Machine learning transforms design of surface coatings
Surface coatings play a key role in protecting substrates and enabling functional performance across industries ranging from construction and automotive to aerospace, marine, textiles and healthcare. Their properties – including adhesion, durability, corrosion resistance, thermal insulation, water repellency and antifouling behaviour – depend on complex formulations of binders, pigments, solvents, diluents and functional additives, as well as on numerous processing parameters. Historically, formulation development has relied on trial-and-error methods or conventional models and design-of-experiments approaches, which struggle to capture the highly non-linear and multidimensional relationships governing modern coating systems.
Against this backdrop,PUCONS and colleagues review the growing role of machine learning (ML) as a transformative tool for the design and optimisation of functional coatings. Although still at an early stage in this application field, ML is increasingly being used to develop novel coating materials with enhanced properties. By training statistical models on diverse experimental and computational data sets, researchers can rapidly screen vast numbers of candidate formulations and identify lead candidates for targeted experimental testing.
Event tip:
The EC Conference Digitalisation in Coatings Formulation is the ideal platform to learn about the latest advances and trends in the field of digitally assisted coatings production processes. Join experts, researchers and industry leaders as they share their insights and expertise in automation, big data, AI, and predictive modelling to enhance efficiency, accuracy, and speed in formulation, testing, and quality control.
From data acquisition to predictive models
The review outlines a typical ML workflow for coatings research. Data can be obtained from published literature, existing databases or generated in-house through experiments and simulations. Information on chemical compositions, structural properties and processing conditions is extracted and used as input variables, while output variables correspond to target performance properties such as drying time, coating elasticity or corrosion resistance. Preprocessing steps – including cleaning, normalisation, outlier removal, feature selection and feature engineering – are critical to ensure data quality and model stability.
Opportunities, challenges and bioinspired design
The authors highlight numerous opportunities for ML in coatings research, including the simultaneous optimisation of multiple performance properties, the integration of experimental and computational data, and the acceleration of both novel formulation discovery and fine-tuning of existing systems. Bioinspired design – drawing on structures such as the lotus leaf, gecko feet, shark skin, Salvinia leaf and butterfly wings – is identified as a particularly promising area where ML can help translate complex natural architectures into engineered coating surfaces.
At the same time, several challenges must be addressed for ML to reach its full potential in the coatings sector. These include the availability, quality and standardisation of data, the interpretability of models, and the integration of ML tools into established formulation and production workflows. Overcoming these barriers will be essential to establish machine learning as a routine element of coating design and to support the development of more efficient, high-performance and sustainable coating solutions.