ALMA CO and VLT/SINFONI H2 observations of the Antennae overlap region: mass and energy dissipation

Abstract

We present an analysis of super-giant molecular complexes (SGMCs) in the
overlap region of the Antennae galaxy merger, based on ALMA CO(3-2)
interferometry and VLT/SINFONI imaging spectroscopy of H2 1-0 S(1) at angular
resolutions of 0.9" and 0.7", respectively. All but one SGMC have multiple
velocity components offset from each other by up to 150 km/s. H2 line emission
is found in all SGMCs and the kinematics of H2 and CO are well matched. H2/CO
line ratios vary by up to a factor of 10 among SGMCs and different velocity
components of the same SGMCs. We also identify the CO counterpart of a bright,
compact source of near-IR H2 line emission, which shows no Brgamma, and was
first identified with SINFONI. This source has the highest H2/CO line ratio,
and coincides with the steepest CO velocity gradient of the entire overlap
region. With a size of 50 pc and a virial mass of a few 10^7 Msun it is perhaps
a pre-cluster cloud that has not yet formed significant numbers of massive
stars. We present observational evidence that the H2 emission is powered by
shocks, and demonstrate how the H2 1-0 S(1) and the CO(3-2) lines can be used
as tracers of energy dissipation and gas mass, respectively. The variations in
the H2/CO line ratio may indicate that the SGMCs are dissipating their
turbulent kinetic energy at different rates. The compact source could represent
a short (~ 1 Myr) evolutionary stage in the early formation of super-star
clusters.

