A typical organic solar cell is characterized by a very thin absorbing active layer consisting of donor and acceptor, often not thicker than 100 nm. However, in the overwhelming majority of investigated systems fullerene-based acceptors are used that contribute only weakly to photon-harvesting - in combination with the low thickness only a limited fraction of incident photons can be absorbed. Therefore, increasing the absorption of the photoactive layer by using an absorbing acceptor is a rational strategy that could give rise to an enhanced photon harvesting and hence an improvement of the photocurrent. Among the various classes used as alternative acceptors are perylene diimides (PDI). They show high molar extinction coefficients, good photo-stability and offer versatile synthetic variations.
An encouraging example for a solar cells making use of PDI is the recently published p DTS(FBTTh2)2:PDI system that shows efficiencies as high as 3% for optimized devices. However, the efficiency depends strongly on the use of a solvent additive, namely diiodooctane (DIO), that pushes the efficiency by a factor of 30. In this work the influence of DIO on the overall performance is investigated by sub-picosecond to millisecond pump-probe transient absorption spectroscopy (TA) in the visible as well as the near-infrared region as well as and time-resolved photoluminescence (TRPL) spectroscopy. The former tracks the dynamics of non-radiative species whereas the latter enables the observation of emissive decay channels.
Striking differences in the processes of charge generation as well as charge recombination are observed that allow offering convincing conclusions on the positive effect of DIO. Sharper peaks point towards a higher degree of crystallinity if DIO is used and processes get more intensity dependent, owing to a more efficient generation of free charges rather than bound charge carriers. Additionally selective excitation PDI allows for further conclusions. A delayed depopulation of p-DTS(FBTTh2)2 can be attributed to diffusion limitation which does not occur in samples without DIO, suggesting smaller domains when DIO is used which, in turn, facilitates geminate recombination. This trend is also continued in charge recombination where fitting of the data reveals a respectable fraction of free charges in the optimized sample, whereas the non-optimized blend suffers from a high share of bound charges. The results of this work contribute to a better understanding which fundamental processes are connected to the improvement of solar cell devices.