Diffraction Limited Imaging Spectroscopy of the SgrA* Region using OSIRIS, a new Keck Instrument
ABSTRACT We present diffraction limited spectroscopic observations of an infrared flare associated with the radio source SgrA*. These are the first results obtained with OSIRIS, the new facility infrared imaging spectrograph for the Keck Observatory operated with the laser guide star adaptive optics system. After subtracting the spectrum of precursor emission at the location of Sgr A*, we find the flare has a spectral index of -2.6 +- 0.9. If we do not subtract the precursor light, then our spectral index is consistent with earlier observations by Ghez et al. (2005). All observations published so far suggest that the spectral index is a function of the flare's K-band flux. Comment: paper accepted for publication in ApJ Letters
arXiv:astro-ph/0605253v1 10 May 2006
To appear in ApJ Letters
Diffraction Limited Imaging Spectroscopy of the SgrA* Region
using OSIRIS, a new Keck Instrument
A. Krabbe, C. Iserlohe
I. Physikalisches Institut, Universit¨ at zu K¨ oln, 50937 K¨ oln, Germany
J. E. Larkin, M. Barczys, M. McElwain, J. Weiss, S. A. Wright
Division of Astronomy, University of California, Los Angeles, CA, 90095-1562, USA
Leiden Observatory, P.O. Box 9513, NL-2300 RA Leiden, The Netherlands
We present diffraction limited spectroscopic observations of an infrared flare
associated with the radio source SgrA*. These are the first results obtained with
OSIRIS, the new facility infrared imaging spectrograph for the Keck Observatory
operated with the laser guide star adaptive optics system. After subtracting the
spectrum of precursor emission at the location of Sgr A*, we find the flare has a
spectral index (F(ν) ∝ να) of α = −2.6±0.9. If we do not subtract the precursor
light, then our spectral index is consistent with earlier observations by Ghez et
al. (2005). All observations published so far suggest that the spectral index is a
function of the flare’s K-band flux.
Subject headings: Spectroscopy: infrared, imaging — Telescopes: Keck — In-
struments: OSIRIS — individual(Galactic Center, SgrA*)
– 2 –
The low luminosity of the Supermassive Black Hole (SBH) in the center of our Galaxy
is a standing puzzle and challenges our understanding of the mechanisms of mass accretion
as well as the physics close to the event horizon. Infrared variability of SgrA*, first reported
by Genzel et al. (2003) and Ghez et al. (2004), has become an important observable. The
timescales (few minutes) imply that emission arises from the immediate environment outside
of the SBH. Due to the accretion’s favorable duty cycle, flares are not only fairly frequent, but
they are also readily observable with adaptive optics at 8-10m class telescopes (Cl´ enet et al.
2005; Ghez et al. 2005). The infrared variability is closely linked to the X-ray variability
detected a few years earlier (Baganoff et al.
al. 2003). Simultaneous infrared/X-ray observations have demonstrated that the flares are
related multiwavelength phenomena (Eckart et al. 2004). The spectral index of the flare and
its possible variation from flare to flare and during a single flare are important parameters
that will allow us to determine the emission process of the radiation (Yuan et al. 2003, 2004;
Liu et al. 2004). We observed SgrA* with OSIRIS, a new Keck facility instrument, during
its commissioning time (Larkin et al. 2003; Krabbe et al. 2004; Quirrenbach et al. 2003;
Weiss et al. 2002). These data are the first laser guide star (LGS) assisted spectra ever taken
of the Galactic Center region.
2001; Porquet et al.2003; Goldwurm et
2. Observations and Data reduction
The OSIRIS (OH Suppressing InfraRed Imaging Spectrograph) instrument is a new fa-
cility near infrared (NIR) Z- to K-band imaging spectrograph designed for the Keck Obser-
vatory’s Adaptive Optics (AO) system. It utilizes an array of micro lenses and a HAWAII-2
detector (2048×2048 pixels) to simultaneously obtain more than 1020 spectra over a rectan-
gular field of view with about 16×64 spatial positions. Each spectrum in this mode covers
about 1700 channels at a spectral resolution of R = 3700. OSIRIS achieves high sensitivity
with good instrumental throughputs, low backgrounds per AO resolution element and high
enough spectral resolution to work between the night sky lines. First light was achieved
on February 22, 2005, and commissioning will be completed in 2006. A detailed account of
OSIRIS will be given by Larkin et al. (in preparation).
