Materials Informatics Research Articles

Architected materials informatics: Construction and application to cellular-structured heat sink optimization

Architected materials informatics: Construction and application to cellular-structured heat sink optimization

Predicting Reactivity and Passivation of Solid-State Battery Interfaces.

In this work, we build a computationally inexpensive, data-driven model that utilizes atomistic structure information to predict the reactivity of interfaces between any candidate solid-state electrolyte material and a Li metal anode. This model is trained on data from ab initio molecular dynamics (AIMD) simulations of the time evolution of the solid electrolyte-Li metal interfaces for 67 different materials. Predicting the reactivity of solid-state interfaces with ab initio techniques remains an elusive challenge in materials discovery and informatics, and previous work on predicting interfacial compatibility of solid-state Li-ion electrolytes and Li metal anodes has focused mainly on thermodynamic convex hull calculations. Our framework involves training machine learning models on AIMD data, thereby capturing information on both kinetics and thermodynamics, and then leveraging these models to predict the reactivity of thousands of new candidates in the span of seconds, avoiding the need for additional weeks-long AIMD simulations. We identify over 300 new chemically stable and over 780 passivating solid electrolytes that are predicted to be thermodynamically unfavored. Our results indicate many potential solid-state electrolyte candidates have been incorrectly labeled unstable via purely thermodynamic approaches using density functional theory (DFT) energetics, and that the pool of promising, Li-stable solid-state electrolyte materials may be much larger than previously thought from screening efforts. To showcase the value of our approach, we highlight two borate materials that were identified by our model and confirmed by further AIMD calculations to likely be highly conductive and chemically stable with Li: LiB13C2 and LiB12PC.

Ionic species representations for materials informatics

High-dimensional representations of the elements have become common within the field of materials informatics to build useful, structure-agnostic models for the chemistry of materials. However, the characteristics of elements change when they adopt a given oxidation state, with distinct structural preferences and physical properties. We explore several methods for developing embedding vectors of elements decorated with oxidation states. Graphs generated from 110 160 crystals are used to train representations of 84 elements that form 336 species. Clustering these learned representations of ionic species in low-dimensional space reproduces expected chemical heuristics, particularly the separation of cations from anions. We show that these representations have enhanced expressive power for property prediction tasks involving inorganic compounds. We expect that ionic representations, necessary for the description of mixed valence and complex magnetic systems, will support more powerful machine learning models for materials.

Preface

The series “Hungarian Conference and Exhibition on Materials Science, Testing and Informatics” (later simply “Hungarian Conference on Materials Science”) was intended to provide a forum where Hungarian and foreign scientists and research groups interested in metals and alloys, silicates, polymers and composites get an opportunity to exchange and publish ideas and establish new and integrated partnerships. The 14th Hungarian Conference on Materials Science was held at the Balaton lakeshore in Balatonalmádi, October 8-10, 2023. The winner of the photo competition of the 14th Hungarian Conference on Materials Science was Kovács Árpád Győző, University of Miskolc The title of the picture: “Flower, my sweet flower”. The photo shows a fragment of a glass-fiber-reinforced polymer component (GF30), a degraded polymer matrix. The examination technique: ZEISS EVO MA 10 scanning electron microscope (HV mode, 20 kV, gold-coated surface, BSE image) The organisers present this volume as a peer-reviewed selection of papers submitted to publication by participants The organisers express their gratitude to all sponsors. Thanks are due to organising societies, as the Hungarian Materials Science Society and to other Hungarian scientific and engineering societies for their support. List of Scientific Committee, Organising Committee, Conference Secretariat, Diamond Sponsors of the Conference and Sponsor of the Conference are available in this Pdf.

