Ammonium Paratungstate Production: Digestion to Crystal

Last updated: 2026-10-31

Ammonium paratungstate production turns one tonne of 65% WO3 wolframite concentrate into roughly 0.71 t of APT, and the spread between a well-run and a poorly run digestion stage is about six points of WO3 recovery — 97% against 91%. That spread is worth more than the reagent bill, so the autoclave, not the crystallizer, decides the margin.

Key Takeaways

• Digestion recovery sets economics: 95-98% WO3 is routine, 91% means grind or alkali is wrong.
• Scheelite needs a finer grind and 1.5-2.0x stoichiometric alkali; wolframite tolerates coarser feed.
• Molybdenum co-dissolves completely and must be removed before crystallization, never after.
• APT crystal size of 30-80 µm d50 comes from supersaturation control, not the centrifuge.
• Mother liquor purge of 5-15% sets sodium, sulfur, and molybdenum carryover into crystal.

Where Does Ammonium Paratungstate Production Start?

Ammonium paratungstate production starts at the autoclave, not the crystallizer. The flowsheet is six steps, each removing what the next cannot tolerate: gangue, phosphorus and arsenic, molybdenum, sodium, and water.

Leaching explained

Alkaline pressure digestion is a hydrometallurgical operation that dissolves tungsten from concentrate into sodium tungstate liquor inside an autoclave held above atmospheric boiling point. Sodium tungstate purification is a sequence of solid-liquid separation, ion exchange or solvent extraction, and sulfide treatment that turns a dirty leach liquor into a clean ammonium tungstate solution. An evaporative crystallizer is a vacuum vessel that boils water from that solution at 55-80 °C so APT grows as discrete crystals rather than a fines storm.

Digestion leaves 100-180 g/L WO3 in liquor with most impurities dissolved. Dilution and filtration remove residue; ion exchange or amine extraction concentrates tungsten while rejecting sodium and sulfate; sulfidation or selective extraction removes molybdenum. The eluate is then evaporated, crystallized, centrifuged, washed, dried at 100-120 °C, and for oxide customers calcined to tungsten trioxide at 600-800 °C.

Why not chlorination. Direct chlorination to tungsten hexachloride serves special grades, but runs at 500-900 °C with chlorine-class reagents, needs gas-tight graphite or nickel hardware, and adds a reduction step. For a refinery selling APT that is a detour through the metal.

Why not soda-ash roast. The legacy 800-900 °C roast sinters the charge, consumes carbonate in side reactions with silica, and loses recovery on low-grade feed. Most Chinese wolframite and CIS scheelite plants use pressure digestion, which reaches 95-98% in 2-4 hours at 150-180 °C; a soda autoclave at 200-225 °C remains the variant for low-grade scheelite. For leaching fundamentals see leaching; for sector context, tungsten.

What Feed and Reagent Conditions Does Alkaline Digestion Need?

Digestion is a caustic leach above atmospheric boiling point, and the two feed families sit in different parts of the window. Wolframite, (Fe,Mn)WO4, dissolves readily: iron and manganese report to residue as hydroxides while tungsten leaves as sodium tungstate. Scheelite, CaWO4, is harder, because calcium stays in the system as Ca(OH)2 and re-precipitates tungsten as CaWO4 the moment caustic concentration or temperature falls. That is why the scheelite column runs hotter, stronger, and finer.

ASME VIII Div 1 vs Div 2 overview

Alkali is arithmetic before it is art. One tonne of WO3 consumes 345 kg of NaOH at exact stoichiometry, since 1 mol WO3 takes 2 mol NaOH. At the 1.5x excess typical of wolframite that becomes about 520 kg, and at 2.0x for stubborn scheelite about 690 kg, before silica and phosphate take their share. Excess above roughly 2.0x buys almost no extraction while raising reagent cost, silica dissolution, and neutralization load.

Grind and temperature. Wolframite leaches fast enough at P80 75-100 µm; scheelite is surface-reaction limited and wants P80 40-55 µm. Pressure keeps the liquor liquid, with 0.6-1.0 MPa(g) covering 150-170 °C and 1.0-1.5 MPa(g) covering 170-180 °C. Above 180 °C the extra kinetics rarely justify more silica dissolution and faster seal and agitator wear.

