Interweaving Synthesis, Theory, Experiment, Devices, Data Tools and Machine Learning for Exploring Materials—Example of Halide Perovskites

Every material or material class comes with its individual properties and functions that may be useful for a certain type of application. Understanding the respective peculiarities, is particularly interesting from a fundamental point of view. Employing first-principles theory, one can explore the relevant interactions on the electronic scale. A broad variety of available experimental probes allows for in-depth characterization. Bringing both together, often highlight discrepancies between computational results, typically obtained for perfect crystals, and measurements performed on real materials. Moreover, uncertainties may arise coming from the preparation and/or treatment of the investigated samples. On top of all this, every technique – be it experimental or theoretical – comes with its own subtleties. Taking the halide perovskites as an example, for instance, a reliable calculation of the electronic structure even with forefront methodology is still a remarkable challenge [1]. This makes it even more challenging to relate these results to real-world halide perovskites used in state-of-the-art devices. This is also in part due to the sample quality, concerning structural disorder, instability issues, and complex compositions.<br/><br/>Augmented datasets that combine knowledge derived from all perspectives – sample synthesis, experiments, devices, and <i>ab initio</i> calculations – have the power of overcoming this barrier and answering scientific questions that cannot be resolved from their individualistic perspectives. Within the FAIRmat consortium [2] (https://fairmat-nfdi.eu), the NOMAD Laboratory [3] (https://nomad-lab.eu), originally developed for computational materials science, is now adapting towards managing experimental material-science data and their various applications, one of them being solar cells.<br/><br/>In this presentation, we demonstrate with the example of organic-inorganic halide perovskites how NOMAD enables the exploration of materials in an interoperable manner from both, the theoretical and the experimental perspective, including synthesis, characterization, and device properties. This enables not only scientific insight through in-depth comparison and analysis but provides also a basis for employing novel artificial-intelligence based workflows and tools. For example, users may train their machine-learning models with theoretical results or experimental data either coming from NOMAD or from their own labs, or from curated datasets like the Perovskite Database Project [4], or with a combination of all of them.<br/><br/>Work carried out in collaboration with J. Márquez Prieto, E. Unger, T. Unold, L. Himanen, M. Scheidgen and the entire FAIRmat team.<br/><br/>[1] C. Vona, D. Nabok, and C. Draxl, Electronic structure of (organic-)inorganic metal halide perovskites: the dilemma of choosing the right functional, Adv. Theory Simul. 5, 2100496 (2022).<br/>[2] M. Scheffler, M. Aeschlimann, M. Albrecht, T. Bereau, H.-J. Bungartz, C. Felser, M. Greiner, A. Groß, C. Koch, K. Kremer, W. E. Nagel, M. Scheidgen, C. Wöll, and C. Draxl, FAIR data enabling new horizons for materials research, Nature 604, 635 (2022).<br/>[3] C. Draxl and M. Scheffler, The NOMAD Laboratory: From Data Sharing to Artificial Intelligence<br/>J. Phys. Mater. 2, 036001 (2019).<br/>[4] Eva Unger and T. Jesper Jacobsson, The Perovskite Database Project: A Perspective on Collective Data Sharing, ACS Energy Lett. 7, 1240 (2022).