Data-anchored multistage mechanistic modelling of Ni-BTC- modified TiO₂ aerogel photoanodes for dye-sensitised solar cells

Dye-sensitised solar-cell photoanodes exhibit coupled optical, interfacial and charge-transport processes, making it difficult to assign photovoltaic improvement to a single material descriptor. Here, a data-anchored multistage modelling framework was developed for TiO₂ aerogel photoanodes containing 0–0.75 wt% Ni-BTC metal–organic framework (MOF), using the experimental dataset of Ramasubbu et al. as locked calibration and validation anchors. The workflow combines structural interpolation, band-gap and optical-generation descriptors, N719 injection energetics, electrochemical-impedance-derived transport and recombination descriptors, a reduced device-level J–V model, composition-specific digital twins, continuous composition optimisation, sensitivity analysis and Monte Carlo uncertainty propagation. The experimentally measured maximum power-conversion efficiency (PCE) of 8.846% at 0.50 wt% Ni-BTC coincides with the lowest interfacial resistance R₂ (9.81 Ω), a 1.1617 ms electron lifetime and the highest normalised transport descriptor τₙ/R₂. In contrast, 0.25 wt% exhibits the highest modelled injection index but a lower J sc , demonstrating that charge generation alone does not determine device output. At 0.75 wt%, the lifetime increases to 1.4601 ms, but R₂ and R₃ rise to 54.3 and 96.1 Ω, respectively, producing a severe transport penalty and reducing PCE to 4.353%. Monte Carlo propagation retained 0.50 wt% as the PCE optimum in all 10,000 realisations under the stated uncertainty assumptions. The framework therefore identifies interfacial resistance and coupled transport as dominant design variables and provides a reproducible route for prospective experimental validation of optimised MOF–TiO₂ photoanodes.