Recently, all-polymer solar cells consisting of donor- and acceptor- conjugated polymers have become a promising next-generation energy source with various advantages such as, flexibility, lightweight, high thermal stability, and cost-saving by solution processes.  Despite of these advantages, all-polymer solar cells still lag behind compared to conventional silicon solar cells or perovskite solar cells in terms of power conversion efficiency (PCE).  For further improvement in PCE, short-circuit current density (J_SC) should be increased.  It is therefore essential to fabricate all-polymer solar cells with a thick active layer.  Unfortunately, however, it is known that as the active layer thickness increases, bimolecular charge recombination becomes dominant rather than charge collection, resulting in a decrease in the fill factor (FF).  In other words, there is a trade-off relationship between J_SC and FF.  Here, we have focused on all-polymer solar cells based on crystalline donor polymers with different side chains (J51, J61, and J71) and a naphthalene diimide-based acceptor polymer (N2200) to discuss how charge collection and recombination impact on FF in the all-polymer solar cells with different active layer thicknesses.  For the J51:N2200 and J61:N2200 devices, FF was drastically degraded down to 0.4 with increasing active layer thickness.  For the J71:N2200 device, on the other hand, FF was only slightly decreased and maintained as high as 0.6 even for thicker active layers.
  To get further insight into the origin of such a different tendency in FF, we analyzed the charge recombination dynamics measured by transient photovoltage/photocurrent (TPV/TPC) and charge extraction (CE) for these devices.  On the basis of experimental data obtained, we estimated the reduction factor for charge recombination, which is defined by ζ = k_rec/k_L where k_rec is bimolecular recombination rate constant and k_L is the diffusion-limited Langevin recombination rate constant.  For all the devices, ζ was evaluated to be about 10⁻² order, suggesting suppressed bimolecular recombination compared to the diffusion-limited Langevin recombination.
  Subsequently, we simulated the J–V curve using the recombination kinetic parameters obtained in order to examine how recombination dynamics impacts on FF.  Assuming that the bimolecular recombination is a dominant current loss, the J–V curve can be represented by J(V) = J_GEN(V) + J_BR(V) where J_GEN(V) is photogenerated current density and J_BR(V) is the current density loss caused by bimolecular recombination. The simulation result shows that only J71:N2200 device is in good agreement with the experimental J–V curve, indicating that the decrease in FF is mainly derived by bimolecular recombination.  In stark contrast, for the J51:N2200 and J61:N2200 devices, the simulated J–V curve is not in agreement with the experimental one.  This is because the charge carrier density is underestimated in these devices.  Interestingly, the J–V curve is well reproduced by considering additional charge carriers, suggesting that there is an additional recombination driven by isolated charge carriers.  On the basis of the figure of merit α, proposed by Neher et al.,^[1] we conclude that high FF observed for the J71:N2200 device is due to charge collection 2500 times faster than charge recombination.