Catalytic innovations and Machine Learning – assisted approaches for CO2 utilisation: Current status and future prospects

• Acceleration of catalyst screening and design of advanced structure materials using ML • Optimisation of reaction conditions on CO 2 hydrogenation through ML algorithms • Integrate ML with LCA and technoeconomic analysis for process scale-up and sustainability • Future directions and limitations are discussed for data-driven CO 2 hydrogenation The development of carbon capture and utilisation (CCU) technologies is a promising approach for the transformation of carbon dioxide (CO 2 ) from a waste product into a valuable chemical feedstock. Herein, this review focuses on thermocatalytic CO 2 hydrogenation routes, outlining the conventional and recently emerged routes regarding catalyst development. Monometallic, bimetallic and supported systems as well as advanced structured materials are thoroughly discussed, focusing on their performance based on their structure-activity relationships. Machine learning (ML) techniques are also explored highlighting how correlations and patterns can be identified in large datasets, enabling rapid catalyst screening for the identification of the most suitable catalyst among thousands of possible candidates, saving resources, time and manual labour. Operating conditions also significantly affect process performance and product yield, therefore, are discussed, while ML is also introduced as a powerful approach to optimise reaction conditions for CO 2 hydrogenation. Finally, techno-economic aspects are outlined, evaluating the main barriers and opportunities for commercialisation. ML-assisted life cycle assessment (LCA) and techno-economic analysis (TEA) is also discussed for automating and increasing the predictive capability of these techniques by leveraging ML models. By emphasising on the convergence of catalyst innovation and optimum operating conditions, this review aims to guide future strategies toward efficient, scalable, and sustainable CO₂ conversion processes that support global decarbonisation goals, while bridging the gap between labour-intensive tasks by enabling automated testing and training and taking it a step further for industrial deployment.