Perovskite-based solar cells have been extensively studied by the scientific community over the past decade and they are currently a very promising technology to be integrated into tandem PV module, for example associated with silicon solar cells [1]. However, one of the challenges lies in the up-scaling of the production of perovskite solar cells from small laboratory-scale cells (< 1 cm²) to larger modules [1,2]. In this context, there is considerable interest in extending the analysis previously conducted on a micrometer or millimeter scale [3-6] to a larger scale. Our work introduces, for the first time, full-sample size hyperspectral absolutely calibrated photoluminescence (PL) imaging applied to 16 cm² perovskite semi-transparent mini-modules.

We conducted HI photoluminescence acquisitions on semi-transparent perovskite mini-modules composed of glass/FTO/TiO2/triple-cation-perovskite/PTAA/ITO and followed their evolution over time for two weeks. In parallel, IV measurements were performed on the same semi-transparent module to track the corresponding evolution of electrical properties. We observe the presence of two categories of cells: cells that have local PL drops and relatively low PL intensity on average, and cells with relatively high PL intensity. The spectral information shows that their PL peak position is also different. By fitting the absolute spectra, we obtain the maps of local quasi-Fermi level splitting (QFLS), band gap value (E_g) and voltage loss (V_loss = E_g/q-QFLS), shown in the TOC graphic. This allows us to distinguish PL intensity variations caused by bandgap fluctuations from those due to defect-induced voltage losses.

In low-PL-intensity cells, local PL drops suggest probable shunt pathways. We found that these cells exhibit higher quasi-Fermi level splitting (QFLS) due to a better light-induced defect passivation but greater voltage losses due to larger bandgaps. In overall, they caused 2% of QFLS losses as compared to a theoretically homogeneous module without shunts. However, as supported by numerical simulations, they had the benefit of delaying the ageing of the device by screening ion migration.

When following the evolution over time of the mini-module, we observe a decrease in the bandgap and thus voltage loss, as the sum of the QFLS of the cells remains relatively stable (as shown in the TOC graphic). By comparing to the open-circuit voltage (V_oc) extracted from IV measurements, we could estimate that initially, the voltage loss due to the perovskite absorber and its interfaces (difference between blue and black curves) represented 92% of the total loss (difference between the blue and the red curves), while the remaining 8% are due to the collection properties and band alignment (difference between black and red curves). Over the two weeks of experiment the total V_loss of the module was reduced by 330 mV, with 220 mV (67%) coming from the passivation of the bulk and its interfaces, and 110 meV (33%) resulting from improvement in other layers properties and/or their band alignment with the perovskite absorber.

This work paves the way for a wide range of characterization studies to address for example the stability or scalability of many kinds of large size thin film devices. Probing compositional inhomogeneities and device performance at each step of the fabrication process of perovskite-based modules will be a powerful tool to accelerate their industrialization. It also opens the door for hyperspectral electroluminescence, which will provide a better understanding of charge transport properties at the module scale. Finally, it can also be applied to tandem characterization, in which additional layers complicate charge transport and may induce additional strain potentially leading to phase segregation.

This work has just been published in Solar RRL [7].