Ferrosilicon Production: Raw Materials and Furnace Control

What Ferrosilicon Is and Why the Smelting Route Matters

Ferrosilicon is an iron-silicon alloy in which silicon is the dominant functional element and iron acts as the carrier that lowers the melting range and stabilises the melt. Commercial material is almost never a single compound: it is a heterogeneous product containing free silicon, iron silicide phases such as FeSi and FeSi2, plus controlled residuals of aluminium, calcium, carbon, sulphur and phosphorus. That phase structure is why two lots with identical silicon assay can behave very differently in a ladle, in an induction furnace or in a grinding mill.

For a buyer, the practical consequence is that the production route must be understood before the assay is accepted, because the route fixes three things that no certificate can retrofit: the size and porosity of the solid, the type and quantity of inclusions, and the residual element profile. All commercial ferrosilicon is made by carbothermic reduction of a silica source in a submerged arc furnace. Variants such as low-carbon grades, calcium-bearing inoculants and densified media grades are downstream treatments of that same base smelt, not separate metallurgies.

This article walks the process from quartz to screened lump, and marks the control points where purchasers should attach an acceptance condition rather than a general statement of quality.

Raw Materials and Their Specifications

Four material streams enter the furnace, and each one carries a specification that shows up later in the certificate of analysis.

Silica source. Quartzite, quartz rock and other silica-rich minerals are the silicon carrier. Smelting-grade material is judged on SiO2 content, lump size distribution and mechanical strength rather than on colour. A workable furnace feed typically runs 96-99% SiO2 with tightly bounded Al2O3, Fe2O3, CaO and P2O5, because aluminium and phosphorus are not removed by the smelting reaction and pass almost completely into the alloy. Lumps below about 20 mm and above about 100 mm both cause problems: fines choke the gas path and cut permeability, while oversized lumps sit unreacted in the burden and destabilise the electrode position. Thermal shock resistance matters as much as chemistry, since a friable quartzite generates fines inside the furnace where they cannot be seen.

Carbon reductant. Metallurgical coke, gas coke, low-temperature carbonised coke and charcoal are all used, and most plants blend them. The reductant must supply both the carbon for reduction and the physical permeability of the charge column. Fixed carbon, ash, volatile matter, moisture and reactivity are the controlling parameters; ash brings aluminium, and high volatile matter steals carbon through the off-gas. Charcoal offers high reactivity and low ash but poor mechanical strength, so it is normally a minority component in the blend.

Iron source. Iron enters as steel scrap, mill scale, iron turnings or, in some plants, as iron ore. Clean, low-phosphorus scrap is preferred because phosphorus cannot be corrected downstream. Copper, tin and other tramp elements from mixed scrap are equally permanent, so scrap sourcing is an alloying decision rather than a purchasing convenience.

Electrode and paste. Self-baking electrodes consume paste continuously; prebaked electrodes are consumed as units. In either case the electrode column is a consumable with a direct cost per tonne of alloy and a direct effect on carbon pick-up in the melt.

Input stream	Controlling parameters	Typical furnace feed window	Effect if out of window
Quartzite / quartz	SiO2, Al2O3, P2O5, lump size, strength	96-99% SiO2; 20-100 mm lump	Fines cut permeability; high Al and P pass into alloy
Metallurgical coke	Fixed carbon, ash, volatile matter, moisture	High fixed carbon, low ash, screened size	Carbon imbalance shifts Si assay and raises power
Charcoal (blend component)	Reactivity, ash, moisture	Minority fraction of reductant	Poor strength collapses charge column
Steel scrap	Fe content, P, Cu, Sn, oil and coating	Clean, low-phosphorus, sized	Tramp elements become permanent residuals

Submerged Arc Furnace Smelting: Reactions and the Control Window

The burden is charged around three electrodes into a circular submerged arc furnace, usually in the 12.5-33 MVA class for ferrosilicon duty. The arc sits buried in the charge, so the reaction zone is shielded from the atmosphere and the off-gas leaves through the column above it. Silica is reduced by carbon at temperatures above roughly 1700 degrees C, and the overall reduction can be written as SiO2 + 2C producing Si + 2CO. In practice the reduction proceeds through intermediate silicon carbide and through silicon monoxide vapour, which is why silicon is partly transported upward in the gas phase and why a tight, permeable charge is essential to recovering it.

