The ability of a matter to fall into a glassy state upon cooling differs greatly among metallic alloys. It is conventionally measured by the critical cooling rate R_c, below which crystallization inevitably happens. There are a lot of factors involved in determining R_c for an alloy, including both elemental features and alloy properties. However, the underlying physical mechanism is still far from being well understood. Therefore, the design of new metallic glasses is mainly by time- and labor-consuming trial-and-error experiments. This considerably slows down the development process of metallic glasses. Nowadays, large-scale computer simulations have been playing a significant role in understanding glass formation. Although the atomic-scale features can be well captured, the simulations themselves are constrained to a limited timescale. To overcome these issues, we propose to explore the glass-forming ability of the modeled alloys from computer simulations by supervised machine learning. We aim to gain insights into the key features determining R_c and found that the non-linear couplings of the geometrical and energetic factors are of great importance. An optimized machine learning model is then established to predict new glass formers with a timescale beyond the current simulation capability. This study will shed new light on both unveiling the glass formation mechanism and guiding new alloy design in practice.

The escalating adoption of high-throughput methods in applied materials science dramatically increases the amount of generated data and allows for the deployment and use of sophisticated data-driven methods. To exploit the full potential of these accelerated approaches, the generated data need to be managed, preserved and shared. The heterogeneity of such data calls for highly flexible models to represent the data from fabrication workflows, measurements and simulations. We propose the use of a native graph database to store the data instead of relying on rigid relational data models. To develop a flexible and extendable data model, we create an ontology that serves as the blueprint of the data model. The Python framework Django is used to enable seamless integration into the virtual materials intelligence platform VIMI. The Django framework relies on the Object Graph Mapper neomodel to create a mapping between database classes and Python objects. The model can store the whole bandwidth of the data from fabrication to simulation data. Implementing the database into a platform will encourage researchers to share data while profiting from rich and highly curated data to accelerate their research.

Aiming at the comprehensive understanding of the single sulfur poisoning effect and, eventually, the multiple impurities poisoning phenomena on the SOFC (Solid Oxide Fuel Cell) cathode materials, the sulfur poisoning effect on the (La_0.6Sr_0.4)_0.95Co_0.2Fe_0.8O₃ (LSCF-6428) has been investigated in the presence of 10 ppm SO₂ at 800, 900, and 1,000 °C, respectively, with a combined computational and experimental approach. The good agreement between the CALPHAD (Computer Coupling of Phase Diagrams and Thermochemistry) simulations and the XRD (X-Ray Diffraction), SEM (Scanning Electron Microscopy), and TEM (Transmission Electron Microscopy) characterization results support the reliability of the CALPHAD approach in the SOFC field. Furthermore, comprehensive simulations were made to understand the impact of temperature, P(SO₂), P(O₂), and Sr concentration on the threshold of SrSO₄ stability. Results showed that the formation of SrSO₄ is thermodynamically favored at lower temperatures, higher P(SO₂), higher P(O₂), and higher Sr concentration. Finally, comparisons were also made between LSCF-6428 and LSM20 (La_0.8Sr_0.2MnO₃) using simulations, which confirmed that LSCF-6428 is a poor sulfur-tolerant cathode, in agreement with the literature.

High-entropy alloys (HEAs) are an emerging class of alloys with multi-principal elements that greatly expands the compositional space for advanced alloy design. Besides chemistry, processing history can also affect the phase and microstructure formation in HEAs. The number of possible alloy compositions and processing paths gives rise to enormous material design space, which makes it challenging to explore by traditional trial-and-error approaches. This review highlights the progress in combinatorial high-throughput studies towards rapid prediction, manufacturing, and characterization of promising HEA compositions. This review begins with an introduction to HEAs and their unique properties. Then, this review describes high-throughput computational methods such as machine learning that can predict desired alloy compositions from hundreds or even thousands of candidates. The next section presents advances in combinatorial synthesis of material libraries by additive manufacturing for efficient development of high-performance HEAs at bulk scale. The final section discusses the high-throughput characterization techniques used to accelerate the material property measurements for systematic understanding of the composition-processing-structure-property relationships in combinatorial HEA libraries.

