Red Mud Settling: PAM for Bayer Process Washers

Red mud settling is the hardest flocculation duty in industrial minerals. The slurry leaving a Bayer digester is caustic soda at 150–250 g/L expressed as Na₂O, sitting at 95–105°C by the time it reaches the settler, carrying sub-micron goethite, hematite, sodalite and titania. Almost nothing about that environment resembles the neutral-pH, ambient-temperature water where a normal anionic polyacrylamide works, and a grade that settles copper tailings beautifully can do very little in a primary settler.

This page is written for alumina refinery process engineers and for the procurement people who have to buy flocculant for the circuit. It covers why the Bayer environment attacks conventional polyacrylamide, which chemistry belongs at each stage from primary settler through the washer train to residue disposal, what dosage and settling rates to expect, and — the part most supplier pages skip — where polyacrylamide is the right answer and where it is not. We supply high molecular weight anionic PAM. That product has a real and useful place in this circuit, and it also has stages where it is the wrong tool, and saying so is more useful to you than a claim we cannot support.

Why the Bayer Circuit Breaks Ordinary Polyacrylamide

Three properties of Bayer liquor act on the polymer at once, and they compound.

Caustic hydrolysis of the backbone. Polyacrylamide is a chain of amide groups. In strong caustic at high temperature those amide groups convert to carboxylate, which is the same reaction used deliberately and under control to make partially hydrolysed anionic PAM in the first place. In a primary settler the reaction is neither deliberate nor controlled: a polymer dosed at 30% hydrolysis keeps hydrolysing in service, drifting toward a fully charged polyacrylate. Higher charge in a liquor of this ionic strength does not help — it does not extend the chain the way it would in fresh water — and the amide groups being consumed are the ones responsible for hydrogen-bonding adsorption onto mineral surfaces. The polymer loses its grip as it charges up.

Extreme ionic strength. At 150–250 g/L Na₂O the double layer around every particle is compressed to almost nothing and charge on the polymer is fully screened. Anionic PAM in fresh water works partly because its carboxylate groups repel each other and hold the chain extended, giving it bridging reach. In Bayer liquor there is no repulsion left to do that, so the chain is coiled and its effective reach is a fraction of what a jar test in tap water would suggest. This is the single biggest reason a bench screening on cold dilute liquor overstates performance.

Temperature. Primary settlers run at 95–105°C, and washers step down from there. Every degree accelerates both the hydrolysis above and ordinary thermal chain scission. A polymer solution that would hold its viscosity for a day at ambient loses molecular weight in minutes at settler temperature, which sets a hard limit on how much of the dose survives to do work — and means make-up water temperature and dosing residence time matter far more here than in a water treatment plant. The general mechanics of that loss are in our guide to polyacrylamide viscosity control.

Which Chemistry Belongs Where

The circuit is not one duty, it is four, and the correct flocculant changes as liquor strength and temperature fall through the train. Treating it as a single polymer decision is the most common sourcing mistake we see from refineries buying on price.

The primary settler row is the one to be clear about. Since the late 1980s the reference chemistry for that duty has been hydroxamated polyacrylamide, in which some amide groups are converted to hydroxamate. Hydroxamate chelates iron on goethite and hematite surfaces, so adsorption is a genuine coordination bond rather than hydrogen bonding, and it survives conditions that strip a conventional polyacrylamide. We do not manufacture hydroxamated polymer, and no amount of molecular weight or hydrolysis adjustment on a standard anionic grade substitutes for it. If your primary settler is the problem, the honest answer is that you need a hydroxamated product from a supplier who makes one, or a starch-based programme.

Causticised starch remains in real use, either alone on ores that respond to it or as a blend with synthetic polymer, and it has advantages worth knowing: very low cost per tonne, no caustic-hydrolysis failure mode, and it flocculates the finest sodalite fraction that synthetics often miss. Its drawbacks are high dose rates in the kilograms per tonne range, organic carbon loading into the liquor, and variable performance batch to batch. Many refineries run starch in the primary settler and synthetic polymer down the washer train, which is exactly the split in the table.

The Washer Train Is Where Anionic PAM Earns Its Keep

Counter-current decantation washing exists to recover soda and dissolved alumina from the mud before it leaves the plant, and its economics are driven by soda loss rather than by throughput. Every tonne of residue that leaves with entrained liquor takes caustic with it, and caustic is the largest single consumable cost in a refinery. Underflow density is therefore the number that matters: a washer running at 45% solids instead of 52% is sending measurably more soda to the residue area on every tonne of mud.

Liquor strength drops stage by stage down the train, which changes the polymer job as you go. At the top of the train, caustic is still strong enough that hydrolysis is the dominant risk and dose needs to be high. By the last washer the liquor is dilute enough that a high-hydrolysis high-MW anionic grade behaves much closer to how it does in mineral processing, and the bridging mechanism described in our mining tailings guide starts to apply properly. Practical consequence: one polymer specification across the whole train is a compromise at both ends. Refineries that split the train — a more robust product on the first one or two washers, a cheaper high-MW anionic on the last stages — usually come out ahead on total cost.

