[eng] The application of electron diffraction to crystallographically characterize all kinds of materials has experienced new developments that have attracted some attention in recent years. A large number of structural analyses from different compounds have been already carried out with the help of 3D electron diffraction data that were not possible with the available X-ray methods. The use of a transmission electron microscope as an electron nano-diffractometer has proved to be advantageous when diffraction data from single nanocrystals are required, for instance in phase mixtures. In this way, the individual phases do not have to be purely synthesized, which always involves the risk of structural changes. The work presented here includes the development and implementation of a novel and universal routine for the accurate and reliable acquisition of electron diffraction data. It contains the alignment of the transmission electron microscope in order to set a quasi-parallel and precessed beam suitable for the acquisition of this kind of data, and the description and comparison of the developed technique to the other methods already available. This novel acquisition method has been called fast and automated diffraction tomography (Fast-ADT) and it is based on two consecutive tilt scans of the goniometric stage of the microscope; one to image the crystal and generate a crystal tracking file, and a second one to acquire the diffraction patterns while the beam is automatically shifted to follow the crystal at the different tilt angles. Such technique has been implemented in different experimental setups to test different detectors as well as different samplings of the diffraction space to point out their advantages and hindrances. The potential of this new data collection strategy to solve various crystallographic problems is illustrated using three known materials: barite, an inorganic salt stable under the electron beam used to check the level of accuracy that can be obtained from electron diffraction, RUB-5, a layered silicate with disordered features to show the advantage of high frame-rate cameras, and the pi-ferrosilicide, a crystalline phase from a commercial alloy measured to demonstrate how electron diffraction can reliably determine absolute structures. In addition, two unknown crystal structures from commercial products are fully determined and refined. The first one is the dehydrated DRED1 molecular crystal, an inorganic dye used in textile applications, which has been solved ab initio revealing the 46 non-hydrogen atoms of the two independent molecules in P-1. Furthermore, the structure refinement based on intensities calculated according to the dynamical diffraction theory has proved the position of some of the hydrogens, confirming the square-like H-bond network initially found from a Rietveld refinement based on a X-ray powder diffractogram. Finally, the incommensurate modulated structure of one of the polymorphs of dicalcium silicate (belite), a major constituent of cement, has been comprehensively characterized only by means of electron diffraction. Such crystal structure has been known since 1971 but the observed modulation has never been fully studied. Here, the Fast-ADT technique is presented as a robust acquisition method to systematically identify and properly characterize different crystallographic phases and features of powder mixtures. In this particular case, the precise knowledge of the different crystal structures in cement clinkers, such as the alpha’H-C2S, enables the exact phase analysis of these industrial phase mixtures directly from the manufacturing process, and paves the way for its CO2 emissions reduction that, at the moment, adds up to 5% of the global anthropogenic emission. In this context, a novel technique for the collection of 3D electron diffraction datasets has been successfully developed and implemented to increase the range of reliable available tools for the characterization of atomic structures from different materials., [cat] L’aplicació de la difracció d’electrons per caracteritzar cristal·logràficament tot tipus de materials ha experimentat nous desenvolupaments que han cridat certa atenció durant aquests últims anys. Un gran nombre d’anàlisis estructurals en diferents compostos ja s’han dut a terme amb l’ajut de dades tridimensionals de difracció d’electrons que no eren possibles amb els habituals mètodes de raigs X. L’ús d'un microscopi electrònic de transmissió com a nano-difractòmetre d’electrons ha demostrat ser molt més beneficiós quan es requereixen dades de difracció de nanocristalls individuals, per exemple en mescles de fases cristal·logràfiques. D’aquesta manera, les fases individuals no s’han de sintetitzar en estat pur, fet que sempre comporta el risc de canvis estructurals. El treball presentat aquí inclou el desenvolupament i implementació d’una nova rutina universal per l’adquisició precisa i fiable de dades de difracció d’electrons. El potencial d’aquesta nova estratègia de recopilació de dades per resoldre diversos problemes cristal·logràfics s’il·lustra mitjançant tres materials coneguts. A més a més, dues estructures cristal·lines desconegudes de productes comercials han estat determinades i refinades completament; un colorant orgànic de baixa simetria i una estructura modulada incommensurada d’un component principal del ciment. En particular, el coneixement precís de les diferents estructures cristal·lines dels clinkers de ciment, com ara la alpha’H-C2S, permet l’anàlisi exacta de les fases d'aquestes mescles industrials extretes directament del procés de fabricació, i facilita el seu estudi per reduir les emissions de CO2.