Origine du champ magnétique en cosmologie et formation des galaxies

In this dissertation, I present a theoretical work addressing two fundamental questions of modern cosmology. First, galaxy formation models suffer from problems of which the <strong>over-cooling</strong> is presumably the most serious. In order to propose a solution to it, I explore the physics of the biphasic magnetised Interstellar Medium (ISM) of our Galaxy. I show how interferences of Alfvén waves within a cloud of the ISM strongly enhance the internal magneto-turbulent energy density. I then study numerically the energy transport and injection mechanisms in a system of magnetised clouds. On the one side, the motions of clouds connected by a magnetic field spread energy over the whole system. On the other side, they inject energy into the clouds by generating internal hydromagnetic waves. The latter provide a non-thermal pressure which opposes external compression and lengthens the life time of the system. I finally show that the time of energy dissipation is several orders of magnitude longer than the one of the corresponding homogeneous system. <strong>The energy is effectively stored on several dynamical time scales</strong>. Considering a toy model of a biphasic protogalaxy, I show that the cloud systems in the ISM and the protogalaxy are dynamically equivalent. To solve the over-cooling, I suggest to transpose the results obtained for the ISM to the protogalaxy case. Second, the <strong>origin of magnetic fields in cosmology</strong> eluded any attempt of explanation. Yet its role in structuring the ISM and its involvement in galaxy formation are evident. Here I propose a <strong>new magnetogenesis mechanism operating on large scales</strong> at the reionisation of the Universe. On galaxy scales, the magnetic field produced is roughly eight orders of magnitude stronger than in usual models. Furthermore, it is generated preferably on large scales, small scale fields being largely suppressed. These two properties allow the model to account for fields detected in large scale structures.