Pulp density trades energy against mixing. At 30-45% solids each tonne of slurry carries less water to heat, but viscosity and settling rise; above 45% the agitator, not the chemistry, is the limit.

Impurities. Phosphorus enters as phosphate, stays soluble, competes for caustic, and later loads resin; magnesium or aluminum sulfate knocks it out, and arsenic follows the same route. Silicon dissolves as sodium silicate, typically 2-8 g/L SiO2 at 170 °C, then polymerizes on cooling into a gel that blinds cloth and fouls resin, which is why plants age the diluted liquor at 70-80 °C instead of quenching it. Molybdenum dissolves completely and does not fractionate in the autoclave.

These are typical values, to be confirmed by test work on your concentrate; a feed with 5% phosphate will shift the alkali balance outside the table. A 2-20 L unit reproducing these conditions is described under laboratory pressure leaching autoclave.

Which Equipment Sits on Each Step?

Nearly every item on the train is specified against hot caustic and chloride.

Autoclave. A horizontal multi-compartment or vertical agitated vessel, jacket or live-steam heated, with an agitator sized for 30-45% solids and a flash letdown. Working volume is slurry flow times residence over a 0.75 fill factor: a 3,000 t/y line at 3.5 m³/h needs 14 m³ for 3 hours, so two 8 m³ vessels.

Flash and dilution tank. Letdown flashes vapor and dilutes 150-180 g/L liquor to the 100-150 g/L band, setting the cooling rate that decides whether silica stays dissolved or gels.

Filtration. A recessed-plate press or rotary drum filter removes residue. Presses win on fine residue where wash ratio matters; drums suit higher tonnage. Pilot units generate the cake and wash data first: pilot-scale plate and frame filter presses.

Purification. Fixed-bed ion exchange on strong-base resin at 15-25 g/L WO3 and pH 2-3, or mixer-settlers with a tertiary amine; columns give cleaner eluate, mixer-settlers tolerate solids and higher flow. See packed extraction column and solvent extraction.

Material selection is where projects get caught. Type 316L is not enough in the autoclave circuit: 300-series stainless suffers caustic stress-corrosion cracking above roughly 50-80 °C once NaOH passes about 5%, and chloride accelerates it at weld heat-affected zones. Nickel 200 resists caustic beyond 300 °C and is the default wetted material, solid or clad on carbon steel, with Alloy 600 for the hottest nozzles, agitator shafts, and instrument pockets.

How Is Tungsten Separated From Molybdenum?

Molybdenum is the impurity that ruins APT quietly. It dissolves as sodium molybdate at essentially 100% extraction, follows tungsten through dilution and filtration, and ends up in the crystal. In the customer’s plant it alters carbide grain growth, embrittles tungsten wire, and volatilizes as MoO3 during sintering, so a buyer finding 200 ppm against a 20 ppm limit rejects the lot.

Sulfide precipitation. Molybdate reacts with NaHS or Na2S to form insoluble MoS3 while tungstate stays in solution. The window is pH 8-9 at 60-80 °C for 30-60 minutes, with sulfide at 3-6 times the molar molybdenum requirement. Below pH 8 the reagent degrades to H2S; above pH 10 precipitation is incomplete. The MoS3 is colloidal, needing flocculant and a fine-cloth polish filter, and residual sulfide must be destroyed with hydrogen peroxide before the resin sees it. Expect 90-97% removal per stage, with two stages common above 0.3% Mo in concentrate.

Selective solvent extraction. A tertiary amine at 5-10 vol% in kerosene with 5% isodecanol extracts molybdenum strongly from acidic sulfate at pH 2-3, where 50-100 g/L sulfate suppresses tungsten extraction. Tungsten is scrubbed with dilute ammonia, molybdenum stripped with stronger ammonia. This route also purifies tungsten, so it can replace ion exchange rather than supplement it.

Ion exchange selectivity. Strong-base resin adsorbs both elements and separates them poorly alone. The workable variant reduces molybdenum to Mo(V) with sulfur dioxide before loading, changing its complexation enough for selective ammonia elution; the alternative is to put removal on the eluate. It is economic only at low Mo:W ratios, because loaded molybdenum consumes capacity tungsten is paying for.

