A&A 448, L57-L60 (2006)
DOI: 10.1051/0004-6361:200600006
M. R. Pestalozzi1 - V. Minier2,3 - F. Motte3 - J. E. Conway4
1 - School of Physics, Astronomy and Mathematics, University of
Hertfordshire, AL10 9BS Hatfield, UK
2 -
Service d'Astrophysique, DAPNIA/DSM/CEA Saclay,
91191 Gif-sur-Yvette, France
3 -
AIM, UMR 7158, CEA-CNRS-Université Paris
VII, CEA/Saclay, 91191 Gif sur Yvette, France
4 -
Onsala Space Observatory, 439 92 Onsala, Sweden
Received 12 October 2005 / Accepted 19 January 2006
Abstract
Context. NGC 7538 is known to host a 6.7 and 12.2 GHz methanol maser cospatial with a Ultra Compact (UC) H IIregion, IRS 1.
Aims. We report on the serendipitous discovery of two additional 6.7 GHz methanol masers in the same region, not associated with IRS 1.
Methods. Interferometry maser positions are compared with recent single-dish and interferometry continuum observations.
Results. The positions of the masers agree to high accuracy with the 1.2 mm continuum peak emission in NGC 7538 IRS 9 and NGC 7538 S. This clear association is also confirmed by the positional agreement of the masers with existing high resolution continuum observations at cm and/or mm wavelengths.
Conclusions. Making use of the established strong relation between methanol masers and high-mas star formation, we claim that we have accurately positioned the high-mass protostars within the regions where they are detected. The variety of objects hosting a 6.7 GHz methanol maser in NGC 7538 shows that this emission probably traces different evolutionary stages within the protostellar phase.
Key words: star formation - massive stars - interstellar medium - masers
The NGC 7538 nebula is a furnace of intense massive star formation located
at some 2.7 kpc from the Sun. At least 11 infrared sources have been
identified embedded within it (Fig. 1, Ojha et al. 2004 and
references therein). These IR sources lie in four regions hosting a large
range of different types of young stellar objects (YSOs) with signs of
decreasing age moving from the North-West to the South-East. In the far
North-West the first star formation region contains the IR sources IRS 5, 6, 7
which are associated with a developed
H IIregion of 3 pc in size. IRS 6 has been proposed as its
exciting source (Ojha et al. 2004). A further star formation region can be
identified with the IR sources IRS 1-3, where infrared emission at K-band and
long-ward is dominated by IRS 1 (De Buizer & Minier 2005). IRS 1 is known to power a
Ultra Compact (UC) H IIregion, an early stage of massive star
formation. Masers of different species have also been detected toward this
source (OH, H2O, CH3OH, see Minier et al. 1998 and references therein). A
third star-forming region is centered on NGC 7538 S and also contains
IRS 11, 10
to its northwest (Fig. 1). Coincident
with NGC 7538 S OH and H2O masers have been detected
(Kameya et al. 1990). Finally, furthest to the South-East, the fourth known region of
star formation contains the deeply embedded IR source IRS 9, believed to be in
a stage prior to the formation of a UC H IIregion. H2O masers were also
detected toward this source (Kameya et al. 1990; Davis et al. 1998).
![]() |
Figure 1:
General view of the NGC 7538 region. Left: the underlying colour
image is a RGB image of J, H, K NIR bands made from the 2MASS archive. The
grey contours (2, 5, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90% of peak
emission, ![]() |
Open with DEXTER |
Methanol masers are indicative of the earliest stages of massive star
formation. They arise from cold, embedded molecular clumps and in some cases
from dark infrared clouds (Purcell et al. 2005; Hill et al. 2005). These clumps are often
interpreted to be protoclusters of high mass YSOs, i.e. precursors of OB
associations (e.g. Minier et al. 2005; Motte et al. 2003). High angular resolution
observations of methanol maser environments reveal that these masers are
associated with hot cores, UC H IIregions and very often do not
coincide with radio continuum emission from bright Ultra Compact (UC) H II
regions (Walsh et al. 1998; Minier et al. 2001). The detection of a methanol maser in such
regions is
a clear indication of the presence of a deeply embedded hot core that is
characterised by large methanol abundance and 100-200 K gas and dust
temperature. These physical conditions are in agreement with those derived
from maser modelling, as e.g. Sobolev et al. 1997. The high astrometric accuracy
of VLBI observations of methanol masers gives thus the best
estimate of the position of the high-mass protostellar core at a scale of
10 AU. One clear example
of this fact is given by the main component of the CH3OH maser in
NGC 7538 (component A in Fig. 2): the maser pinpoints
the massive protostellar core within the IRS 1 submm clump to
milliarcsecond accuracy
(1 mas at
3 kpc is
3 AU, see Pestalozzi et al. 2004).
In this letter we report on the serendipitous detection of 6.7 GHz methanol masers in the third and fourth star formation regions in NGC 7538, cospatial with NGC 7538 S and IRS 9. Supported by what exposed above, we argue that with this discovery we have accurately positioned the two massive protostars in those regions. This is then confirmed by the coincidence of the new masers with the peak of the 1.2 mm dust emission coming from the thick cocoon surrounding them.
