Deep learning (DL) is the artificial intelligence (AI)-based approach which enables the extraction of complex imaging patterns from the process data and allows for a more accurate prediction in process metallurgy. Compared with the conventional machine learning (ML), this approach enhances the process control efficiency and improves prediction quality. The current study focused on classification of non-metallic inclusions (i.e. Al2O3 and MnS) in stainless steel as a case study. Several semantic segmentation models were applied to predict the morphology and elemental distribution of MnS and Al2O3 inclusions using images characterized by scanning electron microscope (SEM). The obtained prediction results were compared with inclusion morphology and composition data obtained from energy dispersive spectroscopy (EDS). Finally, OpenCV was used to predict the particle size of MnS and Al2O3 inclusion. The prediction results show that the Kernel-based Network (K-Net) model can achieve the highest accuracy for classifying oxide and sulfide inclusions, with a mean intersection over union (mIoU) of 0.8615 and average accuracy (aAcc) of 0.9889 for three-dimensional (3D) images, and an mIoU of 0.8863 and aAcc of 0.9955 for two-dimensional (2D) images. The OpenCV also demonstrated high accuracy in predicting the particle size of inclusions compared with experimental data. The successful implementation of DL approach in non-metallic inclusion classification in steels paves the way for developing intelligence assisted process metallurgy route.

To satisfy the high performance requirements of new generation materials in application of aerospace and automotive industries, a series of TiZrHf-based entropic (i.e., entropy-stabilized) alloys with α+β dual-phase microstructure were designed using the CALPHAD (CALculation of PHAse Diagram) methodology. The present work aimed to achieve an balanced performance between mechanical properties, corrosion resistance, and high-temperature oxidation stability by tailoring Zr/Hf ratios. The alloys were comprehensively characterized using X-ray diffraction (XRD), electron channeling contrast imaging (ECCI), and high angle annular dark field scanning transmission electron microscope (HAADF-STEM). Due to the effect of transformation-induced plasticity (TRIP), the homogenized and cryogenic alloys exhibit balanced mechanical properties (i.e., high strength and ductility). Electrochemical tests in 3.5 wt% NaCl solution demonstrated good corrosion resistance, and the stability of the passive film was slightly compromised by both cryogenic treatment and Zr/Hf additions. Moderate high-temperature oxidation tests at 500 and 600 °C showed that the alloys have good oxidation resistance result from the formation of protective scales dominated by TiO2 and Al2O3. However, the formation of less-protective Zr/Hf-oxides at higher temperatures (700 °C) was found to be detrimental. This work provide a CALPHAD-guided design strategy for developing Ti-based entropic alloys with a well-balanced properties for applying in different severe environments.

This study systematically investigates the effect of trace hafnium (Hf) addition (0.069 wt%) on the microstructure evolution and inclusion characteristics in V-alloyed Co-based dual-phase high-entropy alloys. The trace amount of Hf addition was found to mainly form HfOx inclusion in the matrix. This phase replaces the conventional oxide formation and could stabilize the V element in the matrix and against its oxidation. Compare with the Hf-rich HEAs, where Hf segregates at grain boundaries within an FCC-dominant matrix and forms lamellar HCP precipitates. However, Co46Cr30Mn7.5Fe7.5Ni7.5V1.5 with a trace amount of Hf enhances the blocky HCP phase formation and increases the HCP fraction from 15 % to 26 %. Non-metallic inclusion analysis here reveals that V-HEA contains HfOx alongside MnS and HfOx-MnS. In situ observations by high-temperature confocal laser scanning microscopy demonstrate that inclusion motion in the liquid matrix is governed by liquid surface flow and fluctuations. When inter-inclusion distances approach a critical value, attractive capillary force triggers the increase in velocity and acceleration and leads to a rapid collision of inclusions. Smaller inclusions exhibit a weaker attraction and are obviously influenced by the surface flow of liquid. While the larger inclusions are more prone to coarsening through agglomeration. This work provides theoretical and experimental insights into tailoring inclusion characteristics and microstructure evolution in HEAs through trace amounts of critical element addition. 

The agglomeration of inclusions in continuous casting blooms destroys the continuity and compactness of the steel matrix, which seriously restricts the fatigue life and corrosion resistance of the steel. A novel traceability method for the distribution of inclusions was introduced to reveal the evolution of inclusion agglomeration. This method demonstrated the position evolution of inclusions when they passed through different planes by assigning colors for inclusions. In this study, a mathematical model coupled with electromagnetic field, flow, heat transfer, solidification, and non-metallic inclusion movement was developed to study the agglomeration behavior of inclusions under dual-mode electromagnetic field control modes (edge-to-center flow mode and the coupled mode of center-to-edge flow and edge-to-center flow). Numerical simulation revealed that the coupled mode significantly enhanced inclusion distribution uniformity in the solidified shell, with a 63.7 % reduction of the number in the localized agglomeration zone compared to the edge-to-center flow mode. Experimental measurements demonstrated a 46.7 % decrease in inclusion number density near the quarter position of the loose side under coupled mode compared to edge-to-center flow mode. Under coupled mode, the flow of molten steel at the center and the edge of the mold with the opposite directions helped to disperse the inclusions and promote the uniform distribution of inclusions in the cross-section. This study provides a new strategy to suppress inclusion agglomeration in continuous casting blooms by electromagnetic metallurgy technology.