For a concentrate above about 0.3% Mo, two-stage sulfide precipitation ahead of ion exchange is preferred because it removes the bulk of the molybdenum at low capital cost and leaves the resin to do what it does well. For a plant already running amine extraction on acidic sulfate liquor, adding a scrub-and-strip section is preferred because the separation needs no new solid-liquid step and the molybdenum reports as a saleable by-product. For a wolframite feed below 1:1,000 Mo:W, a single sulfidation stage with tight pH control is preferred because two stages cost more than the residual molybdenum is worth.

How Is APT Crystallized and What Controls Crystal Size?

Crystallization preserves or throws away all the impurity work upstream. APT is crystallized by evaporating water under vacuum at 55-80 °C, a deliberate window: below 55 °C the solubility curve is too flat for economic boil-off, above 85-90 °C ammonia volatilizes and the liquor drifts toward ammonium metatungstate.

Hold the supersaturation ratio at 1.05-1.15 and existing crystals grow, keeping the population coarse. Run higher — faster boil-off, cold feed, a vacuum break — and nucleation dominates, delivering 10-20 µm fines that centrifuge badly and dust in the customer’s plant.

Target 30-80 µm d50 by laser diffraction on a volume basis, under 10% below 10 µm, and tapped bulk density around 1.8-2.5 g/cm³ (indicative).

A forced circulation crystallizer with an elutriation leg is the usual choice. High internal circulation keeps supersaturation low at the growth zone; an elutriation leg is a vertical classifying section that returns fine crystals to the boiling zone while coarse product leaves; and the external exchanger can be cleaned mid-batch. For the ammonium paratungstate production step specifically, that classification converts a 25 µm product into a 55 µm one at the same yield.

Mother liquor recycle is the quiet constraint: sodium, chloride, sulfate, molybdenum, and phosphorus concentrate three- to tenfold before steady state. A purge of 5-15% holds the balance; recovery options sit in mother liquor recovery.

1. Charge and analyze. Fill with eluate at 120-180 g/L WO3, confirm Na, Mo, S, and Cl against the batch card, and reject an out-of-window batch.
2. Pull vacuum and heat. Reach 55-65 °C at 15-25 kPa absolute, with the external exchanger at a 5-10 °C approach to limit wall scaling.
3. Concentrate to the metastable limit. Evaporate at 3-6% of batch volume per hour to 1.05-1.10 saturation, verified by density, not by clock.
4. Seed. Add 3-8% of batch mass as 20-40 µm seed at the circulation inlet, then hold supersaturation by trimming boil-off, never by adding water.
5. Grow at controlled boil-off. Hold 55-80 °C and a ratio below 1.15 for 4-8 hours at 15-25% magma density, logging boil-off, density, and temperature every 30 minutes.
6. Classify and discharge. Draw product from the elutriation leg at set-point density so only classified coarse crystals leave and fines stay in the vessel.
7. Dewater, wash, and dry. Centrifuge, displace mother liquor with two stages of demineralized water at 1.5-2 m³/t APT, dry at 100-120 °C, and return wash water to the circuit.

Design detail on circulation ratio and classification is in crystallizer design and evaporator crystallizer.

Troubleshooting: Why Does APT Miss Specification?

Seven failure modes account for most off-spec APT. Each has a signature that appears before the assay does.

