Résumé : Microscopic approaches enable one to study nuclear bound states as well as nuclear collisions in a unified framework.

At non-relativistic energies, all physical quantities are determined by the solutions of the many-body Schrödinger equation based on an interaction potential between nucleons.

The difficulty of solving this equation for collisions and taking the antisymmetrization principle into account restricts these approaches to light nuclei and requires the development of nuclear models based on some simplifying assumptions.

One of these assumptions, which is done in this work, is to consider that the nucleons are aggregated in clusters in the nuclear systems.

Another major problem of the microscopic description is the difficulty of determining a reliable interaction potential between nucleons.

In spite of many years of efforts to establish such potentials, none has yet been proved to accurately describe both the spectroscopic properties of nuclei and the reactions between light nuclei.

For this reason, many effective NN interactions, adapted to the model space and to the studied collision, have been built and used in microscopic models.

In parallel, for a few years, some efforts have been done to use in the microscopic models more realistic NN interactions, adjusted to reproduce the two-nucleon properties.

However, this requires solving much more accurately the Schrödinger equation by relaxing, for instance, the cluster assumption.

These approaches therefore need large computational times, which limits the size of the systems that can be studied.

In this work, a two-body realistic interaction has been adapted to the simple microscopic cluster model by using the Unitary Correlation Operator Method. This new realistic effective interaction has been adjusted so that the α+α elastic phase shifts obtained with the microscopic cluster model agree rather well with the experimental data.

This interaction has been used to study α+N and α+3He scattering.

The calculated phase shifts give a rather good agreement with experimental data without additional adjustment, without three-body interactions and with simple basis functions.

Besides this study of elastic scattering between light nuclei, this work deals with the nucleus-nucleus bremsstrahlung.

Previous microscopic models of nucleus-nucleus bremsstrahlung were based on a photon-emission operator fully neglecting the meson-exchange currents.

In this work, a microscopic cluster model of bremsstrahlung is developed, which implicitly takes them partially into account by using an extension of the Siegert theorem.

Then, the photon-emission operator can be deduced from the charge density rather than from the current density.

Although this extension of the Siegert theorem does not fully remove the nuclear-current dependence, the effects of the meson-exchange currents should be largely reduced, especially at low photon energy.

The microscopic cluster model of nucleus-nucleus bremsstrahlung developed in this work has been applied to the α+ α and α+N systems. This model is based on an effective NN interaction, which enables a good reproduction of the elastic phase shifts for the α+ α and α+N systems.

The agreement with experimental bremsstrahlung cross sections is rather good but the comparison between theory and experiment requires more numerous and more accurate data to be conclusive. With an extension to the p shell, the present model could also describe heavier cluster systems such as 12C+p and 16O+p for which experimental data exist at low energies.