In this thesis we investigated the separation of two endocrine disrupting chemicals (EDC), bisphenol-A (BPA) and nonylphenol (NP) from water over the defect free silicalite zeolite. Two force-fields were investigated, the OPLS-AA force-field which is an all-atom one, and the OPLS-UA force-field which is a united atom one. In order to be able to simulate BPA, we simulated and studied the diffusion of different molecules in silicalite. We compared two famous bulk water models, the non-rigid TIP3P modified for CHARMM model and the rigid SPC model, to literature and simulated the diffusion of these water molecules at temperatures from 300K to 600K. We found that these models coupled with our parameters for silicalite compared poorly with literature except for values calculated by Yazaydin et al. The mean-square displacements (MSDs) were more important in the x-direction (sinusoidal channel) than in the expected y-direction (straight channels) for both models resulting in small self-diffusion coefficient values. Results tended to improve as temperature increased. We believe that the high number of hydrogen bonds, implying the presence of clusters of water molecules, is responsible for the poor self-diffusion coefficient. The charges chosen to describe our silicalite zeolite, +2.05, may also be a reason of our small self-diffusion coefficient. We then investigated the self-diffusion of aromatic molecules at 300 and 400K. Benzene, phenol and toluene were studied. We found self-diffusion coefficients for benzene that did not compare well to experiments but that was close to simulation work done by Rungsirisakun et al. Our diffusion coefficients for benzene were several orders of magnitude bigger than the experimental values found in literature for both force-fields. The diffusion patterns for both phenol and toluene did not allow us to calculate self-diffusion coefficients for both investigated force-fields. We believe that the jumps in the MSDs of these molecules are due to the rotation that they undergo in the nanopores. Phenol anchors to the framework by hydrogen-bonds between the hydrogen of its alcohol group and the oxygen of the framework. The diffusion seems to happen when the alcohol group is in a line with one channel. The same diffusion phenomenon was seen for toluene molecule but was related to the methyl group attached to its benzene ring. When this group is in front of a channel, the energetic barrier is reduced and the molecule can diffuse through it. Finally bigger molecules were simulated and studied. Neopentane seemed to have a very low self-diffusion coefficient in silicalite if it could move at all. We report values of self-diffusion of 1.3 10-14 m2.s-1 at both 300K and 400K. This value seems a little high compared to benzene experimental self-diffusion coefficient values that are in the same order of magnitude at both temperatures. The linear nonylphenol molecule that we simulated seemed to diffuse through silicalite with patterns that were close to the one seen for phenol. The hydrogen bonding between its alcohol group and the framework slows down its diffusion in silicalite. With the same reasoning as for phenol we decided not to calculate diffusion coefficient for NP. The last molecule investigated was bisphenol-A (BPA). We found that BPA almost did not diffuse through silicalite. The size of the molecule can explain why it did not diffuse, but we believe that the angle between the two phenol groups should be able to bend enough for it to diffuse, slowly, through silicalite. Our conclusion is that the two phenol groups at both ends of the molecules are the most important factor in its very slow diffusion. Hydrogen bonding is taking place at both ends making it very hard for the molecule to move in the framework. We decided to generate self-diffusion coefficients for this molecule because the diffusion process did not have jumps. We found self-diffusion coefficient that are 3.10-15 m2.s-1 and 15. 10-15 m2.s-1 at 300 and 400K respectively for the OPLS-AA force-field, and 11.6.10-15 m2.s-1 and 6.68.10-15 m2.s-1 at 300 and 400K respectively for the OPLS-UA force-field. The last result was unexpected as we thought that the self-diffusion coefficient was going to increase with temperature. We believe that running much longer simulations for every molecule that we studied should give more reasonable and reliable results as the self-diffusion coefficients values are very small.