Formation and emergence of stellar clusters

Thesis defense of Yann Bernard - November, Monday 9th at 1.00 pm - Manuel Forestini room IPAG

Stars are social celestial objects. We now know that they form in groups within molecular clouds and spend their early lives interacting with their peers. Although we understand the major processes at work, constructing a theory that describes their formation, early evolution, and explains the properties of more evolved stars that we can observe with our instruments, is much more difficult.

The early dynamics of star clusters embedded within their natal regions are very complex to model. These dynamics depend heavily on the interactions between the stars and their native gaseous environment, such as gravity and stellar feedback. These interactions involve a strong stochasticity, which complicates predictions about the properties of star formation. On the other hand, observations struggle to produce sufficiently complete and accurate samples of these clusters due to the constraints imposed by their optically thick environment. However, understanding these early stellar dynamics would help bridge the gap between more precise observations of emerged clusters, such as those from Gaia, and theories of star formation.

This project aims to compensate for the stochasticity of early stellar dynamics by performing statistical analyses on large samples of simulations. In this thesis, I have developed simulations, based on the sc Phantom code, capable of addressing this ambition. The key aspect of developing these simulations was to optimise computation time as much as possible in order to track the complete evolution of star-formation events. To achieve this, we imagined a new star-formation scheme that mimics the fragmentation process of prestellar cores to limit the resolution. It was also necessary to update the code’s stellar dynamics to efficiently track the orbits of stars within clusters. The inclusion of photoionisation by massive stars completed the set of processes essential for a convincing early evolution of embedded clusters.

Once these developments were done, I tested the model by simulating three collapses of turbulent molecular clouds. These three simulations produced very different final clusters. It appears that the emergence of the first ionHII regions has a significant impact on the future evolution of the formation region. These initial tests also demonstrated the model’s performance, with improvements of up to a factor of 80 thanks to the new stellar dynamics. They also made it possible to eliminate the remaining bottlenecks that were still affecting performance.

Finally, I carried out a parametric study aimed at characterising the stellar mass distributions (IMF) emerging from these simulations. To this end, I calculated a sample of 30 simulations to test the influence of the parameters of our new star formation scheme. I was able to demonstrate the influence of these parameters on the peak and the slope at high masses of the IMF. It is possible to adjust these parameters to align the simulated mass distributions with a canonical IMF. However, there is some degeneracy in this choice. A more detailed study of morphology and multiplicity within clusters is needed to complement this characterisation and determine the optimal parameterisation for these simulations.

This simulation model must continue to be refined, in particular by incorporating primordial binary systems and modelling protostellar jets. Nevertheless, at the conclusion of this thesis, it appears entirely feasible to characterise the evolutionary trajectories of buried clusters using a large sample of simulations.

Thesis direction

• Estelle Moraux

The jury will consist of:

• Cathie Clarke, Full Professor Cambridge U., Examiner
• Yueh-Ning Lee, Associate Professor, Taïwan U., Examiner
• Catherine Dougados, Senior researcher IPAG CNRS, Examiner
• Matthias Gonzalez, Full Professor, UPC, Referee
• Christian Boily, Full Professor, Strasbourg U., Referee
• Daniel Price, Full Professor Monash U., guest