Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~marel/abstracts/psdir/lapal01_ngc6240_4.ps Äàòà èçìåíåíèÿ: Fri Jun 22 18:14:47 2001 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 17:17:36 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 11

The Central kpc of Starbursts and AGN
ASP Conference Series, Vol. xxx, 2001
J. H. Knapen, J. E. Beckman, I. Shlosman, and T. J. Mahoney
HST observations of NGC 6240
Joris Gerssen, Roeland P. van der Marel
Space Telescope Science Institute, Baltimore, USA
David Axon
University of Hertfordshire, Hatfield, UK
Chris Mihos
Case Western Reserve University, Cleveland, OH, USA
Lars Hernquist
Harvard University, Cambridge, MA, USA
Joshua E. Barnes
Institute for Astronomy, Honolulu, HI, USA
Abstract. WFPC2 images and STIS spectroscopic observations are
presented of the double nucleus in the merger system NGC 6240. We
find that: (a) the kinematics of the ionized gas is similar to that of the
molecular gas, despite a di#erent morphology; (b) the gaseous and stellar
kinematics are quite di#erent, suggesting an early merger stage; (c) nei­
ther the gaseous nor the stellar kinematics show an obvious sign of the
supermassive black hole believed to be responsible for the X­ray emission
of NGC 6240; and (d) the steep o#­nuclear velocity gradient is not due
to a # 10 11 M# black hole, in contrast to earlier suggestions.
1. Introduction
The luminous IR galaxy NGC 6240 has both the tidal tails and a double nucleus
that are characteristic of a merging system. Since its identification by the IRAS
satellite as one the most nearby ULIRGs, NGC 6240 has been the subject of
numerous studies at virtually all wavelengths.
One of the key results emerging from these studies is that NGC 6240 har­
bors both a starburst and an AGN. Several lines of evidence indicate the pres­
ence of young stars, including e.g. the observed strength of the CO bandhead
(Rieke et al. 1985). However, observations with various X­ray satellites show a
strong X­ray component that can only originate from an AGN (e.g., Vignati et
al. 1999). Spectroscopic line diagnostics in the mid­IR support the view of a
1

2 Gerssen et al.
B V
I K
Halpha
Figure 1. (a; left) Mosaic of HST images in the B, V , I and K bands
from WFPC2 and NICMOS (each frame is 8 by 8 arcsec). (b; right)
Narrow­band WFPC2 H # +[NII] image (20 by 20 arcsec). The cross
marks the position of the hypothesized black hole that is discussed in
Section 3. The orientation is the same in all panels, with north in the
top left corner.
composite AGN/starburst source, but indicate that most of the IR luminosity
of # 5 â 10 11 L# must be powered by the starburst (Genzel et al. 1998).
To improve our understanding of the double nucleus we have studied this
region at high resolution using the Hubble Space Telescope (HST; GO Programs
6430 and 8261; PI: van der Marel). We have obtained WFPC imaging and
longslit STIS spectra to determine both the gaseous and the stellar kinematics.
An additional set of emission line spectra was obtained some 12 arcsec from the
double nucleus to examine the nature of a large velocity gradient that exists at
this location (Bland­Hawthorn, Wilson & Tully 1991).
2. Double Nucleus
2.1. Imaging
The two nuclei of NGC 6240 are separated by 1.8 arcsec or about 0.8 kpc. Fig­
ure 1a shows close­ups of our WFPC2 observations in the B, V and I bands
(filters F450W, F547M and F814W, respectively), together with a K­band NIC­
MOS image retrieved from the HST Data Archive (filter F222M, GO program
7219, PI: Scoville). There is a clear change in the morphology of the double
nucleus as the wavelength increases, due to strong dust absorption, especially
on the southern nucleus. This is consistent with the previous finding from lower­
resolution ground­based data that the distance between the flux­centroids of the
two nuclei appears to decrease with wavelength (Schulz et al. 1993).

