Machine learning-assisted exploration of novel hybrid organic-inorganic perovskites for naval applications
Abstract
Approved for Public Release. Hybrid organic-inorganic perovskites (HOIPs) have garnered substantial attention in the field of materials science due to their exceptional dielectric, electronic, and piezoelectric properties. These materials, which typically consistof organic cations and inorganic metal halides, exhibit a unique combination of traits that make them particularly promising for naval applications. HOIPs display notable piezoelectric properties like oxide perovskites. These materials can convert mechanical stress into an electric charge and vice versa, making them suitable for sensors, actuators, and acoustic transduction materials. However, these properties of HOIPs are influenced by multiple factors, including crystal structure, chemical composition, choice of organiccations, the presence of defects or impurities, and the processing conditions during material fabrication. Thus, it is challenging to forecast how changes in composition would impact the properties of HOIPs because of the complicated and non-linear interactions between their material properties and composition. However, understanding the relationships between the material properties and theirunderlying molecular structure is crucial for designing and optimizing these hybrid perovskites for naval applications. As a result, designing and developing novel HOIPs materials with specific properties requires time-consuming, resource-intensive, and arduous experimental methods. The proposed project aims to streamline the design and development of novel HOIPs for naval-related applications using artificial intelligence and machine learning (ML) models. The project will predict properties and identify quantitative structure-property relationships by developing hybrid ML models, thus reducing the time and resources required for materials discovery of HOIPs. This transformative endeavor includes developing a HOIP database that is comprehensive, designing hybrid ML and computational models, and generative design of novel HOIPs with tailored properties, ultimately advancing research in the acoustic transductionmaterials/devices and enabling the rapid development of advanced materials for challenging naval applications. The project specifically studies the structure-property relationships of the HOIPs using validated theoretical approaches, high-throughput computationalmodeling, and advanced AI/ML techniques. The four critical pillars of this research are 1) data collection, pre-processing, and representations, 2) high-throughput modeling and screening of HOIPs, 3) designing hybrid ML models and identifying key features and quantitative structure-property relationships, 4) generative design of novel HOPIs with tailored properties. This project will yield accelerated materials discovery, improved performance in harsh environments, new HOIP compositions that perform over a larger temperature-strain-field range, etc. Developing high-performance materials with the desired properties for specific applications, such as high-efficiency electronic devices, will advance existing Navy SONAR and make innovative systems feasible. The outcome can contribute to the development of advanced materials for underwater sonar technologies, benefiting various sectors and creating new efficient solutions.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 15, 2024
- Source ID
- N000142412220
Entities
People
- Ying Li
Organizations
- Office of Naval Research
- United States Navy
- University of Wisconsin System