Understanding microstructural evolution occupies a central position in the discipline of materials science and engineering. As stated by Carter et al., microstructural evolution involves complex, coupled, and often nonlinear processes even the description of the dynamics for isolated microstructural evolution processes can be quite complicated. It would be desirable to enrich the microstructural evolution theory by introducing a powerful mathematical tool, which could enable describing and predicting the rich intertwining phenomena such as diffusive or displacive phase transformation, grain growth, generation, or annihilation of defects (vacancy, dislocations, etc.) in a straightforward manner. There have been continuing efforts along this front, and I will restrict myself to the issues in the development and application of the thermodynamic variational principle. Although being reviewed by various authors recently, we hope to redraw attentions to some valuable papers and provide our understanding and viewpoints. It is our opinion that the most appealing feature about the principle is the nature that it could give approximate solutions with tunable accuracy. The other feature is its role as a basic principle in deriving the new models. It is hoped that this paper could promote the development and application of the variational principle even further in materials science.

In recent years, interdisciplinary research has become increasingly popular within the scientific community. The fields of materials science and chemistry have also gradually begun to apply the machine learning technology developed by scientists from computer science. Graph neural networks (GNNs) are new machine learning models with powerful feature extraction, relationship inference, and compositional generalization capabilities. These advantages drive researchers to design computational models to accelerate material property prediction and new materials design, dramatically reducing the cost of traditional experimental methods. This review focuses on the principles and applications of the GNNs. The basic concepts and advantages of the GNNs are first introduced and compared to the traditional machine learning and neural networks. Then, the principles and highlights of seven classic GNN models, namely crystal graph convolutional neural networks, iCGCNN, Orbital Graph Convolutional Neural Network, MatErials Graph Network, Global Attention mechanism with Graph Neural Network, Atomistic Line Graph Neural Network, and BonDNet are discussed. Their connections and differences are also summarized. Finally, insights and prospects are provided for the rapid development of GNNs in materials science at the atomic scale.

This article proposed a design strategy that integrated machine learning models based on random forest and genetic algorithm (GA) for the rapid screening of hardness in the AlCoCrCuFeMoNiTi high-entropy alloys system. Through feature engineering and modeling, valence electron concentration, atomic size difference (δr), Pauling electronegativity difference (Δ χ), geometric parameters (Λ), and the Cr content were identified as the five key features in the database. The GAwas employed to search for alloys with superior hardness and guided synthesis. After three iterations, the HEA Al₁₈Co₂₁Cr₂₃Fe₂₃Mo₁₅ exhibiting the highest predicted hardness (868.8 HV) was identified. The alloy was predominantly composed of BCC, ordered B2, and σ phases, with an experimental hardness of 899.8 ± 9.9 HV, which as approximately 5.38% greater than the maximum hardness observed in the original dataset. The design strategy can also solve other regression problems and pave the way for optimizing material performance in various engineering applications.

Finding high temperature superconductors (HTS) has been a continuing challenge due to the difficulty in predicting the transition temperature (T_c) of superconductors. Recently, the efficiency of predicting T_c has been greatly improved via machine learning (ML). Unfortunately, prevailing ML models have not shown adequate generalization ability to find new HTS, yet. In this work, a graph neural network model is trained to predict the maximal T_c (T_c^max) of various materials. Our model reveals a close connection between T_c^max and chemical bonds. It suggests that shorter bond lengths are favored by high T_c, which is in coherence with previous domain knowledge. More importantly, it also indicates that chemical bonds consisting of some specific chemical elements are responsible for high T_c, which is new even to the human experts. It can provide a convenient guidance to the materials scientists in search of HTS.

The corrosion susceptibility of magnesium (Mg) alloys presents a significant challenge for their broad application. Although there have been extensive experimental and theoretical investigations, the corrosion mechanisms of Mg alloys are still unclear, especially the anodic dissolution process. Here, a thorough theoretical investigation based on ab initio molecular dynamics and metadynamics simulations has been conducted to clarify the underlying corrosion mechanism of Mg anode and propose effective strategies for enhancing corrosion resistance. Through comprehensive analyses of interfacial structures and equilibrium potentials for Mg(0001)/H₂O interface models with different water thicknesses, the Mg(0001)/72 H₂O model is identified to be reasonable with −2.17 V vs. standard hydrogen electrode equilibrium potential. In addition, utilizing metadynamics, the free energy barrier for Mg dissolution is calculated to be 0.835 eV, enabling the theoretical determination of anodic polarization curves for pure Mg that aligns well with experimental data. Based on the Mg(0001)/72 H₂O model, we further explore the effects of various alloying elements on anodic corrosion resistance, among which Al and Mn alloying elements are found to enhance corrosion resistance of Mg. This study provides valuable atomic-scale insights into the corrosion mechanism of magnesium alloys, offering theoretical guidance for developing novel corrosion-resistant Mg alloys.