APS Conf. Ser. Style 3
Figure 1b shows our narrow band WFPC2 H # +[NII] image. The double
nucleus is visible also in ionized gas. The large­scale morphology of the gas has a
highly filamentary structure that is characteristic of a starburst wind (Heckman,
Armus & Miley 1987).
2.2. Gas Kinematics
To map the velocity field of the ionized gas in the nuclear region we obtained
a set of STIS H # +[NII] emission line spectra with the slit parallel to the line
that connects the nuclei. We used slit widths of 0.1 ## and 0.2 ## to cover an area
with a full width of 1.1 ## . The spectra were analysed by fitting Gaussians to
the emission lines. The best­fitting Gaussian parameters yield the flux, radial
velocity and velocity dispersion. The results are shown as contour maps as a
function of position in Figure 2a--c.
The flux distribution corresponds well with that in the H # +[NII] narrow
band image. Although the northern nucleus is less luminous than the southern
nucleus in continuum emission, it is more centrally concentrated and has a higher
peak intensity in the emission lines.
The velocity field does not show clear signs of rotation around either of
the two nuclei. However, it does show a steep velocity gradient of 400 km s -1
arcsec -1 between the two nuclei, as shown in Figure 2d. The velocity field of
molecular CO gas obtained by Tacconi et al. (1999) yields a similar picture, with
the same velocity gradient that we observe in H # +[NII] (same peak to peak
velocity amplitude and turn­over radius). While Tacconi et al. observe that
the CO velocity field is highly complex, they are able, with some simplifying
assumptions, to fit it with a model of a rotating disk between the two nuclei.
This is not unreasonable, given that the observed CO flux peaks between the
two nuclei. However, the ionized gas emission that we observe clearly does not
peak between the two nuclei, cf. Figure 2a. It is therefore surprising that the
ionized gas and the molecular gas do have very similar kinematics.
Current N­body simulations do not yet have su#cient resolution to follow
the gaseous and stellar kinematics inside the central kpc of mergers in much
detail. However, they do indicate that the dynamics can be complex and out of
equilibrium, so the interpretation of the gas velocities in NGC 6240 as due to
organized disk rotation should be viewed with some caution. Note that gas disks
are often observed around the nuclei of ULIRGs, but not generally in between
the nuclei.
The velocity dispersion map of the ionized gas shows peaks that roughly
coincide with the two nuclei. However, the di#erence between the maximum and
minimum velocity dispersions is small, less than 25 percent. To first order, the
velocity dispersion of the ionized gas is consistent with # 250 km s -1 over the
entire nuclear region. One of the goals of the gas kinematical observations was to
find evidence for broad emission lines that could identify the AGN responsible
for the X­ray emission. However, we see no sign of broad emission lines or
strongly increasing line widths towards either of the nuclei.
2.3. Stellar Kinematics
We also obtained STIS spectra at the near­IR Ca triplet to study the stellar
kinematics in the region of the double nucleus. Due to the lower S/N of absorp­

4 Gerssen et al.
Figure 2. Kinematics of the central region of NGC 6240 derived from
HST STIS spectroscopy. The panels on the left are two dimensional
contour plots. The horizontal direction corresponds to the position an­
gle that connects the nuclei. The northern nucleus is on the left. The
panels on the right are kinematical profiles along the line that con­
nects the nuclei. (a; top left) contour plot of the flux distribution of
H # +[NII]. The double nucleus is clearly visible. (b; middle left) veloc­
ity field of the ionized gas; dotted contours indicate negative velocities.
(c; bottom left) contour map of the velocity dispersion of the ionized
gas. The dispersion peaks near the two nuclei, and is lower in between.
(d; top right) mean line­of­sight velocity of the gas along the line that
connects the nuclei. There is no obvious sign of rotation around either
of the two nuclei, but there is a strong velocity gradient between the
two nuclei. (e; middle right) mean line­of­sight velocity of the stars.
(f; bottom right) the stellar velocity dispersion. Arrows indicate the
positions of the nuclei.
tion lines it was not possible to make a two­dimensional map of the kinematics,
so instead only a single spectrum was obtained with the slit along the line con­
necting the nuclei. The inferred mean velocities and velocity dispersions of the
stars are shown in Figures 2e,f.
The stellar kinematics di#er considerably from the gas kinematics. The
large velocity gradient seen in the ionized gas is absent in the stellar kinematics
and there appears to be very little rotation around the nuclei. However, a ground
based stellar velocity field derived from the CO bandhead (Tezca et al. 2000)
does show rotation around both nuclei but at large angles to the position angle
of the line that connects the nuclei.
The stellar velocity dispersion (derived after binning rows along the spec­
trum to increase signal­to­noise) peaks at the nuclei. At both nuclei the mea­
sured velocity dispersion is around 200 km s -1 , well below earlier, spatially