A novel methodology to control the size and chemistry of TiN particle in an inclusion-engineered steel is manufactured using spark plasma sintering (SPS) in combination of post heat treatment. The chemical composition of TiN particles keeps being stable after sintering with metallic Fe–C–Mn particles, and a very slight deviation of TiN size could be observed. In addition, an attempt to further classify the particle size in different narrow ranges has been presented here for future work. This feasibility study paves the way for a further development of the fabrication methodology for the next generation inclusion-engineered steels with a homogenous distribution of desired phase of particles.

Complex phase (CP) steels are widely used in automotive components such as frame rails, rocker panels, and tunnel stiffeners owing to their high strength and good local formability. The subtle hardness difference between microstructures allows CP steels to exhibit excellent hole expansion performance, with the high-hardness martensite-austenite (MA) constituents being the critical structure. The distribution of MA constituents is crucial to the mechanical properties of the product. This study aims to improve the hole expansion property by constructing a continuous distribution of MA constituents along the rolling direction at the thickness center. Microstructures and hole expansion behavior were investigated using CLSM, SEM, EBSD, and hole expansion tests. Results indicate that after thermodynamic treatment, the MA constituents were aggregated at the thickness center in a continuous distribution along the rolling direction with a long axis of approximately 1.25 μm, and an average distance of less than 1.0 μm. Microhardness quantification of the plastic damage on the punching edge suggests that the advanced steel exhibits the highest hardening at the thickness center with a 41% hardness increase after punching, which is higher than the 31% hardening in the maximum hardening burr zone of the base steel. The advanced steel, despite suffering severe punching damage, exhibited a hole expansion ratio of approximately 43%, higher than the 34% of the base steel. Quasi in situ interrupted hole expansion tests indicate that at the thickness center of the advanced steel, the circumferential cracks formed through a multiple void interaction mechanism which promotes the stress release. In the matrix, pit-like damage is caused by a void coalescence mechanism. Both mechanisms lead to the mechanical instability and eventual failure of the steel. The damaging position of the hole edge had a decisive impact on the fracture mode.

摘 要 复相钢兼具高强度和良好的局部可成形性，被广泛应用于汽车车架导轨、摇臂板和隧道加强件等典型汽车零部件。复相钢中各类微观结构之间较小的硬度差异使其具有优异的扩孔性能，其中高硬度马氏体-奥氏体(MA)组元是影响复相钢扩孔性能的关键组织，其分布对扩孔性能的影响至关重要。本工作提出了构造厚度中心沿轧向连续分布MA组元以提升复相钢扩孔率的方法，利用CLSM、SEM、EBSD手段和扩孔实验，研究了构造MA组元特征分布前后复相钢的微观结构和扩孔行为特性。结果表明，基准钢的MA组元均匀分布，长轴为0.98 μm，平均中心间距为1.2 μm。构造特征组织后的实验用钢MA组元聚集在厚度中心，长轴约1.25 μm，沿轧向连续分布，平均间距小于1.0 μm。微观硬度量化冲裁边的塑性损伤结果表明，冲孔损伤后实验钢板厚度中心处硬化最高，较损伤前硬化41%，高于基准钢最大硬化区(毛刺区，31%)。冲孔损伤更高的实验用钢的扩孔率约43%，高于基准钢(约34%)。利用准原位中断扩孔实验分析了扩孔行为与显微组织特征的关系。实验用钢通过多孔隙相互作用机制在厚度中心处形成环状裂纹促使应力释放，同时在基体中通过单一孔隙机制形成坑状损伤，导致材料局部失稳并最终失效。受损质点处于孔缘位置对断裂方式具有一定程度的影响。关键词 复相钢，马氏体-奥氏体(MA)组元，扩孔率，断裂

Cobalt-based entropic alloys doped with Ti/Zr/Hf have been investigated in the present work. Thermodynamic calculations have been conducted to predict the phase evolution. The effect of two types of processes (homogenization and cryogenic treatment) on microstructures and properties have been comprehensively analyzed. The compositions and microstructures of the designed alloys in different states have been investigated using multiple techniques. Electrochemical corrosion behaviors at room temperature, high-temperature oxidation behaviors at 600 °C, 800 °C, and 1000 °C, as well as the hardness and compression tests, have been systematically performed. The Ti-doped cobalt-based entropic alloy demonstrated excellent overall properties, including strong electrochemical corrosion resistance, high-temperature oxidation resistance, and a combination of high strength and ductility. The phase map from electron backscatter diffraction (EBSD) indicated that Ti has weaker stability for the formation of the C14-Laves phase compared to the alloying effects of Zr and Hf. The characterization results align with the thermodynamic calculations. This work paves a way for establishing material design strategies to develop advanced alloys with superior performance.

