The internal redistribution of mobile ions in halide perovskites has long been considered a potential source of performance degradation in solar cells and other devices.  Here we report evidence that electric fields directed across the plane of the substrate, induced by electrode edges such as the boundary of metallization or the transparent conductive oxide, frequently dictate patterns of degradation across a range of stability tests. These edge effects are especially apparent in small-area laboratory-scale devices wherein they manifest under forward- and reverse-bias stressing, maximum power-point tracking, thermal stability testing, and even simple aging in storage. These edge effects directly illustrate the detrimental effects of ionic accumulation and/or depletion in solar-cell and similar device types.  

Prior studies have already established that bias-induced material degradation occurs in test devices made with laterally spaced electrodes, and generally attributed the observed effects to ion migration, with some of these observing the implied compositional changes directly [1]. However, the relevance of specifically lateral ion migration has not been previously discussed with reference to solar-cell and LED-type devices to our knowledge, where ion migration is generally considered in only 1-dimension perpendicular to the substrate.

We begin by using model calculations to show that in principle lateral redistribution can alter ion concentrations in active areas by factors of 20 or more, depending on the device architecture and initial conditions, and on timescales that are relevant for stability testing. We then present experimental evidence that in mixed-halide perovskites (Cs_xFA_yMA_1-x-yPb(I_zBr_1-z)₃), there is at least one ionic species with high lateral mobility (D>10^-7cm²s^-1), which causes significant edge-effects in some p-i-n solar cells and LEDs. The manifestation and severity of these edge effects are found to depend sensitively on the perovskite stoichiometry. Effects range from the relatively subtle (e.g. small modulations in photoluminescence intensity near edges), to catastrophic failure that propagates inwards from active area boundaries. We shall also show evidence that point degradation, a common mode of failure in which degradation spreads from one or more isolated defects, is mediated by an ionic species instead of a neutral one (e.g. I₂). Connections between edge degradation and reverse-bias damage will also be discussed.

Our observations imply that device size should be considered as a significant factor in stability studies, and shed new light on important modes of perovskite degradation.