1. Low WO3 extraction from digestion. Symptom: residue assays 3-8% WO3 and recovery sits at 90-93%. Cause: coarse grind leaving unliberated scheelite, alkali below 1.4x, temperature 10-15 °C under set point from a fouled jacket, or a silica and Ca(OH)2 coating passivating the surface. Action: check P80 against the liberation curve, raise excess in 0.1x steps, verify the thermocouple with a second probe, and run a 2-hour caustic strip on residue.
2. Silica gel blinding the filters. Symptom: cycle time doubles within a week and cloth washes do not restore it. Cause: over-digestion of silicate, rapid quenching, or no aging before filtration. Action: age 2-4 hours at 70-80 °C, dose magnesium or aluminum sulfate, and drop autoclave temperature 5 °C on high-silicate feed.
3. Sodium carryover above the contract limit. Symptom: Na at 40-80 ppm against a limit in the low tens. Cause: short displacement wash, wash water short-circuiting through the cake, or sodium accumulating in recycled mother liquor after the purge was cut. Action: wash at 2 m³/t APT in two stages, check the distribution plate, and sample final wash filtrate before blaming the centrifuge.
4. Molybdenum above limit. Symptom: 100-250 ppm in crystal while sulfidation looks fine. Cause: pH drifting below 8, NaHS dosed on feed rather than analyzed liquor, short reaction time, or MoS3 passing the polish filter as colloid. Action: automatic pH control at 8.5-9, 4:1 S:Mo, 60 minutes at 70 °C, flocculant, and a filtrate Mo check before releasing the batch.
5. Off-size crystals and fines. Symptom: d50 at 18 µm with 30% below 10 µm. Cause: supersaturation spike from fast boil-off or a vacuum break, late or missing seed, excessive shear, cold eluate charge. Action: cut boil-off by a third, seed at 1.05-1.10 saturation, preheat feed to within 5 °C of vessel temperature, and check the elutriation flow.
6. Low bulk density. Symptom: 1.2-1.5 g/cm³ tapped against a 1.8 target, with poor flow in the customer’s feeder. Cause: fine population, agglomerates from inadequate washing, sodium and sulfate inclusion, or over-fast drying. Action: shift the size distribution coarser first; never mill or classify your way out of a crystallization problem.
7. Mother liquor losses taking WO3 out of the circuit. Symptom: plant balance closes at 94-95% while every unit assays above 97%. Cause: purge set too high for sodium control, centrifuge carryover, cloth passing solids, or wash water to effluent. Action: run a daily WO3 balance by stream, recover tungsten from purge and wash water by ion exchange, and treat any stream above 1 g/L WO3 as product.

The symptom appears at the back end, the cause two or three operations upstream, and the fix is a control loop rather than new equipment.

What Does Ammonium Paratungstate Production Cost?

Concentrate dominates and it is not close. At a 65% WO3 feed and 97% recovery, one tonne of APT consumes about 1.41 t of concentrate (dry basis) — 0.888 t of WO3 divided by 0.65 and by 0.97. That multiplier sets the cash-cost floor and typically accounts for 75-85% of it.

Alkali consumption follows stoichiometry plus side reactions, 350-600 kg NaOH per tonne of APT on a 100% basis, the upper half belonging to high-silica scheelite and to plants running 1.8-2.0x excess. Energy splits into digestion steam and evaporation, with evaporation the larger and more variable term. A multiple-effect train needs roughly 3-5 t of steam per tonne of APT; a mechanical vapor recompression unit needs 350-550 kWh/t APT and almost no steam, which is why MVR evaporation crystallization units pay back fastest where steam is expensive and power is firm.

These are indicative planning values excluding capital charges, freight, assaying, and effluent treatment; replace them with vendor quotations and site rates. Two cautions: the power figure assumes the crystallizer and vacuum system are the only large loads, and the alkali figure assumes under 1% phosphate, which can push consumption past 700 kg/t.

Field Data: 3,000 t/y APT Line, Recovery and Crystal Size

One anonymized line, sized at 3,000 t/y of APT from wolframite; all figures are indicative, not guarantees.

Feed assayed 65.2% WO3 (dry basis) with 0.18% Mo, 0.9% SiO2, and 0.4% P, ground to P80 88 µm. In the first campaign, at P80 120 µm and 1.3x alkali, extraction settled at 92.4% and residue ran 4.6% WO3. Regrinding to 88 µm and raising excess to 1.55x moved extraction to 96.8% and residue to 1.9% WO3 — a 4.4-point gain that repaid the extra mill power in about five months.

Molybdenum was the harder fix. With one sulfidation stage at pH 7.6, APT carried 210 ppm against a 25 ppm contract limit and batch rejection exceeded 20%. Automatic pH control at 8.6 plus a second stage at 4:1 S:Mo brought molybdenum to 18 ppm; sodium fell from 45 to 14 ppm once centrifuge wash rose from 0.8 to 1.9 m³/t.

Crystal size moved only after reseeding and a 30% cut in boil-off, taking d50 from 24 to 58 µm; the centrifuge was never the constraint. Specific energy fell from 610 to 480 kWh/t APT and steam from 5.2 to 0.9 t/t when the batch pan was replaced by a forced circulation unit with an MVR compressor. Tapped bulk density stabilized at 2.1 g/cm³.

