Basic Oxygen Furnace (BOF) Steelmaking

The BOF slag performs two functions simultaneously: it absorbs oxidised impurities from the metal bath, and it provides the thermodynamic conditions required for phosphorus retention. Phosphorus in the metal is oxidised to P₂O₅, which is stabilised in the slag by combination with CaO to form tricalcium phosphate (3CaO·P₂O₅). This compound is only stable at high slag basicity (CaO/SiO₂ > 3.0) and moderate slag temperature — conditions that are most favourable in the final stages of the blow, when basicity is highest and bath temperature is still below approximately 1,630 °C. Above this temperature, phosphorus reversion — the return of phosphorus from slag back into the metal — becomes thermodynamically significant.

For high-phosphorus hot metals (>0.10% P), plants use a double-slag technique: the converter is tilted at mid-blow and approximately 60–70% of the phosphorus-rich initial slag is poured off, fresh lime is added, and blowing resumes with a clean slag for the second stage. This dramatically improves final phosphorus removal — achieving <0.008% P in the final steel — but adds 8–12 minutes to the tap-to-tap time and increases lime consumption by 15–25 kg/t.

BOF slag volume is typically 100–180 kg/t steel for single-slag practice, rising to 200–250 kg/t for double-slag. Granulated BOF slag has potential use as road aggregate or a cement substitute, but its highly variable composition and the presence of free lime — which expands on contact with water — restrict current valorisation rates. Zero-waste slag valorisation remains an active area of process development across the industry.

Slopping is the uncontrolled overflow of foamy slag from the converter mouth during the blow — one of the most operationally disruptive events in BOF steelmaking. A slopping event can spill 5–20 t of slag in seconds, damaging equipment below the converter, creating a safety hazard for operators, and causing direct metallic yield loss as iron is lost with the ejected slag.

Slopping is caused by a combination of rapid CO evolution (from the carbon boil) into a slag that cannot contain the foam. The specific conditions that promote slopping are: high FeO content in a slag that has not dissolved sufficient lime (FeO acts as both an oxidant and a foaming promoter), a high lance position that generates surface-skimming gas rather than deep penetration, and the critical carbon range of 1.5–2.5% where CO generation rate peaks. The foamy slag at this stage consists of a dense emulsion of slag droplets and CO bubbles — if the foam height exceeds the vessel freeboard (typically 3–4 m above the bath surface), it overflows.

Mitigation strategies include: dynamic lance height control to maintain deeper oxygen penetration during the foaming phase (physically breaking bubbles), reducing oxygen flow rate momentarily when off-gas CO/CO₂ ratio spikes (a sign of intense foaming), and using sublance-assisted slag foaming models that predict slopping risk from off-gas composition trends. Some plants also add anti-slopping agents — coarse iron ore or sinter pellets — to the converter mouth when slopping begins, providing nucleation sites that collapse the foam.

CO bubbles nucleate on MnO and FeO particles dissolved in the slag. In normal (non-slopping) operation, this foaming is actually desirable: the foam increases the slag-metal interfacial area by 10–20 times compared to a flat slag, dramatically accelerating phosphorus removal kinetics in the second half of the blow. Controlled foaming is the mechanism that makes single-slag dephosphorisation possible in 15–18 minutes — without foaming, the interfacial area would be too small for adequate P removal at these rates.

The sublance is a disposable measuring probe deployed through a guide tube alongside the oxygen lance at approximately 90% of blow completion. A disposable thermocouple measures bath temperature directly; a carbon sensor (typically a saturated iron membrane in contact with the bath for 4 seconds) measures bath carbon via the iron-carbon liquidus relationship. The combined result — temperature and carbon — is returned to the process computer within 45 seconds of insertion.

The sublance result allows the end-point control model to calculate the remaining oxygen needed to reach tap carbon target, or to confirm that the blow can be cut immediately. If the sublance result shows the bath is already at target carbon but below target temperature, a small "thermal pad" blow continues at low oxygen flow to raise temperature without further decarburisation. If carbon is above target, the remaining oxygen is calculated and the blow continues at normal rate.

When the sublance result disagrees with the model by more than 0.02% C or 10 °C, it indicates a model calibration issue — typically caused by variable hot metal composition or an unusual scrap mix. In this case the sublance measurement takes precedence.

Modern process control systems at the most advanced BOF plants now use neural-network or machine-learning models that predict tap C and T at oxygen cut without a sublance, by integrating the off-gas mass balance (total oxygen consumed, CO/CO₂ ratio history, lime dissolution model) from blow start. These models, trained on thousands of heats, can achieve one-shot accuracy comparable to sublance-assisted blowing — reducing the additional 45 seconds of cycle time and the cost of disposable sublance probes. However, the sublance remains the primary endpoint verification system at most plants, as it provides a direct physical measurement rather than a model-estimated value.