![]() |
Figure 2:
Cross correlated spectrum on the KN-DA baseline averaged over the
whole observing run (dashed line) and
autocorrelation spectrum from the DA antenna (solid line), both from
run1. Minimal fringe spacing on the KN-DA baseline is ![]() |
Open with DEXTER |
The data used for this paper was taken in two runs, with the principal aim of studying the known 6.7 GHz methanol maser emission toward NGC 7538 (i.e. spectral components A-E, see Fig. 2). As explained below, the results presented here refer to the serendipitous discovery of two new maser features, unseen until now in any interferometry experiment.
The first data set (run1, November 4th-5th 2004) was obtained during
a joint experiment including the European VLBI Network and the three antennas of the MERLIN
array which were then operational at
5 cm wavelength, Cambridge (CM), Knockin (KN) and Darnhall (DA). The spectral
resolution was 1.9 and 3.9 kHz/chan (or 0.087 and 0.175 km s-1 per channel,
respectively) over 512 channels for the joint EVN and the MERLIN only data,
respectively. Only the MERLIN portion of this data set has so far been
reduced and is presented here. The second data set (run2, December
11th-12th 2004) was a 13 hour MERLIN only experiment,
including 5 antennas: CM, KN, DA, the 25-m Jodrell Bank Mk2 antenna
(JB) and Tabley (TA). The spectral resolution was 0.976 kHz/chan
(0.044 km s-1per channel) over 512 channels. In both runs the telescopes were
pointed, and the data correlated, at the position of the brightest methanol
maser feature in IRS 1, feature A in Fig. 2 (see also
Table 1).
Table 1: Position, line of sight (LOS) velocity (v) and offset (r) from feature A of the two maser components F and G. The uncertainties in the positions and in the velocity widths are 50 mas (135 AU) and 0.044 km s-1 respectively.
First order calibration and translation into fits
format was performed
with tdproc
at Jodrell Bank. Corrections to the primary
amplitude calibration, phase calibration (self calibration on the
brightest emission channel) and image production were performed in AIPS.
To locate all the maser emission, large
arcsec spectral cubes were made
spanning the full observed velocity range.
In the cross correlation data from run1 (confirmed then by the analysis of the data from run2) emission was detected at the velocities of the known spectral features but also unexpectedly in some other spectral channels. The latter was large non noise-like side-lobe emission, indicating that maser emission at these velocities came from sources located outside of the map region. Maps were then iteratively re-centered until the new features, were brought close to the image centre.
The large position offsets of the new features from the correlated position
cause large variations of the visibility phase over the integration time, and
hence introduce amplitude errors on the visibilities. These do not affect
feature position determination, but prevent us from making reliable
images. Features F & G showed maximal phase changes of 0.6 and
1.8
over the integration time of 7 s, producing mean amplitude losses
of
1 and 8% respectively, on the baseline in Fig. 2.
The positions of the masers were compared to those of 1.2 mm dust peak
emission that were obtained in a MAMBO-2/IRAM-30m map. The image was obtained
in January 2004 using the MAMBO-2 bolometer array (Kreysa et al. 1998) installed at
the IRAM 30 m telescope in the on-the-fly mode. The resulting angular
resolution is HPBW
and the absolute pointing is
accurate to within
.
The comparison of the cross correlated spectrum on the shortest baseline with
the integrated spectrum of the maser emission shows discrepancy in
the peak flux of feature F & G of 20%
(Fig. 2). Unfortunately, the amplitude errors mentioned in
Sect. 2 and the unreliable baseline subtraction in the
integrated spectrum introduce errors in the amplitude that can be estimated to
15%. These prevent us to invoque resolution as responsible for the missing
flux. Our best estimate for the size of F & G (upper limit) is about
half the fringe spacing of the shortest baseline, i.e.
85 mas, or
260 AU, which is to be taken as an upper limit. Note that in the higher
spectral resolution data of run2 (not shown here), F & G show clear
multiplicity (at least two and three components for F &
G respectively). Observations pointed at the new masers are necessary to
obtain reliable and accurate maps of all features.
A review of archival EVN VLBI autocorrelation data for all antennas except Effelsberg shows that both F and G are clearly present in all experiments since 1997. These features always have similar flux ratios to spectral feature A(within a variation of the latter of some 10%). This indicates that the emission from F and G is probably non-variable. The new features are not detected in the autocorrelations from the Effelsberg 100 m antenna because they fall outside the HPBW of the primary beam at 5cm (i.e. 1.6 arcmin). Note that despite the reduction in amplitude due to the primary beam attenuation a hint of feature G was present in the discovery spectrum for NGC 7538 taken with the 140 foot Green Bank telescope (Menten 1991).
