scattered light to study the emission from the central source as viewed by the southern nebula, when compared to viewing the emission directly from region A. In effect, this allows us to study the line-profiles as a function of viewing angle. This is a technique we have previously used to study the wind geometry of the evolved massive star IRC +10420 (Davies et al., 2007).
We explore the spatial variations of Br and CO in Figs. 9 and 10, where we plot the integrated spectra of the five bright knots of emission identified in Fig. 3 as A-E. In respect to the Br emission, there is very little variation across the five regions. The regions show evidence for P Cygni absorption, and perhaps an extended red wing. The equivalent widths (EWs) of the profiles in the southern nebula are slightly larger than those of regions A and B. As the Br emission in the southern nebula is most likely scattered light, these small changes in EW are possibly due to anisotropy in the ionized circumstellar gas.
The CO emission shows clear variations across the observed field. To highlight this in Fig. 11 we plot the spectra in the vicinity of the v=2-0 bandhead. The blue edge appears shallower in regions A and B, while the edge is much steeper in the three southern regions. Also, the transitions between 2.30-2.305m appear to be narrower and more pronounced in the southern knots (C, D and E) than in those close to region A. This suggests that the CO emission, when viewed directly through the bright central source at region A, has a broadened velocity profile with respect to the profile reflected from the southern nebula. As the southern nebula’s viewing angle to the central source is different from our direct line of sight to the same region, this suggests that the CO profile’s broadening is a function of the viewing angle. Such behaviour is typically attributed to the emission forming in a disk (Bik & Thi, 2004a; Blum et al., 2004). As the material in the southern nebula ‘sees’ less rotational broadening than our direct line-of-sight, the orientation of the disk is consistent with being aligned perpendicularly to the larger-scale outflow.
3.2.5 Spectro-astrometry of Br
Here we investigate the small-scale geometry of the Br emission by performing a 3-D spectro-astrometric analysis of the Br line. At each wavelength channel of the datacube we extracted a 2-D image and determined the astrometric centroid of the emission at region A. This method can give spatial information on extremely small scales – the astrometric accuracy on the centroid is roughly equivalent to . So, for the spatial resolution of our observations (0.13) and a SNR at region A of several hundred, in principle we can expect to achieve accuracies of order of a tenth of a milliarcsecond. In order to verify this we performed spectro-astrometric tests on the data from the telluric standards. We found that indeed the r.m.s. on the centroid position was of order 0.1mas. We tried different methods of centroiding, including flux-weighted centroiding, finding the location at which the derivatives go to zero and fitting 2-D gaussian profiles. All techniques were found to give similar results within the errors. In addition we observed no artifacts around the telluric absorption lines, and no excursions across the telluric standard star’s intrinsic Br absorption, as is to be expected (see Appendix A).
In the left panel of Fig. 12 we plot the position of the centroid as a function of wavelength in the region of Br . The spectro-astrometric profiles have had any large-scale gradients removed by fitting a 3rd-degree polynomial to the continuum. Firstly, we note that the astrometric accuracy of the centroid, as judged by the continuum regions, is comparable to that predicted (0.1-0.2mas), corresponding to a physical scale of 0.4-0.8AU for a systemic distance of 3.8kpc. Secondly, the centroid shows a clear shift in the region of Br . In the north-south direction (labelled ‘DEC’ in the figure), the centroid appears to shift south slightly in the region of the P Cygni absorption, before shifting northwards by 1 mas. Similarly, in the east-west direction (labelled ‘RA’, where the positive direction is eastward), there is a slight shift to the east coincident with the P Cygni absorption, followed by an excursion to the west of 0.5mas.
This behaviour is illustrated again in the right panel of Fig. 12. This time, the centroid is plotted in RA-DEC space with colour-coded symbols to indicate the velocity of each point. The location of the continuum centroid is illustrated by the cross. The plot shows that the emission with the largest red-shifted velocities (400 km s) is located to the north-west of the continuum location, while the emission blue-shifted by a similar amount is located to the south-east of the continuum, indicating expansion. From analysis of those points with speeds between 300-400 km s we find that the statistical significance of the red- and blue-shifted points being spatially separated from the continuum is 2.8 and 4.1 respectively. In addition to this behaviour there is a large excursion across the centre of the line with km s.