Full text

Page 1

arXiv:1201.2942v1 [astro-ph.CO] 13 Jan 2012
Astronomy & Astrophysics manuscript no. herrera˙antennaeb
January 17, 2012
c ? ESO 2012
Letter to the Editor
ALMA CO and VLT/SINFONI H2observations of the Antennae
overlap region: mass and energy dissipation⋆
C. N. Herrera1⋆⋆, F. Boulanger1, N. P. H. Nesvadba1, and E. Falgarone2
1Institut d’Astrophysique Spatiale, UMR 8617 CNRS, Universit´ e Paris-Sud 11, 91405 Cedex Orsay, France
2LERMA, UMR 8112 CNRS, Ecole Normale Sup´ erieure and observatoire de Paris, Paris, France
Preprint online version: January 17, 2012
ABSTRACT
We present an analysis of super-giant molecular complexes (SGMCs) in the overlap region of the Antennae galaxy merger, based on
ALMA CO(3−2) interferometry and VLT/SINFONI imaging spectroscopy of H21−0 S(1) at angular resolutions of 0.9′′and 0.7′′,
respectively. All but one SGMC have multiple velocity components offset from each other by up to 150 km s−1. H2line emission
is found in all SGMCs and the kinematics of H2and CO are well matched. H2/CO line ratios vary by up to a factor of 10 among
SGMCs and different velocity components of the same SGMCs. We also identify the CO counterpart of a bright, compact source
of near-IR H2line emission, which shows no Brγ, and was first identified with SINFONI. This source has the highest H2/CO line
ratio, and coincides with the steepest CO velocity gradient of the entire overlap region. With a size of 50 pc and a virial mass of a
few 107M⊙it is perhaps a pre-cluster cloud that has not yet formed significant numbers of massive stars. We present observational
evidence that the H2emission is powered by shocks, and demonstrate how the H21−0 S(1) and the CO(3−2) lines can be used as
tracers of energy dissipation and gas mass, respectively. The variations in the H2/CO line ratio may indicate that the SGMCs are
dissipating their turbulent kinetic energy at different rates. The compact source could represent a short (∼ 1 Myr) evolutionary stage
in the early formation of super-star clusters.
Key words. Galaxies: individual: Antennae – Galaxies:ISM – Radio lines: ISM – Infrared: ISM – Turbulence
1. Introduction
Major gas-rich mergers are important sites of star formation and
galaxy evolution in the Universe. The Antennae galaxy merger
(NGC4038/4039) is an ideal target for studying in detail how
galaxy interactions affect the interstellar medium and star for-
mation. Most stars in the Antennae form in super-star clusters
(SSCs) with stellar masses up to a few 106M⊙(Whitmore et al.
2010) located where the two galaxies permeate each other, the
’overlap region’. Super-giant molecular complexes (SGMCs)
with masses of several 108M⊙and sizes of ∼500 pc have been
identifiedinCO(1−0)intheoverlapregionwiththeOVROinter-
ferometer (Wilson et al. 2000). Ueda et al. (2012) have recently
reported higher resolution (∼ 100 pc), CO(3-2) observations of
the Antennae obtained with the SMA.
The formation of SSCs involves a complex interplay of
merger-driven gas dynamics, turbulence fed by the galaxy in-
teraction, and dissipation of the kinetic energy of the gas.
Hydrodynamic simulations suggest that massive complexes of
cold gas, akin to SGMCs, form where gas flows trigger com-
pression,coolingand gravitationalfragmentation(Teyssier et al.
2010). Within SGMCs, a hierarchy of structures must form
including clouds that are sufficiently massive to form SSCs
(Weidner et al. 2010).
Recent VLT/SINFONI imaging spectroscopy of the peak of
pure-rotationalH2emission in the overlap region previously ob-
served with Spitzer (Herrera et al. 2011, H11 hereafter)revealed
⋆Based on ALMA Science Verification data and observations with
the VLT/SINFONI, Program IDs 383.B-0789 and 386.B-0942.
⋆⋆supported by a CNRS-CONICYT grant.
bright diffuse H2line emission associated with an SGMC and a
compact(∼0.6′′, ∼50pc) source.H11 proposedthat the compact
source may be a massive cloud on its way to form a SSC within
the next few Myr. The H2lines are powered by shocks and trace
energy that is being dissipated and radiated away as the cloud
complex, and a pre-cluster cloud (PCC) within, grow through
gas accretion.
Herrera et al. (2011) hadsub-arcsecondresolutionSINFONI
data of shocked gas, but lacked CO observations at similar res-
olution, which are required to probe the bulk of the gas on the
relevant scales of ≤100 pc. In this letter, we take advantage of
the recently released ALMA science verificationobservationsof
CO(3−2) in band 7 (∼345 GHz) at ∼1′′angular resolution to
compare the morphology and kinematics of CO(3−2) and H2
line emission in the Antennae overlap region. We show that to-
gether the ALMA and VLT observations provide the comple-
mentary (mass and energetics) data needed to characterize the
dynamical state of SGMCs and to search for pre-cluster clouds.
2. Comparison of ALMA and SINFONI observations
Our analysis relies on two data sets. First, CO(3−2) line emis-
sionwasobtainedduringALMAscienceverification.Thesedata
are part of a mosaic of the Antennae obtained in 10 hr of ob-
serving time in band 7 (345 GHz) between May and June 2011,
with 10 to 13 antennae and baselines from 25 to 200 m. This
gives an angular resolution of 0.′′6×1.′′1 (66 pc×115 pc, at a dis-
tance of 22 Mpc) and covers the entire overlap region. The data
have an intrinsic spectral resolution of 0.85 km s−1and were
binned into channels of 10 km s−1. We used the reduced data
1

