A
12
CO J = 6–5 MAP OF M82: THE SIGNIFICANCE OF WARM MOLECULAR GAS
John S. Ward
1
and Jonas Zmuidzinas
Department of Physics, 320-47, California Institute of Technology, Pasadena, CA 91125;
john.ward@jpl.nasa.gov, jonas@submm.caltech.edu
and
Andrew I. Harris and Kate G. Isaak
2
Astronomy Department, University of Maryland, College Park, MD 20742;
harris@astro.umd.edu, isaak@mrao.cam.ac.uk
Received 2001 November 17; accepted 2002 December 23
ABSTRACT
We present a
12
CO J ¼ 6–5 map of the nuclear regions of the starburst galaxy M82 at a resolution of 14
00
taken at the Caltech Submillimeter Observatory (CSO). Hot spots were found on either side of the dynamical
center. We compare our results with a high-resolution
12
CO J ¼ 2–1 interferometer map, and present a
12
CO
J ¼ 6–5/
12
CO J ¼ 2–1 line ratio map obtained using a novel deconvolution technique. This line ratio is
highest at the two J ¼ 6–5 integrated intensity peaks, reaching 0.4 and 0.5 in the northeast and southwest
peaks, respectively, and is typically 0.2 elsewhere in the nuclear region. We also present measurements of
12
CO
J ¼ 4–3,
12
CO J ¼ 3–2, and
13
CO J ¼ 3–2, and an upper limit for
13
CO J ¼ 6–5. We analyze these
observations in the context of a two-component large velocity gradient (LVG) excitation model. Likelihood
density curves were calculated for each of the model parameters and a variety of related physical quantities for
the northeast and southwest peaks based on the measured line intensities and their associated uncertainties.
This approach reveals in an unbiased way how well various quantities can be constrained by the CO observa-
tions. We find that the beam-averaged
12
CO and
13
CO column densities, the isotopomer abundance ratio, and
the area filling factors are among the best constrained quantities, while the cool component H
2
density and
pressure are less well constrained. The results of this analysis suggest that the warm gas is less dense than the
cool gas, and that over half of the total molecular gas mass in these nuclear regions is warmer than 50 K.
Subject headings: galaxies: individual (M82) — galaxies: ISM — galaxies: starburst — submillimeter
1. INTRODUCTION
M82, also known as NGC 3034, is a nearby irregular star-
burst galaxy in Ursa Major, and is seen nearly edge-on. Dust
obscures the nuclear region at optical wavelengths. M82 is
the brightest IR galaxy in the sky, and has a total IR
luminosity of 3:8 10
10
L
(Colbert et al. 1999). It is
believed that the starburst was triggered by a close encounter
with M81 about 10
8
yr ago (Yun, Ho, & Lo 1993). The bright
far-infrared (FIR) fine-structure lines seen in the inner kilo-
parsec of M82 imply a far-UV radiation field about 10
3
times
larger than the local solar value (Colbert et al. 1999). This
strong UV field is generated by a population of massive
young stars that also heat the dust which produces the large
IR luminosity. The nucleus has a large molecular gas mass, a
few 10
8
M
, giving rise to strong CO emission (Wild et al.
1992; Weiss et al. 2001). At an estimated current star forma-
tion rate of 1 M
yr
1
, this gas will be consumed in about
2 10
8
yr (Lord et al. 1996). Maps of both molecular line
emission and thermal dust continuum show a double-peaked
structure in the central kiloparsec (Neininger et al. 1998;
Hughes, Gear, & Robson 1994).
The CO J ¼ 6–5 line was first detected toward M82 by
Harris et al. (1991), who showed that molecular gas in star-
burst galaxies is substantially warmer than in typical disk
clouds. The CO J ¼ 6–5 transition probes higher excitation
temperatures (116 K above the ground state) than the more
accessible CO J ¼ 1–0 (5.5 K), CO J ¼ 2–1 (17 K), and CO
J ¼ 3–2 (33 K) trans itions. The higher J spectral lines thus
provide important information needed to understand the
large mass of T 100 K molecular gas that is heated by
massive young stars in a starburst galaxy. Recent improve-
ments in receiver techn ology now allow high-quality maps
to be made of the
12
CO J ¼ 6–5 and J ¼ 7–6 rotational lines
in nearby galaxies. Mao et al. (2000) recently mapped the
12
CO J ¼ 7–6 line. We present here a map of
12
CO J ¼ 6–5.
