Photoinduced Dynamic Defects Responsible for the Giant, Reversible, and Bidirectional Light-Soaking Effect in Perovskite Solar Cells

Abstract

Perovskite solar cells (PSCs) exhibit large, reversible, and bidirectional light-soaking effects (LSEs); however, these anomalous LSEs are poorly understood, limiting the stability engineering and commercialization. We present a unified defect theory for the LSEs in lead halide perovskites by reconciling their defect photochemistry, ionic migration, and carrier dynamics. We considered typical detrimental defects (IPb, Ii, VI) and observed that two atomic configurations were favored, where the carrier lifetime of one configuration was nearly 1 order of magnitude longer than that in the other. First-principles calculations showed that light illumination promotes ion-diffusion-assisted transitions from energetically stable configurations to metastable configurations, which are converted back to stable configurations in the dark. Fermi-level-dependent formation energies of stable/metastable configurations were used to rationalize contradictory experimental results of anomalous LSEs in PSCs observed in various studies, thus providing insights for minimizing the LSE to achieve high-performance stable PSCs.

Revisiting the Iodine Vacancy Surface Defects to Rationalize Passivation Strategies in Perovskite Solar Cells

“…The differences in formation energy and electronic structures of V I + , V I , and V I − at the first and second layers provide the atomic basis for the light-soaking effect in PSCs, 54 which involves the defect diffusion under illumination. The climbing image-nudged elastic band (CI-NEB) method was then used to calculate the kinetic energy barriers of ion migration at the different charged states, and the results are shown in Figure 4a.…”

“…Therefore, it is essential to revisit the V I defect near the surface of perovskite MAPbI 3 , such as the subsurface second, third, and fourth layers as shown in Figure a. For these three atomic layers, the stable defect configurations of the 0 and +1 charged V I mainly are nondimer structures (except for V I 0 in the second layer because its dimer structure has a lower total energy than the 0.12 eV nondimer, and this behavior is the opposite of that of the bulk system), and these results are shown in the Table . For the −1 charged state, the stable configuration of V I is the dimer structure.…”

Defect-Polaron and Enormous Light-Induced Fermi-Level Shift at Halide Perovskite Surface

“…Our proposed picture is based on some theoretical predictions, − which found that a V I defect can be deformed into the so-called Pb-dimer when it captures an additional electron. We note that Pb-dimer formation in the bulk has been disputed as a computational artifact in some recent works in which a more accurate HSE06 hybrid functional with spin–orbit coupling (SOC) is used. , However, another computational work in which the HSE06+SOC is also employed has shown that Pb-dimer formation can be energetically feasible at the surface or in regions where two V I defects are clustered .…”

“…Theoretical studies − have shown that when the V I has a charge state of 0 (i.e., a single electron is bound by the positive V I defect), the defect level is slightly below the CBM, which corresponds to a shallow electron trap. However, if the defect captures an additional electron (the charge state changes to −1), the lattice deforms and a Pb dimer is formed (Figure e).…”

“…However, if the defect captures an additional electron (the charge state changes to −1), the lattice deforms and a Pb dimer is formed (Figure e). The lattice deformation lowers the energy of the defect level, which results in the formation of a deep electron trap. − The additional electron can be originated from optical excitation, which can explain the observed light-induced effect in MAI-deficient samples. Although some recent simulation works that employ a more accurate functional have discarded Pb dimer formation in the bulk as a computational artifect, , Pb dimer formation is still energetically feasible near the surface or around clustered V I defects .…”