Frequently Asked Questions About Ammonium Paratungstate Production

What WO3 recovery should a new APT plant guarantee?

Plan on 95-97% from a well-controlled caustic autoclave on wolframite, and 94-97% on scheelite where grind is fine enough. Guarantees above 98% usually rest on a laboratory leach of a selected sample rather than plant feed, and leave no allowance for silica coating, scale, or turndown. Confirm with locked-cycle pilot work before signing.

Can one autoclave treat both wolframite and scheelite?

Only with a compromise that suits neither. Scheelite needs 165-180 °C, 140-180 g/L NaOH, and P80 near 45 µm; wolframite runs best at 150-170 °C and 100-160 g/L. A single vessel can run the scheelite condition for both, but the wolframite campaign then pays extra alkali.

Why crystallize APT under vacuum instead of at atmospheric pressure?

Because boiling point is the control variable. At 15-25 kPa absolute the liquor boils at 55-65 °C, which keeps ammonia in solution, holds the metastable zone wide, and lets the exchanger run at a small approach. Atmospheric boiling pushes past 100 °C, where ammonia loss changes liquor composition and crystal habit.

How much molybdenum can the process tolerate in the feed?

Sulfide precipitation handles roughly 0.5-1.0% Mo in concentrate economically; above that the reagent cost per tonne of APT and the MoS3 handling load grow faster than the value recovered. Beyond about 1%, a dedicated molybdenum circuit, usually selective amine extraction with a saleable by-product, is the better answer.

How much mother liquor should be purged, and where does it go?

Typically 5-15% of circulating mother liquor, set by the sodium and sulfate balance rather than a fixed percentage. The purge is not waste: it carries 20-60 g/L WO3 and goes to ion exchange or a small recovery crystallizer. Purging to drain to save capital sells tungsten at the effluent tariff.

What crystal size do carbide and wire customers want?

Usually 30-80 µm d50 with a low fines fraction, good flow, and tapped bulk density above 1.8 g/cm³. Carbide grades tolerate size variation better than fines, because fines dust and segregate in press feed; wire grades care more about sodium and molybdenum. Ask for the customer’s sieve or laser analysis first.

What test work is needed before equipment sizing?

Batch autoclave tests across a grind and alkali matrix, a locked-cycle test to establish impurity build-up, filtration tests on the actual residue, and crystallization trials at the intended vacuum and boil-off. Each is small against the equipment it sizes, and skipping the locked-cycle test means discovering purge requirements during commissioning.

Conclusion

Ammonium paratungstate production is decided in four places: grind and alkali excess at the autoclave, the pH window at the sulfidation reactor, the supersaturation band in the crystallizer, and the purge rate on mother liquor. Get those right and 96-97% WO3 recovery with 20 ppm molybdenum and a 55 µm crystal is ordinary rather than exceptional; get one wrong and the plant chases the symptom at the centrifuge for a year. The most useful step before committing capital is a locked-cycle pilot campaign on your own concentrate — send the assay, grind curve, and target specification through the request a quote page and our process group will return an indicative flowsheet, reagent balance, and equipment list. More background sits in the resources library.

References

1. U.S. Geological Survey, Mineral Commodity Summaries, tungsten chapter, annual series. https://www.usgs.gov/centers/national-minerals-information-center
2. International Tungsten Industry Association, technical notes on tungsten processing and APT. https://www.itia.org/
3. Weiss, N.L. (ed.), SME Mineral Processing Handbook, Society of Mining Engineers of AIME, 1985.
4. Lassner, E. & Schubert, W.-D., Tungsten: Properties, Chemistry, Technology of the Element, Alloys, and Chemical Compounds, Kluwer Academic / Plenum Publishers, 1999.
5. ASM International, ASM Handbook, Volume 2: Properties and Selection: Nonferrous Alloys and Special-Purpose Materials.
6. Habashi, F., Handbook of Extractive Metallurgy, Wiley-VCH, 1997.
7. Gupta, C.K. & Mukherjee, T.K., Hydrometallurgy in Extraction Processes, CRC Press, 1990.

Changelog

• 2026-10-31: First publication.