The evaluation and prediction of fatigue properties are crucial for metallic materials. Although the determination of S-N curves represents the most important methods for evaluating such properties, its fatigue testing is costly and time-consuming. Furthermore, fatigue testing involves different test conditions, thereby complicating the evaluation of the fatigue properties. This study develops a transfer convolutional neural network (TR-CNN) framework, in which the prediction of the reversed torsion S-N curves of steels is transferred from rotating bending S-N curves. In the TR-CNN framework, the source CNN models for rotating-bending curve prediction are first trained based on the composition and process conditions. Subsequently, based on the source models, the reversed torsion S-N curves are estimated by training the TR-CNN models based on only a small dataset. After proving the rationality of the framework, its universality with respect to different amounts of data is further investigated. The reversed torsion curves under small-sample conditions (22 samples) are predicted accurately by the TR-CNN. Additionally, the TR-CNN models remain accurate under varying amounts of data (22-112 samples), showing excellent generality for different amounts of fatigue data. The predictive capability of the TR-CNN models is improved by introducing tensile properties into the source models. The proposed TR-CNN framework can significantly reduce the cost of evaluating fatigue properties, and the prediction of S-N curves can be optimized by combining the transfer framework and low-cost properties related to fatigue.
To improve the precision of machine-learning predictions, we investigate various techniques that combine multiple quality sources for the same property. In particular, focusing on the electronic band gap, we aim at having the lowest error by taking advantage of all available experimental measurements and density-functional theory calculations. We show that learning about the difference between high- and low-quality values, considered a correction, significantly improves the results compared to learning on the sole high-quality experimental data. As a preliminary step, we also introduce an extension of the MODNet model, which consists of using a genetic algorithm for hyperparameter optimization. Thanks to this, MODNet is shown to achieve excellent performance on the Matbench test suite.
Thermodynamic databases are essential prerequisites for developing advanced materials, such as Ni-based superalloys. The present work collects a large amount of experimental and first-principles calculation data concerning the thermodynamics and phase diagrams of the Ni-Mo-Re system, based on which the thermodynamic properties of the ternary and its binary sub-systems Ni-Mo and Mo-Re are assessed by means of the CALculation of PHAse Diagrams (CALPHAD) approach. The thermodynamic database containing all model parameters is established and most experimental data are reproduced satisfactorily. The present work demonstrates the use of the CALPHAD method as a practical appliance in the toolbox of materials informatics to analyze and discriminate various types of data by thermodynamic modeling and then produce more useful data in wider ranges of compositions and temperatures by computational predictions.
Additive manufacturing (AM) is a disruptive technology with a unique capability in fabricating parts with complex geometry and fixing broken supply chains. However, many AM techniques are complicated with their processing features due to complex heating and cooling cycles with the melting of feedstock materials. Therefore, it is quite challenging to directly apply the materials design and processing optimization method used for conventional manufacturing to AM techniques. In this viewpoint paper, we discuss some of the ongoing efforts of high-throughput (HT) experimentation, which can be used for materials development and processing design. Particularly, we focus on the beam- and powder-based AM techniques since these methods have demonstrated success in HT experimentation. In addition, we propose new opportunities to apply AM techniques as the materials informatic tools contributing to materials genome.
To achieve specific applications, it is always desirable to design new materials with peculiar topological properties. Herein, based on a D2h B2Cu6H6 molecule with the unique chemical bonding of planar pentacoordinate boron (ppB) as a building block, we constructed an infinite CuB monolayer by linking B2Cu6 subunits in an orthorhombic lattice. The planarity of the CuB sheet is attributed to the multicenter bonds and electron donation-back donation, as revealed by chemical bonding analysis. As a global minimum confirmed by the particle swarm optimization method, the CuB monolayer is expected to be highly stable, as indicated by its rather high cohesive energy, absence of soft phonon modes, and good resistance to high temperature, and thus is highly feasible for experimental realization. Remarkably, this CuB monolayer is metallic and predicted to be superconducting with an estimated critical temperature (Tc) of 4.6 K, and the critical temperature could be further enhanced by tensile strains (to 21 K at atmospheric pressure).
Machine learning has promoted the rapid development of materials science. In this review, we provide an overview of recent advances in machine learning for inorganic phosphors. We take two aspects of material properties prediction and optimization based on iterative experiments as entry points to outline the applications of machine learning for inorganic phosphors in terms of Debye temperature prediction and luminescence intensity and thermal stability optimization. By analyzing the machine learning methods and their application objectives, current problems are summarized and suggestions for subsequent development are proposed.
As promising next-generation candidates for applications in aero-engines, L12-strengthened cobalt (Co)-based superalloys have attracted extensive attention. However, the L12 strengthening phase in first-generation Co-Al-W-based superalloys is metastable, and both its solvus temperature and mechanical properties still need improvement. Therefore, it is necessary to discover new L12-strengthened Co-based superalloy systems with a stable L12 phase by exploring the effect of alloying elements on their stability. Traditional first-principles calculations are capable of providing the crystal structure and mechanical properties of the L12 phase doped by transition metals but suffer from low efficiency and relatively high computational costs. The present study combines machine learning (ML) with first-principles calculations to accelerate crystal structure and mechanical property predictions, with the latter providing both the training and validation datasets. Three ML models are established and trained to predict the occupancy of alloying elements in the supercell and the stability and mechanical properties of the L12 phase. The ML predictions are evaluated using first-principles calculations and the accompanying data are used to further refine the ML models. Our ML-accelerated first-principles calculation approach offers more efficient predictions of the crystal structure and mechanical properties for Co-V-Ta- and Co-Al-V-based systems than the traditional counterpart. This approach is applicable to expediting crystal structure and mechanical property calculations and thus the design and discovery of other advanced materials beyond Co-based superalloys.