Page 2

C. N. Herrera, F. Boulanger, N. P. H. Nesvadba, and E. Falgarone: CO and H2observations of the Antennae overlap region
Fig.1.Left. ALMACO morphologyshownontopofourCFHT K-bandcontinuumimage(H11).Dottedboxesmarkthetwo ALMA
mosaics, the solid box marks the overlap region. Right. H21−0 S(1) morphology as seen with SINFONI. Boxes mark individual
SINFONI fields-of-view, contours show CO(3−2) from 2 to 42 Jy km s−1beam−1in steps of 8 Jy km s−1beam−1. The inset at
the bottom left of the right panel shows the ALMA beam. We also mark massive and young SSCs (asterisks), and the compact H2
source PCC discussed in § 3.
cubes publicly available on the ALMA website1, which we cor-
rectedfor the primarybeamattenuation.Lowspatial frequencies
were filtered out because of missing short spacings that cause a
loss of extended structures >4′′and negative sidelobes adjacent
to bright emission. To measure fluxes we used a clipped cube
where all pixel values <2σ (6 mJy beam−1) were set to zero. To
quantify the missing flux, we constructedspectra at the center of
the overlap region with the angular resolution of the single-dish
JCMT and HHT CO(3−2) observations of Zhu et al. (2003) and
Schulz et al. (2007), respectively. We found about half the total
flux of the single-dish data. The total flux of the SGMCs in the
overlap region agrees with the SMA observations (Ueda et al.
2012). Second, we used VLT/SINFONI imaging spectroscopy
of ro-vibrationalH2lines in the near-infraredK-band at R=3000
in four regions each 8′′×8′′in size (Fig. 1). They were observed
in February2011with on-sourceintegrationtimes of 40 minutes
per pointing. Our previous observations of SGMC 2 have been
discussed by H11. We obtained and reducedthe data of the three
additional fields in a similar way.
Fig. 1 shows the spatial distribution of the H2and CO emis-
sion in the Antennae overlap region at a spatial resolution of
≤1′′(100 pc). The right panel displays the color image of the
H21−0 S(1) emission with CO(3−2) contours for comparison.
Intensity maps were obtained from the SINFONI and ALMA
data cubes by fitting Gaussian profiles to the spectra in each spa-
tial pixel, using one component for SINFONI and up to three
components for the higher spectral resolution ALMA data. In
Fig. 2 we compare the velocity fields of these two lines. For the
CO velocitieswe computedthe first momentmap,andfor H2we
constructed the velocity map from Gaussian fits.
The ALMA channel maps are very similar to the SMA ob-
servations in Ueda et al. (2012). We identified all SGMCs dis-
1http://almascience.eso.org/alma-data/science-verification
covered by Wilson et al. (2000) except for SGMC 3, which is
not covered by ALMA. The new data show two velocity com-
ponents in SGMC 1 and 2. We also decomposedSGMC 4/5 into
two smaller complexes at positions (−3,−9.5) and (−3,−3.5) in
Fig. 1, each of which has two velocity components. We kept the
name SGMC4/5 for the first complex, and labeled the northern
extension SGMC 6. CO and H2spectra of each SGMC are dis-
played in Fig. 2. The CO spectra were integrated over each box
in Fig. 1 using the clipped cube.
Table 1 lists the CO line properties of each SGMC and each
velocity component, named a and b. Fluxes, velocities and line
widths of all components are measured from Gaussian fits to the
spectra in Fig. 2. Error bars include only fit uncertainties, not
the systematic errors owing to the missing short spacings. We
estimated the R3−2/1−0 = ICO(3−2)/ICO(1−0)ratios of the SGMCs
by comparing the ALMA and OVRO data, after smoothing to a
common resolution. ICOis the integrated intensity in K km s−1.
This ratio varies from source to source between 0.3−0.8, with
a mean value of 0.5, consistent with what Schulz et al. (2007)
foundfrom single-dish data for the entire overlapregion,as well
as the peak line ratios measured with the SMA for each individ-
ual SGMC (Ueda et al. 2012).
We estimated gas masses from the CO fluxes, where the XCO
factor is the main source of uncertainty. To be consistent with
previous studies, we used the same XCOfactor for CO(1−0) as
Wilson et al. (2000), XCO=3 × 1020H2cm−2(K km s−1)−1, and
adopted the scaling R3−2/1−0=0.59 of Schulz et al. (2007). Our
mass estimates (Tab. 1) are comparable to those of Wilson et al.
(2000). Similar to H11, our data show that most of the H2
emission away from SSCs is powered by shocks, not UV heat-
ing in PDRs. The observed H2 2−1/1−0 S(1) ratios, 0.1−0.2
(Tab. 1) can be accounted for by PDR models, but only for
high UV fields (χ>104, in units of the mean value in the solar
neighborhood) and high densities (nH>105cm−3, Le Petit et al.
2

