Building models to rationalize the outcome of experiments, predict the outcome of future measurements and design new materials and molecules is at the heart of chemistry. There is no understanding and progress in chemistry without models but the nature of chemical models is constantly evolving.
Electronic structure models are frequently coupled with virtual high-throughput screening to identify materials with desirable properties (Phys. Chem. Chem. Phys. 2020, Adv. Funct. Mater. 2020, Energ. Envir. Sci. 2019). The data produced by such models can be analysed to identify patterns and search for novel promising compounds via machine learning methods (Mater. Horiz. 2010).
Properties and functions are often related to dynamical properties, like in porous materials (Nature 2019), interfaces (Nat. Nanotech. 2018, JACS 2018, JACS 2020) and optical properties (Adv. Funct. Mater. 2016). Such dynamics often involve different scales and suitable methodology to bridge such scales. Furthermore, the correct interpretation of the results, e.g. from electronic structure, is often as important as their accuracy (Molecules 2020, Chem. Eur. J. 2018).
Some models of critical importance are entirely based on data (from cheminformatics to machine learning and AI), like those employed to guide our medicinal chemistry research (PNAS 2019, Nat. Comm 2019) and, increasingly, our research related to energy and materials (Chem. Sci 2021, Chem. Mater 2020).
We see the future of chemical models as based on the integration of multiple modelling methods and the development of suitable software (e.g. MOLPRO). We are proactively exploring completely novel approaches through our interactions with computer science and the development of modelling tools for robotics applications in chemistry.