The observed behaviour can be explained if we make the assumption that the Br emission line is made up of two or more components. The broad high-velocity emission originates in an axisymmetric structure, most likely a bi-polar jet, while the narrow low velocity emission is formed in an extended structure such as a small, dense Hii-region. The nature of this narrow emission is difficult to disentangle, but it is possible that it is formed in an aspherical clumpy medium with much velocity-dependent extinction and self-absorption. This is similar to the picture deduced by Drew et al. (1993) and Bunn et al. (1995) from velocity-resolved line-profile ratios.
In addition to the detection of an outflow, the spectro-astrometric data also indicate that the bright flux maximum observed in the integrated image is indeed the location of the central star.
For the broad-line component of the emission we can measure the PA of this structure by taking the data at and taking the mean positions of the red- and blue-shifted components. We measure the PA of this structure to be 11325° (see right panel of Fig. 12), and so is aligned with the large-scale outflow seen in Fig. 1 (PA=1355°). From the alignment of these angles it is likely that in this small-scale structure we are seeing the base of a bi-polar ionized wind or jet. In addition, the blue-shifted portion of the Br emission is oriented towards the south-east, consistent with the hypothesis that the outflow seen in the wide-field image is the lobe coming towards us, while the red-shifted lobe to the north-west is largely obscured.
The outflow velocity, as measured from Fig. 12, appears to be around km s. Assuming an inclination of 60° (de Wit et al., 2009 submitted), this gives an outflow velocity measured from Br of 600 km s. This is likely an underestimate of the outflow’s terminal velocity, as a large fraction of the Br emission may form in the accelarating part of the wind. This value is consistent with MYSO jet velocities measured by radio proper motion studies (Marti et al., 1998; Curiel et al., 2006).
3.2.6 The CO absorption
The low-J transistions of the CO v=2-0 state are observed in absorption in W33A’s spectrum (Sect. 3.2). The population of these states requires much cooler conditions (30 K) than that of the high-J states seen in emission (few 1000 K) (Chandler et al., 1993, 1995; Bik & Thi, 2004a). Therefore these two distinct spectral features must arise in different regions – the CO bandhead emission likely originates in material close to the central star; while the absorption lines are formed in the outer envelope at several thousand AU (see later).
The right-hand panel of Fig. 9 shows that the CO absorption lines vary in strength between the bright central source (A) and the southern regions (C, D and E). Specifically, the absorption is stronger towards region A than the southern nebula. To explore this behaviour further, sub-cubes were extracted from the primary data-cube, each containing the full spatial information of one of these absorption lines. We observed a behaviour which was similar for each line and so all lines were combined to increase the signal-to-noise.
In Fig. 13 we show the image obtained when the datacube is integrated over these features. The image shows that the CO absorption is concentrated around region A in a band extending roughly east-west, and is perpendicular to the large-scale outflow. Significantly less absorption is seen in the southern nebula, indicating that the material in the southern regions sees less obscuring material than our direct line-of-sight to region A. This suggests that the envelope has the morphology of a torus, with a physical radius of 2000 AU, and with an opening to the south through which the outflow escapes.
Figure 14 shows the measured velocity centroid of the CO absorption at each spatial position across the extinction structure seen in Fig. 13. The image has been smoothed by a boxcar filter of width 5 pixels to highlight any velocity gradient. At negative (blue-shifted) velocities, the opacity appears to be concentrated in a region about 0.2-0.3 east of the centre of the bright source (i.e. region A). At positive (red-shifted) velocities however, there is a clear shift in the location of the opacity centre to a region 0.1-0.2 west of region A. No apparent trends with velocity are seen in the southern regions of the nebula, though the data are very noisy and are not shown in the figure. We determined the PA of rotation to be 6010° by fitting the blue and red flux peaks in Fig. 14. This is therefore consistent with being perpedicular to the larger scale outflow of PA=1355°. The PA of this feature is perpendicular to the large scale outflow observed in Fig. 1, as well as to the small-scale outflow seen in the spectro-astrometric analysis of the Br emission. As with the broadening of the CO bandhead emission described in the previous section, the observed behaviour is indicative of an extended rotating structure seen edge-on. The kinematics of this feature are explored further in Sect. 4.
