The Question We Asked

Most research on green steel focuses on temperature and hydrogen partial pressure. Our team at the Max Planck Institute für Eisenforschung asked a different question: what happens when you change the absolute gas pressure, not just the hydrogen concentration? This matters for two routes: hydrogen-based direct reduction (HyDR) in shaft furnaces, and hydrogen plasma smelting reduction (HyPSR), where an electric arc melts and reduces ore simultaneously.

The literature had scattered data but no systematic picture. We decided to build one—from thermodynamic first principles and controlled experiments.

Two Routes, Two Pressure Effects

For HyDR (solid-state reduction), Le Chatelier tells you that absolute pressure should not shift equilibrium when the number of gas moles is equal on both sides: Fe₂O₃ + 3H₂ → 2Fe + 3H₂O. That is correct but incomplete. What matters is partial pressure of H₂ relative to H₂O. In a countercurrent shaft furnace, water builds up, lowering p_H₂/(p_H₂+p_H₂O). Increasing total pressure while keeping gas composition fixed raises the absolute H₂ partial pressure. That shifts the operating point on the Baur-Glässner diagram from a thermodynamically stalled condition (point A) to a region where wustite reduction can proceed (point B) without raising temperature. Our reading of the literature—Sato, Kawasaki, Habermann—shows that pure H₂ under 5–36 bar accelerates kinetics considerably. The mechanism is simply more H₂ molecules per volume, higher collision probability, and a larger driving force against back-diffusing water vapor.

For HyPSR, the story is entirely different. Here we work with thermal plasmas (electrons and heavy particles at similar temperatures, 2000–20000 K). The reducing power comes not just from H₂ but from dissociated H atoms and even H⁺ ions. The dissociation reaction H₂ ⇌ 2H is strongly pressure-dependent. Lower total pressure shifts equilibrium toward atomic hydrogen at a given temperature. That is the central insight.

Thermodynamic Simulations: What We Calculated

We used ThermoCalc with the SSUB5 database (SGTE substances) to compute equilibrium molar fractions of H₂, H, H⁺, and electrons in Ar-H₂ mixtures from 500 K to 30000 K. Three pressures: 50 mbar, 450 mbar, 900 mbar. Three H₂ fractions: 0.1, 0.2, 0.9.

The key numbers: at 2500°C—a realistic temperature near the reaction interface between the plasma arc and the melt—the H atom molar fraction at 900 mbar is 0.024. At 450 mbar, it rises by 40% to 0.033. At 50 mbar, it jumps to 0.08, 2.3 times higher. The H⁺ species only appear above 5500°C, far from the reaction zone (which sits at 2000–2500°C). So the relevant active species for HyPSR is atomic H, not H⁺.

We then calculated hydrogen utilization efficiency during reduction of a liquid Fe-O melt (Fe-28.9 wt% O) at 2000°C. When reducing with H₂ molecules, efficiency starts around 80% at low reduction degrees but drops to 40% as the melt approaches FeO stoichiometry. When reducing with H atoms, efficiency stays at 98% across the entire reduction range. The reason: H atoms react exothermically with FeO (FeO + 2H → Fe + H₂O), while H₂ reacts endothermically. The atomic pathway is both faster and more hydrogen-efficient.

Experimental Validation: What We Measured

We built a simple but controlled test. Hematite pieces (15–18 g, composition in Table I) were placed on a water-cooled copper hearth (anode) under a tungsten electrode (cathode) in an 18 L chamber. Arc current: 200 A. We varied total pressure (450 vs 900 mbar) and H₂ fraction (10% vs 20% in Ar). After reduction, samples were hammered, magnetically separated, and the oxide fraction analyzed by XRD with Rietveld refinement to quantify Fe, FeO, Fe₃O₄, and Fe₂SiO₄.

First pressure comparison (10% H₂, 10 minutes). At 900 mbar: 92% metallization, 45% H₂ utilization, 20% Fe loss (evaporation). At 450 mbar: only 60% metallization after 10 minutes, but H₂ utilization climbed to 73%. Why slower but more efficient? Because at 450 mbar we put only half the moles of H₂ into the fixed 18 L volume. Fewer reducing agents, slower kinetics. But the H atoms that do form are used much more efficiently. We ran an additional experiment at 450 mbar to full conversion: it took 13 minutes and achieved 67% H₂ utilization—still far better than 45% at 900 mbar. This confirms our thermodynamic prediction: lower pressure increases the H/H₂ ratio near the reaction interface.

