Phosphoric acid plant fluorine recovery: fluosilicic acid from two scrubbers

A wet-process phosphoric acid plant releases the phosphate rock's fluorine as gas in two places: the attack tank, as silicon tetrafluoride, and the vacuum concentrator, where the acid's fluosilicic acid breaks down to SiF4 and HF about 1:2 (EPA AP-42 §8.9 puts 20-40% of the rock's fluorine into the concentrator vapour). Both streams are scrubbed with water to fluosilicic acid, H2SiF6, sold to make aluminium fluoride and cryolite, as EFMA/Fertilizers Europe BAT Booklet No. 4 (2000) §2.4 describes them. The fume scrubber takes the attack-tank gas of the wet-process-phosphoric-acid-dihydrate example (72.6 mol/s CO2 and 1.50 mol/s SiF4) through two absorbers in series at 90% each, the booklet's figure for one atmospheric absorber: 99.0% of the fluorine is removed, the booklet's 'better than 99%', and 1.14 g/s of fluorine leaves with the stack gas. With no HF in that gas the SiF4 hydrolyses, 3 SiF4 + 2 H2O -> 2 H2SiF6 + SiO2, so a third of the silicon drops out as 107 kg/h of silica, and the acid leaves at the booklet's 22% H2SiF6, 2.33 t/h of it. The hot humid gas condenses 11.56 mol/s of water into the liquor on its way to 320 K, and the 128 kW reaction heat is part of the 678 kW the liquor cooler removes. The vacuum scrubber takes the concentrator vapour at 0.12 bar in a single stage at 85%, inside the booklet's 83-86% recovery for a vacuum system: SiF4 + 2 HF -> H2SiF6 uses all the SiF4 with the HF, so no silica forms, and the acid leaves at 20% (the booklet's 17-23%), 7.72 t/h of it; 59.8 g/s of fluorine goes on with the water vapour to the barometric condenser. Representative inputs, labelled: the concentrator vapour (21 mol/s of fluorine, half the DH example's acid fluorine, with the 30 kg/s of water a 28 to 54% P2O5 concentration evaporates), the scrub water rates and the scrubber temperatures. Honesty notes: (1) removal per absorber is an input, not predicted from the contactor; the H2SiF6 vapour pressure (Munter et al. 1947) that limits product strength could not be obtained, so no equilibrium cap is applied; (2) the gas leaves water-saturated at the scrubber temperature, the liquor's water activity taken as 1; (3) the energy balance is on formation enthalpies, with no sensible heat for the aqueous H2SiF6 and HF and no heat of dilution; (4) the reactor's own fluorine split is an input of that example, and this flowsheet takes its off-gas as fixed numbers, not as a live link.