Research Centre for Theoretical Astrophysics, Machine VP
School of Physics,
University of Sydney
Co-Investigator Sue Byleveld
Research Centre for Theoretical Astrophysics, School of Physics, University of Sydney
The Influence of Magnetic Fields on Star Formation
Stars, even our own sun, were born deep within giant molecular clouds, which are large clouds of hydrogen gas and dust inhabiting the space between stars. The dust shrouds the stars as they are born, obscuring the light they emit and making observations difficult. With the recent advances in speed and memory capabilities of computers, numerical simulations have become a valuable tool with which to probe the physical processes at work as a star is formed.
We are particularly interested in the influence of magnetic fields on star formation. Just as the Earth possesses its own magnetic field, so too do giant molecular clouds. What is the effect of this field on the dynamical processes at work within these clouds which have been proposed as resulting in star formation? Specifically, we are investigating the supersonic collisions which occur between `sub-clouds', or regions of enhanced density inside giant molecular clouds. As the `sub-clouds' collide they compress the gas between them which then may collapse under its own gravitational forces and form a dense, disk-like structure, the precursor to a star. A large number of particles (~20000) must be used so that our three dimensional simulations can be as realistic as possible. Not only do we require many particles, but our consideration of hydrodynamic, gravitational and magnetic effects using a vectorized code, means that only very powerful computers, such as the VP2200 are capable of running our simulations.
What are the basic questions addressed?
In this project we are investigating what influences magnetic fields have on star formation via dynamic processes occurring within giant molecular clouds. Although many studies have considered the influence of magnetic fields on the quasi-static, slow collapse of single molecular clouds leading to the formation of protostellar disks, less is know about dynamic star formation. In particular, we are interested in the formation of protostellar disks resulting from density enhancements in shock compressed material. The shock compression arises from the supersonic collision of `sub-clouds' within the giant molecular clouds themselves.
Questions we are addressing about the influence of magnetic fields on this system include:
Do magnetic fields (of a magnitude comparable with the galactic field) slow the formation of `protostars'?
What magnetic field strength is required to arrest their formation entirely?
How do these collisions of `sub-clouds' affect the morphology of the magnetic field?
How do magnetic fields influence the formation of binary or multiple systems?
How do the effects of ambipolar diffusion influence the formation of stars via this mechanism?
What are the results to date and the future of the work?
To date we have successfully run simulations with a range of initial conditions including various magnetic field morphologies and strengths. In all cases where the field resembled the non-random component of the galactic magnetic field collapse to a protostellar disk still occurred, but took ~ 10-20 percent longer than in the corresponding unmagnetized system. However, for a magnetic field strength an order of magnitude greater than the galactic field no collapse occurs. The `sub-clouds' collide, compressing material between them, but then re-expand before collapse under self-gravity can occur.
For a particular simulation presented at the conference on `Circumstellar Disks, Outflows and Star Formation' held in Cozumel, Mexico in December 1994, we found the following results. For an off-centre collision with a constant magnetic field (representing the non-random component of the galactic field), and with the magnetic field anchored at the boundary, the sub-clouds dragged the field lines inwards and twisted them as they collided asymmetrically. The field morphology displayed the expected flux-freezing behaviour. Following the evolution of the system for 1.4 Myrs, we found the density to be enhanced by 7 orders of magnitude. A dense structure formed which accreted matter, rotating as a single entity for 3-4 revolutions in the last 0.02 Myrs of the simulation. As it rotated, spiral arms of matter formed around the disk removing of angular momentum. The structure formed could be identified with a protostellar disk. It had a diameter of ~ 800 AU, a scale height of ~ 150 AU and a mass of ~ 5 solar masses, supporting a maximum magnetic field strength of ~ 60 micro T.
An integral aspect of this project is to compare tests of standard shock tubes and the propagation of MHD shock waves to the expected shock behaviour. Our early work in this area has shown that, excluding boundary effects, our results are similar to those of other authors with the exception of the smoothing of the shocks (an effect which is inherent in SPH). New boundary conditions have been developed (continuous boundary conditions) and these should remove the spurious effects of the boundary mentioned above. These tests will continue as an important part of our future work.
We envisage that another aspect of our future work will be to investigate the system further and to consider the formation or otherwise of binary and multiple stellar systems.
All the systems considered thus far in this project have been assumed to be fully ionized. Star forming regions, however, are only partially ionized leading to the effect of ambipolar diffusion which has been shown to be of great importance in quasi-static theories of star formation. We are intending to develop our code further to incorporate the physics of partially ionized plasmas. SPH is well suited to such a problem in that, being a particle method, it can keep track of the neutral and ionized particles as they interact with each other as magnetic fields can diffuse because of the presence of the neutral particles. We will then re-examine the collisions of `sub-clouds' using a more realistic code.
What computational techniques are used and why is a supercomputer required?
The code is three dimensional and uses a magnetic version of smoothed particle hydrodynamics with tree-code gravity. A supercomputer is required because:
* The code is well vectorized (it was developed to run on a CRAY and has been adapted to run on a VP2200) and runs more efficiently on a vector machine than local Sparc stations and Dec-Alpha machines. The magnetic fields have been included in the code in such a way as to retain the vectorized structure.
* With a supercomputer more particles can be used which is necessary to represent `sub-clouds' and the surrounding medium in three dimensions.
* We require a large amount of memory. (There are 6 arrays for the magnetic variables alone).
* We require a realistic turn-around time for large jobs which cannot be obtained using a local workstation.
The Influence of Magnetic Fields on Star Formation, S E Byleveld and H Pongracic, Proc. A.S.A. (1994), submitted.