Dual DRT-based deconvolution of electrochemical signatures in perovskite solar cells

We introduce a dual distribution of relaxation (DRT) based approach for analyzing electrochemical impedance spectroscopy (EIS) data in perovskite solar cells (PSCs), combining regression and classification with Bayesian model selection and Havriliak-Negami (HN) modeling to resolve spectra into discrete, Lorentzian-like peaks. This time-domain decomposition offers a powerful alternative for identifying underlying physical processes, such as charge transfer, trap-assisted recombination, and ionic migration by directly extracting characteristic relaxation times (τ). In contrast to traditional equivalent circuit fitting or conventional DRT methods, which often yield broad and overlapping Gaussian-like peaks, our method enables sharper resolution of individual electrochemical signatures. Furthermore, we validated the framework using simulated EIS spectra for two distinct system types, determining the optimal number of peaks (Q) through statistical model selection. Applied to experimental PSC data under varying bias conditions, the approach helps to identify the voltage-dependent relaxation processes, including fast charge transfer (τ ∼10⁻⁶ s), intermediate trap-mediated recombination (τ ∼10⁻² s), and slow ionic motion (τ ∼1 s). Lower-Q models fail to capture low-frequency features such as polarization and charge accumulation, while optimal Q yields accurate, physically meaningful representations of device behavior. This data-driven methodology highlights time-domain DRT as a rigorous and insightful tool for dissecting the complex kinetics that govern PSC performance.