PHysics- and dAta-driven multiscale modelling desigN of layered lead halide perovskiTe mAterials for Stable phoTovoltaICs – PHANTASTIC

The overarching objective of PHANTASTIC is to develop a multiscale computational materials engineering approach combining data- and physics-driven models for the design, and experimental validation, of multi-layered lead halide perovskites with improved stability.
Hybrid perovskites are at the forefront of photovoltaics research. Combining large optical absorption cross-section, long-range and balanced charge transport, resilience to deep trap formation and small exciton binding energy, 3D lead halide perovskite solar cells have reached certified Power Conversion Efficiency, PCE, close to ~25.2%. There is, however, one caveat, namely 3D lead halide perovskites feature limited long-term stability, which has both intrinsic (as the enthalpy of reaction associated with decomposition is close to zero) and extrinsic (exposure to moisture or oxygen speeds up degradation) origins. This has triggered increasing interest for lower-dimensionality hybrid lead halide perovskites, such as Ruddlesden-Popper 2D layered materials, displaying improved stability and reduced hysteresis behaviour, likely because of lower ionic diffusion in the vertical device direction. 2D lead halide perovskites also turn into smaller PCEs, which has prompted extensive experimental and modelling efforts dedicated to the design of layered materials incorporating ?-functional organic molecules. These works aim at improving charge transport in the direction normal to the inorganic layers as this, together with increased excitonic effects associated with dielectric and spatial confinements, appears to be a key bottleneck. In parallel, surface defect passivation by means of various coatings on top of the photoactive material has been shown to be a successful approach to boost PCEs and improve on stability. In this context, a natural choice for coating of the 3D lead halide perovskite is the use of layers of parent 2D lead halide perovskites that would simultaneously act as a hole or electron transport/extraction layers. The broad range of organic cations that can be incorporated in these vertical hetero-structures with piled up, tailored, 3D, 2D or quasi-2D layers offers an immense versatility in material and device architectures.
However, the progress thus far was largely driven by trial and error with the chemical-physics understanding lagging behind. The full exploitation of the potential of the materials and device architectures requires a systematic, first-principles based, exploration of the chemical space based on physical grounds and guided by data-driven models. Such an approach also needs to encompass multiple length scales, as a detailed description of the photophysics of these multi-layered structures is out of reach even to the most efficient atomistic computational models. On the other hand, while being amenable to include the salient physics of complex architectures, device models can only become predictive if they rely on input data provided by microscopic ab initio simulations. This is where PHANTASTIC stands. Built around the unique and truly complementary expertise of the partners, we will devise a bottom-up materials engineering approach to functioning lead halide based photovoltaic cells that articulates around three main, closely intertwined, pillars: (i) The training of Machine Learning (ML) algorithms against state-of-the-art ab initio molecular dynamics and electronic structure calculations to develop classical potentials for ions and tight-binding models for electrons, respectively; (ii) The implementation of a numerical solver to coupled drift-diffusion Poisson equations parameterized against microscopic simulations of Density of States (DOS) and dielectric constant vertical grading; and (iii) The assessment using advanced experimental characterization tools of the structural rearrangements with time and exposure to environmental factors and light irradiation of (quasi)2D, 3D lead halide perovskites and their vertical heterojunctions.