Second comparison (450 mbar, H₂ fraction 10% vs 20%). With 20% H₂, reduction was faster: 30% metallization in 2 minutes. But Fe evaporation skyrocketed to 50% of total iron in the same 2 minutes. After 5 minutes, the sample was almost pure Fe, but 70% of the input iron had evaporated. Why? Hydrogen increases the thermal conductivity and voltage drop of argon plasmas. The arc becomes narrower, hotter, and penetrates deeper into the melt. Temperatures exceed the boiling points of Fe (2860°C) and FeO (∼3400°C) locally. You get reduction, but you lose your product to the gas phase.

What This Means for Reactor Design for Green Steel Production

Our work gives three actionable rules for HyPSR, which we are now testing in scaled-up setups.

First, do not use high H₂ fractions. Above 10–15%, the thermal penalty outweighs the kinetic benefit. Evaporation becomes uncontrollable. The optimal appears to be in the 5–10% range, but this depends on arc geometry and power.

Second, operate at moderately reduced pressure. Our experiments at 450 mbar show clear improvement in hydrogen utilization (73% vs 45%) with acceptable Fe loss (20%, which appears to be the thermodynamic minimum at these temperatures). Going to 50 mbar would give even more H atoms, but the reactor volume would need to increase proportionally to keep the same H₂ throughput. There is an economic optimum.

Third, separate the arc physics from the melt chemistry. The high-temperature zone near the cathode produces H atoms. These diffuse toward the cooler melt interface (2000–2500°C) where they react with FeO. If you raise total pressure, you suppress H atom formation. If you raise H₂ fraction, you raise arc temperature and cause evaporation. The two parameters are coupled but not independent. We now have a quantitative map of that coupling.

Where We Stand Compared to the Literature

Previous work by Kamiya (1984) and Weigel (1985) observed that higher H₂ fractions reduced reduction efficiency. They attributed this to slower dissociation. We confirm that thermodynamically (Figure 7 in our paper) and add the evaporation mechanism as an equally important constraint. Badr (2007) studied HyPSR at atmospheric pressure and reported moderate utilization. Our 450 mbar experiments are, to our knowledge, the first systematic demonstration of pressure-enhanced H utilization in this context.

Our thermodynamic calculations of H₂ vs H as reducing agents (Figure 4) provide a quantitative benchmark: using H atoms instead of H₂ improves hydrogen efficiency from ∼40% to ∼98% during the FeO→Fe step. That is not a small correction. It is a factor of 2.5. The engineering challenge is to deliver those H atoms to the reaction interface without overheating the melt. Lower pressure helps. So does arc configuration—a topic we are now exploring with optical emission spectroscopy and Langmuir probes.

The Open Questions We Are Still Working On

We do not yet have direct, in-situ measurements of H atom density near the reaction interface. Our efficiency calculations are indirect—inferred from gas consumption and metallization. We are building a reactor with a quartz window and a calibrated Hα emission line measurement (656.3 nm) to quantify dissociation fractions as a function of pressure and current.

We also do not know how pressure affects the reduction of gangue-containing ores. Our hematite was relatively pure (Table I). Industrial ores contain SiO₂, Al₂O₃, CaO, MgO. These form silicates and spinels that may alter the local oxygen potential and the wetting behavior between melt and arc. That changes evaporation rates and hydrogen transport. We have started a series of experiments with four different ore grades.

Finally, the scale-up question: our chamber is 18 L. An industrial EAF is 100–200 m³. Maintaining reduced pressure (450 mbar) in such a volume requires massive vacuum pumps and energy. The trade-off between hydrogen savings (from 45% to 73% utilization) and pumping energy is not yet quantified. Our rough back-of-envelope suggests that for hydrogen costs above 3 €/kg, reduced pressure becomes economically favorable. Current green hydrogen prices in Europe are 5–8 €/kg. So the case is real.

Summary for the Practicing Metallurgist

We show that absolute gas pressure is not a neutral parameter in hydrogen plasma reduction. Lower pressure (450 mbar vs 900 mbar) increases the concentration of atomic hydrogen at the reaction interface by approximately 40%, raising H₂ utilization from 45% to 73% at the cost of slower kinetics (13 min vs 10 min to full conversion). Higher H₂ fractions (20% vs 10%) accelerate reduction but cause severe Fe evaporation (70% loss in 5 minutes) due to increased arc temperature. The sweet spot for HyPSR appears to be 5–10% H₂ in Ar or N₂, at 400–500 mbar total pressure, with arc currents and gaps optimized to deliver H atoms without overheating the melt.

These are early-stage findings. Our team is now building a 50 kW pilot reactor with pressure control and spectroscopic diagnostics to validate these trends at larger scale and with industrial ores. We invite collaboration from the community—especially on the fluid dynamics of arc-melt interaction under reduced pressure, which is poorly understood.