Over the past decade, we are witnessing a rapid growth of power conversion efficiencies of the perovskite solar cells approaching the conventional silicon single-crystal performances. Nowadays, the short circuit currents J_SC almost reached the fundamental limit. Therefore, the biggest space for improvement lies in the open-circuit voltage V_OC and fill factor FF of a solar cell. Both properties are directly affected by active defects in the active material. V_OC is reduced by non-radiative recombination on local defects, and FF is proportional to the charge carrier’s lifetime, which is also affected by defect density. A detailed understanding of the defect formation process is therefore crucial for its reduction and improvement of the final photovoltaic properties.

In our study, we used a simultaneous measurement of in-situ photoluminescence (PL) and grazing-incidence wide-angle X-ray scattering (GIWAXS) to observe the perovskite layer crystallization process in real-time. We show that the layer formation process can be divided into three stages:

1. The growth starts with nucleation and free growth of perovskite grains. According to the high PL intensity, the defect density remains at a very low level during this stage.

2. In the second stage, the PL intensity decreases by order of magnitude, although the perovskite GIWAXS signal proves continual film growth. It indicates the formation of defective grain boundaries, and the non-radiative charge carriers recombination takes place. This corresponds to an increase of non-radiative losses in V_OC by more than 100 mV. Direct passivation of these defects during this growth stage will significantly increase the final solar cell performance.

3. During the final part, the perovskite film slowly decomposes into PbI₂. This process passivates the surface, and the photoluminescence signal is partially recovered. This implies that although post-growth defect passivation is possible, it cannot cure all active recombination sites. [1]

Our study provides direct control of perovskite growth and, more importantly, defect formation and eventual passivation. The most significant advantage of such an approach is that the addition of passivating agents can be precisely timed, and the effect immediately measured. With the correct timing, the formation of defects at grain boundaries may be suppressed. This will lead to a significant improvement of V_OC and FF in the finalized photovoltaic device.