APS Conf. Ser. Style 5
Figure 3. The ionized gas velocity field at the position of the hy­
pothesized o#­nuclear black hole (cross; see also Figure 1b). (a; top)
Model velocity field for a Keplerian disk around a 10 10 M# black hole.
(b; bottom) Velocity field observed with HST at # 0.5 ## resolution (the
model in the top panel was smoothed to the same resolution). The ob­
served kinematics are not consistent with the hypothesized black hole.
The x­axis of both panels lies in the direction of position angle 155 # .
unresolved, observations of # 350 km s -1 (Lester & Ga#ney 1994; Doyon et
al. 1994), but consistent with the more recent observations by Tezca et al. The
signal­to­noise ratio between the two nuclei is too low, even after rebinning, to
reliably determine the stellar velocity dispersion. Tezca et al. find that in their
data the velocity dispersion actually peaks between the nuclei. However, the dy­
namical importance of this observation is unclear because N­body simulations
show that projection e#ects can lead to spurious peaks in the observed velocity
dispersions (Mihos 2000).
Stars are collisionless objects, unlike gas which can be influenced by shocks,
infall and starburst winds. In general stars therefore trace the gravitational
potential of a galaxy more reliably than gas. One of the goals of the stellar
kinematical observations was to find evidence for an increasing velocity disper­
sion towards either of the nuclei, which would have provided evidence for the
gravitational influence of the black hole that is responsible for the X­ray emis­
sion. However, we do not observe the large velocity dispersions that would
unambiguously identify a black hole.

6 Gerssen et al.
3. The Hypothesized O#­Nuclear Black Hole
Bland­Hawthorn et al. (1991) reported the presence of a steep velocity gradient
in a ground­based H # velocity field of NGC 6240 at a projected distance of 6 kpc
from the double nucleus (marked by a cross in Figure 1b). This was interpreted
with a model in which the velocity gradient is due to rotation around a black hole
of # 10 11 M# . This would be quite remarkable, given that the most massive black
hole ever convincingly detected is only # 3â10 9 M# (e.g., Kormendy & Gebhardt
2001). Also, no counterpart is seen at the position of the hypothesized black
hole in optical continuum, ionized gas, radio emission or X­ray wavelengths.
We have mapped the H # velocity field around the hypothesized black hole
using STIS spectra obtained with the # 0.5 ## wide slit placed at several parallel
positions (Figure 3b). While the resulting spatial resolution is admittedly low for
HST standards, it is still a few times better than for the best available ground­
based data. Although a steep gradient is observed in the HST data, the gradient
is not steeper than the measured ground based value. This is not consistent with
expectation if there were indeed a # 10 11 M# black hole present. Then the gradi­
ent should be steeper when observed at higher spatial resolution. If the velocity
gradient observed with HST were interpreted with a model of gas orbiting a
black hole, the implied black holes mass would only be # 10 10 M# . However,
such a model is inconsistent with the observed two­dimensional structure of the
velocity field. The predicted velocity field at the observational resolution for
an inclined Keplerian disk is shown in Figure 3a. The velocity gradient quickly
flattens perpendicular to the kinematical major axis, by contrast to the observed
gradient. Hence, the observed velocity field is inconsistent with the signature
expected from a black hole. The location of the observed velocity gradient along
the extension of a filament (see Figure 1b) suggests instead that it is due to
kinematic gradients in the starburst wind, or possibly projection e#ects.
References
Bland­Hawthorn, J., Wilson, A. S., & Tully, R. B. 1991, ApJ, 371, L19
Doyon, R., et al. 1994, ApJ, 437, L23
Genzel, R., et al. 1998, ApJ, 498, 579
Heckman, T. M., Armus, L., & Miley, G. K. 1987, AJ, 93, 276
Kormendy, J., & Gebhardt, K. 2001, in `Proc. 20th Texas Symposium', AIP, in
press [astro­ph/0105230]
Lester, D. F., & Ga#ney, N. I., 1994, ApJ, 431, L13
Mihos C., in `Dynamics of Galaxies: from the Early Universe to the Present',
Combes F. et al., eds., ASP Conference Series, Vol. 197, 275, 2000
Rieke, G. H., et al. 1985, ApJ, 290, 116
Schulz, H., Fried, J.W., R˜oser, S., & Keel, W. C., 1993, A&A, 277, 416
Tacconi, L. J., et al. 1999, ApJ, 524, 732
Tecza, M., et al. 2000, ApJ, 537, 178
Vignati, P., et al. 1999, A&A, 349, L57