As improving receiver technology provides astronomers
with ever higher quality data, advances in digital computers
open up possibilities for analyzing observations in new and
better ways. For instance, CO excitation analyses in the
large velocity gradient (LVG) approximation can now be
calculated in a few milliseconds on a typic al modern desk-
top computer. This allows us to evaluate multicomponent
models with many parameters, examining huge volumes of
parameter space to find all combinations of the model
parameters consistent with the measured data rather than
only calculating some narrow range of possible solutions.
Using measured line intensities, we have calculated likeli-
hood density functions for each of the parameters of a two-
component LVG model for the northeast and southwest
CO emission peaks. We have also calculated likelihood
functions for a variety of physical quantities derived from
these parameters, such as the pressure and the beam-
averaged column density. These likelihood curves were used
to find median likelihood estimators and 95% confidence
intervals for the quantities of interest.
We have also developed a novel deconvolution
technique to compute line ratio maps when a high-quality,
1
Current address: Mail Stop 168-314, Jet Propulsion Laboratory, 4800
Oak Grove Drive, Pasadena, CA 91109-8099.
2
Current address: Cavendish Astrophysics, University of Cambridge,
Cambridge CB3 0HE, UK.
The Astrophysical Journal, 587:171–185, 2003 April 10
# 2003. The American Astronomical Society. All rights reserved. Printed in U.S.A.
171
high-resolution map is available for the denominator and
both maps are of the same species. This technique uses all
spatial and velocity information in both maps, and can
calculate the line ratio at a higher spatial resolution than the
numerator map. With this technique, information is not
thrown away, as with the common method of integrating
away the velocity information and smoothing the higher
resolution map before dividing.
2. OBSERVATIONS
All observations were performed at the Caltech Sub-
millimeter Observatory (CSO) in 1998 January and 1999
January using SIS receivers operating in double-sideband
mode. The
12
CO J ¼ 6–5 map was made in excellent
weather (
225 GHz
d0:04), with typical single-sideband
system temperatures (including the atmosphere) around
3000 K. The receiver consisted of a quasi-optical SIS mixer
with an integrated HEMT low noise amplifier (LNA); an
upgraded version of this receiver was described by Ward et
al. (2000). The use of a broadband tunerless receiver
reduces calibration errors associated with receiver sideband
imbalance. A pair of 1024 channel, 500 MHz acousto-
optical spectrometers were used as back ends. The chopping
secondary was switched at a 1 Hz rate with a 60
00
throw in
azimuth. Spectral intensities were calibrated on an ambient-
temperature chopper wheel and the sky periodically during
the observations. Beam maps of Saturn were used to deter-
mine the beam FWHP of approximately 14
00
, and the main
beam efficiency of 30%.
We do not have information about the error beam, which
would contain most of the missing power; as a result, we
are using a conservative estimate of 30% for the total
calibration uncertainty.
The map of M82 consists of 36 positions on 7>5 centers,
and covers 70
00
along the major axis of the galaxy and 50
00
along the minor axis (Fig. 1). At a distance of 3.6 Mpc
(Freedman et al. 1994), this corresponds to an area of
1200 880 pc mapped with a 250 pc FWHP beam. The
coordinates of the J ¼ 6–5 map were corrected for system-
atic pointing offsets by comparing line velocity structure to
12
CO J ¼ 2–1 and C
18
O J ¼ 1–0 Plateau de Bure Interfer-
ometer data cubes from Weiss et al. (2001), according to the
following procedure. A line connecting the antenna temper-
ature peaks of the bright hot spots on either side of the
dynamical center was found. This line was then drawn on
top of a contour map of line center velocity,
v
c
¼
R
T
A
vdv
R
T
A
dv
: ð1Þ
The contours of constant velocity crossed the line at nearly
right angles. A reference point was chosen where the line
connecting the hot spots crossed the 200 km s
1
velocity
contour. The coordinat es of the
12
CO J ¼ 6–5 map were
then adjusted to match the reference point to the coordi-
nates of a point found the same way in the interferometer
maps. The correction was 4>5 in right ascension and 7>2in
declination, for a total shift of about 0.6 of the FWHP
beam. This shift may have arisen because the telescope fixed
pointing offsets were not determined immediately preceding
the observations. The total pointing uncertainty is about 4
00
in each axis. Maps of integrated intensity and peak antenna
temperature are shown in Figures 2 and 3, respectively.
