Nanosecond Recombination Dynamics in Perovskite Solar Cells with Bayesian Inference

Metal halide perovskites have emerged as an exciting class of optoelectronic materials, achieving a record-breaking 26% power conversion efficiency in single-junction solar cells in just over a decade. These materials are also showing great potential in applications such as light-emitting diodes and X-ray detectors. However, many advances in efficiency have been achieved empirically—through alterations in chemical composition and processing—without a complete understanding of the underlying photophysics.
In recent years, spectroscopic studies have significantly improved our understanding of how chemical and electronic compositions affect charge carrier recombination, a key factor in dictating solar cell efficiency. For mixed-halide perovskites, chemical heterogeneity has been shown to cause photodoping effects[1] and to funnel charges towards pristine, high-efficiency recombination regions, away from deep nanoscale trap clusters that form poor-quality, efficiency-detrimental regions[2,3].
To advance the commercialization of perovskite solar cells, it is crucial to explore how interfaces between the absorber material, charge transport layers, and electrodes influence recombination dynamics in complete devices. Previous studies have highlighted the detrimental effects of some common transport layers, such as C60[4], though much of this work has focused on steady-state behaviour. More understanding is needed through time-resolved studies.
In our work, we conduct transient absorption and photoluminescence measurements on a series of cell components, ranging from neat perovskite layers to complete devices. By examining recombination dynamics in the presence of various components, we aim to identify factors that enhance charge extraction and reduce recombination rates. We extend these studies to operating devices to investigate whether the presence of an external electric field alters recombination behaviour under "real-life" conditions.
Additionally, we demonstrate how Bayesian Inference can be employed to deconvolute experimental decay curves derived from complex recombination models. This method is gaining traction across the field, offering a more accurate representation of recombination parameters and avoiding the local minima that traditional fitting techniques often encounter.
By deepening our understanding of these complex systems, we aim to develop smarter, more efficient perovskite solar cells, offering a viable alternative to traditional silicon-based technology.