The Galactic Center was observed on April 29, 2005 at Keck II during the first deploy-
ment of OSIRIS with the LGS AO system (Wizinowich et al., in prep., van Dam et al., in
prep.). Two consecutive K-band frames, each with 300 s integration time were obtained at
an airmass of 1.65 with SgrA* in the field of view (FOV). Overhead between the frames was
about 37 s. The frames were followed by an empty sky field obtained immediately after the
– 3 –
2ndframe. The angular scale of OSIRIS was set to 20 mas/pixel.
A dedicated OSIRIS data reduction pipeline has been produced, and was used to iden-
tically reduce both exposures. After sky subtraction, the individual spectra in each frame
were extracted from the raw frame, using a special map of the point spread function of each
lenslet at all wavelengths. This allows the removal of crosstalk from adjacent spectra and
correct assignment of flux to each field position over the 2-dimensional field. Arc line spec-
tra are used to calibrate the wavelength scale of each field point. Atmospheric differential
dispersion effects were also corrected by tracing the peak emission of the stellar continuum
through the wavelength slices of the cube. Telluric and instrumental transmission, as well as
foreground extinction to the Galactic Center, were corrected in both data cubes by dividing
all spectra by the average spectrum of star S2. All spectra were finally multiplied by a
black-body curve of T = 30000 K, representing the approximate flux density of S2 which
is assumed to be between spectral classes O8V and B0V based on spectra from Ghez et al.
(2003). Since we didn’t attempt to model the stellar absorption lines, e.g., Brγ at 2.166 µm,
the final spectra do not contain valid information about emission or absorption lines at those
spectral positions and were ignored in our analysis.
Figure 1 has two panels, hereafter referred to as frame 1 and frame 2, that display the
image produced by collapsing all of the wavelength channels from 2.02 µm through 2.38 µm
for each of the two frames. Some of the stars are labeled according to Eisenhauer et al.
(2005). Frame 2 was observed 0.3′′south with respect to Frame 1. The angular resolution
on the sky is 60 mas at a pixel scale of 20 mas. The angular resolution achieved is worse
than the diffraction limit of 46 mas at 2.18 µm, probably due to the low telescope elevation.
The staggered edge along the sides of each data cube is standard for OSIRIS and is due to
the complex mapping of lenslets onto the detector. Due to constraints in the commissioning
schedule, a global flat was not available for these observations. This means that individual
spatial locations have a well calibrated spectrum, but based on comparisons of known stellar
fluxes in the field, the relative intensity of one spectrum to another is uncertain at the 20%
level (reflected in the flux errors in Table 1). Star S2 has a well determined flux and was
used to flux calibrate both frames individually.
The flare at the position of SgrA* is apparent in frame 2. It is located 145±5 mas south
and 22±5 mas west of S2 at the epoch of the observations: MJD53489.624680. Inspecting
the identical position in frame 1 reveals a weaker but still notable emission at the location
– 4 –
of Sgr A*, which we will term the precursor. In both frames, the emission is close to other
stars making a proper background subtraction important and slightly difficult. The levels
and spectra of the backgrounds at SgrA* and S2 are certainly not identical and had to be
treated individually. In the end the spectrum of SgrA* was determined from frame 2 using
a 5 pixel × 4 pixel box centered on the position of the flare. The general background was
determined by measuring the average pixel value in the 22 pixels that form a circumference
around the 5×4 aperture. The spectrum of S2 was determined in an identical fashion from
the same frame, in order to reduce systematic errors that could result from using different
fractions of the point spread function or by measuring the background in a different manner.
An identical procedure was applied to frame 1 to extract the precursor spectrum, except a
5×5 pixel aperture was used due to a fractional shift of the lenslet grid on the sky between
frames 2 and 1. Again, the size and geometry of the box was identical to that used for the
extraction of the S2 spectrum discussed earlier. Assuming that the flare is unresolved in our
data, the extraction regions for SgrA* and star S2 then cover the same fraction of the PSF
for each source.
The resulting spectra are shown in Figure 2. The lowest spectrum represents the pre-
cursor emission in frame 1. The middle spectrum is the extracted spectrum of the flare from
frame 2, and has has been shifted vertically by one unit for clarity. The upper spectrum is
the difference between the spectrum in frame 2 and the spectrum in frame 1, and we will
refer to this as the flare spectrum. Again for clarity, it has been shifted vertically by 4 units.