Position-velocity plots are shown in Figure 4.
Fig. 1.—Spectra of
12
CO J ¼ 6–5 in M82. The map has been rotated such that the horizontal offsets are approximately along the major axis. Offsets are in
arcsec from an arbitrary center. The vertical scale ranges from T
MB
of 1 to 4.5 K, and the horizontal scale ranges from 80 to 520 km s
1
.
172 WARD ET AL. Vol. 587
In addition to the
12
CO J ¼ 6–5 map, we have also
observed 17 positions of the
12
CO J ¼ 4–3 line, nine
positions of
12
CO J ¼ 3–2, and five positions of
13
CO
J ¼ 3–2, all with the CSO facility SIS receivers. Most of
these positions were observed as cuts along the major axis of
the galaxy. The beam sizes were 16>5, 24>4, and 25>5
FWHP at 461, 346, and 331 GHz, respectively, with corre-
sponding main beam efficiencies of 0.45, 0.50, and 0.50. The
CO J ¼ 3–2 observations were made with a 4 GHz wide-
band analog correlator spectrometer (WASP) similar to
that described by Harris & Zmuidzinas (2001). The
13
CO
J ¼ 3–2 spectra were scaled up by 14% to account for the
difference in atmospheric transmission in the two sidebands.
We also integrated on the
13
CO J ¼ 6–5 line for a single
Fig. 2.—Integrated intensity of
12
CO J ¼ 6–5 in M82. Contours are 50, 100, 150, 200, 250, 300, 350, and 400 K km s
1
.
Fig. 3.—Peak antenna temperature of
12
CO J ¼ 6–5 in M82. Contours are for T
MB
of 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, and 4.0 K.
No. 1, 2003
12
CO J = 6–5 MAP OF M82 173
pointing in 1998 January and 1999 January for a total of 96
minutes unde r conditions similar to those for the
12
CO
J ¼ 6–5 observations, but did not detect the line.
Zero-order baselines were subtracted from all reported
measurements except
12
CO J ¼ 4–3, which was corrected
for a first-order baseline, and all reported results (including
both the table s and the figures) are given on a main beam
brightness temperature scale (T
MB
).
3. RESULTS
Figure 2 shows two peaks in integrated intensity, the
southwest hot spot being stronger than the northeast. Com-
paring the integrated intensity to Figure 3, we see that the
two hot spots are much more distinct in peak antenna tem-
perature than in integrated intensity. This is due in part to a
spatial smearing effect caused by the antenna beam width
being comparable to the spacing between the hot spots, i.e.,
a14
00
FWHP beam compared to 19
00
peak separation in
Figure 3. The locations of the hot spots in the peak antenna
temperature plot are less affected by smearing, since the hot
spots differ in center velocity by 200 km s
1
.
Figure 5 shows the
12
CO J ¼ 6–5 integrated intensity
plotted over the
12
CO J ¼ 2–1 map from Weiss et al. (2001).
The hot spots in the CO J ¼ 6–5 map are closer together
than in the lower J map. This is consistent with the finding
of Mao et al. (2000) that the angular separation of the lobes
is smaller for a ‘‘ high CO excitation component ’’ than is
seen in the low-J CO lines. It can be seen in the figure that
the J ¼ 6–5 southwe st hot spot is located between the lower
J southwest and center hot spots. This suggests that the
southwest hot spot of the J ¼ 6–5 map includes unresolved
emission from both of these regions. The southwest hot spot
appears to be somewhat extended in the direction of the
minor axis in the J ¼ 6–5 map. It is unlikely that this shape
is real because it is not seen in any other published observa-
tions, including HCN J ¼ 1–0 (Brouillet & Schilke 1993)
and CO J ¼ 7–6 (Mao et al. 2000) maps. It is also unlikely
to be a calibration effect since the weather and receiver were
very stable during the observations. It is possible that the
telescope pointing drifted a few seconds of arc during the
observations, causing this artifact.
