A&A 462, L1-L4 (2007)
DOI: 10.1051/0004-6361:20066590
LETTER TO THE EDITOR
R. Schödel1 - A. Eckart1,2 - K. Muzic1 - L. Meyer1 - T. Viehmann1 - G. C. Bower3
1 - I. Physikalisches Institut, Universität zu Köln,
Zülpicher Str. 1, 50937 Köln, Germany
2 -
Max-Planck-Institut für Radioastronomie, Auf dem
Hügel 69, 53121 Bonn, Germany
3 -
Astronomy Department and Radio Astronomy Laboratory,
University of California, Berkeley, CA 94720, USA
Received 18 October 2006 / Accepted 16 November 2006
Abstract
Context. Sagittarius A* (Sgr A*) at the center of the Milky Way is a black hole accreting at extremely sub-Eddington rates. Measurements of its emission in the infrared and X-ray domains are difficult due to its faintness and high variability.
Aims. The Galactic center was observed at 8.6 m in order to detect a mid-infrared (MIR) counterpart to Sgr A*, parallel to NIR observations. The goal was to set constraints on possible emission mechanisms.
Methods. Imaging data were acquired with the adaptive-optics assisted NIR instrument NACO and the MIR instrument VISIR at the ESO VLT.
Results. We present MIR imaging data of an unprecedented quality in terms of spatial resolution and sensitivity. An extended ridge of emission is found to be present in the immediate vicinity of Sgr A* thereby rendering any detection of a point source difficult. No MIR point source related to Sgr A* was detected during the observations. We derive a tight upper limit of mJy (dereddened) on any possible point source present during the observations in the night of 4/5 June 2006. The absence of a flare in simultaneous observations at 2.2
m and the low limits on any possible variability in the MIR strongly suggest that Sgr A* was in a quasi-quiescent state during this night. During the night from 5 to 6 June 2006, Sgr A* was found to be variable on a low level at 3.8
m. No point source at 8.6
m was detected during the simultaneous MIR observations. Due to the poorer atmospheric conditions, a higher upper limit of
mJy was found for Sgr A* at 8.6
m during the second night.
Conclusions. The observations are consistent with theoretical predictions. If the published models are correct, the observations demonstrate successfully that a 8.6 m counterpart of Sgr A* can be easily detected in its flaring state. Spectral indices derived from simultaneous observations of flaring emission from Sgr A* at NIR and MIR wavelengths will enable us to distinguish between different kinds of flare models.
Key words: Galaxy: center - Galaxy: nucleus - accretion, accretion disks
The center of the Milky Way harbors a supermassive black hole of
(e.g., Ghez et al. 2003; Schödel et al. 2002; Eisenhauer et al. 2005).
The non-thermal source related to this supermassive black hole,
Sagittarius A* (Sgr A*), radiates at only
10-9-10-10 times its
Eddington luminosity from radio wavelengths to the X-ray domain. Its
low luminosity is consistent with emission from so-called radiatively
inefficient accretion flows, a jet, or a combination of the two models
(e.g., Bower et al. 2004; Shen et al. 2005; Yuan et al. 2003,2002).
Table 1: Summary of observations with VISIR and NACO at the ESO VLT used in this work.
X-ray and near-infrared (NIR) counterparts to Sgr A* were only
discovered with the availability of sensitive, high-resolution
instruments for these wavelengths. It was found that Sgr A* is highly
variable at these wavelengths, showing flaring emission on time scales
of 60-100 min, with flux increases up to 100 at X-rays and up
to 10 in the NIR
(e.g., Genzel et al. 2003; Baganoff et al. 2001). The variability at
X-ray and IR wavelengths appears to be simultaneous
(Eckart et al. 2006a,2004). At MIR wavelengths, only upper
limits to the flux of Sgr A* have been reported so far
(Eckart et al. 2006a; Telesco et al. 1996; Stolovy et al. 1996; Cotera et al. 1999).
