1. Atomic Gas associated with GMCs in the LMC
Thank you very much (chairperson).
My name is Annie Hughes -- I’m a postdoc at the MPIA in Heidelberg. Today I’m going to be talking about what we have learnt from the kinematics of the giant molecular clouds in our neighbour galaxy, the Large Magellanic Cloud, and of the atomic gas that is associated with that cloud population.
The collaborators that I’ve listed here on my title slide -- Tony Wong, Juergen Ott, Erik Muller and Jorge Pineda -- are the core team responsible for the Mopra survey of the CO emission in the LMC that I’ll be discussing today.
Annie Hughes (MPIA), T. Wong (U. Illinois), J. Ott (NRAO), E. Muller (NAOJ), J. Pineda (JPL) & the MAGMA team

2. ⇒ GMCs strongly self- gravitating
A Simple Question
Are GMCs dynamically decoupled from their environment?
Yes?
Bolatto et al. 2008
Pcl ~ 0.47 αvir G Σ2
The analysis that I’m going to present is aimed at answering one simple question:
Are giant molecular clouds decoupled in a kinematic sense from the atomic gas around them?
In recent years, I think that the common answer among the extragalactic astronomers -- to the extent that they think about this question at all -- would be yes.
The key piece of evidence that is usually cited to support this picture of dynamically isolated GMCs is that GMCs seem to be significantly overpressured with respect to the average interstellar pressure.
We can see this from Larson’s Third Law -- illustrated here in a plot by Alberto Bolatto and his collaborators -- which suggests that GMCs in many different environments have approximately uniform surface densities of 100 solar masses per square parsec.
For a GMC with this mass surface density that manages to achieve approximate virial balance between its gravitational and kinetic energies, the internal pressure is given by this equation, and this turns out to be larger than the pressure in the ISM, roughly by a factor of 5.
So, we conclude, GMCs must be strongly self-gravitating and therefore impervious to what’s going on in the surrounding atomic medium.
In particular, we only have a rough idea of the physical processes that might dominate the formation of GMCs, the timescales on which GMC formation occurs, or -- I would also say -- whether GMCs are gravitationally bound.
Even if you don’t care about GMCs very much, the answers to these questions are interesting because they will influence how we explain the discrepancy between the supply of molecular gas and the observed star formation rate in galaxies.
More specifically, If clouds are bound and they collapse roughly the free-fall time, then the rate of star formation in the MW should be a few hundred solar masses per year, whereas the observed star formation rate is of order unity.
To get around this, you can either assume that molecular clouds are long-lived and that there is some process -- such as magnetic fields -- that slows their collapse. The other approach is to think that molecular cloud assembly is relatively fast, but that there is some mechanism that makes star formation very inefficient, e.g. because GMCs are easily dispersed by stellar feedback or external perturbations.
for Σ ~ 100 M⨀/pc2,
Pcl/kB ~ 105 K cm-3
vs. PISM/kB ~ 2 x 104 K cm-3
(Cox 2005)
⇒ GMCs strongly self- gravitating

3. A Simple Question No? * in the LMC In the LMC,
Are GMCs* dynamically decoupled from their environment?
No?
In the LMC,
ΣH2 ~ 30 M⨀/pc2 & Rmol < 1
Pext ~ Pkin,ISM + Pgrav,(HI+H2)
Pcl,kin~ Pext ~ 2 x 104 K cm-3
In the LMC, however, I’m beginning to think that the answer might be no.
What started me thinking about this, was redoing the calculation on the previous slide using numbers that are more appropriate to conditions in the LMC.
In particular, we know that the LMC clouds are somewhat different from GMCs in other nearby galaxies in the sense that their mass surface densities are low, and the fraction of molecular gas along any line of sight in the LMC is also very low compared to normal spirals.
In the case where most of the interstellar gas is atomic, we also need to take the weight of the atomic layer into account when we estimate the confining pressure at the molecular cloud surface. And it turns out, that in the LMC the kinetic pressure from the internal turbulent motions that act to disperse the GMCs are roughly the same magnitude as the pressure terms that act to confine the cloud.
So while the GMCs studied by Bolatto and his collaborators may seem to be strongly overpressured with respect to the ISM, this conclusion would not seem to hold for GMCs in the LMC.
For comparison to M33& M51, see poster #14 by Dario Colombo
* in the LMC

