Periodic Reporting for period 1 - AdvancedStarForm (Global and local star formation with state-of-the-art physics)
Reporting period: 2015-09-01 to 2017-08-31
Stars are the building blocks, the fundamental units of luminous matter in the universe, and they are responsible, directly or indirectly, for most of what we see when we observe it. It is thus of central importance to understand how stars form and what determines their properties. Many observational campaigns are today focusing on proto-planetary systems, and it is of paramount importance to develop new generations of theoretical models of star/planet formation and evolution, which can predict observable properties of proto-planetary systems, with obvious consequences for the understanding of our own solar system.
New stars form within large turbulent complexes, harboring several thousands of solar masses of cold gas: the giant molecular clouds. The gas fragments and collapses gravitationally, then heats up via compression until temperatures exceed the nuclear fusion ignition point and the star is born. Computational modeling is very challenging, as large-scale environmental factors couple to small-scale processes close to the star, connecting many physical mechanisms, including magnetic fields, gravity, radiation, and chemistry. The project aims to construct a unified description of star formation, from large to small scales, using the world’s most advanced numerical physics.
A two-way approach was used. First, I performed global simulations of giant molecular clouds to study star formation in a global context. I also worked to devise a new model for representing unresolved stars in global models, based on a large set of stellar structure calculations. The local approach focussed on an individual proto-planetary system, resolving all the scales down to the stellar surface (see image “The scales of star formation”). The fast rotation of the star and its envelope creates an accretion disk around the star, and it is inside this disk that planets eventually form. One needs to carefully incorporate magnetic fields, radiation and chemistry, which are all linked together, as the absence of one or the other can mean that no disk or planets form at all.
New stars form within large turbulent complexes, harboring several thousands of solar masses of cold gas: the giant molecular clouds. The gas fragments and collapses gravitationally, then heats up via compression until temperatures exceed the nuclear fusion ignition point and the star is born. Computational modeling is very challenging, as large-scale environmental factors couple to small-scale processes close to the star, connecting many physical mechanisms, including magnetic fields, gravity, radiation, and chemistry. The project aims to construct a unified description of star formation, from large to small scales, using the world’s most advanced numerical physics.
A two-way approach was used. First, I performed global simulations of giant molecular clouds to study star formation in a global context. I also worked to devise a new model for representing unresolved stars in global models, based on a large set of stellar structure calculations. The local approach focussed on an individual proto-planetary system, resolving all the scales down to the stellar surface (see image “The scales of star formation”). The fast rotation of the star and its envelope creates an accretion disk around the star, and it is inside this disk that planets eventually form. One needs to carefully incorporate magnetic fields, radiation and chemistry, which are all linked together, as the absence of one or the other can mean that no disk or planets form at all.
Magnetic fields play a dominant role in the dynamics of interstellar matter. Their implementation in computer models is however non-trivial in systems where the gas is partially ionised. Only charged atoms feel the presence of magnetic fields, and collisions between ions and neutrals lead to magnetic diffusion. An early prescription for the coupling between gas dynamics and magnetic fields (known as magneto-hydrodynamics; MHD) ignored this detail (the ‘ideal’ approximation). Its use in simulations of star formation is now known to have pathological side-effects, preventing the formation of proto-planetary disks, and hence planets. This project makes use of a state-of-the-art module for diffusive MHD. The strength of the diffusion is quantified by “resistivities”, that depend on the temperature and abundances of the chemical species. The first task was to compute new tables of resistivities using a chemical network. The tables were published (Marchand et al. 2016, A&A, 592, A18) and were made available publicly.
I then carried out two simulations evolving a large molecular cloud to assess the exact importance of magnetic diffusion; the first used ideal MHD, while the second included magnetic diffusion (see image “Large-scale star formation”). Strong magnetic fields inhibit fragmentation, and many more isolated structures form in the second (middle panel) simulation. This affects the number of stars that are formed and their masses; more massive objects are present in the ideal MHD run (right panel), while the protostellar distribution in the diffusive framework is a far better match to recent telescope observations represented by the purple curve.
The aim of the third task was to refine present models used to represent stars (“star particles”) in global simulations. The end goal is to run stellar evolution software inside each star particle, but this requires an initial starting point, corresponding to the birth of the protostar. We needed to determine which are the starting conditions, depending on the mass, size and temperature of each star’s parent body. We created a large set of star formation models, describing the formation of the stellar embryo, and found this seed’s properties to be independent of the initial conditions, implying all stars have the same starting point. The results were published (Vaytet & Haugbølle 2017, A&A, 598, A116) and presented during an international conference. The raw data from the simulations were also made publicly available.