The detection of Sgr A* at MIR wavelengths is difficult due to the
lower spatial resolution compared to NIR wavelengths, the general
difficulties of imaging in the thermal IR regime, and the presence of
warm dust near Sgr A*. Warm dust is associated with the mini-spiral
gas streamers that pass close to Sgr A*. Therefore, Sgr A* is no
isolated point source in the MIR, and its detection requires high
image quality, above all high spatial resolution, in order to achieve
a sufficiently high contrast. Here, we report on new MIR observations,
using the European Southern Observatory's MIR imager and spectrograph
VISIR at the Very Large Telescope (VLT) on Cerro Paranal in Chile.
Although Sgr A* was not detected, the acquired images are - in terms
of sensitivity and spatial resolution - the highest-quality
8.6 m-images of the Galactic center (GC) region published up
to now, allowing us to report the so far tightest upper limit on the
8.6
m flux of Sgr A*. Even more interesting is that we can
show - via simultaneously acquired adaptive-optics NIR imaging data
- that the infrared flux of Sgr A* agrees well with models of the
quiescent/low activity emission. This information allows us to
conclude that - according to currently accepted theoretical models -
Sgr A* can be easily detected at MIR wavelengths during a bright
flare. Such an observation will allow the NIR-to-MIR spectral slope of
Sgr A* during flares to be derived and thus allow us to distinguish
between different flare models.
The GC was observed in the MIR N-band with the ESO VLT unit telescope
3 (UT 3) using the MIR imager/spectrograph VISIR
(Lagage et al. 2004,2003) in the nights of 4/5 and 5/6
June 2006. The pixel scale was 0.075'' per pixel. The PAH 1 filter with a central wavelength of 8.59 m and half-band width of
0.42
m was used. The standard nodding (
east of
north) and chopping (with a chop throw of 30'') technique was used.
Standard data reduction was applied, i.e. the sky background acquired
during the chopping and nodding observations was subtracted from the
images of the target and individual frames were combined with a simple
shift-and-add technique. Dithering was applied during the
observations, leading to an FOV in the combined mosaic images of
.
The seeing during the observations on 5 June
2006 was good enough to result in a PSF FWHM of
0.3'', close
to the diffraction limit of the VLT at
8.6
m.
In this work, we also use NIR-imaging data that were obtained parallel
to the MIR data with the adaptive optics imager/spectrograph NACO at
the ESO VLT. On June 5, the data were obtained in the Ks-band in
polarimetric mode with a pixel scale of 0.013'' per pixel. The data
reduction and flux calibration were identical to the procedures
described in Eckart et al. (2006b) and in Meyer et al. (2006, A&A,
in press). The L'-band (3.8 m) imaging data were obtained with
NACO/VLT on 29 May and 6 June 2006 with a pixel scale of 0.027'' per
pixel. The reduction of the L'-band data was standard and identical to
the data-reduction procedures described in Muzic et al. (2006, A&A,
submitted). Details of the NIR and MIR observations are listed
in Table 1. The total integration time of the
observations corresponds to
,
as given in the
table. The total integration time during both MIR observations was
thus 36 min.
Flux calibration was achieved by observations of the standard star
HD 178345 (14.32 Jy in the PAH 1 filter, see ESO VISIR web
site and Cohen et al. 1999). The isolated standard star was also used
the PSF reference for image deconvolution.
![]() |
Figure 1:
Sgr A* and its immediate environment at
8.6 ![]() ![]() ![]() ![]() |
Open with DEXTER |
The astrometric reference frame could be established via the positions
and proper motions of the stars IRS 3, IRS 7, IRS 9, IRS 14NE,
IRS 12N, IRS 2L, IRS 6E, and IRS 29, which were published by
Ott (2004). All of these stars were detected on the VISIR
image as point sources. The positions of these eight stars were used
to solve transformation equations up to the second order. The accuracy
of the astrometric solution could be checked by choosing sub-groups of
seven out of the eight stars and repeating the transformation. Also,
we compared the measured astrometric position of the stars IRS 16NW,
IRS 16C, and IRS 29 to their predicted positions. The position of
Sgr A* could thus be established with a
uncertainty of
0.03'' (less than half a pixel) in the MIR data. This is about a
factor of 10 better than in previous work
(Stolovy et al. 1996). There are two main reasons for this increased
astrometric accuracy: on the one hand, the high sensitivity and
spatial resolution of the VISIR data and, on the other, the improved
IR position of Sgr A* due both to precise positions and proper motions
of stars in the NIR (e.g., Genzel et al. 2000; Ott 2004) and
to the well-known position of Sgr A* in the IR reference frame via SiO
maser sources (e.g., Reid et al. 2003).
