Organic photovoltaic (OPV) cells are attractive for flexible, low-cost, large-area, and lightweight solar conversion applications. Despite this demand, robust and efficient devices have been limited by the quality of organic semiconductor materials and by the poor understanding and control of their interfaces. Interface modification using self-assembled monolayers or conducting polymers can be leveraged to tune the composition and phase segregation in binary bulk heterojunction photovoltaic cells. In this work, the interface composition of a 1:1 mixture of poly(3-hexylthiophene):[6,6]-phenyl-C61-butyric-acid-methyl-ester (P3HT:PCBM) was characterized using near-edge x-ray absorption fine structure (NEXAFS) spectroscopy as a function of surface energy. The substrates consisted of a low surface energy Nafion-based copolymer, 4-phenylbutyltrichlorosilane or octyltrichlorosilane self-assembled monolayers on SiO2, or high surface energy native SiO2. It was observed that while the free surface of the film was always P3HT-rich (7:3 P3HT:PCBM), the bottom interfacial composition varied from P3HT-rich (4:1 P3HT:PCBM) to PCBM-rich (1:4 P3HT:PCBM) as the surface energy of the substrate increased from 20 mN/m2 to 80 mN/m2. These observations were further supported by electrical characteration of bulk heterojunction films deposited on thin-film transistor structures where the surface energy of the gate dielectric was modified with self-assembled monolayers. The transistor performance exhibited higher hole mobility at P3HT enriched organic-dielectric interfaces (low surface energy substrates), while ambipolar transport was observed in devices with a PCBM enriched interface (high surface energy substrates). These observations of surface energy dependant interfacial composition should have clear implications for optimizing photovoltaic cell design in regards to "conventional" and "inverted" device architectures. However, P3HT:PCBM bulk heterojunction solar cells constructed on low and high surface energy substrates in conventional and inverted device structures exhibit nominally identical performance. Early efforts at modelling the effect of compositional gradients on photovoltaic performance suggest that this is expected given that current densities increase in constricted percolation pathways to maintain constant overall current. Although this may have little impact on initial device performance, the effects of higher current densities in the constricted interfacial regions on device lifetime are currently being investigated.