Green steel from red mud through climate-neutral hydrogen plasma reduction

The study addresses whether highly alkaline red mud, a large-scale hazardous by-product of alumina refining, can be directly converted into metallic iron in a simple, carbon-neutral process. Red mud storage poses severe environmental and economic challenges, with limited recycling and costly neutralization. Previous approaches to extract iron often rely on carbon-based reduction or require energy- and CO2-intensive pre- and post-treatments (roasting, milling, pelletizing, magnetic separation). The authors propose using a lean hydrogen thermal plasma in an electric arc furnace to directly reduce untreated red mud into liquid iron and simultaneously neutralize the residual oxides’ alkalinity. The aim is to create a sustainable nexus between aluminium and steel production by valorizing red mud as a feedstock for green iron, thereby mitigating CO2 emissions and reducing waste hazards.

Prior work explored carbothermic smelting and replacement of carbon with hydrogen (molecular or plasma) to reduce iron oxides in red mud, including hydrogen-plasma smelting during bauxite refining to suppress red mud formation. However, reported routes typically require substantial preprocessing (roasting, milling, pelletizing, wet magnetic separation) and still entail CO2 emissions from intermediate steps. Other electric arc furnace or pyrometallurgical schemes remain carbon-based, shifting rather than solving environmental impacts. The present work builds on hydrogen plasma reduction literature for iron ores and addresses gaps by eliminating pre/post-treatments, using a single-step, hydrogen-plasma process targeted specifically at untreated red mud.

• Feedstock: 15 g batches of untreated red mud of characterized composition and crystallography (Extended Data Figs. 1–2; Tables 1–2).
• Reactor/process: Electric arc furnace; lean hydrogen-containing thermal plasma (Ar–10% H2) ignited at 200 A. Reduction times spanned 1–15 minutes. In some high-energy tests, current was increased to 800 A (Supplementary information).
• Operation: Red mud powder was directly exposed to the hydrogen plasma, producing a molten oxide phase in which metallic iron nucleates and coalesces into liquid nodules that separate from the more viscous/lower-density oxide melt. Samples were cooled to solidify metal nodules within a glassy oxide matrix.
• Separation and quantification: Metallic and oxide portions were mechanically separated after solidification. Masses of each fraction were measured. Phase quantification and tracking of reduction kinetics, oxygen removal, and iron partitioning were performed by X-ray diffraction (XRD), with composition normalized by specimen mass to account for volatilization effects. Metallization and theoretical yield were compared to thermodynamic calculations (Supplementary information: Thermodynamic calculations; Theoretical limits of Fe extraction).
• Microstructure/chemistry: Cross-sections examined by scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDX) for samples reduced 1 min and 10 min, probing iron droplet formation, oxide phase evolution (titanomagnetite, hercynite, Ti- and Ca-rich oxides), and chemical partitioning between top and bottom regions.
• pH measurement: Residual oxide pH tracked over reduction time to assess neutralization (initial pH ~10.5; time-resolved pH after 5 and 15 min).
• Control experiment: 15 g red mud processed for 5 min in pure Ar plasma (non-reducing) in the same apparatus to decouple thermal/plasma effects from hydrogen chemistry. XRD and mass loss compared to the hydrogen plasma case.
• Data interpretation: Phase/chemistry trends analyzed to infer reaction pathways (direct reduction from titanomagnetite to Fe with limited wüstite formation), oxygen transfer among oxides, and water formation via hydrogen-based reduction.