Hydrolysis degree is the lever to get right. Our field starting point for washer and residue duty is 18–20 million daltons at 35–45% hydrolysis, which is higher hydrolysis than we recommend for almost any other mineral application. The reason is that some hydrolysis is going to happen in service whether you choose it or not, so it is better to dose a polymer already near its equilibrium charge than to dose a 25% product and let the liquor take it to 45% while it is trying to work. Below about 30% hydrolysis the polymer under-performs from the start in this liquor; above roughly 50% adsorption falls away because too few amide groups remain. The general tradeoff is covered in our charge density guide, but red mud sits at the extreme end of it.

Dosage runs 100–300 g per tonne of dry solids, with 1–2 m/h settling flux typical, and bauxite is the highest polymer consumer per tonne of any ore we supply. Clay-type minerals with very high surface area are the reason: goethitic and lateritic bauxites consume 3–5 times the polymer of a coarse ore because there is simply more surface to cover. Gibbsitic bauxite processed at lower digestion temperature is generally the easier case; boehmitic and diasporic ores digested at 240–270°C arrive at the settler hotter and harder to flocculate.

Dosage and Performance Targets by Ore Type

Starting points for washer and residue duty, not primary settler. Treat them as the centre of a jar test range rather than a specification — bauxite mineralogy varies more between deposits than almost any other ore we supply, and the reactive silica and goethite fractions move the answer more than the alumina grade does.

Note the direction of the goethitic row: higher dose and slower settling at the same time. High-goethite ores are the worst case on both counts because goethite is fine, high-surface-area and poorly crystalline, and it holds water in the underflow as well as consuming polymer. Refineries switching to a higher-goethite ore blend routinely find washer capacity constrained before anything else in the plant, and no polymer optimisation fully recovers what the mineralogy took away — the honest answer there is a dose increase plus a capacity review, not a different grade.

Underflow density figures assume conventional thickeners. Deep cone and paste thickeners reach considerably higher solids and need the polymer selection re-run, because the mechanism shifts from settling rate to yield stress and rake torque. If the mud has to be pumped to a dry stack, yield stress becomes the constraint rather than percent solids, and a floc structure that maximises settling rate is not the one that pumps best.

Bench Testing Hot Caustic Liquor Without Fooling Yourself

A cold-liquor jar test on red mud is worse than no test, because it produces confident numbers that do not survive the plant. Two errors account for most of the gap.

The first is temperature. Screening at 25°C on liquor that runs at 90°C removes the entire thermal degradation mechanism and most of the caustic hydrolysis, so the polymer looks far more effective than it will be. Tests need to be run hot, in a heated jacketed vessel or at minimum in a water bath at process temperature, with the polymer contact time matched to the plant's dosing residence time. A polymer that performs at 90°C with 30 seconds of contact and fails with 4 minutes of contact is telling you something important about your dosing point location.

The second is dilution. Diluting liquor to make it safe or convenient to handle drops ionic strength, un-screens the polymer charge, and lets the chain extend — so bridging reach in the beaker is nothing like bridging reach in the settler. Test at plant liquor strength. Where that is impractical for safety reasons, at least run a dilution series so you can see which direction the performance curve is moving and by how much, rather than extrapolating from one dilute point.

Make-up water is the third thing to control, and it is often overlooked because it seems trivial next to the liquor chemistry. Prepare stock in clean water at 0.1–0.3%, hydrate for the full 45–90 minutes a high-MW grade needs, and do not prepare stock in process liquor or condensate — hydrolysis starts in the make-up tank and the dose you inject is then not the product you specified. Standard procedure is in our PAM dissolving method, and the general test discipline in our jar test procedure applies with the temperature and ionic-strength caveats above.

Measure four things, in this order of usefulness: interface descent rate over the first 5 minutes, supernatant clarity at 30 minutes, settled bed solids at 2 hours, and — if you can — soda concentration in the settled bed liquor, which is the number that actually maps to money. Per ICSOBA and International Aluminium Institute practice on residue management, settling and washing performance should be evaluated against soda recovery and residue disposal characteristics together rather than on settling rate alone. A polymer that settles faster but leaves a looser bed can cost more in soda loss than it saves in throughput.

Residue Disposal: Where the Duty Is Changing

Bauxite residue disposal has moved decisively away from wet impoundment. Ajka in 2010 and the wider re-examination of tailings facilities after Brumadinho pushed the industry toward dry stacking and filtered residue, and the Global Industry Standard on Tailings Management now frames how new facilities are assessed. For flocculant selection this changes what you are optimising for.

