Question 6 of 7: Magnesium Production (20 marks)

Question 6 — Magnesium Production (20 marks)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Magnesium is produced by two fundamentally different routes: thermal reduction of magnesium oxide with silicon, and electrolysis of molten magnesium chloride. Both are answered below. It is worth noting at the outset that the Pidgeon process is a Canadian invention — Lloyd Pidgeon developed it at the National Research Council in Ottawa in the 1940s, and it was first operated at scale at Haley, Ontario.

(a) The silicothermic (Pidgeon) process. The process reduces calcined dolomite with ferrosilicon under vacuum, and produces magnesium directly as a condensed vapour of very high purity.

Dolomite is first calcined at 1100–1200 °C in a rotary or shaft kiln to drive off both carbonates and leave a mixed oxide known as dolime:

The dolime is ground, blended with ferrosilicon containing about 75–80 % Si (with a little fluorspar as a catalyst to break down the oxide film), and briquetted so that the charge has adequate density and gas permeability. The briquettes are loaded into horizontal nickel–chromium alloy retorts, roughly 300 mm in diameter and 3 m long, which are then evacuated to 10–100 Pa and externally heated to 1150–1200 °C. The reduction is:

Two features of this reaction deserve emphasis. First, the magnesium is produced as a vapour at the operating temperature, well above its 1090 °C boiling point at atmospheric pressure, so it leaves the reaction zone continuously and travels to a water-cooled condenser at the cold end of the retort, where it deposits as solid “crown” crystals of 99.95–99.98 % purity. Second, the reaction as written is not thermodynamically favourable at atmospheric pressure at 1200 °C: it is driven forward only because the vacuum keeps the magnesium partial pressure low, continuously removing the product. This is Le Châtelier's principle used as a process design tool, and it is the single most important point to make in the answer. The role of the lime is chemical rather than incidental — binding the silica as dicalcium silicate, Ca₂SiO₄, lowers the activity of SiO₂ and makes the reduction very much easier than it would be with pure magnesia.

A cycle takes 8–12 hours, after which the retort is cooled, the crown is removed and the spent briquette cake of dicalcium silicate is discharged. The magnesium crowns are remelted under a protective flux or SF₆-free cover gas and cast into ingot. The virtues of the route are simplicity, low capital cost, tolerance of small scale and exceptional product purity; its faults are that it is a batch operation with heavy manual handling, that retort life is limited by creep and oxidation, and that it is energy-intensive at roughly 10–11 kg of coal equivalent per kilogram of magnesium — which is why the process migrated to regions with very cheap coal.

(b) The Magnetherm process. The Magnetherm process, developed in France in the 1960s, is silicothermic reduction rescued from the limitations of the retort. The same chemistry is used, but the reduction is carried out in a liquid slag bath rather than in a solid briquette charge:

Alumina is added to the charge so that the calcined dolomite, the silica formed by the reaction and the added Al₂O₃ together form a molten calcium aluminosilicate slag at about 1550–1600 °C — some 400 °C above Pidgeon practice. Because the slag is liquid, it can be heated internally by electrical resistance, passing current between a carbon electrode and the conducting bath, instead of through a retort wall; the vessel is a refractory-lined furnace rather than an alloy tube, so the temperature is no longer limited by the creep strength of a metal casing. The furnace operates under a vacuum of roughly 3–5 kPa, and the magnesium again leaves as vapour and is condensed externally, in this case as a liquid because the condenser runs above the melting point.

The advantages over Pidgeon practice are substantial: the unit is semi-continuous, with slag tapped periodically and fresh charge fed without cooling the vessel; capacity per unit is far larger; and internal resistance heating is thermally much more efficient than conducting heat through a retort wall, cutting energy consumption to roughly half. The offsetting drawbacks are a higher capital cost, the need for reliable large-scale vacuum equipment, and a magnesium product slightly less pure than the Pidgeon crown because of entrainment from the bath. The higher temperature also means the alumina addition is not optional — without it the slag would be too viscous to conduct current or to tap.

(c) The electrolytic process. The electrolytic route reduces magnesium chloride rather than the oxide, and it is the dominant process wherever seawater, brine or carnallite is the raw material.

The feed is prepared as anhydrous magnesium chloride, which is the hard part of the process. Magnesium may be precipitated from seawater or brine as Mg(OH)₂ with dolime, filtered, and converted to the chloride with hydrochloric acid; the resulting solution must then be dehydrated without hydrolysing, since heating hydrated MgCl₂ in air simply produces MgO and HCl. Industrial practice therefore dehydrates in stages in an HCl atmosphere, or uses the Norsk Hydro route of prilling followed by fluidised-bed drying, or feeds carnallite (KCl·MgCl₂·6H₂O) directly.

The anhydrous chloride is dissolved in a molten salt electrolyte — typically 10–20 % MgCl₂ in a mixture of NaCl, KCl and CaCl₂ — held at 680–720 °C in a refractory-lined cell. The alkali chlorides are there to lower the melting point, raise the conductivity, and adjust the density of the bath so that magnesium floats. The overall decomposition is:

with the half-reactions $\mathrm{Mg^{2+} + 2\,e^- \rightarrow Mg}$ at the steel cathode and $\mathrm{2\,Cl^- \rightarrow Cl_2 + 2\,e^-}$ at the graphite anode. Cells run at 5–7 V and 100–300 kA. Molten magnesium, less dense than the electrolyte, rises and is collected behind a partition wall that keeps it away from the chlorine, then vacuum-ladled out periodically; the chlorine is captured and recycled to the dehydration step or sold, which is a genuine advantage of the route since the reagent is regenerated rather than consumed.

Energy consumption is about 12–18 kWh per kilogram of magnesium, which compares favourably with the thermal routes where electricity is cheap. The process is continuous and readily scaled, and it was the basis of the Norsk Hydro operation at Bécancour, Québec. Its disadvantages are the difficulty and cost of producing genuinely anhydrous feed, the corrosive chlorine handling duty, and a product typically 99.8–99.9 % pure — adequate for alloying but below Pidgeon crown quality — requiring flux refining before casting.