Page 3

C. N. Herrera, F. Boulanger, N. P. H. Nesvadba, and E. Falgarone: CO and H2observations of the Antennae overlap region
Fig.2. Left panel: Velocity map of H21−0 S(1) (left) and first-moment map of CO(3−2) (right). Dotted lines mark the position-
velocity cut shown in the right panel. Mid panel: Integrated line profiles of CO(3−2) (red) and H21−0 S(1) (black) for each SGMC.
Blue spectra show the CO(3−2) lines convolved to the spectral resolution of SINFONI. Black bars correspond to 5 ×10−16erg s−1
cm−2Å−1for H21−0 S(1) and red bars to 1.5 Jy for CO(3−2), for all SGMCs. Offsets between the black and blue spectra indicate
variations in the CO-to-H2ratio between velocity components. Right panel: H21−0 S(1) position-velocity diagram of SGMC 2
(H11), with CO(3−2) emission shown as contours in steps of 0.02 Jy beam−1starting at 0.03 Jy beam−1.
2006). These conditions exist near massive stars embedded in
molecular clouds, but the extended H2 emission is generally
not observed to peak near the brightest SSCs. The notable ex-
ception is the embedded cluster in SGMC4/5, which, however,
has a higher line ratio (Gilbert et al. 2000). In addition, the
[CII]157µm/[OI]63µm line ratio indicates that the mean radi-
ation field in the overlap region is χ∼103(Schulz et al. 2007),
an order of magnitude smaller than that required to account for
the H22−1/1−0 S(1) ratios. The other luminous SSCs have al-
ready dispersed most of their gas and dust. Thus, the brightest
UV-heated gas would come from clouds away from the clusters.
Outside the clusters the intensity of the radiationfield is not very
high. For a 5×106M⊙SSC, the UV field is χ>104only out to
a distance of <100 pc from the cluster and less if we include
extinction.
Our analysis has three main results. (1) All SGMCs have
H2 1−0 S(1) line emission and the H2 1−0 S(1) kinematics
matchthose of CO(3−2)well. Zhu et al. (2003) andSchulz et al.
(2007) found that the CO is emitted from gas at temperatures
with T∼30−150 K. The excitation temperature of the warmer
H2 gas emitting in the near-IR is ∼1000−2000 K (H11). The
similarity of the gas kinematics of CO and H2 indicates that
warm and cold gas are closely associated. (2) Figs. 1 and 2 show
largevariationsintheH2/COratiobetweenSGMCs andbetween
individual velocity components in the same SGMC (Tab. 1).
Extinction is an unlikely cause, because the near- and mid-IR
H2pure-rotationalemission-line regionshave similar morpholo-
gies (at least at the 5′′scales resolved by Spitzer-IRS; see Fig. 4
in Brandl et al. 2009, and the discussion in H11). In H11, we
related the H2emission to the dissipation of kinetic energy.With
this interpretation, the H2/CO ratio traces differences in the en-
ergy dissipation rate per unit mass, which must be related to the
dynamical evolutionof the gas. (3) All clouds have two spatially
separated velocity components as seen in the channel maps of
SGMC 2 in Fig. 3. In the other clouds, the spatial offsets be-
tween velocity components are less obvious, possibly because
of projection effects. The velocity difference between compo-
nents within an individual SGMC is up to 150 km s−1(Fig. 2).
Given the size and mass of the SGMCs, this is too large to be
accountedfor by the gas self-gravity.The gas kinematics is most
likely driven by the galaxy interaction. Single-dish observations
foundsimilar velocitygradientsin the extendedemission around
the SGMCs, which further supports this idea. The two compo-
nentsinSGMC 2arespatiallyseparatedandresolvedbyALMA.
We estimated their sizes andfoundtheir virial masses to be com-
parable to the molecular masses derived from the CO fluxes.
3. The compact H2source
The ALMA maps also give new insight into the nature of the
bright, compact H2emitter associated with SGMC 2, PCC, re-
cently discovered by H11. PCC is the brightest H2line emitter
in the overlap region. It is not detected in the lower-resolution
CO(1−0) data, but is an emission peak in the SMA and ALMA
maps (Fig. 1). Isolating the PCC CO counterpart from the sur-
rounding extended emission is difficult with an algorithm like
CLUMPFIND becauseof the largevelocitygradientacross PCC
(see right panel in Fig. 2). This may explain why the source is
not specifically listed in Table B.1 of Ueda et al. (2012). There
is no other CO peak in the SGMCs with an obvious counterpart
in the near-IR, except for the embedded SSC in SGMC 4/5.
Fig. 3 shows a comparison of the CO and CRIRES high-
resolution (6 km s−1) H21−0 S(1) spectra of PCC (H11). The
CO spectrumis thepeakemission at thepositionof thePCC cor-
rected for the surrounding emission, measured over an annulus
outside the source, to isolate the CO velocity component asso-
ciated with the PCC. CO(3−2) and H21−0 S(1) spectra are re-
markablysimilar.Table1lists theparametersofourGaussianfit.
The H2-luminousPCC has a velocitydispersion (40km s−1) sig-
nificantly higher than those of GMCs with the same sizes in the
Milky Way (5 km s−1; Falgarone et al. 2009; Heyer et al. 2009).
Using the velocity dispersion of CO(3−2), we obtain a virial
mass of Mvir = 5Rσ2/G = 4.6 × 107× (σv/40 km s−1)2M⊙.
3