The total luminosity of the
12
CO J ¼ 6–5 line was found
to be 9:0 10
5
L
. This was calculated from the spatially
and spectrally integrated intensity of 2:8 10
5
Kkms
1
arcsec
2
after zeroth-order baseline subtraction, assuming
spherically symmetric emission and a distance of 3.6 Mpc.
The
13
CO J ¼ 6–5 line was not detected after 96 minutes
of integration time near the central pointing. The 3 upper
limit of the integrated intensity for a 14
00
beam calculated
from the rms channel noise is 7.9 K km s
1
. A reliable upper
limit may be somewhat larger, however, since uncertainties
in zeroth-order baseline subtraction and a small amount of
baseline ripple could dominate over the channel noise.
Nonetheless, with an implied J ¼ 6–5 antenna temperature
ratio of I(
12
CO)/I(
13
COÞe40, it is clear that
13
CO J ¼ 6–5
is very weak. Although
13
CO J ¼ 6–5 emission was not
detected, 450 lm continuum was observed at a level of
2:4 0:6Jybeam
1
. Thi s is consistent with the bolometer
value of 3.5 Jy beam
1
measured at the same location with a
narrower 9
00
FWHP beam by Hughes et al. (1994).
3.1. A
12
CO J ¼ 6 5/
12
CO J ¼ 2 1 Line Ratio Map
Figure 6 shows a map of the
12
CO J ¼ 6–5/
12
CO J ¼ 2–1
line ratio. The J ¼ 2–1 data were observed with the Plateau
de Bure Interferometer and include short-spa cing correc-
tions based on obs ervations with the IRAM 30 m telescope
(Weiss et al. 2001). The ratio map was calculated using a
Fig. 4.—Position-velocity diagram along the major axis of M82 for
12
CO J ¼ 4–3 and
12
CO J ¼ 6–5. The southwest hot spot peaks at different velocities in
the two transitions, and the overall velocity gradient is steeper in CO J ¼ 6–5. Note that the broad, lower intensity spectrum between the two hot spots explains
the difference in integrated intensity and peak temperature maps. The positions are offset from arbitrary centers along the major axis of the galaxy. Contours
are for T
MB
from 0.5 to 4 K in steps of 0.5 K.
174 WARD ET AL. Vol. 587
novel deconvolution technique that takes advantage of
velocity information in the J ¼ 6–5 map along with the high
spatial resolution of the J ¼ 2–1 map to improve the resolu-
tion and accuracy of the ratio map. A Lagrange multiplier
was used to set the relative weighting between minimizing
the resulting
2
and maximizing the smoothness of the
resulting ratio map. Details of the calculation are given in
the Appendix. It is immediately apparent that the brightness
temperature ratio is highest at the two integrated intensity
peaks, reaching 0.4 and 0.5 in the northeast and southwest
peaks, respectively, and is typically 0.2 elsewhere in the
nuclear region.
Fig. 5.—M82
12
CO J ¼ 6–5 integrated intensity contours superimposed on
12
CO J ¼ 2–1 integrated intensity from Weiss et al. (2001). Contours are 50,
100, 150, 200, 250, 300, 350, and 400 K km s
1
.
Fig. 6.—Deconvolved
12
CO J ¼ 6–5/
12
CO J ¼ 2–1 brightness temperature ratio map. Contours are from 0.1 to 0.45 by 0.05. The squares indicate the
locations of the J ¼ 6–5 intensity peaks, and the triangles the locations of the J ¼ 2–1 integrated intensity peaks.
No. 1, 2003
12
CO J = 6–5 MAP OF M82 175