In this study, a small dataset of 370 datapoints of Mg alloys are selected for machine learning (ML), in which each datapoint includes five rare-earth-free alloying elements (Ca, Zn, Al, Mn and Sn), three extrusion parameters (extrusion speed, temperature and ratio), and three mechanical properties (yield strength [YS], ultimate tensile strength [UTS] and elongation [EL]). The ML algorithms, including support vector machine regression (SVR), artificial neural network, and other three methods, are employed, and the SVR has the best performance in predicting mechanical properties based on the components, and process parameters, with the mean absolute percentage error of YS, UTS, and EL being 6.34%, 4.19%, and 13.64% in the test set, respectively. The SVR model combined with multi-objective genetic algorithm are successfully used to optimize mechanical properties of four extruded alloys from Mg-Ca, Mg-Ca-Zn, Mg- Ca-Mn-Sn, and Mg-Ca-Al-Zn-Mn series alloys, respectively, and the corresponding experimental results are in good agreement with the designed ones. Furthermore, new composition schemes are proposed from a wider range of elements and processing features to match the objectives of high-strength, strength-ductility balanced, and high-ductility Mg alloys, and the four-, five- and six-element alloying schemes are provided for the candidates of new-generation wrought Mg alloys.

For a long time, the phase-field method has been considered a mesoscale phenomenological method that lacks physical accuracy and is unable to be closely linked to the mechanical or functional properties of materials. Some misunderstandings existing in these viewpoints need to be clarified. Therefore, it is necessary to propose or adopt the perspective of “unified phase-field modeling (UPFM)” to address these issues, which means that phase-field modeling has multiple unified characteristics. Specifically, the phase-field method is the perfect unity of thermodynamics and kinetics, the unity of multi-scale models from microto meso and then to macro, the unity of internal or/and external driving energy with order parameters as field variables, the unity of multiple physical fields, and thus the unity of material composition design, process optimization, microstructure control, and performance prediction. It is precisely because the phase-field approach has these unified characteristics that, after more than 40 years of development, it has been increasingly widely applied in materials science and engineering.

In this work, eight Mn-RE (RE = Ce, Pr, Sm, Tb, Er, Tm, Lu, and Y) binary systems were reassessed thermodynamically by the CALPHAD method based on the reported optimizations and experimental information. Self-consistent thermodynamic parameters to describe Gibbs energies of various phases in eight Mn-RE binary systems were obtained. The calculated phase equilibria and thermodynamic properties of eight Mn-RE binary systems are in good accordance with the experimental results. Furthermore, phase equilibria and thermodynamic properties of 13 Mn-RE (RE = La, Ce, Pr, Nd, Sm, Gd, Tb, Dy, Ho, Er, Tm, Lu, and Y) binary systems were discussed systematically in combination with the present calculations and the reported optimizations. A trend was found for the variation of phase equilibria and thermodynamic properties of the Mn-RE binary systems. In general, as the RE atomic number increases, the enthalpies of mixing of liquid alloys as well as the enthalpies of formation of the intermetallic compounds become increasingly negative, and the formation temperatures of the intermetallic compounds become higher. The results provide a complete set of self-consistent thermodynamic parameters for the Mn-RE binary systems, and a thermodynamic database of 13 Mn-RE binary systems was finally achieved.

The 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.

Accurately and quickly predicting hydrogen embrittlement performance is critical for the service of metal materials. However, due to multi-source heterogeneity, existing hydrogen embrittlement data are missing, making it impractical to train reliable machine learning models. In this study, we proposed an ensemble learning training strategy for missing data based on the Adaboost algorithm. This method introduced a mask matrix with missing data and enabled each round of training to generate sub-datasets, considering missing value information. The strategy first trained a subset of features based on the existing dataset and a selected method and continuously focused on the combination of features with the highest error for iterative training, where the mask matrix of the missing data was used as the input to fit the weights of each base learner using a neural network. Compared with directly modeling on highly sparse data, the predictive ability of this strategy was significantly improved by approximately 20%. In addition, in the testing of new samples, the predicted mean absolute error of the new model was successfully reduced from 0.2 to 0.09. This strategy offers good adaptability to the hydrogen embrittlement sensitivity of different sizes and can avoid interference from feature importance caused by filling data.

