The work described here was undertaken as part of my PhD thesis at the Dublin Institute for Advanced Studies , supervised by Dr. Andrew J. Lim. This page is a copy of the page originally at with some minor changes. I presented the work in a poster at the European Week of Astronomy and Space Science (University of Hertfordshire, April 20-23, 2009) which was the subject of a press release, and this page was written to accompany the press release.

Read the paper!   arXiv:0912.1499 |  ADS reference |  Download a PDF preprint


Pillars of Creation in Eagle Nebula (From HST)

Massive stars are born in the densest parts of Giant Molecular Clouds: dense, cold and dusty regions of the galaxy. These stars are very luminous and emit huge quantities of ultra-violet (UV) radiation which ionises and heats the surrounding gas, generating a rapidly expanding bubble known as an H II region.

This process does not happen very evenly, leading to strange and sometimes beautiful structures forming on the walls of this bubble. The famous Eagle Nebula pillars (right) are examples of these structures, but there are many others, often called "Elephant Trunks" by astronomers because of their elongated appearance.

There are a number of possible formation mechanisms for these pillars. Instabilities in the bubble's expansion could lead to corrugations forming in the expanding ionisation front, which could develop into elephant trunks. Observations show that molecular clouds are very clumpy though, so it is also possible that the densest clumps act as seeds for the formation of pillars. This is the idea we are investigating with our computer simulations.

2D Simulations

Real space is 3D, but often a lot can be learned from constructing simple 2D models of physical processes. As well as being simpler to visualise and analyse, they are also far less costly to simulate in terms of time and computing power. A desktop computer can run very detailed 2D models of star forming regions in a matter of hours.

We first looked at the effect of ionising radiation from a single massive star on dense clumps of gas placed in a lower density ambient gas. The idea is that if we put a number of dense clumps near each other, in a random distribution, some of them might move into the shadowed regions and organise themselves into something like a pillar.

Photo-ionised clumps are accelerated by the Rocket Effect, where hot gas "exhausts" away from illuminated surfaces in a photo-evaporation flow, giving an equal and opposite "kick" to the dense neutral gas (from Newton's 3rd Law). If a clump is fully exposed to radiation then it is pushed away from the star, but if it is partially shadowed then the direction of the kick will be partially away from the star and also away from the illuminated side and into the shadow. In this way it is possible that much more gas could get added to the shadowed region quite quickly, and a pillar could be built up more easily than with just a single dense clump. This is indeed what we see in our 2D models.

The best way to see it is with some figures, shown below (or in an mpeg movie - 10MB download). These show the neutral gas density on a log scale, where black is lowest, greeen is medium, and yellow to red is high density. We start off with 10 clumps in the first figure, and a star is "switched on" off the domain to the left. The next figures show the effects of the star's UV radiation at intervals of 100,000 years. The three bottom clumps merge into a dense trunk-like feature, and the 5 central clumps end up as a dense pillar-like structure also.

2D simulations
      of 10 clumps ionised by star 100,000 years 200,000 years 300,000 years

This was one of the more successful simulations we ran, out of about 200 models with different numbers and sizes of clumps. The key to this model's success was that it had a few clumps near each other, and also sufficiently large low density surrounding regions to allow the star's radiation to penetrate far beyond the clumps. All figures can be viewed here, shown at time intervals of 10,000 years.

3D Simulations

2D simulations have a number of limitations, and bigger 3D simulations are needed to check the results. We have done a number of 3D simulations at low resolution (168x128x128 grid cells along each of the 3 directions), and we resimulated the most promising one at higher resolution (256^3 grid cells, and then with 512x384x384 cells). This model has a background gas density of 100 atoms per c.c., and 30 dense clumps ranging from 3-10 times the mass of the sun. We placed a radiation source off the grid about 6 light-years away and allowed the gas to evolve due to the UV radiation.

From the simulation outputs we have generated 3D volume-rendered images of the gas density and also 2D slices through parts of the simulation where pillars form. The 3D images are put together as an mpeg movie for the 256^3 run, and below we show images from the 512x384x384 run from snapshots taken at 100, 200, and 400 thousand years. In the images below, red shows the densest gas, with yellow and green progressively less dense, and very low density gas made transparent.

100,000 years 200,000 years 400,000 years

You can see there is an elephant trunk near the front of the box in the central panel, and in the right-hand panel there is one towards the back right and another to the left. The front pillar forms rapidly and is not long-lived, but the two late-forming elephant trunks take 300kyr to form and survive at least 200kyr after that until the end of the simulation. The full sequence of TIF images can be viewed in this directory (note physical times are indicated on the figures in seconds, hence the large numbers).

Cuts through both directions of the front pillar are shown below, again in neutral gas density (at 10, 100, 150, and 150 thousand years). It forms in a way we didn't expect initially: A massive clump is shadowed by two smaller clumps. These are accelerated past the massive clump and wrap around it forming a dense pillar. The massive clump protrudes because it has been shielded by the smaller clumps and it takes a long time for the rocket effect to accelerate it. This elephant trunk is about 0.7pc long. The arrows on the figures show the direction of gas motion (click a figure to see a larger version), with a longer arrow indicating a higher speed.

10,000 years 100,000 years 150,000 years 250,000 years

10,000 years 100,000 years 150,000 years 250,000 years

This modelling only considers radiation and fluid dynamics. We are currently running similar models with various magnetic field configurations. We would also like to include the effects of gravity which would allow us to track the evolution of the elephant trunks much further, to the point where the densest parts may start to collapse under their own weight to form new stars. The chemistry and thermal physics in our models is relatively crude, however, so improving this is probably the next stage in our work.


JM was funded by the Irish Research Council for Science, Engineering and Technology on an Embark Initiative Postgraduate Scholarship. AJL was funded by a Schrödinger Fellowship from the Dublin Institute for Advanced Studies. JM and AJL acknowledge the SFI/HEA Irish Centre for High-End Computing (ICHEC) for the provision of excellent computational facilities and support. Figures were produced using the VisIt visualisation tool.

Valid HTML 4.01 Transitional