3.2.7 Molecular hydrogen
The light of H 2.12m shows a very different morphology to that of the continuum and the Br and CO emission (see Fig. 15). The emission is concentrated to the south-east of region A, while very little H emission is seen at region A itself. The other H lines observed in the spectrum display similar behaviour, though have much poorer signal-to-noise and so are not shown. The orientation of this emission, extending roughly south-eastwards of region A, is approximately coincident with that of the base of the large-scale outflow seen in the wide-field image of Fig. 1.
4 Discussion – CO kinematics, and the mass of the central object
The pieces of evidence presented above are all consistent with the qualitative picture of a massive forming star in which the central object is surrounded by a rotating disk from which it is accreting. A rotationally-flattened envelope surrounds the central star and its disk, while the star drives a fast bi-polar wind.
In this section we present a kinematic analysis of the CO bandhead emission and the low-J CO absorption lines and use them to derive the mass of the object about which they rotate.
4.1 The bandhead broadening
As shown in Fig. 11, the CO bandhead emission lines appear to be broader towards region ‘A’ than those observed in the southern nebula. This behaviour can be attributed to a flattened rotating structure (i.e. a disk) which is seen at lower inclination when viewed directly than when seen reflected by the nebula to the south. If we assume the extreme case that (a) the line-of-sight ‘A’ views the disk edge-on, and (b) that the southern nebula views the disk pole-on, by convolving the profile seen from the southern nebula until it matches that at ‘A’ we can estimate a lower limit to the disk rotation speed. By assuming Keplerian rotation we can then estimate the mass of the central object about which the disk rotates.
In Fig. 16 we plot the CO bandhead profile seen at ‘A’. We also plot the profile seen in the southern nebula (region ‘C’) which has been convolved with a rotational broadening function of the form , where is velocity and is the disk rotational velocity. The best match to the blue-edge of the bandhead, as well as the lower-J line profiles, is found when the rotational velocity is 505 km s. If the direct line-of-sight is not exactly edge-on, and/or if the orientation of the reflected light is not exactly pole-on, the disk rotation speed will be greater than this value. By taking the inclination angle recently derived by de Wit et al. (submitted) of 6020° (i.e. 30° away from being edge-on), we can further constrain the disk rotation speed to 64 km s.
Assuming that the disk rotation is Keplerian, and that the CO emission arises in a narrow annulus at a fixed radius from the central object we can now derive a mass of the central object from,
4.2 The absorption-line velocity structure
In Sect. 3.2.6 we showed that the absorption lines of CO appear to be formed in a rotating structure coincident with region A, and measured the PA of this structure to be 6010°. In Fig. 17 we show a position-velocity slice across the averaged CO datacube at a PA of 60°. The data shows the classical signature of a Keplerian rotating structure. To help illustrate this, overplotted are models of Keplerian rotation for three different central masses and with an assumed distance of 3.8kpc (Faúndez et al., 2004). Each curve has been convolved with the spatial resolution of our observations (0.13). A formal least-squares fit to the data yields a central mass of , a result which is insensitive to the uncertainty in PA. If we again assume the system inclination is 6020° we find that the outer envelope sees a central mass of . Note that the central mass seen by the envelope will include the mass of the envelope itself. This number is very similar to that derived for the central mass enclosed by the circumstellar disk, and suggests that the majority of the system’s mass is contained within the central star.
If the central object has arrived on the main-sequence, this mass would correspond to a spectral type earlier than B1 and a zero-age min-sequence luminosity of (e.g. Meynet & Maeder, 2000). This is in good agreement with the latest estimation of W33A’s bolometric luminosity from MIPS photometry, (Mottram et al. 2009, in prep). The derived spectral-type and luminosity may also explain the detection of the Mg ii emission lines. Such lines are typically seen in the spectra of early-to-mid B supergiants, and are primarily formed in the wind (Hanson et al., 2005). The ionization emission lines that we see in W33A (Br and Mg ii) are probably formed in the bi-polar outflow, the density of which may be similar to that of a typical B supergiant wind.
5 Summary and conclusions
We have presented a study of adaptive-optics assisted near-infrared integral-field spectroscopy of the young stellar object W33A. The results can be summarized as follows:
In the first 3-D spectro-astrometric study of its kind, we have resolved the Br emission on sub-milliarcsecond scales (physical scales 1 AU), and find that the emission geometry is that of a fast bipolar wind which is aligned with the large-scale outflow seen in wide-field images. The H 2.12m emission also traces an outflow with the same orientation out to 1.