Aluminum-lithium (Al-Li) alloy is one of the most promising lightweight structural materials in the aeronautic and aerospace industries. The key to achieving their excellent mechanical properties lies in tailoring T₁ strengthening precipitates; however, the nucleation of such nanoparticles remains unknown. Combining atomic resolution HAADF-STEM with first-principles calculations based on the density functional theory (DFT), here, we report a counterintuitive nucleation mechanism of the T₁ that evolves from an Eshelby inclusion with unstable stacking faults. This precursor is accelerated by Ag-Mg clusters to reduce the barrier, forming the structural framework. In addition, these Ag-Mg clusters trap the free Cu and Li to prepare the chemical compositions for T₁. Our findings provide a new perspective on the phase transformations of complex precipitates through solute clusters in terms of geometric structure and chemical bonding functions.

Additive friction stir deposition (AFSD) provides strong flexibility and better performance in component design, which is controlled by the process parameters. It is an essential and difficult task to tune those parameters. The recent exploration of machine learning (ML) exhibits great potential to obtain a suitable balance between productivity and set parameters. In this study, ML techniques, including support vector machine (SVM), random forest (RF) and artificial neural network (ANN), are applied to predict the mechanical properties of the AFSD-based AA6061 deposition. Expect for the stable parameters (temperature, force and torque) in situ monitored by the self-developed process-aware kit during the AFSD process and the other factors (rotation speed, traverse speed, feed rate and layer thickness) are also set as input variables. The output variables are microhardness and ultimate tensile strength (UTS). Prediction results show that the ANN model performs the best prediction accuracy with the highest R² (0.9998) and the lowest mean absolute error (MAE, 0.0050) and root mean square error (RMSE, 0.0063). Furthermore, analysis suggests that the feed rate (24.8%/24.1%) and layer thickness (25.6%/26.6%) indicate a higher contribution that affects the mechanical properties.

Generative adversarial networks (GANs), as a powerful tool for inverse materials discovery, are being increasingly applied in various fields of materials science. This review provides systematic investigations on the applications of GANs from a group of different aspects. The basic principles of GANs are first introduced; then a detailed review of GANs-based studies regarding distinct scenarios across composition design, processing optimization, crystal structure search, microstructure characterization and defect detection is presented. At the end, several challenges and possible solutions are discussed and outlined. This overview highlights the efficacy of GANs in materials science, and may stimulate the further use of GANs for more intriguing achievements.

Enhancing the interpretability of machine learning methods for predicting material properties is a key, yet complex topic in materials science. This study proposes an interpretable convolutional neural network (CNN) to establish the relationship between the microstructural evolution and mechanical properties of non-uniform and nonlinear multisystem dual-phase steel materials and achieve an inverse analysis of the elastic-plastic mechanism. This study demonstrates that the developed CNN model achieves an accuracy of 94% in predicting the stress-strain curves of dual-phase steel microstructures with different compositions and processes, with the mean absolute error not exceeding 50 MPa, representing merely 5.26% of the average tensile strength of dual-phase steels in the dataset. The reverse visualization results of the CNN model indicate that, during tensile deformation, the grain boundaries maintain deformation coordination within the grains by impeding dislocation slip. This results in a significant stress concentration at the grain boundaries, with stresses at the boundaries being higher than those borne by the martensitic phase and minimal stresses in the ferrite phase. Moreover, compared with traditional crystal plasticity models, the CNN model exhibits a substantial improvement in computational efficiency. This method provides a generic plan for improving the interpretability of machine learning methods for predicting material properties and can be easily applied to other alloy systems.

The emerging photovoltaic (PV) technologies, such as organic and perovskite PVs, have the characteristics of complex compositions and processing, resulting in a large multidimensional parameter space for the development and optimization of the technologies. Traditional manual methods are time-consuming and laborintensive in screening and optimizing material properties. Materials genome engineering (MGE) advances an innovative approach that combines efficient experimentation, big database and artificial intelligence (AI) algorithms to accelerate materials research and development. High-throughput (HT) research platforms perform multidimensional experimental tasks rapidly, providing a large amount of reliable and consistent data for the creation of materials databases. Therefore, the development of novel experimental methods combining HT and AI can accelerate materials design and application, which is beneficial for establishing material-processing-property relationships and overcoming bottlenecks in the development of emerging PV technologies. This review introduces the key technologies involved in MGE and overviews the accelerating role of MGE in the field of organic and perovskite PVs.