I. The Physical Nature of Metallic Inclusions and the Evolution of Classification Systems
Metallic inclusions in steel, as "microscopic markers" of the metallurgical process, not only reflect the complete history of the smelting process but also become "invisible killers" restricting the application of high-end steels. In the nearly century-long development of metallurgy, the understanding of inclusions has undergone a cognitive shift from "harmful and must be removed" to "controllable and optimizable for use." Research in modern clean steel technology shows that completely eliminating inclusions is neither economical nor practical; the scientific goal is to control them within safe size and favorable morphology ranges.
According to the modern classification system based on formation mechanisms, metallic inclusions have developed into a four-dimensional system encompassing "endogenous-exogenous-interface reaction-secondary precipitation." Exogenous metallic fragments, as the most typical macroscopic defects, have a formation process filled with process variables. When high-melting-point alloy additives (such as ferrotungsten, ferromolybdenum) are added to molten steel, a eutectic molten film of Fe-W or Fe-Mo forms on the surface of the block. The thickness of this film determines the melting rate. Studies show that when the alloy block size exceeds a critical dimension (Dc = 30mm), the heat transfer rate of the surface molten film is lower than the internal thermal conduction rate, creating a "cold core" phenomenon with a temperature gradient exceeding 200°C/cm. This unmelted core retains its original crystal structure during subsequent solidification, with a lattice constant mismatch of 7-12% compared to the matrix, forming a natural source of stress concentration.
Welding process inclusions are a microscopic-scale recurrence of the metallurgical process. In the TIG welding process, when the welding current density falls below a critical value (120A corresponding to a current density of 85 A/mm²), the molten droplet formed at the tungsten electrode tip is constrained by the balance between surface tension and gravity. Computational fluid dynamics simulations show that droplets smaller than 1.5mm in diameter exhibit unstable oscillatory trajectories in the argon shielding gas flow field. Some droplets deviate from the main flow direction into the weld pool boundary layer and are captured by the rapidly solidifying weld metal. These captured tungsten particles have unique micro-features: a surface oxide layer about 50-200nm thick and the presence of metastable β-W phase inside due to rapid cooling, with a hardness up to 1.3 times that of conventional α-W phase.
Cast-specific structures, as products of the solidification process, have more complex formation mechanisms. The formation of "cold shuts" involves the coupling of oxidation kinetics and fluid dynamics. During pouring, the oxide film formed on the steel surface (mainly FeO) ruptures and becomes entrapped due to turbulent flow. Experimental data shows that when the pouring speed exceeds 0.8 m/s, the probability of oxide film fragmentation increases threefold. These oxide fragments undergo complex reduction-dissolution processes within the molten steel. The incompletely reduced parts form oxygen-rich cores, surrounded by composition gradient zones, where the carbon content variation gradient from the core outward can reach 0.5% per 100µm.
II. The Modern Evolution of Inclusion Detection Technology
The limitations of traditional metallographic testing are becoming increasingly evident in the field of advanced materials. Modern detection technology is developing towards "multi-scale, multi-modal, and in-situ dynamic" directions. A major breakthrough in ultrasonic testing technology is the application of phased array technology. Through probe arrays with 64-128 elements, detection resolution can leap from millimeter to sub-millimeter level. Latest research indicates that combining focused probes with synthetic aperture technology improves the detection rate for 100µm-level inclusions from the traditional 30% to 85%, while enabling three-dimensional spatial localization.
Electron microscopy analysis technology has undergone revolutionary changes. Field emission scanning electron microscopy combined with energy dispersive spectroscopy (EDS) mapping can complete elemental distribution analysis over several square millimeters within minutes. The more advanced electron backscatter diffraction (EBSD) technique can reveal the crystallographic orientation relationship between inclusions and the matrix, which is crucial for understanding crack propagation paths. Experiments have found that when specific orientation relationships (such as cube-cube orientation) exist at the inclusion-matrix interface, the interfacial energy decreases by 35%, and the difficulty of crack initiation increases accordingly.
Breakthroughs in atomic-scale characterization technology provide new perspectives for understanding the nature of inclusions. Atom probe tomography (APT) can reconstruct three-dimensional elemental distribution with atomic resolution. Recent APT analysis of the interface between TiN inclusions and the matrix revealed a 2-3nm thick transition zone at the interface. Within this zone, Ti and N concentrations show gradient changes, accompanied by segregation of elements like C and Si. This microstructure explains why certain interfaces exhibit exceptional resistance to crack propagation.
The development of online monitoring technology is changing the traditional post-facto inspection mode. A continuous casting billet surface inspection system based on laser-induced breakdown spectroscopy (LIBS) can analyze surface composition in real-time at a speed of 100 points per second. A line-scan CCD surface inspection system installed during hot rolling uses machine learning algorithms to identify surface anomalies caused by inclusions, with an identification accuracy rate exceeding 95%. These real-time data provide a valuable time window for process adjustments, enabling a shift from "passive detection" to "active control."
III. Physicochemical Principles of Inclusion Control
The core of inclusion control lies in understanding their behavior in molten steel. While Stokes' law describes the floating behavior of ideal spherical particles, the behavior of inclusions in actual molten steel is far more complex. Firstly, the drag coefficient for non-spherical particles is 1.5-3 times that of spherical ones, resulting in a correspondingly lower floating speed. Secondly, velocity gradients caused by molten steel convection generate the Magnus effect, causing lateral displacement of rotating particles. Computational fluid dynamics simulations show that in a tundish, the actual trajectory of a 50µm diameter Al₂O₃ inclusion is 40-60% longer than the ideal path.
The physical basis of electromagnetic purification technology lies in the difference in electrical conductivity between inclusions and molten steel. When an alternating magnetic field (frequency 50-1000 Hz) is applied to molten steel, induced currents are generated differently in the steel and the inclusions. Theoretical calculations show that for oxide inclusions with conductivity less than 1% of molten steel, the differential electromagnetic force can be 10-100 times the gravitational force. A steel mill applying a rotating magnetic field with a frequency of 200 Hz and magnetic flux density of 0.1 T improved the removal rate of 20-50µm inclusions by 40%. It also found a significant fragmentation effect on clustered Al₂O₃, reducing the average cluster size from 150µm to 80µm.
Optimization of deoxidation processes involves a balance between thermodynamics and kinetics. Al₂O₃ generated by traditional aluminum deoxidation is solid and prone to forming clusters. Calcium treatment can transform Al₂O₃ into low-melting-point (<1500°C) calcium aluminates. Experimental data indicates that when the Ca/Al mass ratio reaches 0.12-0.15, the proportion of liquid inclusions exceeds 80%. The more advanced magnesium-calcium composite treatment technology, by forming MgO·Al₂O₃ spinel phase, reduces its contact angle in molten steel by 15° compared to Al₂O₃, making it easier to coalesce and float.
Controlling reoxidation is the core challenge of modern clean steel technology. Contact between molten steel and air for just 0.1 seconds can increase oxygen content by 5-10 ppm. Using a sealing system with a long nozzle and submerged entry nozzle, combined with an Ar gas curtain, can limit reoxidation to within 1 ppm. Recent developments in intelligent control technology involve real-time monitoring of molten steel oxygen activity and temperature to dynamically adjust shielding gas flow. This has reduced argon consumption per ton of steel by 30% while cutting reoxidation products by 50%.