4. The MAGellanic Mopra Assessment
Background: Hα MCELS, Smith et al. 1999
MAGMA Observations
~ 3.6 square degrees
114 NANTEN clouds
~80% of LMC’s CO flux
θfwhm = 45” (11pc)
Δv = 0.53 km/s
But first I’d like to spend a few minutes introducing the CO survey that we’ve used to identify and parameterise GMCs.
This was the Magellanic Mopra Assessment, also called the MAGMA survey. MAGMA is a large mapping project of CO emission from both Magellanic Clouds using the Mopra Telescope in Australia.
In the LMC, MAGMA conducted targeted mapping of ~ 120 bright clouds identified in the lower resolution NANTEN survey of the LMC. The combined fields cover around 3.6 square degrees. These clouds number less than half of the NANTEN catalogue but constitute close to 80% of the CO flux measured by NANTEN.
At 115GHz, the Mopra beam is 45” corresponding to a linear resolution of about 10 pc at the distance of the LMC.
Mopra Telescope

5. MAGMA Background: Hα MCELS, Smith et al. 1999 Tmb [K]
LSR Velocity [km/s]
Tmb [K]
Of course, the area of the LMC is so large compared to the resolution that it is difficult to see the detailed cloud structure.
Here I’ve just zoomed in on a molecular cloud complex in the stellar bar region.
The background here is the 24um image obtained by the Spitzer SAGE team.
And this is an average spectrum for this cloud over here. The intrinsic RMS of individual spectra is about 0.3 K in a 1 km/s channel.
Background: 24um SAGE, Meixner et al. 2006

6. MAGMA Background: Hα MCELS, Smith et al. 1999 Tmb [K]
LSR Velocity [km/s]
Tmb [K]
Here’s another cloud right over on the eastern edge of the LMC. The 24um is shown on the same colour stretch as before.
And again, the average spectrum for this large cloud.
Background: 24um SAGE, Meixner et al. 2006

7. The MAGMA Survey First Public Data Release:
Background: Hα MCELS, Smith et al. 1999
LSR Velocity [km/s]
Tmb [K]
First Public Data Release:
Wong et al, 2011, ApJs, in press, arXiv: )
So I’ve taken a little time to show off these data, because it’s just become public.
The data paper has recently been accepted and is no on astroph,
and as of last week you can now download the whole cube plus moment-maps and cloud catalogues from Tony’s website.

8. “Top Down” GMC formation
Predictions:
i) Velocity gradients should align with Galactic rotation
ii) Angular momentum conserved <<<
OK, so let’s get back to the main point of this talk -- whether the kinematics of MAGMA GMCs and the surrounding atomic material can tell us anything about how GMCs in the LMC have formed.
In a simple “top down” model for cloud formation, GMCs form by condensing out of atomic gas that is rotating with the Galactic disk.
Since angular momentum is conserved quantity:
i) the rotation of GMCs should be prograde with respect to the rotation of the Galaxy, and
ii) secondly, the cloud should spin-up to larger angular velocities as it collapses,
To date, most of the empirical investigations to check this model have been done in the M33 or the Milky Way, and their results don’t exactly validate the basic top down formation model.
In particular, most previous work has found that
i) GMCs show a mixture of prograde and retrograde rotation,
ii) that the magnitude of the GMC velocity gradients is too small by up to an order of magnitude;
Here I’m showing the results from a study of GMCs in M33 by Erik Rosolowsky-- he found a mixture of prograde and retrograde rotation for his clouds, a slight
The fact that the observed gradients are much smaller than predicted by top-down models has led to the notion that GMCs must experience some kind of braking during their formation.
Previous Observations:
MW & M33: Blitz 1990, 1993; Phillips 1999; Rosolowsky et al 2003; Koda et al 2006, Imara et al. 2011a&b
M33
i) Gradients weakly aligned with Galactic rotation (or not at all)
ii) Gradients are ~10 times smaller than model predictions

9. Velocity Gradients of MAGMA GMCs
Linear gradients provide a reasonable fit to velocity field for ~50% of resolved clouds in the MAGMA sample
143 clouds
On this slide you can see maps of the velocity centroid of 6 example clouds in the MAGMA sample... the scales of the panels are slightly different, so I’ve put a red bar indicating a length scale of 100 pc. The velocity gradient for each cloud is estimated by fitting a plane to these maps.
Our first basic result is that we find a linear gradient for roughly half the resolved clouds in the MAGMA sample. For the remaining half, the GMCs have velocity fields look like the cloud in this bottom left panel, where the velocity field appears to show genuine complex structure and a linear gradient is obviously a poor description of the velocity field.
For the GMCs where a linear gradient did provide a reasonable fit to the velocity field, we find that the gradient amplitudes are relatively small -- the average gradient is about 0.1 km/s per parsec with a dispersion about that average value of about 20%. This is quite similar to the gradients that have been observed in M33 and the Milky Way.
Red bar indicates ~100pc