• Single-step iron extraction: Direct exposure of untreated red mud to an Ar–10% H2 plasma (200 A) yields liquid iron that solidifies as macroscopic nodules within a glassy oxide matrix, enabling clean metal/oxide separation.
• Yield and metallization: After 10 min, ~2.6 g Fe obtained from 15 g red mud (sample mass 8.8 g; ~30 wt% metallic Fe in the reduced specimen). This corresponds to 62.4% metallization relative to the initial ~4.17 g Fe in the feed, rising to ~70% when accounting for loss of ignition/evaporation. Thermodynamic prediction was 2.67 g Fe, indicating near-equilibrium extraction.
• Purity and form: The extracted metal is mostly in nodules representing, on average, 98 wt% of the obtained metal fraction, with high purity suitable for steelmaking. Discussion reports average Fe content of ~95 wt% with negligible S, P, and C.
• Kinetics and phase pathway: Rapid formation of micron-sized Fe nodules occurs within 1 min, largely via direct reduction from titanomagnetite to Fe with limited wüstite formation. With time, titanomagnetite reduces and hercynite (Al2FeO4) forms as a key Fe carrier in the oxide portion before further reduction to Fe. Ti-enriched zones develop, and at higher energy exposure (e.g., 800 A, 15 min), the oxide becomes simpler and enriched in thermally stable Ti and Ca oxides.
• Oxygen and iron in oxides: Early in reduction, oxygen in oxides remains ~35 wt% while Fe in oxides initially ~30 wt%, then decreases as metallic Fe forms. Initial 1 min shows ~3 g mass loss from haematite decomposition (Fe2O3 → Fe3O4), clay/water evaporation, and minor Fe formation (~2 wt%).
• pH neutralization: Residual oxides become less alkaline: pH decreases from ~10.5 (as-received red mud) to ~9 after 5 min, and to ~7.5 after 15 min, potentially enabling direct use in construction without costly neutralization.
• Control (Ar plasma): 5 min in pure Ar yields only ~0.27 g Fe (~7% of available Fe) versus ~1.34 g Fe (~33%) with H2 plasma for the same time and mass. Ar-treated samples show higher O content in oxides, retention of clay/water, and lower mass loss (30.1%) compared to hydrogen plasma (40.9%). This demonstrates the dominant role of hydrogen species in rapid reduction and oxygen removal.
• Evaporation and losses: Fe evaporation loss after 10 min is ~7%. Hydrogen plasma exposure selectively removes oxygen (forming H2O) and drives oxygen transfer to more oxygen-affine cations (e.g., Si, Ti, Ca), aiding Fe precipitation.
• Comparative kinetics: Red mud reduces about 20% faster in metallization than a pure haematite sample of similar Fe content under similar hydrogen plasma conditions, attributed to catalytic and viscosity/fluid-dynamics effects of gangue oxides.
• Co-valorization potential: Micrometre-scale Ti-enriched domains indicate potential for Ti recovery. Prior Fe removal may facilitate downstream extraction of rare-earth elements (e.g., Sc, Y) from the oxide residue using established techniques.

The results show that hydrogen plasma enables rapid, selective reduction of red mud’s iron oxides to metallic Fe in a single step, answering the central question of whether untreated red mud can be directly valorized as a green iron feedstock. The process exploits both chemical selectivity (oxygen transfer and hydrogen-driven reduction to H2O) and physical separation (density/viscosity contrasts) to achieve clean partitioning of metal and oxides. Mechanistically, Fe forms directly from titanomagnetite with minimal wüstite, aided by microalloying effects (e.g., Ti, Si) that stabilize magnetite-like structures and facilitate Fe nucleation, thereby accelerating kinetics relative to pure haematite. The high metallization (~70% in 10 min), near-thermodynamic yields, and high Fe purity with negligible S, P, C make the product directly usable in steelmaking. Simultaneously, the residual oxide’s near-neutral pH and glassy, thermally stable composition make it suitable for construction applications, reducing environmental liabilities and neutralization costs. Compared with inert Ar plasma, hydrogen species strongly enhance oxygen removal and Fe formation, underscoring the pivotal role of hydrogen chemistry beyond mere thermal effects. The observation of Ti-rich and complex oxide domains suggests opportunities to co-extract other valuable metals, and prior Fe removal may improve the efficiency of rare-earth-element recovery. Economically, rapid processing, potential scalability (exothermic reactions, improved efficiency at larger scale), and reduced CO2 emissions relative to carbon-based routes could make the method competitive for Fe-rich red mud, establishing a sustainable link between aluminium and steel value chains.

This work demonstrates the first single-step, carbon-free hydrogen plasma process that converts untreated red mud directly into high-purity metallic iron while neutralizing the oxide residue. The method achieves rapid kinetics, near-equilibrium Fe yields, and clean phase separation without pre- or post-treatments, offering a pathway to green steel feedstock and sustainable red mud management. The residual oxides are near-neutral and potentially usable directly in construction materials. The approach also shows promise for co-valorization of other elements (e.g., Ti) and may enhance subsequent extraction of rare-earth elements. Future research should focus on upscaling and reactor engineering to harness exothermicity and mass/heat transfer advantages, optimizing process parameters to further increase Fe recovery from complex Ti/Ca oxides, quantifying energy efficiency at industrial scales, and integrating downstream recovery of Ti and rare-earth elements to maximize overall resource efficiency and economic viability.

• Scale and energy input: Experiments were performed on small 15 g batches; the study notes high energy input for the small processed mass, although upscaling is expected to improve efficiency.
• Composition variability: Red mud composition varies geographically and affects reduction behavior and yields; generalizability across all red mud sources requires further validation.
• Partial extraction: After 10 min, metallization is ~70% with some Fe retained in complex oxides (e.g., Ti- and Ca-rich phases); additional optimization or longer processing may be needed to approach complete recovery.
• Decoupling of effects: While an Ar plasma control isolates thermal/plasma effects, complete decoupling of thermal, evaporation, and hydrogen chemistry in the molten state remains challenging.
• Evaporation/volatilization: Significant mass loss and some Fe evaporation (~7% at 10 min) occur; managing volatilization at scale will be important.
• Economic analysis: Technoeconomic conclusions rely on assumptions and require confirmation under industrial-scale operating conditions and diverse feedstocks.