Aluminium fluoride from a phosphoric plant's own fluorine
What the H2SiF6 a phosphoric plant scrubs is actually for. The 22 wt% fluosilicic acid that phosphoric-acid-fluorine-recovery makes is reacted with aluminium hydroxide -- H2SiF6 + 2 Al(OH)3 -> 2 AlF3 + SiO2 + 4 H2O, with the liquor's free HF going the shorter way, 3 HF + Al(OH)3 -> AlF3 + 3 H2O -- turning a fluorine emission into AlF3 for aluminium smelting. Solved: 2.33 t/h of 22% acid plus near-stoichiometric hydroxide give a cake at 94.3 wt% AlF3 on 95% conversion, 929 kg of Al(OH)3 per tonne of AlF3 (which is just the two molar masses, 78.00/83.98, and nothing else), and 95% of the fluorine fed reporting to the product. Al, F, Si, H and O each close to machine precision. HONESTY NOTES: (1) NO energy balance -- AlF3's own formation enthalpy is held (-1510.4 kJ/mol) but aluminium hydroxide's is in no source bundled here (not the NBS rows, not the chemicals package, and the vendored LLNL file carries only its PHASES block, so gibbsite's dissolution enthalpy cannot be turned into a formation enthalpy without an aqueous Al(3+) datum this tree does not have), so the reaction heat cannot be formed and the block is isothermal with no duty, as superphosphate_den is for the same reason; (2) the conversions, the product moisture and the share of silica reporting to the cake rather than the liquor are all plant measurements and are never defaulted; (3) no crystallisation, so the trihydrate and its calcination to anhydrous AlF3 are a separate dryer the flowsheet draws; (4) no cryolite route, which needs a sodium source and its own stoichiometry; (5) the spent liquor's silica is carried as SiO2 rather than the colloidal and fluorosilicate forms a real liquor holds.
The flowsheet
The solved topology — every unit op's real duty, conversion, or split, read straight off a genuine converged solve.
The stream table
Every stream's flow, temperature, pressure, and composition — real converged numbers, not placeholders.