Towards a sustainable lime production through machine learning molecular dynamics simulations

Lime, the common name for calcium oxide, is a key material for industries ranging from steel manufacturing to construction, chemical production, and paper processing. It is produced through the calcination of calcium carbonate (CaCO3) in large kilns above 1000 °C, a process that is both energy- and carbon-intensive. About one-third of lime production emissions arise from fuel combustion to heat the kilns, with an average energy demand of roughly 4.25 GJ per ton. Improving the efficiency of this process would therefore yield substantial environmental and economic benefits. Within the framework of the European Innovation Council project MOJITO, this project aims to elucidate the atomistic mechanisms governing CaCO3 calcination through large-scale molecular dynamics (MD) simulations based on machine learning potentials (MLPs). The MLPs are trained on first-principles data via an active learning framework, and they will enable MD simulations that combine quantum accuracy with computational efficiency. These simulations will explore the role of temperature, crystalline polymorphism, and chemical environment on the kinetics of CaCO3 decomposition, providing detailed insight into reaction pathways and rate-limiting steps that are inaccessible to experiments. The results will deliver a fundamental, atomistic understanding of the calcination process and guide the optimization of industrial conditions for lime production, supporting the development of next-generation, low-carbon calcination technologies.