Iron present in the burden combines with reduced silicon to form iron silicide, which dissolves in the pool and lowers the activity of silicon, driving the reduction forward. This is the metallurgical reason iron is charged at all: without a metal carrier, silicon would be lost as SiO vapour and as slag.

The operator controls a narrow window defined by four coupled variables:

Charge resistance and electrode immersion. The electrode tip position sets the resistance path and therefore the power input distribution. Deep immersion favours silicon recovery and protects the lining; shallow immersion raises the heat load on the side walls and the shell.

Carbon balance. Too little carbon leaves silica unreduced and inflates slag volume; too much carbon raises carbon in the alloy and can carburise the metal pool. The correct balance is judged from the silicon assay, the slag appearance and the off-gas composition together, never from a single reading.

Temperature and tapping rhythm. Tapping at around 1600-1700 degrees C keeps the alloy fluid enough to separate from slag without overheating the lining. Irregular tapping intervals change the pool level, which changes electrode immersion, which changes the power profile: the disturbance propagates backward through the whole control loop.

Gas handling. Off-gas from a ferrosilicon furnace carries carbon monoxide and fine dust. Capturing and combusting the CO is both an emissions control and an energy recovery measure; uncontrolled leakage at the charge floor is a safety issue as well as an environmental one.

Tapping, Casting and Solidification

Metal is tapped intermittently through a taphole into a ladle or a casting bed. Two casting practices dominate: casting into sand beds or metal moulds to form ingots, and casting into a thin layer that is quenched and then broken. The choice affects the internal structure of the product, not just its shape.

Slow cooling in a thick cast promotes segregation: silicon-rich phases and iron silicide separate over a coarse scale, so a lump broken from the centre of a thick ingot and a lump from its rim can differ measurably in assay and in dissolution behaviour. Faster cooling in a thin bed produces a finer, more uniform structure that dissolves more predictably in a ladle and grinds more evenly. For buyers of powder or of cored-wire fill, cooling practice is a legitimate technical question and belongs in the purchase specification alongside the assay.

Slag is separated during tapping. Ferrosilicon slag is relatively light and is skimmed or left behind in the ladle; slag carry-over into the casting bed shows up as adherent crust on the lump and is a common cause of rejected consignments because it dilutes the effective silicon per tonne delivered.

After solidification the cake is broken, and this is the first point at which the product can be judged visually: a sound fracture surface is dense and metallic, with no excessive porosity, no visible slag skin and no unreacted quartz inclusions.

Crushing, Sizing and Grade Control

Crushed and screened ferrosilicon is sold by size fraction, and the fraction is a functional specification rather than a packaging detail. Fine material dissolves quickly but oxidises and generates fines in handling; coarse lump dissolves slowly and can sit unmelted in a small ladle. The buyer should state the intended melting unit and the desired dissolution behaviour, then accept a defined size window with a stated oversize and undersize tolerance.

Size fraction	Typical designation	Usual application	Handling note
0-3 mm	Fines / powder feed	Cored wire fill, briquetting, powder injection	Susceptible to moisture pick-up; needs dry storage
3-8 mm	Small granule	Foundry inoculant blends, continuous casting additions	Good flow, low dust in automatic feeders
10-50 mm	Standard lump	Electric arc furnace and ladle deoxidation	Default fraction for bulk steelmaking
10-100 mm	Coarse lump	Large ladles, cupola and induction furnace charging	Check for internal porosity before accepting

Screening alone does not fix chemistry. Because a cast cake is not homogeneous, blending across several casts is what delivers a repeatable assay in each shipment, and a supplier who cannot describe the blending step cannot guarantee lot-to-lot consistency even with a compliant certificate.

Refining and Low-Carbon Routes

Standard ferrosilicon from the submerged arc furnace carries carbon at the level set by the carbon balance, typically a few tenths of a percent. Where the steel grade cannot tolerate that carbon, the alloy is refined by oxygen blowing in a ladle or converter: carbon and part of the silicon are oxidised, and the silicon content is then corrected. Oxygen refining lowers carbon substantially but loses silicon yield, which is why low-carbon ferrosilicon carries a cost premium that reflects silicon burned rather than silicon delivered.