Research Methods using Materials Informatics and Molecular Dynamics for the Development of Road Pavement Materials (I)

Research Methods using Materials Informatics and Molecular Dynamics for the Development of Road Pavement Materials (I)

Putting error bars on density functional theory

Predicting the error in density functional theory (DFT) calculations due to the choice of exchange–correlation (XC) functional is crucial to the success of DFT, but currently, there are limited options to estimate this a priori. This is particularly important for high-throughput screening of new materials. In this work, the structure and elastic properties of binary and ternary oxides are computed using four XC functionals: LDA, PBE-GGA, PBEsol, and vdW-DF with C09 exchange. To analyze the systemic errors inherent to each XC functional, we employed materials informatics methods to predict the expected errors. The predicted errors were also used to better the DFT-predicted lattice parameters. Our results emphasize the link between the computed errors and the electron density and hybridization errors of a functional. In essence, these results provide “error bars” for choosing a functional for the creation of high-accuracy, high-throughput datasets as well as avenues for the development of XC functionals with enhanced performance, thereby enabling the accelerated discovery and design of new materials.

Borides as promising M2AX phase materials with high elastic modulus using machine learning and optimization

There is growing interest in novel MAX phase materials for various applications ranging from aircraft/spacecraft and defense to energy and electronics due to their unique combination of metallic and ceramic properties. Traditional materials discovery has mostly relied on human intuition coupled with rigorous experiments; however, this approach has been time-consuming and inefficient. Over the last few decades, advances in fundamental and data-driven approaches such as first-principles modeling, materials informatics, machine learning and optimization, coupled with an exponential rise in computational power, have enabled faster and more efficient materials discovery. Here, we present an exploration of high elastic modulus novel boride-based M2AX phase materials using a combination of the aforementioned methods. Specifically, an ensemble of gradient boosted machine learning models was developed to predict the elastic modulus from informatics-based structural features by leveraging a dataset of Density Functional Theory (DFT)-predicted elastic moduli for 223 M2AX phase materials (carbides and nitrides). Using Bayesian optimization, inverse modeling was carried out to maximize the model-predicted elastic modulus by identifying the optimal features. Finally, model predictions for 1,035 candidate M2AX materials were generated to compare their features with the optimal features to identify potential novel promising materials. We found that Ta2PB, Nb2PB, and V2PB have similar high elastic moduli (371.7, 351.5, and 347.4 GPa) to their carbide counterparts (364.7, 357.7, and 373.5 GPa), and our results support the possibility that borides can be a viable tertiary element for M2AX phases.

Nano-particles size measurement based on semantic segmentation via convolution neural network

Automatic recognition and size measurement of nanoparticles in SEM images have been an evolving field of study within the material informatics domain. These methods have been typically assessed using traditional image processing techniques. This paper presents a robust and fully automated pipeline for all researchers, particularly those in chemical engineering, based on image processing techniques and convolution neural networks. We evaluated our method on a dataset of 1,701 2D particle images, employing a three-fold split for training, validation, and testing. The proposed approach integrates semantic segmentation models (U-Net, LinkNet) with size measurement algorithms, achieving high accuracy even in the presence of noise or blurriness. Our results demonstrate the method effectiveness in precisely estimating particle sizes, with a Pearson correlation coefficient of 0.9633 and mean absolute error of 1.287e-06. The method’s robustness against poor-quality images and its fully automated procedures make it a valuable tool for researches in SEM images field.

Rational Design of Deep Learning Networks Based on a Fusion Strategy for Improved Material Property Predictions.

The success of machine learning in predicting material properties is largely dependent on the design of the model. However, the current designs of deep learning models in materials science have the following prominent problems. First, the model design lacks a rational guidance strategy and heavily relies on a large amount of trial and error. Second, numerous deep learning models are utilized across various fields, each with its own advantages and disadvantages. Therefore, it is important to incorporate a fusion strategy to fully leverage them and further expand the design strategies of the models. To address these problems, we analyze that the main reason is the lack of a new feedback method rich in physical insights. In this study, we developed a feedback method called the Chemical Environment Clustering Vector (CECV) of compounds at different thresholds, which is rich in physical insights. Based on CECV, we rationally designed the Long Short-Term Memory and Gated Recurrent Unit fused with Deep Convolutional Neural Network (L-G-DCNN) to explore the field of structure-agnostic material property predictions. L-G-DCNN accurately captures the interactions between elements in compounds, enabling more accurate and efficient predictions of the material properties. Our results demonstrate that the performance of the L-G-DCNN surpasses the current state-of-the-art structure-agnostic models across 28 benchmark data sets, exhibiting superior sample efficiency and faster convergence speed. By employing different visualization methods, we demonstrate that the fusion strategy based on CECV significantly enhances the comprehension of the L-G-DCNN model design and provides a fresh perspective for researchers in the field of materials informatics.