Rare earth can modify inclusions in non-oriented silicon steel which is harmful to magnetic properties. This study focused on the 3.1% Si non-oriented silicon steel under industrial production conditions. Samples were taken during the stages before and after addition of rare earth ferrosilicon alloy in Ruhrstahl-Heraeus (RH) unit, different pouring time in tundish, and continuous casting slab. This study systematically examined the morphology, composition, and size distribution of inclusions throughout the smelting process of non-oriented silicon steel by scanning electron microscopy with energy-dispersive spectroscopy (SEM/EDS), and thermodynamic analysis at liquid steel temperature and thermodynamic analysis of equilibrium solidification. The research results demonstrated that the rare earth treatment ultimately modifies the original Al2O3 inclusions in the non-oriented silicon steel into REAlO3 and RE2O2S inclusions, while also aggregating AlN inclusions to form composite inclusions. After rare earth modification, the average size of the inclusions decreases. In the RH treatment process, the inclusions before the addition of rare earth ferrosilicon alloy are mainly AlN and Al2O3. After the addition of rare earth ferrosilicon alloy, the inclusions are mainly RES and REAlO3. In the tundish and continuous casting, the rare earth content decreased, and the rare earth inclusions transform into RE2O2S and REAlO3. For the size of inclusions, after adding rare earth ferrosilicon alloy, the average size of inclusions rapidly decreased from 16.15 μm to 2.65 μm and reach its minimum size 2.16 μm at the end of RH treatment. When the molten steel entered the tundish, the average size of inclusions increased slightly and gradually decreased with the progress of pouring. The average size of inclusions in the slab is 5.79 μm. Phase stability diagram calculation indicates the most stable rare earth inclusion is Ce2O2S in molten steel. Thermodynamic calculations indicated that Al2O3, Ce2O2S, Ce2S3, AlN, and MnS precipitate sequentially during the equilibrium solidification process of molten steel.

High-temperature confocal scanning laser microscopy (HT-CSLM) is a potent methodology for investigating various phenomena in the field of metallurgy. Initially applied to the observation of solid phase transformations and solidification, this method has gained traction in the field of non-metallic inclusion in steels in recent years. An overview of the experimental capabilities of HT-CSLM and the most important results of recent investigations regarding the topics of clean steel production are provided. It includes the formation of intragranular acicular ferrite (IAF) from the surface of non-metallic inclusions during the continuous cooling and heat treatment, which can be especially beneficial in the toughness of heat-affected zones of welded pieces. Furthermore, the investigation of agglomeration mechanisms of non-metallic inclusions (NMIs) in liquid steel is discussed to improve the insight into attraction forces between particles and clogging phenomena during continuous casting. Also, the dissolution of NMIs in various steelmaking slags can be observed by HT-CSLM to compare dissolution rates and mechanisms of NMI, where significant influences of temperature and chemical composition of the slag were shown. Last but not least, the experimental work regarding the interface between steel and slag is discussed, where novel techniques are currently being developed. A comprehensive summary of experimental techniques using HT-CSLM equipment to investigate different interactions of NMIs with steel and slag phases is compiled. 

Unwanted ice accumulation can lead to catastrophic disasters or economic losses. Photothermal superhydrophobic surfaces show promise for anti-/de-icing applications, but their effectiveness depends critically on precise micro-nano hierarchical structure design and functionalization. Current approaches face significant limitations: lithography enables ordered patterns but becomes cost-prohibitive for nanoscale features, while disordered micro-nano structures suffer from poor performance tunability and inconsistency. This study develops a high-performance structured micro/nano-crystal array photothermal superhydrophobic metamaterial (SMNA-PSM) for anti-/de-icing. The structured crystal array features abundant micro-nano surfaces, transforming deposited Metal-insulator-Metal (MIM) structures into heterogeneous resonators. These heterogeneous resonators with varying sizes, angles, and thicknesses possess more electromagnetic wave response sites and scattering surfaces, converting the separated absorption peaks of the uniform MIM structure into a continuous absorption band, achieving 96% solar spectrum absorptivity. Moreover, by simply adjusting the deposition material, the surface morphology of the crystal array can be tuned from smooth to rough, thereby enabling a switch from hydrophobicity to superhydrophobicity. Unlike conventional micro-nano hierarchical structures, structured micro-nano crystal arrays can be integrated with film stacked architectures, inheriting film-based advantages: tunable performance, uniformity, substrate-friendliness, and scalability. This approach demonstrates broad application potential in micro-nano structure fabrication, broadband wave absorption, wettability control, photothermal conversion and anti-/de-icing.