The extended nebulosity to the south is seen in the continuum, as well as in the light of Br and CO. We have shown that this emission is light reflected by the outflow lobe, and so can be used as a ‘mirror’ with which to observe the emission from the central object from different angles.
We interpret the CO emission as arising in a disk which has a low inclination with respect to our line-of-sight. In the southern nebula we see light reflected towards us from the disk at a lower inclination angle (i.e. more pole-on), which causes the bandhead profile to be less velocity-broadened. This is consistent with the disk plane being perpendicular to the large-scale outflow. The velocity broadnening implies a lower-limit to the central mass of .
From the low-J v=2-0 transitions of CO at 2.35m we find evidence for a rotationally-flattened cool molecular envelope, or ‘torus’, at a radius of 2000 AU from the central star. The plane of rotation is perpendicular to the small- and large-scale outflows. Analysis of the velocity structure of the torus indicates that it is rotating about the outflow’s axis, and that it is orbiting a central mass of .
Our findings suggest a picture of massive star formation within W33A which is in excellent qualitative agreement with the accretion-disk-plus-bipolar wind paradigm. An accretion disk orbits a massive central star, which is surrounded by a cool molecular envelope which has been rotationally-flattened into a torus. The central star is driving a bipolar wind, seen on small scales in the ionized gas, and on larger scales as molecular / continuum emission. The two measurements of the central mass indicate that at most the accretion disk makes up 30% of the system, and so the system mass is dominated by the central star.
Based on observations obtained at the Gemini Observatory, which is operated by the Association of Universities for Research in Astronomy, Inc., under a cooperative agreement with the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), the Science and Technology Facilities Council (United Kingdom), the National Research Council (Canada), CONICYT (Chile), the Australian Research Council (Australia), Ministério da Ciência e Tecnologia (Brazil) and Ministerio de Ciencia, Tecnología e Innovación Productiva (Argentina).
- Beltrán et al. (2006) Beltrán, M. T., Cesaroni, R., Codella, C., Testi, L., Furuya, R. S., & Olmi, L. 2006, Nature, 443, 427
- Beuther et al. (2007) Beuther, H., Churchwell, E. B., McKee, C. F., & Tan, J. C. 2007, in Protostars and Planets V, ed. B. Reipurth, D. Jewitt, & K. Keil, 165–180
- Beuther et al. (2002) Beuther, H., Schilke, P., Sridharan, T. K., Menten, K. M., Walmsley, C. M., & Wyrowski, F. 2002, A&A, 383, 892
- Bik & Thi (2004a) Bik, A. & Thi, W. F. 2004a, A&A, 427, L13
- Bik & Thi (2004b) —. 2004b, A&A, 427, L13
- Blum et al. (2004) Blum, R. D., Barbosa, C. L., Damineli, A., Conti, P. S., & Ridgway, S. 2004, ApJ, 617, 1167
- Bunn et al. (1995) Bunn, J. C., Hoare, M. G., & Drew, J. E. 1995, MNRAS, 272, 346
- Chandler et al. (1995) Chandler, C. J., Carlstrom, J. E., & Scoville, N. Z. 1995, ApJ, 446, 793
- Chandler et al. (1993) Chandler, C. J., Carlstrom, J. E., Scoville, N. Z., Dent, W. R. F., & Geballe, T. R. 1993, ApJ, 412, L71
- Curiel et al. (2006) Curiel, S., Ho, P. T. P., Patel, N. A., Torrelles, J. M., Rodríguez, L. F., Trinidad, M. A., Cantó, J., Hernández, L., Gómez, J. F., Garay, G., & Anglada, G. 2006, ApJ, 638, 878