Accelerated Exploration of Li-Rich Cathode Materials with High Energy Density for Li-Ion Batteries

Li-rich cathode materials have been widely researched as promising cathode materials for Li-ion batteries due to the abundant numbers of Li+ in their formula units. There are, however, still some problems, such as the fading discharge voltage during cycling, to be overcome before its commercial application can be realized. In recent years, there has been a global expansion of efforts to innovate processes in material development by utilizing data science and materials informatics (hereafter referred to as MI), moving away from the traditional trial-and-error approach based on experiments. These initiatives have been surpassing the limitations of human-designed patterns and the number of feasible experiments, and examples of such endeavors are starting to be reported even in the field of battery materials[1]. In this study, we challenged a MI technology combined with first-principles calculation and database technology to widely search for suitable compositions of Li-rich cathode materials at high speed.Li1.4(Mn0.667Ni0.333)1-x Mx O2.35F0.05 (hereafter referred to as LMNOF, M: a dopant element, 0 ≤ x ≤ 0.01) was fabricated by a conventional method. The cathode was fabricated by mixing powder of the active material, acetylene black as a conducting powder and PTFE as a binder at the ratio of 7:2:1 by weight. The electrochemical performance was investigated using 2032 coin type half cells. The models for predicting battery performance were constructed by machine learning techniques. Composition formula information, synthesis conditions, and cell weight were used as candidate explanatory variables. XenonPy[2] was used for elemental-derived information. Principal component analysis was performed on the candidate explanatory variables, and only the variables determined to be important were used to construct the predictive model. The number of candidate M was narrowed down from approximately 100 to 30 based on the values of the enthalpy change between LMNOF with and without M. The enthalpy was calculated to determine the dopant elements for LMNOF by first-principles calculation. Specific M elements will be reported at the time of the presentation. The oxygen desorption energies of LMNOF with the 30 extracted elements were calculated to evaluate the stability of the LMNOF structure against the reaction in the fully charged state. Ten elements with negative values of the energy were selected as the dopant ones (Ms) for experimental demonstration of this study and the rest were used to predict the performance by using the database of the Materials Project[3]. The battery performance of M-doped LMNOF were predicted by the constructed model based on machine learning techniques with experimental data. This model indicated that the performances of LMNOF with a certain element M1 or M2 tended to be higher than ones with another M dopant elements at the stage where there had been no experimental data for M1 and M2. When the experiment was conducted for M1 and M2, the obtained data was turned out to be close to the predicted values and higher than ones with another M. These experimental demonstrations revealed that the constructed model at the initial stage in this study was suitable to predict the reasonable values for the composition without adequate experimental data. The loop of accumulating the experimental data and constructing the prediction model have been continuously carried out to improve the accuracy of predictions. As the number of experimental trials increased, the performance was gradually improved and the composition including M1 alone, M2 alone, M1 and other multiple Ms together, or M2 and other multiple Ms together, showed relatively higher performance, which, we considered, surpassed the maximum values obtained by the conventional trial-and-error experimental method. As a result, our new MI technology was demonstrated as an effective method to search for an unknown and unexplored composition with high performance due to the prediction accuracy of machine learning. In future, we are going to further improve prediction accuracy to find multiple Ms for even higher performance.

Explainable Prediction of Hydrophilic/Hydrophobic Property of Polymer Brush Surfaces by Chemical Modeling and Machine Learning.