10. Spatial Organization of CO Gradients
prograde
As in M33 and MW, we also don’t find that GMCs in the LMC are preferentially prograde rotators.
We seem to see some regions where neighbouring velocity gradients are aligned, for example:
i) at the north western tip of the stellar bar region and along the top of the bar
but there are also many regions where neighbouring GMCs have almost orthogonal velocity gradients.
This is most obvious here in N11, and also in the molecular ridge.
Kim et al 2003
Some localised coherence, but GMCs show no strong preference for prograde rotation

11. A signature of cloud rotation?
Gradients contributes a significant fraction of GMC linewidth
But what’s very interesting to me is that where we do see a velocity gradient, then
i) this gradient accounts for a large fraction of the cloud’s “turbulent” linewidth, suggesting that the velocity gradients seem to represent such a large share of the kinetic energy.
Here I’ve plotted the GMC velocity dispersion versus the velocity change across the face of the clouds. This line represents equality, and we see that most of the clouds are scattered about this line.
On the other hand, if most of the kinetic energy budget in MAGMA GMCs was allocated to rotation, then one thing I would expect to see would be that the velocity gradients are aligned with the morphological major axis of the clouds. This is because rotation would tend to stabilise the GMC against fragmentation normal to the spin axis, but providiing very little support parallel to the spin axis, with the observational consequence that the position angle of the major axis and the direction of the velocity gradient would come into alignment.
If this were actually occurred, then we would expect to see an excess of clouds in this bin, but in fact the velocity gradient seems to be randomly oriented with respect to the cloud morphology.
Question we must ask ourselves -- are these velocity gradients really a sign of cloud rotation? or are they due to some other process?
But no correlation between gradient direction and GMC morphology

12. Gradients in atomic gas around GMCs
Local HI gradient measured directly from observed HI map (not estimated from galactic rotation).
Racc ~ region required to accumulate MGMC assuming ΣHI = 10,20 M⨀/pc2
2Racc
The next step to determine whether the GMCs in the LMC are consistent with a top-down formation model is to examine whether the magnitude and direction of their velocity gradients are consistent with the rotation that was present in the material from which the GMCs formed.
We can’t measure this directly of course, since the gas that formed the GMC is now part of the GMC itself, but we can make a prediction from the rotation curve of the LMC, or by measuring the velocity gradient in the atomic gas that currently surrounds the GMC.
We used the beautiful survey by Sungeun Kim of the LMC’s HI emission.
We use the same method as for the CO gradients -- that is we fit a plane to the moment-1 map of the HI emission.
But since the HI emission is diffuse, it’s not immediately obvious the size of the region that we should use to measure the gradient. These boxes in
LSR Velocity [km/s]
Tmb [K]
HI Tb [K]
Kim et al 2003

13. Comparison of CO & HI gradients
Describe plots
* middle panel is ratio between specific angular momentum from solid body rotation model, assuming progenitor material with a surface density of 10 Msol/pc2 and the observed angular momentum of the GMCs, assuming that the gradients are a signature of rotation.

14. Summary
GMC gradients show no clear relationship to overall rotation of LMC disk.
Gradients are large relative to total GMC linewidths, but no correlation between gradient direction and cloud morphology
Gradients in CO and surrounding HI have similar magnitude
Gradients in CO & HI show some degree of alignment
Specific angular momentum of GMCs significantly less than prediction from top-down formation model
So on this slide, I’ve summarised the key results of our investigation.
Several of these are consistent with a basic top-down model for GMC formation --
i) certainly we would expect the CO & HI velocity gradients to be aligned in such a model, and
ii) the discrepancy between the theoretic and observed angular momentum of the GMCs might be resolved if magnetic fields provided an efficient braking mechanism to slow down the rotation of the newborn clouds.
On the other hand, we find
that there appears to be no strong preference for prograde rotation of the LMC GMCs,
and that the GMCs don’t show the kind of morphology we would expect if these gradients really represented rotation.
So perhaps, instead of a conclusion I will end with a speculation, and that is that velocity gradients in LMC GMCs aren’t tracing rotating, bound clouds at all, but instead arise from molecular gas that is simply being pushed around in a large scale gas flow. While this hypothesis requires further testing, I think it provides a more consistent explanation of the velocity gradients that we observe in the LMC.
Thank you very much.
Are GMCs in the LMC dynamically decoupled from their environment? No.

16. Properties of LMC GMCs LMC GMCs have: Size ~ 10 to 150 pc
Mass ~ 103 to 106 M⨀
Mass ~ 30 M⨀ pc-2
For comparison to M33& M51, see poster #14 by Dario Colombo