All three of the spectra have been smoothed by a 30 pixel wide boxcar filter. The sky frame
was obtained after both frames 1 and 2 and was a poorer match for frame 1. This was par-
ticularly important in the spectral range between 2.04 µm and 2.06 µm and this region was
masked out in frame 1. Atmospheric OH lines, however, subtracted out well. All three of
the spectra were fit with a power law (F(λ) ∝ λm) and the resulting fits and slopes (m) are
given on Figure 2 along with 1σ errors. These slopes were converted into a frequency power
law index α (Fν∝ να). Table 1 shows all of the derived quantities including the slope and
spectral index. We note here that the spectral index of the flare is quite red (α=−2.6±0.9).
From the spectral fits, the K-band flux of the flare and the precursor were determined
relative to S2 by extrapolating the power law fit from (2.02 µm, 2.38 µm) to the full K-band.
We assumed that S2 has a magnitude of mK = 13.9 mag (24 mJy) based on photometry
from Ghez et al. (2003). The K-band flux of the precursor emission at the location of SgrA*
in frame 1 then becomes 3.5 mJy, or 6.9 times fainter than S2. The corresponding flux at the
location of Sgr A* in frame 2 is 9.6 mJy, or 2.5 times fainter than S2. This is an increase in
the K-band flux of SgrA* by a factor of 2.8 within 5.6 minutes (exposure time plus overhead
– 5 –
4.Discussion and Summary
While frame 2 (Fig. 1b) obviously shows a flare at the position of SgrA*, it is less
clear what is at the same location in frame 1 (Fig 1a). Is it a true quiet state or a mild
flare precursor? A faint source at the location of SgrA* has been found by other observers,
including in high resolution H-band images by Eisenhauer et al.
instrument. Photometric K-band data by Genzel et al (2003) indicate that SgrA* can be
more than 6 times fainter than S2. Eckart et al.
12 times fainter than S2 corresponding to a K-band flux density of 2 mJy. Comparing this
factor with Table 1 suggests that the K-band emission in frame 1 is less than a factor of
2 brighter than the lowest activity level measured and may indeed be dominated by an
underlying constant source. The spectral index of the source in frame 1 (α=2.7±1.3) is also
consistent with a blackbody at a temperature around 3000 K. This suggests the possibility
that the emission at the location of SgrA* during low activity may be dominated by a
different mechanism than the flare itself. One option is a stellar component either in the
background or at the very cusp of the stellar cluster within less than a light day of SgrA*.
If this is the case, then these faint stars are also orbiting in the immediate vicinity of SgrA*.
(2005) with the NACO
(2004) report the lowest activity level
The temporal separation between the two frames is 337 seconds (exposure times of 300
sec with a gap of 37 seconds). Within this time or shorter, the K-band flux increased by more
than a magnitude to 9.6 mJy. This peak flux compares very well with Eckart et al. (2004)
and also with Ghez et al. (2005) for the maximum observed activity. Our observations thus
very likely record the beginning of a flare with a typical level of activity. We also conclude
that the intrinsic K-band flux of the flare is probably best represented by the difference
between the peak value of 9.6 mJy and the lowest background of 2 mJy, yielding (7.6±34%)
mJy. This value has been used in Figure 3.
Eisenhauer et al. (2005) were first to report measurements of the spectral index of the
emission of SgrA* in the NIR. Their spectral indices α for the flux density Fνlie in the range
between -3.3 and -4.8, with an averaged index of αE= −4 ± 1 for fluxes < 2 mJy. Their
value has been obtained in a similar fashion to ours, in that two spectra of different activity
levels were subtracted from each other. However, their result is different from our value of
αK= −2.6 ± 0.9 (see Table 1 and Figure 3).
Figure 3 summarizes all published spectral indices for SgrA*’s K-band flares as a function
of their 2 µm flux density. In a recent paper, Ghez et al. (2005) report spectral indices based
on K′and L′imaging. Their method of determining α is directly based on K′-L′PSF fitting
photometry. However, different from our result and from Eisenhauer et al.
result does not account for flux from the precursor source. Thus their value of α = -0.5 has
a different meaning than our flare index. It is more correctly compared to the slope of the