Calcium-bearing and barium-bearing inoculants are made by adding the alloying element to the melt or by a controlled re-melt, and their performance depends on the calcium-to-aluminium ratio as much as on the calcium content itself. Densified medium grades used in heavy-media separation are a different product family: they are atomised and heat-treated to a fine, rounded, high-density powder with silicon around 15-16%, and they should never be specified against a metallurgical FeSi75 analysis.

Energy, Electrodes and the Real Cost Structure

Electricity is the largest single cost in ferrosilicon, which is why smelters sit where power is cheap and stable, and why the alloy's price tracks energy policy more closely than it tracks ore. Typical industry figures for FeSi75 smelting fall in the range of 8000-9000 kWh per tonne of alloy, with reductant consumption around 1.0-1.2 tonnes and silica feed around 1.7-1.8 tonnes per tonne of alloy; actual values move with furnace size, charge quality and silicon recovery, which is normally in the 85-92% band.

For a purchaser this explains several otherwise puzzling features of the market. A smelter with a low power tariff and a good quartzite source can be structurally cheaper than a plant with better equipment. Silicon recovery, not silicon assay, is the real efficiency metric. And a shift in electrode paste or coke quality shows up as a change in residual aluminium and carbon long before it shows up in price.

Quality Control and Acceptance

Sampling is the weak point in most disputes. Ferrosilicon is lumpy and segregated, so a sample taken from the top of a bag is not representative. Standard practice is to take increments across the consignment according to a documented sampling plan, reduce them by crushing and quartering, and test the reduced sample. GB/T 4010 covers sampling of ferroalloys and GB/T 3650 covers general delivery requirements; ISO 3713 addresses sampling of ferrosilicon and silicon alloys specifically. GB/T 13247 addresses the determination of grain size for ferroalloys and is the reference to cite when a size fraction is disputed.

Chemistry is normally determined by wet analysis or by X-ray fluorescence, with silicon reported as the metal content and residuals reported individually rather than as a sum. For FeSi75, a widely used commercial window is silicon 74.0-80.0%, aluminium 2.0% max, carbon 0.2% max, sulphur 0.02% max and phosphorus 0.04% max, with tighter aluminium and carbon limits available by agreement; GB/T 2272 is the grade reference for FeSi75, FeSi72 and FeSi65. A certificate that reports only silicon cannot be used to reject a lot on aluminium or carbon, so the purchase order should name every element the process is sensitive to.

Physical acceptance covers size distribution against the ordered fraction, moisture, and the absence of visible slag or unreacted quartz. Moisture deserves specific attention: ferrosilicon stored wet and then charged into a hot ladle can generate hydrogen and, in the case of material containing lime-bearing additions, can react exothermically. Dry, covered storage on a raised floor is a technical requirement, not housekeeping preference.

FAQ

Q: What is the reaction that produces ferrosilicon in a submerged arc furnace?
Silica is reduced by carbon at high temperature, overall as SiO2 + 2C forming Si + 2CO, with silicon carbide and silicon monoxide as intermediates and iron dissolving the reduced silicon as iron silicide.

Q: Why is steel scrap charged together with quartzite?
Iron acts as the carrier metal: it combines with reduced silicon and lowers silicon activity in the pool, which improves reduction and reduces silicon loss as SiO vapour.

Q: What silicon range is normal for FeSi75?
Commercial FeSi75 is commonly supplied at 74.0-80.0% Si with aluminium 2.0% max, carbon 0.2% max, sulphur 0.02% max and phosphorus 0.04% max, per the GB/T 2272 grade family.

Q: Why does low-carbon ferrosilicon cost more than standard grade?
Decarburisation is done by oxygen blowing, which oxidises part of the silicon as well as the carbon, so yield falls and more alloy is consumed to deliver the same silicon.

Q: Does lump size affect deoxidation performance?
Yes. Fines dissolve quickly but oxidise and dust in handling, while coarse lump dissolves slowly; the correct fraction depends on ladle size, addition practice and the melting unit.

Q: How should a ferrosilicon consignment be sampled for acceptance?
Increments should be taken across the whole consignment under a documented plan, then crushed and quartered, following GB/T 4010 or ISO 3713 rather than sampling from bag surfaces.