The second half of the research proposal, the local approach, focussed on the describing all the stages involved in the formation of a protostar, with particular emphasis on magnetic diffusion. I performed two 3D simulations of isolated protostellar formation, using ideal and diffusive MHD. In the ideal MHD run, the magnetic field removes all rotation from the protostar, while enough is preserved to form a small disk in the diffusive run. The rotation also affects the topology of the magnetic field around the protostar (see image “Magnetic field topology in protostars”), where wrapped-up field lines are believed to be responsible for launching observed protostellar jets. These results were published (Vaytet et al. 2018, arXiv:1801.08193) and presented at 3 international conferences.
In the fifth task, I coupled the hydrodynamics of a collapsing cloud to a chemical network counting 65 gas and dust species with over 700 reactions. Chemical abundances in the protostellar and its envelope vary by orders of magnitude, which has important consequences for comparison with observations. We are in the process of identifying the most important reactions to help reduce the size of chemical networks before applying them to full 3D models.
The final task involved studying the formation of high-mass stars, which differs significantly from their low-mass counterparts because of extreme luminosities and winds. Computational models to date have ignored magnetic fields, and we carried out new simulations with MHD. We found that magnetic processes dominate the wind launching mechanism, focusing them over great distances. The masses and accretion rates of the star-disk system are strongly affected.
I then carried out two simulations evolving a large molecular cloud to assess the exact importance of magnetic diffusion; the first used ideal MHD, while the second included magnetic diffusion (see image “Large-scale star formation”). Strong magnetic fields inhibit fragmentation, and many more isolated structures form in the second (middle panel) simulation. This affects the number of stars that are formed and their masses; more massive objects are present in the ideal MHD run (right panel), while the protostellar distribution in the diffusive framework is a far better match to recent telescope observations represented by the purple curve.
The aim of the third task was to refine present models used to represent stars (“star particles”) in global simulations. The end goal is to run stellar evolution software inside each star particle, but this requires an initial starting point, corresponding to the birth of the protostar. We needed to determine which are the starting conditions, depending on the mass, size and temperature of each star’s parent body. We created a large set of star formation models, describing the formation of the stellar embryo, and found this seed’s properties to be independent of the initial conditions, implying all stars have the same starting point. The results were published (Vaytet & Haugbølle 2017, A&A, 598, A116) and presented during an international conference. The raw data from the simulations were also made publicly available.
The second half of the research proposal, the local approach, focussed on the describing all the stages involved in the formation of a protostar, with particular emphasis on magnetic diffusion. I performed two 3D simulations of isolated protostellar formation, using ideal and diffusive MHD. In the ideal MHD run, the magnetic field removes all rotation from the protostar, while enough is preserved to form a small disk in the diffusive run. The rotation also affects the topology of the magnetic field around the protostar (see image “Magnetic field topology in protostars”), where wrapped-up field lines are believed to be responsible for launching observed protostellar jets. These results were published (Vaytet et al. 2018, arXiv:1801.08193) and presented at 3 international conferences.
In the fifth task, I coupled the hydrodynamics of a collapsing cloud to a chemical network counting 65 gas and dust species with over 700 reactions. Chemical abundances in the protostellar and its envelope vary by orders of magnitude, which has important consequences for comparison with observations. We are in the process of identifying the most important reactions to help reduce the size of chemical networks before applying them to full 3D models.
The final task involved studying the formation of high-mass stars, which differs significantly from their low-mass counterparts because of extreme luminosities and winds. Computational models to date have ignored magnetic fields, and we carried out new simulations with MHD. We found that magnetic processes dominate the wind launching mechanism, focusing them over great distances. The masses and accretion rates of the star-disk system are strongly affected.
The project has undoubtedly raised awareness in the astrophysics community about the importance of magnetic diffusion and microphysics as a whole. The impact they can have on a system’s global properties such as rotation, gravitational stability, or the star formation efficiency as a whole, is of chief importance. The simulations reported here, both on large and small scales, were among the very first to properly couple gas dynamics, radiation, magnetic fields and gravity. This set of profoundly realistic simulations will help us establish laws on the sizes, masses, multiplicity, accretion rates and luminosities of protostars and their disks, as a function of the environment in which they are formed. This is essential for a better global understanding of the star formation process.
Aside from scientific output, the project also delivered key tools that are now publicly available. This includes tables of magnetic resistivities, a database of 1D star formation simulations, and the creation of a new visualization package for computational astrophysics data called Osiris. We have also made public our diffusive implementation of MHD.
Aside from scientific output, the project also delivered key tools that are now publicly available. This includes tables of magnetic resistivities, a database of 1D star formation simulations, and the creation of a new visualization package for computational astrophysics data called Osiris. We have also made public our diffusive implementation of MHD.