Figure 1 shows a general view of the NGC 7538 region. The CH3OH masers are clearly associated with the most embedded sites of high-mass star formation: IRS 1, IRS 9 and S. Masers in IRS 1 coincide with the 1.2 mm continuum peak emission (within the position accuracy) tracing cold dust emission from massive and deeply embedded protoclusters.
Maser feature F coincides with the 1.2 mm dust continuum emission peak
within
(Fig. 1). This is less than the
position accuracy of the 1.2 mm
data (
), and therefore we can state that they are
cospatial. The methanol maser also coincides with other signposts of high-mass
star formation in NGC 7538 S as reported in the literature, within the
position accuracies
(Sandell et al. 2003; Kameya et al. 1990; Sandell & Sievers 2004). In Sandell et al. (2003) this protocluster/cloud is stated
to be one of the most massive of the region, with 1000
estimated
within a circle of 20
(or 50 000 AU) in diameter. The
authors
of that paper also report of a large rotating torus of some 14000 AU in
radius. Single grey-body Spectral Energy Distribution (SED) fitting of the
cold dust emission in the FIR/(sub)mm/mm emission with an aperture of
60
of this source gives a dust temperature of
35 K and an
emissivity index
.
This source lies some 10
to the south
of IRS 11, where a CO outflow is inferred by the distribution of the magnetic
field orientation (Kameya et al. 1991). On the basis of the above multiple evidence,
we consider the position of the methanol maser to be the most reliable for the
protostellar object in NGC 7538 S.
Maser feature G is coincident within 1
with the peak of the
1.2 mm dust emission. The position of G also agrees, within the
position inaccuracies, with the high resolution positions of the peak emission
of the continuum at 4.8 to 107 GHz (van der Tak et al. 2000; van der Tak & Menten 2005). Furthermore, one
water vapour maser spot as well as one class I methanol maser spot seem also to
coincide, within the positional accuracies, with feature G
(Kameya et al. 1990; Sandell et al. 2005), although with an offset in velocity of some
6 km s-1. Because of the fact that class I
methanol masers are mainly found in outflows (Kurtz et al. 2004), we argue that the
newly detected
maser G marks the driving source of the outflow (together
with the continuum), while the other masers arise in the outflow
shocks. This is supported by the geometry proposed in Sandell et al. (2005) for that
source: we might observe the outflow from IRS 9 pole-on. The protostellar
object in IRS 9 is accurately located both with the class II methanol maser
and the cm/mm continuum.
The presence of three methanol maser sources associated to three different objects in the same star formation region offers a unique opportunity to study the evolutionary details of methanol maser bearing sources.
The methanol maser phase is considered to be relatively short as compared to the lifetime of the massive protostar, as short as a few 104 yr (van der Walt 2005). This implies that the sources hosting methanol masers at 6.7 GHz should show very similar properties. The methanol masers in NGC 7538 (features A, G & F) on the other hand seem to mark massive young stellar objects at different stages of evolution: an IR-quiet protostellar object (NGC 7538 S), an IR-bright protostellar object (IRS 9) and an UCH IIregion (IRS 1). The same distinction seems to be shown by the flux densities of the masers: 300, 15, 5 Jy for A, G, F respectively.
We propose 2 scenarios for this apparent sequence. The first and most intuitive is the evolutionary sequence. Feature F might mark the youngest object in the region, feature A the most evolved but still within the protostellar phase. This scenario is also supported by the presence of discs/outflows in the maser sources: in F a large scale rotating torus is detected but no outflow (Sandell & Sievers 2004); in G a powerful, probably very young, outflow is mapped (Sandell et al. 2005); in A a clear disc/outflow geometry is visible (e.g. Gaume et al. 1995; De Buizer & Minier 2005; Pestalozzi et al. 2004). More evidence has to be collected on the capacity of methanol masers to provide such accuracy in the age the sources hosting them.
The second scenario invokes geometry. The difference in 1.2 mm dust emission peak flux among the methanol maser bearing sources could be due to different viewing angles to the source. IRS 1 is seen edge-on, showing the thick dust in its disc (Gaume et al. 1995; Pestalozzi et al. 2004). IRS 9 is suggested to be seen close to pole-on (Sandell et al. 2005): the disc would be seen almost face-on, not contributing significantly to the 1.2 mm flux. NGC 7538 S is still deeply embedded, so dust is close to uniformly distributed around the source. This interpretation seems to be more difficult to accept, because e.g. IRS 1 suffers of significant contamination by the very close IRS 2 and IRS 3.
From the high spatial resolution MERLIN observations of 6.7 GHz methanol maser emission toward the massive star forming region NGC 7538 we have detected and positioned two new spectral features. We note that this achievement was possible only thanks to the intermediate resolution of the MERLIN with its shortest baselines. We conclude the following:
Acknowledgements
M.P. thanks Tom Muxlow for his help during the data calibration at the Jodrell Bank Observatory. We thank all the participants of the NGC 7538 collaboration for the very fruitful discussions that contributed to the acheivement of this paper.