Organic photovoltaics (OPVs) are flexible, low cost and easily processable, which provides them with a very short energy payback time compared to most PV technologies and makes them strong candidates for industrial mass production. The richness of organic synthesis has afforded a large library of molecular semiconductors, among which the combination of conjugated polymers as electron donors and fullerenes as electron acceptors has been demonstrated to be the best sellers as building blocks of OPV devices. In the past few years, the majority of research focus on OPVs has been devoted to improving their power conversion efficiencies by using new combinations of polymer and fullerene materials. Most devices are based on blend-cast bulk heterojunctions (BHJs), in which a polymer and fullerene are mixed together in a solution that is then used to cast the active layer of the organic solar cell. Because the nm-scale morphology of the film depends on so many of the details of how it is cast, the device performance of blend-cast BHJ solar cells is hypersensitive to the processing kinetics of the active layer. Thus, for any new set of OPV materials, an Edisonian approach involving the fabrication of hundreds of blend-cast devices is needed to find the processing conditions that lead to the optimal morphology and best device performance. In this thesis, I will focus on two main contributions that I have made to help rationally design OPVs. First, our group recently has gone beyond the traditional method of simply blending the donor and acceptor material by developing a new technique to process the active layer of OPVs called sequential processing. This method takes advantage of a pair of quasi-orthogonal solvents to process the two components used in the active layer separately. By studying a series of crystalline polymers with controlled regioregularities and polydispersities, I have found that increasing polymer crystallinity produces the opposite behavior in BHJ solar cells fabricated by sequentially-processing and blend-casting. This suggests that the two processing techniques are complementary and provides guidance on selecting the appropriate processing technique for a given polymer. Second, I have studied the performance and device physics of a new series of controllably tuned fullerene derivatives applied in traditional blend-cast active layers. We obtained a series of carefully designed 1,4-dibenzyl fullerene bisadducts synthesized by our collaborators in Prof. Yves Rubin's group. The fullerenes have methoxy substituents selectively positioned on pendant phenyl rings, which allows us to examine the effect of the subtle molecular changes on both macroscopic solar cell performance and the underlying device physics. Through carrier recombination studies, I have learned that solar cell performance often depends on the material's surface energy and the vertical phase segregation caused by this surface energy in the active layer. The results will allow us to offer new directions on how to select the best device structure with a given new fullerene material. Finally, I have helped to make an interesting discovery during my study of the device physics of as-cast sequentially processed solar cells. I found that the specific type of vertical phase segregation in the as-cast devices gives rise to dark carriers, whose presence can be measured using the charge extraction by linear increasing voltage (CELIV) technique. The dark carriers directly clearly are created by the evaporation of metal electrodes because I found no such carriers when non-metal interfacial layers were inserted between the metal and the organic layer. Through capacitance analysis and transmission electron microscopy studies, we found this n-type doping is caused by metal penetration into the fullerene domain. These findings could have significant impact on determining device performance, explaining device physics and guiding future research directions.