Triple mesoporous layer devices containing a TiO₂ electron transport layer, a ZrO₂ insulating layer and carbon as the hole transporting contact show great promise for scale-up and wide spread implementation. In order to improve these devices and begin to challenge perovskite and inorganic PV record efficiencies a deeper understanding of their operation, and in particular sources of performance loss, is needed.

The current state-of-the-art triple mesoscopic devices use a mixed cation perovskite, consisting of methylammonium and 5-aminovaleric acid (5-AVA). The AVA containing perovskite has been shown to give greater stability and performance – linked to 2D/3D structuring of the perovskite as well as interfacial modifications at the TiO₂ surface. A range of anomalous behaviours have been observed in these cells in response to illumination. They undergo a slow (several minutes) light soaking effect during which time the JV performance of the device is vastly improved. They also show improvement when exposed to a high relative humidity. Both of these processes are reversible, although in the case of the light soaking effect the deterioration in performance is almost instantaneous upon turning off the light (i.e. there is no residual effect when the light is turned back on).

A range of complimentary optoelectronic techniques have been employed in order to study these effects, including transient photovoltage (TPV), differential capacitance and impedance spectroscopy. A striking feature observed using TPV measurements is the presence of a negative photovoltage transient, comparable to that observed in our previous work on planar TiO₂ devices. This behaviour suggests the presence of high rates of interfacial recombination at the TiO₂ surface. In these mesoporous based carbon cells the phenomena is observed at room temperature and is very slow to disappear under continued bias light illumination. In the previous case of the planar devices the negative transient was shown to diminish over time as ions in the perovskite redistributed, leading to a reduction in the recombination rate. For the planar devices this effect was only observed at low temperature (at which the ionic motion was sufficiently slowed) and even then disappeared much quicker than in the mesoporous carbon devices at room temperature. Knowledge of this process has allowed us to assess the impact of different treatments and processing conditions on interfacial recombination, which has helped to guide the improvement in device efficiency.