Factors which affect and limit the performance of substituted tris(bipyridine)cobalt(II) complexes in dye-sensitized solar cells

Abstract

Significant efforts were made towards the development of a novel, solid state dye-sensitized solar cell (DSSC) with a thin layer of cobalt redox polymer on the surface of the photoanode. The intent was to utilize atom transfer radical polymerization (ATRP) to grow a surface-bound polymer based on a cobalt tris(bipyridine) type complex which was only 1-10 monolayers thick. Ultimately, a thin cobalt redox polymer film was successfully grown in a surface-initiated polymerization. However, it is unlikely that the polymerization reaction proceeded in a controlled/living manner or, in fact, that it occurred by way of an ATRP-type mechanism at all. SEM and XPS experiments suggest that the polymerization was initiated at the surface by the dye, but that the active radical was typically not well-deactivated, resulting in isolated regions of uncontrolled polymer growth. Over time (approximately 10 hours), these islands of polymer merged, forming a fairly pinhole-free film on the macroscopic surface of the titania layer. Although reproducible, the film impaired photoanode performance in DSSC iV experiments. Whether it did so because of a reduced photoactive surface area or because of an intrinsic electrical resistance on the part of the polymer remains unclear. A suite of DSSC characterization experiments was developed which could be rapidly performed on a standard two electrode sandwich cell. In combination with a novel technique for mediator replacement without cell disassembly, this suite was utilized to probe the influence of a variety of parameters on DSSC functioning in order to elucidate the processes limiting performance in cells mediated by cobalt complexes. The results of this study clearly implicate diffusion of cobalt(III) to the cathode as frequently being rate-limiting. Strategies to alleviate this condition include increasing the cobalt(III) concentration as well as lowering the solvent viscosity. A somewhat unexpected result of the flow-through study is that TiO2 underlayers or TiCl4 treatments have a surprisingly small effect on the performance of DSSCs, despite their dramatic effect on dark currents. This may be a further indication that diffusion of cobalt(III) - and not recombination across the FTO surface - is performance-limiting in cobalt mediated DSSCs. A novel three electrode DSSC experiment, modified from two existing techniques, was devised and implemented in order to study cathode behavior in cobalt-mediated DSSCs. The new experiment incorporated a split photoanode in conjunction with a customized potentiostat circuit to allow for simultaneous monitoring of cathode potential, photoanode potential and cell current while still maintaining the thin electrode separation of the standard sandwich cell configuration. Interestingly, DSSCs with poor cathodes, characterized by decreased photocurrent and/or fill factor, exhibited only modest charge transfer overpotentials (or none at all) on the cathode. Even more perplexing, mass transfer overpotentials were not observed, even in clearly diffusion-limited cases. Whether these results are due to a deficiency in the experiment or are a consequence of the DSSC system being extremely sensitive to very small overpotentials is still uncertain. The possibility of mass transfer limitations in typical cobalt-mediated DSSCs was examined from a theoretical standpoint. Although conventional electrochemical models do not predict mass transfer of cobalt(III) to be a problem, actual DSSCs are likely more complicated than standard models of relatively simple systems. Two hypotheses were developed and examined. The first supposed that catalytic "islands" on the cathode were responsible for observed photocurrents but were simply not sufficiently numerous to provide adequate surface area for cobalt(III) reduction. However, experiments using cathodes intentionally fabricated to behave in this manner (catalytic gold nanoclusters on relatively inert FTO glass) did not support this hypothesis. The second hypothesis reasoned that diffusion of cobalt(III) through the mesoporous titania layer of the cell was hampered by the layer's small pore size, tortuosity, or both. Preliminary experiments using a modified rotating disk electrode indicate that this is very likely the case. Although further research is required to accurately quantify the effect, initial considerations and results suggest that the effective diffusion coefficient (uncorrected for porosity) of cobalt bipyridine complexes through mesoporous titania films is on the order of 1 x 10-7 cm2s-1 or less.