Abstract One of the main challenges in materials discovery is efficiently exploring the vast search space for targeted properties as approaches that rely on trial-and-error are impractical. We review how methods from the information sciences enable us to accelerate the search and discovery of new materials. In particular, active learning allows us to effectively navigate the search space iteratively to identify promising candidates for guiding experiments and computations. The approach relies on the use of uncertainties and making predictions from a surrogate model together with a utility function that prioritizes the decision making process on unexplored data. We discuss several utility functions and demonstrate their use in materials science applications, impacting both experimental and computational research. We summarize by indicating generalizations to multiple properties and multifidelity data, and identify challenges, future directions and opportunities in the emerging field of materials informatics.
The new descriptors obtain greater accuracy compared to commonly used empirical descriptors in the phase prediction of high entropy alloys, and the generalization ability of these new descriptors has been verified by experiments.
Design of shape memory alloys with large phase transformation strain and low hysteresis is in demand for practical applications that require high output work and high precision. However, this remains challenging due to the competition between these two properties. In this work, we report a method that combines machine learning with multi-objective optimization to assist the rapid design of shape memory alloys. Instead of directly using the predictions from machine learning to guide experiments, this work employs the uncertainty-aware two-objective optimization algorithm to recommend the potential candidates. Such a strategy is beneficial to the case where limited data is available just as the dataset of twenty NiTi-based alloys with hysteresis and phase transformation strain established herein. Key features are screened out from a relatively large feature pool and Gaussian regression models are built for predicting the two properties of unknown alloys. At the end, eight alloys with promise to improve both recoverable strain and hysteresis are recommended, as compared to the alloys in the initial dataset.
Abstract Lattice engineering and distortion have been considered one kind of effective strategies discovering advanced materials. The instinct chemical flexibility of high-entropy pyrochlore oxides (HEPOs) motivate/accelerate to tailor the target properties through phase transformations and lattice distortion. Here, a hybrid knowledge-assisted data-driven machine learning (ML) strategy is utilized to discover the HEPOs with low thermal conductivity (𝜅) through 17 rare-earth (RE = Sc, Y, La ~ Lu) solutes optimized A-site of A 2 B 2 O 7 . A designing routine integrating the ML and high throughput first-principles has been proposed to predict the key physical parameter (KPPs) correlated to the targeted 𝜅 of advanced HEPOs. Among the smart-designed 6188 (5RE 0.2 ) 2 Zr 2 O 7 HEPOs, the best candidates are addressed and validated by the principles of severe lattice distortion and local phase transformation, which effectively reduce 𝜅 by the strong multi-phonon scatting and weak interatomic interactions. Particularly, (Sc 0.2 Y 0.2 La 0.2 Ce 0.2 Pr 0.2 ) 2 Zr 2 O 7 with predicted κ below 1.59 Wm −1 K − 1 is selected to be verified, which match well with the experimental κ = 1.69 Wm −1 K − 1 at 300K and could be further decreased to 0.14 Wm −1 K − 1 at 1473K. Moreover, the coupling effects of lattice vibrations and charges on heat transfer are revealed by the cross-validations of various models, indicating that the weak bonds with low electronegativity and few bonding charge density and the lattice distortion (r*) identified by cation radius ratio (r A /r B ) should be the KPPs to decrease κ efficiently. This work supports an intelligent designing strategy with limited atomic and electronic KPPs to accelerate the development of advanced multi-component HEPOs with properties/performance at multi-scales.
High frequency ultrasound has been widely used for various applications in ophthalmology, dermatology, small animal studies and intravascular (IVUS) imaging as diagnosis and imaging tool. The research and development of high frequency transducer has become one of the most interesting areas in ultrasound technologies. This article briefly reviews the existing technologies for high frequency ultrasound transducer development, then, presents the state-of-the-art piezo composite microfabricated ultrasound transducer (PC-MUT) for the first time. Using this technology, PMN-PT single crystal 1-3 composite has been developed, of which the kerf width is as small as to ~ 4 �m. High frequency (> 40MHz) transducer with advanced performance has been developed for the IVUS application. The bandwidth and sensitivity are almost doubled compared with the conventional ceramic transducer. The features of PC-MUT technology may lead to great imaging quality and more potential medical applications.
The use of oxidation kinetics model can potentially predict oxidation weight gain with time and help understand underlying oxidation mechanism. However, more emphasis is paid on the latter and less on the former as model parameters required for predictions are only available for limited materials. Here, we use machine learning to learn the parameters in oxidation kinetics model from composition-based features to overcome the above challenge. The predicted oxidation weight gain vs. time curves by machine learning augmented oxidation kinetics model agree well with the measurements for alloys outside the training data.
One key challenge in materials informatics is how to effectively use the material data of small size to search for desired materials from a huge unexplored material space. We review the recent progress on the use of tools from data science and domain knowledge to mitigate the issues arising from limited materials data. The enhancement of data quality and amount via data augmentation and feature engineering is first summarized and discussed. Then the strategies that use ensemble model and transfer learning for improved machine learning model are overviewed. Next, we move to the active learning with emphasis on the uncertainty quantification and evaluation. Subsequently, the merits of the combination of domain knowledge and machine learning are stressed. Finally, we discuss some applications of large language models in the field of materials science. We summarize this review by posing the challenges and opportunities in the field of machine learning for small material data.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