The usage of low-grade ferrous scrap has increased over decades to decrease CO₂ emissions and to produce steel products at a low cost. A serious problem in melting post-consumer scrap material is the accumulation of tramp elements, e.g., Cu and Sn, in the liquid steel. These tramp elements are difficult to remove during conventional steelmaking processes. Sn is considered as one of the most harmful tramp elements because, together with Cu, it sometimes induces the liquid metal embrittlement in high-temperature ferrous processing, e.g., continuous casting and hot rolling. Furthermore, the chemical interaction between Fe and Sn plays an important role in the Sn smelting process. The raw material used in the Sn smelting process is SnO₂ (cassiterite), in which Fe₃O₄ is a gangue in the Sn ore. In the process, the reduction of Fe₃O₄ is unavoidable, which results in forming a Fe-Sn alloy (hardhead). The recirculation of the hardhead decreases the furnace capacity and increases the energy consumption in the smelting. The need to efficiently recover Sn from secondary resources is therefore inevitable. The CALculation of PHAse Diagrams (CALPHAD) approach helps to predict the equilibrium state of the multicomponent system. Previously reported studies of the Fe-Sn system show inconsistencies in the calculations and the experimental results. Mainly the miscibility gap in the liquid phase was under debate, as experimental data of the phase boundary are scattered. Experimental study and re-optimization of model parameters were carried out with emphasis on the correct shape of the miscibility gap. Three different experimental techniques were employed: differential scanning calorimetry, electromagnetic levitation, and contact angle measurement. The present thermodynamic model has higher accuracy in predicting the solubility of Sn in the body-centered cubic (bcc), compared to previous assessments. This is achieved by re-evaluating the Gibbs energies of the FeSn and FeSn₂ compounds and the peritectic reaction related to Fe₅Sn₃. Also, the inconsistencies related to the miscibility gap around X_Sn = 0.31-0.81 were resolved. The database developed in the present study can contribute to the development of a large CALPHAD database containing tramp elements.

Al-Si-Mg series alloys are the most common alloys available for additive manufacturing forming with low cracking tendency. However, there is no systematic study on the computational design of SLMed Al-Si-(Mg) alloys due to the huge parameter space of composition and processes. In this paper, a high-quality dataset of SLMed Al-Si-(Mg) alloys containing 176 pieces of data from 50 publications was first established, which recorded the information, including alloy compositions, process parameters, test conditions, and mechanical properties. A threshold value of 35 J/mm³ for energy density (Ed) was then proposed as a criterion to clean the data points with lower ultimate tensile strength (UTS) and elongation (EL). The cleaned dataset consists of a first training/testing dataset with 142 data for model construction and a second testing dataset with 9 data for model verification. After that, four machine learning models were applied to establish the quantitative relation of “composition-processes-properties” in SLMed Al-Si-(Mg) alloys. The MLPReg model was chosen as the optimal one considering its best performance and subsequently utilized to design novel compositions and process parameters for SLMed Al-Si-(Mg) alloys. The UTS and EL of the designed alloy with a maximum comprehensive mechanical property are 549 MPa and 16%, both of which are higher than all the available experimental data. It is anticipated that the present design strategy based on the machine learning method should generally be applicable to other SLMed alloy systems.

A physics-based model is used to predict the melt pool properties in the laser-directed energy deposition of several nickel-based superalloys for different process parameters. The input space is high-dimensional, consisting of a common 19-dimensional composition space for each alloy and the process parameters (laser power and scan velocity). Gaussian Process-based regression frameworks are developed by training surrogates on data generated by a validated analytical model. These surrogates are thereafter used to predict and define relationships between the composition, resultant thermophysical properties, process parameters, and the subsequent melt pool property. The probabilistic predictions are augmented by uncertainty quantification and sensitivity analysis to substantiate the findings further.