Silica in Boiler Feedwater: Comparing Coagulation, Magnesium Dosing, Membranes and Flotation Routes
Every plant that feeds a boiler eventually has to answer for its silica. Both soluble and colloidal SiO2 travel through pretreatment quietly, then deposit as a glassy, insulating scale on the hottest surfaces of the system - and by the time the problem shows up in the blowdown schedule, the removal strategy was decided months earlier. Silica removal therefore deserves to be chosen deliberately, and the available routes differ sharply in how deep they can go.
What coagulation can and cannot do
Coagulation silica removal is a physical-chemical method: metal oxides or hydroxides adsorb or coagulate silica out of the water. It is deliberately described as a non-deep process. Coagulation followed by filtration removes about 60% of colloidal silica; add clarification ahead of the filters and the figure rises to roughly 90%. Deeper than that, coagulation alone will not take you.
Magnesium dosing: powerful, but demanding on conditions
Magnesium reagents are normally dosed with lime, which holds the pH in the optimum window of 10.1-10.3 while also stripping part of the temporary hardness and carbon dioxide. Three operating variables decide the outcome. First, a coagulant is usually added alongside the magnesium reagent - iron salts at 0.2-0.35 mmol/L are typical - because it improves the structure of the magnesium oxide sludge and lifts desilication efficiency. Second, temperature: at 40 °C the residual silica in the effluent can be pushed below 1 mg/L. Third, clarifier residence time, which shrinks as temperature rises - more than 1 h at 30 °C, about 1 h at 40 °C, and only 20-30 min at 120 °C. Raw water matters too: higher hardness helps, while the dosing ratio (mg MgO per mg SiO2-3) generally falls between 5 and 20, dropping as silica content rises and climbing with the colloidal fraction.
Aluminum and iron salts: chemistry worth knowing
Aluminum salt removal of dissolved silica works best at 20 °C with a pH of 8-9, and needs about 30 min of contact to adsorb most of the silica. One subtlety: the precipitate must form inside the solution. Once an aluminum salt precipitate has formed outside - especially after drying - its desilication power largely disappears. Crystal state ranks as AlO(OH) > Al2O3·3H2O > Al(OH)3. Colloidal silica behaves differently, preferring a pH of 4.1-4.7, where a single mole of aluminum salt can capture roughly 40 mol of colloidal silica.
Iron hydroxide takes a different path: it adsorbs dissolved silica most effectively near pH 9, and the amorphous form outperforms the crystalline one. The reagent demand is high - removing 1 mg of SiO2 takes 10-20 mg of iron sulfate - yet ambient-temperature treatment can still bring residual dissolved silica down to 3-5 mg/L. One reported case treated water containing 20 mg/L of silica with suitable doses of ferric chloride, sodium aluminate and calcium oxide and achieved 70-80% removal. Hydrated lime deserves mention in the same breath: at 40 °C it removes temporary hardness and carbon dioxide while trimming residual silica by 30-35%, and at 110-115 °C, as applied for boiler feedwater pretreatment at the Fujian refinery, slaked lime reaches an 80% removal rate through CaSiO3 precipitation.
Membranes: reverse osmosis and ultrafiltration
Since the 1980s, reverse osmosis has been a mainstay ahead of ion exchange for boiler feedwater, cutting the load on the demineralization train. RO handles both colloidal and dissolved silica: overall desalination can reach 93% with SiO2 removal around 80%, and a Japanese patent reports mixed water carrying 5-20 mg/L of silica being reduced to below 1 mg/L by membrane permeation. Ultrafiltration shares the pressure-driven principle but works by sieving. To retain colloidal silica its membrane pore size must not exceed 100 nm, and tests put the operating pressure at no more than 7×104-7×105 Pa. Remember what UF cannot do: it has no desalination capability and almost no effect on dissolved silica.
Microbubble flotation as a niche option
Cassell and co-workers demonstrated that microbubble flotation removes all colloidal substances from water. The recipe: condition the water for 10 min with 1.0×10-3 mol/L Al(NO3)3, dose lauric acid as collector at 25 mg/L and ethanol as frother at 2.5 mL/L. Within the pH 8-11 band, more than 90% of colloidal silica is lifted out after 5 min of flotation. The aluminum salt is not optional - colloidal silica will not float without it. Cassell reduced success to three conditions: a small electrolyte dose and pH adjustment so particles agglomerate, a suitable collector, and a stable froth layer.
Failure modes worth planning around
Magnesium dosing punishes sloppy pH control: let the lime feed drift and the desilication window closes quickly. Aluminum salt systems lose effectiveness if sludge handling lets precipitates dry out and recycle. Iron salt routes consume reagent heavily, so chemical cost scales directly with silica load. And flotation depends on froth stability - a dirty collector or wrong frother dosage shows up immediately as falling removal. None of these are disqualifying, but each one argues for pilot verification on the actual raw water before full-scale commitment.
How to choose
Silica compounds cannot be treated in isolation; suspended matter, colloids, dissolved gases and ions all have to leave together if the system effluent is to meet spec. Coagulation is shallow and struggles to bring effluent below 1 mg/L, so where the water quality requirement is high, arrange reverse osmosis, electrocoagulation or ion exchange downstream of the coagulation stage - and let raw water quality plus technoeconomics make the final call.