Artificial intelligence (AI) and Machine learning (ML) are transforming colloid and interface science by enabling predictive modelling, autonomous experimentation, and accelerated material design. This review highlights recent advances organised in four topics: (1) prediction of basic physical properties; (2) image analysis; (3) process design, monitoring and optimisation; and (4) morphology and phase behaviour prediction. AI models have improved the prediction accuracy of interfacial tension, critical micelle concentration, foam stability, and complex structure–function relationships, in particular, integrated generative AI approaches support the design of new surfactants and emulsifiers. Image analysis has automated microstructural characterisation and enabled real-time quality control, while AI-enhanced process design has delivered digital twins, closed-loop optimisation, and sustainability-oriented workflows. Morphology and phase behaviour prediction has combined simulation-driven neural networks with generative approaches to accelerate material discovery. The future of AI applications in colloids will be shaped by experimental database design and standardisation, hybrid AI methods integrating physics and surrogate modelling, and AI agents leveraging large language models for literature mining, data curation, and experimental optimisation. Together, these developments promise to establish data-rich, physics informed, and increasingly autonomous research ecosystems for colloids and interface science, accelerating material understanding and design.

Artificial intelligence in colloid and interface science: Current research, challenges and future directions / Sridharan, Simha; Bailey, Tom; Marcato, Agnese; Simone, Elena; Watson, Nicholas. - In: CURRENT OPINION IN COLLOID & INTERFACE SCIENCE. - ISSN 1359-0294. - (2025). [10.1016/j.cocis.2025.101965]

Artificial intelligence in colloid and interface science: Current research, challenges and future directions

Agnese Marcato;Elena Simone;
2025

Abstract

Artificial intelligence (AI) and Machine learning (ML) are transforming colloid and interface science by enabling predictive modelling, autonomous experimentation, and accelerated material design. This review highlights recent advances organised in four topics: (1) prediction of basic physical properties; (2) image analysis; (3) process design, monitoring and optimisation; and (4) morphology and phase behaviour prediction. AI models have improved the prediction accuracy of interfacial tension, critical micelle concentration, foam stability, and complex structure–function relationships, in particular, integrated generative AI approaches support the design of new surfactants and emulsifiers. Image analysis has automated microstructural characterisation and enabled real-time quality control, while AI-enhanced process design has delivered digital twins, closed-loop optimisation, and sustainability-oriented workflows. Morphology and phase behaviour prediction has combined simulation-driven neural networks with generative approaches to accelerate material discovery. The future of AI applications in colloids will be shaped by experimental database design and standardisation, hybrid AI methods integrating physics and surrogate modelling, and AI agents leveraging large language models for literature mining, data curation, and experimental optimisation. Together, these developments promise to establish data-rich, physics informed, and increasingly autonomous research ecosystems for colloids and interface science, accelerating material understanding and design.
File in questo prodotto:
File Dimensione Formato  
1-s2.0-S1359029425000718-main.pdf

accesso aperto

Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Creative commons
Dimensione 1.7 MB
Formato Adobe PDF
1.7 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/3003601