Figure 1 presents the flux-calibrated direct shift-and-add image of the central arcseconds around Sgr A* for the night of 4/5 June 2006. Significant signal power was found to be present at the diffraction limit (>60%, estimated by aperture photometry).
![]() |
Figure 2:
The Lucy-Richardson deconvolved and
beam-restored version of the image shown in Fig. 1,
with identical labeling and contour lines. We derive fluxes of
![]() ![]() ![]() |
Open with DEXTER |
A Lucy-Richardson deconvolved and beam-restored version of the image
from Fig. 1 is shown in Fig. 2. The
deconvolved and beam-restored image resembles the original image
closely, but emphasizes some of the finer details. There is a dust
ridge present very close to the position of Sgr A*, where it bends
from an east-west extension to a southeast-northwest direction. In
earlier MIR observations of the GC, Stolovy et al. (1996) found very
similar structures near Sgr A*. Due to their high spatial resolution
and sensitivity, the new data show much richer details, however. Even
some stars, such as IRS 29, IRS 16NW and IRS 16C (labeled in
Fig. 1) can be detected. In spite of the high quality of
the data, no obvious point source is present at the position of
Sgr A*. We conclude that the source coincident with Sgr A* that was
reported by Stolovy et al. (1996) was most probably related to the
background emission of the dust ridge and not to Sgr A* itself. All
the flux appears to originate in the dust ridge. This assumption can
be used to obtain a first, simple flux estimate for Sgr A*. Such an
estimate can be obtained by adding or subtracting point sources of
given flux at the position of Sgr A* and checking whether the shape of
the dust ridge is altered perceptibly in the images. After testing
various flux levels, we estimate that a point source at the position
of Sgr A* that is present in the image cannot be brighter than
5 mJy.
With AO L'-band imaging, Eckart et al. (2006a) and Ghez et al. (2005)
demonstrate the presence of a dust blob between about 0.03'' and
0.15'' southwest of Sgr A*. As discussed by Eckart et al. (2006a),
it is probably this blob that is responsible for most of the MIR
emission near the position of Sgr A* when the latter is in a quiescent
state. This blob is clearly visible in the MIR images presented here
and forms part of the dust ridge. Therefore, a second way of
estimating the flux of a putative point source near Sgr A* can be
obtained by trying to account for the extended dust emission. For this
purpose we used a high-quality L'-band image obtained with NACO in the
night 29/30 May 2006. The StarFinder (Diolaiti et al. 2000)
code was used to extract the stars from the image (including a point
source due to the emission from Sgr A*) and thus to obtain an image of
the diffuse emission at 3.8 m. This image was smoothed,
transformed, and scaled (with the help of the IDL routines POLYWARP
and POLY_2D) to fit the scale, orientation, and resolution of the MIR
data. Flux calibration of the L'-band data was achieved by using the
published fluxes of the sources IRS 16C and IRS 33N
(Blum et al. 1996). Figure 3 shows the diffuse
3.8
m flux superposed as contour lines onto the MIR image from
the right panel of Fig. 1. The median ratio of the diffuse
flux at the two wavelengths in a circular area of 0.75'' around
Sgr A* is
.
This value was used to scale the diffuse flux
present in the L'-band data and to subtract it from the MIR
image. Here it is important to note that the flux density ratio is
much higher in other areas of the mini-spiral. In the northern arm it
can be up to 100. Therefore we obtain a conservative estimate of the
remnant flux. After subtraction of the scaled diffuse 3.8
m
emission, we find an upper limit on the flux of a putative point
source at the position of Sgr A* of
mJy. Following
Lutz et al. (1996), this corresponds to an extinction-corrected flux
of
mJy for the image obtained on 5 June 2006. The lower
data quality on 6 June 2006 leads to a higher upper limit of
mJy (extinction corrected) on any possible Sgr A*
counterpart.