Page 4

C. N. Herrera, F. Boulanger, N. P. H. Nesvadba, and E. Falgarone: CO and H2observations of the Antennae overlap region
Table 1. CO(3−2) properties of the SGMCs (top) and PCC (bottom). We include the integrated H2 1−0 S(1) fluxes, and the
H21−0/2−1 S(1) and H2/CO flux ratios. For PCC, the line parameters are measured with a single-componentGaussian fit.
Source
Velocity
component
a
b
a
b
a
b
a
b
VLSR
km s−1
1417±2
1521±1
1469±1
1613±2
1505±1
1586±3
1506±2
1561±5
1534±2
∆v
SCO
Mmol
M⊙
FH21−0 S(1)
erg s−1cm−2
FH22−1 S(1)
FH21−0 S(1)
H21−0 S(1)
CO(3−2)
km s−1
105±6
63±3
90±2
114±5
58±2
126±6
50±8
113±6
93±4
Jy km s−1
380±24
289±16
289±8
176±10
171±9
256±12
48±12
216±15
7.6 ±0.4
SGMC 1
5.6±0.3 × 108
4.2±0.2 × 108
4.2±0.1 × 108
2.6±0.1 × 108
2.5±0.1 × 108
3.7±0.2 × 108
0.7±0.2 × 108
3.2±0.2 × 108
1.0±0.1 × 107
1.6×10−14
0.22.1
SGMC 2
8.3×10−15
0.21.6
SGMC 4/5
1.3×10−14
0.3 2.6
SGMC 6
6.7×10−16
<0.5a
0.2
H2source (PCC)
−
7.6×10−16
0.1 8.7
aH22−1 S(1) flux corresponds to an upper limit estimated from the noise.
Fig.3. (left) CO(3−2) emission from SGMC 2, where emission
between 1350−1540 km s−1and between 1540−1750 km s−1
are shown in blue and red, respectively. Contours show the
H21−0 S(1) morphology. (right) CO(3−2) (red) and CRIRES
H2(blue) spectrum of PCC. The H2spectrum is smoothed to 18
km s−1resolution. The bar corresponds to 77 mJy beam−1for
CO(3−2) and 4.4 ×10−17erg s−1cm−1Å−1for H21−0 S(1).
Since ALMA does not spatially resolve the PCC, we instead
used the 50 pc size measured with SINFONI. The exceptionally
high H21−0 S(1)-to-Brγ line ratio (>15, H11) provides unam-
biguous evidence that the H2emission of the PCC is powered
by shocks (Puxley et al. 1990). The H2/CO ratio, i.e. the energy
dissipation rate per unit mass, is also exceptionallyhigh,a factor
5 higher than that of the SGMC 2 complex overall.
PCC appears to be located at the interface between blue and
redshiftedgas (Fig. 3) where CO shows a steep velocitygradient
(∼1 km s−1pc−1in the position-velocity diagram in Fig. 2). The
observedproperties of PCC are consistent with a scenario where
the formation of SSCs is triggered by interactions between two
gas flows. In SGMC 2, depending on the full three-dimensional
geometry, the flows could either be colliding or creating a large
velocity shear, and most likely a combination of both. In either
case, the interaction drives a turbulent energy cascade in which
kinetic energy is being dissipated. This is where we would ex-
pect the highest energy dissipation rate.
The bolometric luminosity of the PCC is ∼107L⊙.
Observations of, e.g., NGC 1333 (Maret et al. 2009) show that
the bolometric luminosity of protostellar outflows is on the or-
der of 105L⊙× ˙ Mwind, where ˙ Mwindis the stellar mass loss rate
in M⊙yr−1. The small embeddedstellar mass of M∗=4×104M⊙
(H11) makes protostellar winds an implausible energy source.
