Search for: All records

Abstract Decarbonizing concrete production is a critical step toward achieving carbon neutrality by 2050. This paper highlights the advancements in artificial intelligence-assisted design of low-carbon cost-effective concrete, focusing on integrating machine learning-based property prediction with multi-objective optimization. Data collection and processing techniques, such as automatic data extraction, artificial data generation, and anomaly detection, are first discussed to address the importance of dataset quality. Strategies that capture physicochemical information of ingredients, including by-product supplementary cementitious materials and recycled aggregates, are then examined to enhance model generalizability. Various machine learning models—from individual regression approaches to heterogeneous ensemble methods—are compared for their predictive accuracy and robustness. Methods for hyperparameter tuning, model evaluation, and interpretation to ensure reliable and interpretable predictions are reviewed. Design optimization approaches are then highlighted, ranging from grid/random searches to more sophisticated gradient-based and metaheuristic algorithms, aimed at minimizing carbon footprint, embodied energy, and cost. Future research avenues encompass (1) application-specific design frameworks that integrate critical objectives—mechanical performance, durability, fresh-state behavior, and time-dependent properties such as creep and shrinkage—tailored to specific structural and environmental requirements; (2) holistic design optimization that simultaneously refines mixture design and structural parameters; and (3) probabilistic approaches to systematically manage uncertainties in materials, structural performance, and loading conditions systematically. 

Large Language Models (LLMs) are reshaping many aspects of materials science and chemistry research, enabling advances in molecular property prediction, materials design, scientific automation, knowledge extraction, and more. Recent developments demonstrate that the latest class of models are able to integrate structured and unstructured data, assist in hypothesis generation, and streamline research workflows. To explore the frontier of LLM capabilities across the research lifecycle, we review applications of LLMs through 34 total projects developed during the second annual Large Language Model Hackathon for Applications in Materials Science and Chemistry, a global hybrid event. These projects spanned seven key research areas: (1) molecular and material property prediction, (2) molecular and material design, (3) automation and novel interfaces, (4) scientific communication and education, (5) research data management and automation, (6) hypothesis generation and evaluation, and (7) knowledge extraction and reasoning from the scientific literature. Collectively, these applications illustrate how LLMs serve as versatile predictive models, platforms for rapid prototyping of domain-specific tools, and much more. In particular, improvements in both open source and proprietary LLM performance through the addition of reasoning, additional training data, and new techniques have expanded effectiveness, particularly in low-data environments and interdisciplinary research. As LLMs continue to improve, their integration into scientific workflows presents both new opportunities and new challenges, requiring ongoing exploration, continued refinement, and further research to address reliability, interpretability, and reproducibility. 

Abstract Large language models (LLMs) are reshaping many aspects of materials science and chemistry research, enabling advances in molecular property prediction, materials design, scientific automation, knowledge extraction, and more. Recent developments demonstrate that the latest class of models are able to integrate structured and unstructured data, assist in hypothesis generation, and streamline research workflows. To explore the frontier of LLM capabilities across the research lifecycle, we review applications of LLMs through 32 total projects developed during the second annual LLM hackathon for applications in materials science and chemistry, a global hybrid event. These projects spanned seven key research areas: (1) molecular and material property prediction, (2) molecular and material design, (3) automation and novel interfaces, (4) scientific communication and education, (5) research data management and automation, (6) hypothesis generation and evaluation, and (7) knowledge extraction and reasoning from the scientific literature. Collectively, these applications illustrate how LLMs serve as versatile predictive models, platforms for rapid prototyping of domain-specific tools, and much more. In particular, improvements in both open source and proprietary LLM performance through the addition of reasoning, additional training data, and new techniques have expanded effectiveness, particularly in low-data environments and interdisciplinary research. As LLMs continue to improve, their integration into scientific workflows presents both new opportunities and new challenges, requiring ongoing exploration, continued refinement, and further research to address reliability, interpretability, and reproducibility.