During the simultaneous observations at 2.2 m on June 5, no
counterpart to Sgr A* was detected. The upper limit for the flux of
Sgr A* was determined to
mJy. Although the quality of the
L'-band observations from June 6 was fairly low (due to bad seeing,
poor AO correction and electronic noise on the detector), a weak
variable counterpart of Sgr A* could be detected. Its variability was
measured to be a factor
2 and its extinction
(Lutz et al. 1996) corrected flux to be
mJy.
![]() |
Figure 3:
Contours of diffuse L'-band emission
overplotted onto the VISIR 8.6 ![]() ![]() ![]() |
Open with DEXTER |
A central question is whether Sgr A* was really in a state of
quasi-quiescence when the MIR observations were taken or whether
Sgr A* is generally too faint to be detected at MIR wavelengths, even
when a flare occurs. We argue that Sgr A* was in a state of
quasi-quiescence during the observations on 5 June 2006 for three
reasons. (a) The simultaneous observations with NACO at 2.2 m
show that Sgr A* was not detected at 2.2
m, i.e. no NIR
flare was observed. (b) Models predict the emission of Sgr A* at
8.6
m during a flare to be up to 10 times stronger than the
upper limit reported by us (see Fig. 4). (c)
Differential imaging allows us to derive tight upper limits on the
variability of Sgr A* at 8.6
m during the observations. From a
difference image between the data from 5 June and 6 June an upper
limit of
mJy can be derived on any possible variability of
Sgr A* between these two days. The high quality data from 5 June
allowed an estimate of the upper limit to the variability of Sgr A*
during the observations from differential imaging between individual
images (
30 s of integration time, with 1.5 to 2.5 min time
between them) taken during this night to
mJy. Our
conclusion is that Sgr A* was indeed in a state of quasi-quiescence
during the MIR observations at 8.6
m.
The lower quality of the June 6 data leads to higher limits of
mJy for variability between the individual images. The
L'-band observations show that Sgr A* was continuously variable during
the observations on 6 June. The simultaneously acquired upper limit on
the average flux of Sgr A* at 8.6
m and the limits on shorter
timescale variability at this wavelength are consistent with the
measurements at 3.8
m.
![]() |
Figure 4:
Emission models for Sgr A*
(Yuan et al. 2004,2003). The upper limit on the flux of
Sgr A* at 8.6 ![]() ![]() ![]() |
Open with DEXTER |
Measurements of the Sgr A* quiescent and flaring emission are
indicated in Fig. 4 along with theoretical models
(data and models taken from Yuan et al. 2004,2003). The
simultaneous measurements at 2.2, 3.8, and 8.6 m derived
in this work are indicated in the figure. Two conclusions can be
drawn. (a) The acquired upper limits on the IR flux density of Sgr A*
at 2.2 and 8.6
m set tight constraints on the
quasi-quiescent emission and are consistent with the theoretical
models shown in the figure. The measurements at 3.8 and
8.6
m are consistent with the models, too. The measurements
also agree with the synchrotron/SSC models presented by
Eckart et al. (2006a). Published jet models predict significantly
higher fluxes in the MIR
(see Falcke & Markoff 2000; Yuan et al. 2002), but can probably be
easily adapted to account for the new upper limits. (b) VISIR at the
ESO VLT is sensitive enough to detect flaring emission from Sgr A*
during a bright flare without difficulty and during a fainter flare
during periods of good and stableatmospheric conditions. When
combined with simultaneously-acquired NIR measurements, the MIR-to-NIR
spectral index will allow a flare model where the X-ray emission is due to
acceleration of electrons that produce NIR synchrotron radiation that
is up-scattered into the X-ray domain (steep MIR-to-NIR slope) to be
distinguished from a model where the NIR emission is due to
synchrotron emission from electrons heated to a higher temperature,
while the X-ray emission is due to synchrotron emission from some
electrons that were accelerated into a hard power law distribution
(flat MIR-to-NIR slope).
Acknowledgements
Part of this work was supported by the German Deutsche Forschungsgemeinschaft DFG Sonderforschungsbereich project number SFB 494. We thank the referee for her/his helpful comments.