- Davies et al. (2007) Davies, B., Oudmaijer, R. D., & Sahu, K. C. 2007, ApJ, 671, 2059
- de Wit et al. (2007) de Wit, W. J., Hoare, M. G., Oudmaijer, R. D., & Mottram, J. C. 2007, ApJ, 671, L169
- Drew et al. (1993) Drew, J. E., Bunn, J. C., & Hoare, M. G. 1993, MNRAS, 265, 12
- Faúndez et al. (2004) Faúndez, S., Bronfman, L., Garay, G., Chini, R., Nyman, L.-Å., & May, J. 2004, A&A, 426, 97
- Gibb et al. (2000) Gibb, E. L., Whittet, D. C. B., Schutte, W. A., Boogert, A. C. A., Chiar, J. E., Ehrenfreund, P., Gerakines, P. A., Keane, J. V., Tielens, A. G. G. M., van Dishoeck, E. F., & Kerkhof, O. 2000, ApJ, 536, 347
- Hanson et al. (2005) Hanson, M. M., Kudritzki, R.-P., Kenworthy, M. A., Puls, J., & Tokunaga, A. T. 2005, ApJS, 161, 154
- Hoare et al. (2007) Hoare, M. G., Kurtz, S. E., Lizano, S., Keto, E., & Hofner, P. 2007, in Protostars and Planets V, ed. B. Reipurth, D. Jewitt, & K. Keil, 181–196
- Krumholz et al. (2009) Krumholz, M. R., Klein, R. I., McKee, C. F., Offner, S. S. R., & Cunningham, A. J. 2009, Science, 323, 754
- Lucas et al. (2008) Lucas, P. W., Hoare, M. G., Longmore, A., Schröder, A. C., Davis, C. J., Adamson, A., Bandyopadhyay, R. M., de Grijs, R., Smith, M., Gosling, A., Mitchison, S., Gáspár, A., Coe, M., Tamura, M., Parker, Q., Irwin, M., Hambly, N., Bryant, J., Collins, R. S., Cross, N., Evans, D. W., Gonzalez-Solares, E., Hodgkin, S., Lewis, J., Read, M., Riello, M., Sutorius, E. T. W., Lawrence, A., Drew, J. E., Dye, S., & Thompson, M. A. 2008, MNRAS, 391, 136
- Marti et al. (1998) Marti, J., Rodriguez, L. F., & Reipurth, B. 1998, ApJ, 502, 337
- McGregor et al. (2003) McGregor, P. J., Hart, J., Conroy, P. G., Pfitzner, M. L., Bloxham, G. J., Jones, D. J., Downing, M. D., Dawson, M., Young, P., Jarnyk, M., & Van Harmelen, J. 2003, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 4841, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, ed. M. Iye & A. F. M. Moorwood, 1581–1591
- Meynet & Maeder (2000) Meynet, G. & Maeder, A. 2000, A&A, 361, 101
- Parkin et al. (2009) Parkin, E. R., Pittard, J. M., Hoare, M. G., Wright, N. J., & Drake, J. J. 2009, MNRAS, 1372
- Patel et al. (2005) Patel, N. A., Curiel, S., Sridharan, T. K., Zhang, Q., Hunter, T. R., Ho, P. T. P., Torrelles, J. M., Moran, J. M., Gómez, J. F., & Anglada, G. 2005, Nature, 437, 109
- Rengarajan & Ho (1996) Rengarajan, T. N. & Ho, P. T. P. 1996, ApJ, 465, 363
- Taban et al. (2003) Taban, I. M., Schutte, W. A., Pontoppidan, K. M., & van Dishoeck, E. F. 2003, A&A, 399, 169
- Torrelles et al. (2007) Torrelles, J. M., Patel, N. A., Curiel, S., Ho, P. T. P., Garay, G., & Rodríguez, L. F. 2007, ApJ, 666, L37
- van der Tak & Menten (2005) van der Tak, F. F. S. & Menten, K. M. 2005, A&A, 437, 947
- Yorke & Sonnhalter (2002) Yorke, H. W. & Sonnhalter, C. 2002, ApJ, 569, 846
Appendix A Spectro-astrometry of the telluric standard star
In Fig. 18 we show a spectro-astrometric analysis of the telluric standard star following a similar methodology to that presented in Sect. 3.2.5. The intensity spectrum (bottom panel) shows the Br absorption line, as well as the many telluric features. The top panel shows the astrometry of the star’s flux peak as a function of wavelength, illustrating that no discernable variations are seen across neither the Br line nor the telluric lines. This of course is to be expected, and the Figure serves to show that the technique we employ produces no spurious artifacts greater than the noise level (0.2mas) which could be misinterpreted.