Polymer informatics has attracted increasing attention as a specialized branch of material informatics. Hydrophilicity/hydrophobicity is one of the most important properties of interfaces involved in antifouling, self-cleaning, antifogging, oil/water separation, protein adsorption, and bioseparation. Establishing a quantitative structure-property relationship for the hydrophilicity/hydrophobicity of polymeric interfaces could significantly benefit from machine learning modeling. In this study, we aimed to construct machine learning models that could predict the static water contact angle (CA) as an indicator of hydrophilicity/hydrophobicity based on a data set of polymer brushes. The features of the polymer brush surfaces were numerically described using their grafted structures (thickness) and molecular descriptors derived from their chemical structures. We achieved accurate prediction and understanding of important parameters by employing appropriate molecular descriptors considering the Pearson correlation and machine learning models trained with nested cross-validation. The model interpretation by Shapley additive extension analysis indicated that the amount of partial polar/nonpolar structure in the molecule as well as the averaged hydrophobicity represented by MolLogP plays an important role in determining the CA. Moreover, the model can predict the CAs of polymer brushes composed of chemical structures that are not present in existing databases. The CA values of the hypothetical polymer brushes are predicted.

Materials Informatics Approach to Cu/Nb Nanolaminate Microstructure Correlations with Yield Strength and Electrical Conductivity

Materials Informatics Approach to Cu/Nb Nanolaminate Microstructure Correlations with Yield Strength and Electrical Conductivity

Prediction of nanocomposite properties and process optimization using persistent homology and machine learning

Physical property prediction and synthesis process optimization are key targets in material informatics. In this study, we propose a machine learning approach that utilizes ridge regression to predict the oxygen permeability at fuel cell electrode surfaces and determine the optimal process temperature. These predictions are based on a persistence diagram derived from tomographic images captured using transmission electron microscopy (TEM). Through machine learning analysis of the complex structures present in the Pt/CeO2 nanocomposites, we discovered that l2 regularization considering diverse structural elements is more appropriate than l1 regularization (sparse modeling). Notably, our model successfully captured the activation energy of oxygen permeability, a phenomenon that could not be solely explained by the geometric feature of the Betti numbers, as demonstrated in a previous study. The correspondence between the ridge regression coefficient and persistence diagram revealed the formation process of the local and three-dimensional structures of CeO2 and their contributions to pre-exponential factor and activation energies. This analysis facilitated the determination of the annealing temperature required to achieve the optimal structure and accurately predict the physical properties.

Design of advanced steels by integrated computational materials engineering

AbstractThe integrated computational materials engineering (ICME) has achieved great success in accelerating the rational design and deployment of new materials. It is a new route of designing new materials and processes and highlighted by Materials Genome Initiative/Engineering that stresses the high‐throughput computation in addition to high‐throughput experimentation and materials informatics. This article presents a brief review on the basic theories and multi‐scale computational tools of ICME to design advanced steel grades, including the first‐principles calculations, the CALPHAD method (i.e., computational thermodynamics) fueled by dedicated databases, diffusion and phase‐field simulations, as well as finite analysis methods and machine learning. In the ICME scheme to deal with steels, the CALPHAD method is considered as the core to readily consider multi‐component systems and integrated to link the microscopic simulations (such as diffusion and phase field method to predict microstructure evolutions in response to external conditions) and macroscopic finite analysis method to deal with mechanical properties. Two applications are also presented to address the new routes to carry out materials design, especially for advanced steels.

Pathway to a fully data-driven geotechnics: Lessons from materials informatics

This paper elucidates the challenges and opportunities inherent in integrating data-driven methodologies into geotechnics, drawing inspiration from the success of materials informatics. Highlighting the intricacies of soil complexity, heterogeneity, and the lack of comprehensive data, the discussion underscores the pressing need for community-driven database initiatives and open science movements. By leveraging the transformative power of deep learning, particularly in feature extraction from high-dimensional data and the potential of transfer learning, we envision a paradigm shift towards a more collaborative and innovative geotechnics field. The paper concludes with a forward-looking stance, emphasizing the revolutionary potential brought about by advanced computational tools like large language models in reshaping geotechnics informatics.

Partnerships and collaboration drive innovative graduate training in materials informatics.

