The Nuclear Magnetic Resonance (NMR) is an extraordinary tool to understand molecular structures. Its main limitation resides in its relatively low sensitivity when compared with other physico-chemical methods such as Mass Spectrometry (MS) or Optical Spectroscopy. Dynamic Nuclear Polarization (DNP), invented in the 50's but only widely applied in the last ten years, allows, by transferring the high polarization of paramagnetic centers to surrounding nuclei at low temperature, to enhance NMR signals by several orders of magnitude. This technique can be applied to solids at 100 K using Magic Angle Spinning (DNP-MAS) or in liquid phase through Dissolution-DNP as described by Ardenkjaer-Larsen in 2003. The efficiency of the polarization transfer is strongly dependent on the nature of the paramagnetic centers and on the way they are mixed with the target molecules. Part of this work will consist in the study of new ways to integrate paramagnetic centers into the matrix and also to limit their negative effects on NMR signals in liquid phase after dissolution. Because DNP-NMR has reached a stage where it should become a standard chemical method, a second objective of this thesis is to propose new chemical applications such as kinetic measurements that may be of interest in other domain of chemistry.