La fission nucléaire consiste en la brisure d'un noyau lourd, généralement un actinide, en deux noyaux plus légers (ou trois dans quelques rares cas). Ce phénomène a été découvert par Hahn et Strassman en 1938. Très rapidement Meitner et Frisch proposèrent une explication théorique pour ce processus à l'aide du modèle de la goutte liquide. Depuis les modèles n'ont cessé d'évoluer et de se complexifier à travers l'ajout de nouveaux mécanismes et l'observation de nouveaux phénomènes. L'amélioration des modèles est un enjeu important à la fois pour la compréhension fondamentale du processus de fission mais aussi pour les applications. En effet, le dimensionnement des réacteurs futurs s'appuie de plus en plus sur des simulations numériques. Il devient dès lors primordial de réduire les incertitudes associées aux données utilisées. Cela passe alors par la validation des hypothèses sous-jacentes des modèles de fission nucléaire.Dans le cadre de cette thèse, on s'intéresse à deux aspects de la fission nucléaire qui permettront de tester la robustesse des théories. L'un des aspects concerne l'étude des fragments de fission issus de la région de la symétrie à travers la mesure des rendements et des distributions en énergie cinétique. L'autre aspect étudié est le moment angulaire des fragments de fission.Afin d'accéder au moment angulaire des fragments de fission, l'une des possibilités est d'analyser les propriétés des particules promptes, qui est l'une des ambitions du projet FIPPS (FIssion Product Prompt gamma-ray Spectrometer). Une partie de ce travail a été de caractériser les propriétés des spectromètres magnétiques gazeux à travers des mesures expérimentales et le développement d'une simulation Monte Carlo.La seconde partie de ce travail a consisté en la mesure de rapports isomériques et en l'extraction de la distribution du moment angulaire des fragments de fission à l'aide d'un code de désexcitaiton nucléaire. La mesure d'un noyau doublement magique ($^{132}$Sn) permet de mettre en lumière les limites actuelles des modèles de fission.Enfin la dernière partie de ce travail porte sur la mesure des rendements et des distributions en énergie cinétique des fragments de fission. Certains modèles prédisent l'existence de modes dans la fission nucléaire. La région des masses symétriques est dès lors un lieu de choix pour vérifier la validité de ces affirmations.Il est à noter qu'en parallèle de ces études, un accent fort a été mis sur le développement de méthodes d'analyse s'appuyant sur des outils statistiques afin notamment d'améliorer l'évaluation des incertitudes expérimentales. / Nuclear fission consists in splitting a nucleus, in general an actinide, into smaller nuclei. Despite nuclear fission was discovered in 1939 by Hahn and Strassman, fission models cannot predict the fission observables with an acceptable accuracy for nuclear fuel cycle studies for instance. Improvement of fission models is an important issue for the knowledge of the process itself and for the applications. To reduce uncertainties of the nuclear data used in a nuclear reactor simulation, a validation of the models hypothesis is mandatory.In this work, two features of the nuclear fission were investigated in order to test the resistance of the theories. One aspect is the study of the symmetric fission fragments through the measurement of their yield and kinetic energy distribution. The other aspect is the study of the fission fragment angular momentum.Two techniques are available to assess the angular momentum of a fission fragment. The first one is to look at the properties of the prompt $gamma$. The new spectrometer FIPPS (FIssion Product Prompt gamma-ray Spectrometer), is currently under development at the ILL and will combine a fission filter with a large array of $gamma$ and neutron detectors in order to respond to these issues. The first part of this work is dedicated to the study of the properties of a Gas Filled Magnet (GFM) which is the type of fission filter considered for the FIPPS project.The second part of this work deals with the measurement of isomeric yields and evaluations of the angular momentum distribution of fission fragments. The study of the spherical nucleus $^{132}$Sn shed the light on the current limits of fission models.Finally, the last part of this work is about the measurement of the yields and kinetic energy distributions of symmetric fission fragments. Since models predict the existence of fission modes, the symmetry region is a suitable choice to investigate this kind of prediction.In parallel with all these studies, an emphasis on the development of new methods derived from statistical tools is achieved in order to better control the uncertainties and estimate the biases.

Two long-lived halogen radionuclides (36Cl, T1/2 = 301 kyr, and 129I, T1/2 = 15.7 Myr) have been studied by means of Accelerator Mass Spectrometry (AMS) at the Uppsala Tandem Laboratory. The 36Cl measurements in natural samples using a medium-sized tandem accelerator (~1 MeV/amu) have been considered. A gas-filled magnetic spectrometer (GFM) was proposed for the separation of 36Cl from its isobar, 36S. Semi-empirical Monte-Carlo ion optical calculations were conducted to define optimal conditions for separating 36Cl and 36S. A 180° GFM was constructed and installed at the dedicated AMS beam line. 129I has been measured in waters from the Arctic and North Atlantic Oceans. Most of the 129I currently present in the Earth's surface environment can be traced back to liquid and gaseous releases from the nuclear reprocessing facilities at Sellafield (UK) and La Hague (France). The anthropogenic 129I inventory in the central Arctic Ocean was found to increase proportionally to the integrated 129I releases from these reprocessing facilities. The interaction and origin of water masses in the region have been clearly distinguished with the help of 129I labeling. Predictions based on a compartment model calculation showed that the Atlantic Ocean and deep Arctic Ocean are the major sinks for the reprocessed 129I. The variability in 129I concentration measured in seawater along a transect from the Baltic Sea to the North Atlantic suggests strong enrichment in the Skagerrak–Kattegat basin. The 129I inventory in the Baltic and Bothnian Seas is equal to ~0.3% of the total liquid releases from the reprocessing facilities. A lake sediment core sampled in northeastern Ireland was analyzed for 129I to study the history of the Sellafield releases, in particular the nuclear accident of 1957. High 129I concentration was observed corresponding to 1990 and later, while no indication of the accident was found. The results of this thesis research clearly demonstrate the uniqueness and future potential of 129I as a tracer of processes in both marine and continental archives.