The cloud luminosity may be accounted for by the dissipa-
tion of the cloud kineticenergyfora cloudmass of a few 107M⊙
– a value comparableto the virial mass 5×107M⊙– and a dissi-
pationtimescale of 1 Myr. This is comparableto the cloudcross-
ing time, and also the dynamical time scale associated with the
velocity gradient of 1kms−1pc−1at the position of the cloud.
The similarity of both timescales indicates that the cloud may
still be forming by accreting gas, and therefore that a significant
partof the cloudluminositymay be poweredbygas accretion.In
any case, the time during which the PCC is a bright H2emitter
is short, about 1 Myr.
This short timescale may explain why we do not find more
bright compact sources in our new H2 data. Whitmore et al.
(2010) list five massive (>105M⊙), young (< 5 Myr) SSCs over
the part of the overlap region covered with both ALMA and
SINFONI (Fig. 1). This is consistent with finding only a single
bright PCC with the SINFONI data if the H2-luminous phase
does not last longer than a few Myr.
Our analysis gives a foretaste of the power of combining
mass and energy tracers to study the dynamical state of molecu-
lar gas in galaxymergersandthe early stages of the formationof
SSCs. In the future, this approach can be extended with ALMA
andVLT using additionaltracers ofenergydissipationand mass.
Acknowledgements. We wish to thank the staff at ALMA and the VLT for
making these observations and are particularly grateful to the ALMA SV
team for making the fully reduced and calibrated ALMA data available
(ADS/JAO.ALMA#2011.0.00003.SV). We thank R. Kneissl for helping us in
analyzing the ALMA SV maps, C. Wilson for providing us with her CO(1−0)
OVRO data cube, and P. Guillard for his useful comments. We thank the referee
for comments that improved our paper.
References
Brandl, B. R., Snijders, L., den Brok, M., et al. 2009, ApJ, 699, 1982
Falgarone, E., Pety, J., & Hily-Blant, P. 2009, A&A, 507, 355
Gilbert, A. M., Graham, J. R., McLean, I. S., et al. 2000, ApJ, 533, L57
Herrera, C. N., Boulanger, F., & Nesvadba, N. P. H. 2011, A&A, 534, A138
Heyer, M., Krawczyk, C., Duval, J., & Jackson, J. M. 2009, ApJ, 699, 1092
Le Petit, F., Nehm´ e, C., Le Bourlot, J., & Roueff, E. 2006, ApJS, 164, 506
Maret, S., Bergin, E. A., Neufeld, D. A., et al. 2009, ApJ, 698, 1244
Puxley, P. J., Hawarden, T. G., & Mountain, C. M. 1990, ApJ, 364, 77
Schulz, A., Henkel, C., Muders, D., et al. 2007, A&A, 466, 467
Teyssier, R., Chapon, D., & Bournaud, F. 2010, ApJ, 720, L149
Ueda, J., Iono, D., Petitpas, G., et al. 2012, ApJ, 745, 65
Weidner, C., Bonnell, I. A., & Zinnecker, H. 2010, ApJ, 724, 1503
Whitmore, B. C., Chandar, R., Schweizer, F., et al. 2010, AJ, 140, 75
Wilson, C., Scoville, N., Madden, S., & Charmandaris, V. 2000, ApJ, 542, 120
Zhu, M., Seaquist, E. R., & Kuno, N. 2003, ApJ, 588, 243
4