Holistic and intentional training prepares next-generation materials informatics leaders and workforce for expedited materials discovery and design.

Prediction of hydrogel swelling states using machine learning methods

AbstractIn the field of material informatics, artificial neural networks (ANNs) contribute to the investigation of the processing‐structure‐properties‐performance relationship of materials. This inspires us to leverage the capabilities of ANNs to decode properties of hydrogels, thereby customizing these active materials for sensors or actuators. In the current work, we introduce an approach to predict discrete swelling states of temperature‐responsive hydrogels, especially PNIPAAm, based on their synthesis parameters, utilizing ANN models. To build the database, we analyze literature on temperature‐responsive hydrogels and compile essential synthesis parameters. The corresponding data points related to these synthesis parameters are then extracted. We propose different variants of ANN models and compare their accuracy on the acquired dataset. The selected model can predict the swelling states of hydrogel samples within the test dataset with relative prediction error of 0.11. This approach is applied to predict the expected properties. Subsequently, the hydrogels can be synthesized, and their properties can be experimentally verified. Our approach can be extended to other types of hydrogels and in the prediction of additional properties. The identified synthesis parameters serve as a valuable foundation for the expansion of the database with further literature resources. An enriched database will enhance the performance of the data‐driven model, thereby improving its predictive capabilities.

The significance of materials informatics on material science

In 2011, U.S. President Barack Obama proposed the Material Genome Project. Want to high high-speed and low-cost methods to develop material science, which promoted the rapid development of materials informatics. Materials informatics is an interdisciplinary field that employs the principles of informatics to enhance the comprehension, utilization, and advancement of materials within the realm of materials science. Materials informatics can take multiple approaches and influence many aspects of new material development. Material informatics will cause great changes in the material industry and promote the rapid development of materials. A variety of materials informatics schemes have been developed to analyze materials, and there have been many successful cases worth applying. This article introduces the basic concepts and main research fields of materials informatics and describes four main steps in materials informatics: data collection, data processing, the establishment of a database, construction of a model. The main applications of materials informatics include simulation and prediction of material properties, development of new materials, material optimization and performance improvement. Thinking about the difficult problems in the field of materials informatics for application.

Multimodal learning of heat capacity based on transformers and crystallography pretraining

Thermal properties of materials are essential to many applications of thermal electronic devices. Density functional theory (DFT) has shown capability in obtaining an accurate calculation. However, the expensive computational cost limits the application of the DFT method for high-throughput screening of materials. Recently, machine learning models, especially graph neural networks (GNNs), have demonstrated high accuracy in many material properties’ prediction, such as bandgap and formation energy, but fail to accurately predict heat capacity(CV) due to the limitation in capturing crystallographic features. In our study, we have implemented the material informatics transformer (MatInFormer) framework, which has been pretrained on lattice reconstruction tasks. This approach has shown proficiency in capturing essential crystallographic features. By concatenating these features with human-designed descriptors, we achieved a mean absolute error of 4.893 and 4.505 J/(mol K) in our predictions. Our findings underscore the efficacy of the MatInFormer framework in leveraging crystallography, augmented with additional information processing capabilities.

Practical application of quantum neural network to materials informatics

Quantum neural network (QNN) models have received increasing attention owing to their strong expressibility and resistance to overfitting. It is particularly useful when the size of the training data is small, making it a good fit for materials informatics (MI) problems. However, there are only a few examples of the application of QNN to multivariate regression models, and little is known about how these models are constructed. This study aims to construct a QNN model to predict the melting points of metal oxides as an example of a multivariate regression task for the MI problem. Different architectures (encoding methods and entangler arrangements) are explored to create an effective QNN model. Shallow-depth ansatzs could achieve sufficient expressibility using sufficiently entangled circuits. The “linear” entangler was adequate for providing the necessary entanglement. The expressibility of the QNN model could be further improved by increasing the circuit width. The generalization performance could also be improved, outperforming the classical NN model. No overfitting was observed in the QNN models with a well-designed encoder. These findings suggest that QNN can be a useful tool for MI.