In a wet mud lake the polymer's job was clarity in the decant and enough consolidation to keep the pond operating. In dry stacking the job is a filterable, stackable cake at 55–65% solids or above, and the properties that matter are filtration rate and yield stress rather than free settling velocity. A very high molecular weight polymer producing large open flocs settles fast and filters badly — the floc structure holds water and blinds the filter cloth. Plants moving from lake disposal to filtration usually have to drop molecular weight and often reduce dose, which is counter-intuitive to operators who have spent years pushing dose up for settling rate.

Dust from dried stack surfaces is a separate polymer duty and one where we get more enquiries than the process circuit itself. Residue dust is alkaline and fine, and it is the most visible community-facing issue at most refineries. Anionic PAM applied as a dilute surface spray binds the crust; the mechanism is the same soil-crusting behaviour covered in our soil stabilization guide, with the caveat that residual alkalinity on an un-neutralised stack shortens the life of an anionic polymer on the surface, so re-application intervals are shorter than on neutral soil. Where the stack has been carbonated or seawater-neutralised, a nonionic grade holds up better than anionic.

Return water from the residue area closes the loop back into the plant, and it accumulates dissolved organics and soda over time. Clarity duty on return water is the easiest polymer job in the whole circuit — low dose, high molecular weight, fast settling — and it is often the stage where a refinery first trials a new supplier, precisely because getting it wrong has limited consequence. That is a sensible place to start a qualification, and it is where most of our alumina-sector volume begins.

Sourcing for an Alumina Refinery

Refinery polymer consumption is large and continuous, which changes the buying problem from grade selection to supply reliability. A 2 million tonne per year refinery treating high-goethite ore at 250 g/t can run into hundreds of tonnes of polymer a year on washer duty alone, and a stockout stops the residue circuit before it stops anything else.

What we bring to that is capacity and documentation rather than exotic chemistry: 100,000 tonnes per year across three lines from a 15,000 m² plant in Zhengzhou with 70+ staff, three-stage testing with solid content ≥92% and residual monomer ≤0.05% on every batch, ISO 9001/14001/45001, and 30,000+ tonnes a year exported to 45+ countries. MOQ is 500 kg for trial work and 7–10 days is the standard lead time on common grades, which means a washer-train trial can be run on a pallet rather than a container.

Practical advice on qualification, from refineries that have done it well. Start on return water or the last washer stage where the consequence of a poor result is contained. Run the bench work hot and at plant liquor strength as described above. Ask for retained-sample results on the specific batch you trial, so that a repeat order can be checked against it rather than against a generic data sheet. And keep the primary settler out of scope until the rest is proven — that stage has its own chemistry and mixing a supplier trial into it confuses two variables at once.

On grade selection our starting recommendation for washer and residue duty is the high molecular weight end of our anionic range taken to 35–45% hydrolysis, and for neutralised-stack dust or acid-side polishing duty a nonionic grade. Broader mineral-processing grade mapping is in our APAM mining supplier guide, and if you are cross-referencing against an incumbent brand specification, the published-grade comparisons in our BASF Magnafloc and Zetag equivalents guide set out what can and cannot be matched from a data sheet. For alumina specifically that guide reaches the same conclusion as this page: Bayer-process duty needs liquor conditions, not a grade number.

Frequently Asked Questions

What flocculant is used for red mud settling in the Bayer process?

It depends on the stage. Primary settlers on full-strength hot liquor use hydroxamated polyacrylamide or causticised starch, or a blend, because those survive the caustic and temperature. The washer train and residue thickening use high molecular weight anionic polyacrylamide at 35–45% hydrolysis, and mud lake decant or return water uses high-MW anionic at much lower dose. One product across the whole circuit is always a compromise.

Why does normal anionic polyacrylamide fail in a primary settler?

Three mechanisms at once. Caustic hydrolyses the amide groups the polymer needs for adsorption, converting it toward polyacrylate in service. The very high ionic strength screens the polymer charge so the chain coils and loses bridging reach. And 95–105°C accelerates thermal chain scission, so molecular weight falls within minutes. Hydroxamated grades work there because hydroxamate chelates iron on goethite and hematite surfaces, which is a coordination bond rather than hydrogen bonding.

What is the typical polyacrylamide dosage for red mud?

100–300 g per tonne of dry solids on washer duty, which is the highest polymer demand of any ore we supply. Gibbsitic bauxite with low goethite sits at the bottom of that range, goethitic and lateritic ores at the top. Residue thickening before disposal runs 80–150 g/t and mud lake decant only 20–60 g/t. Expect settling rates of 1–2 m/h and underflow of 38–55% solids on conventional thickeners.

Do you supply hydroxamated polyacrylamide for alumina refineries?

No. We manufacture anionic, cationic and nonionic polyacrylamide, and hydroxamated grades are outside our product line. If your primary settler needs hydroxamated chemistry, buy it from a producer who makes it. We can supply the washer train, residue thickening, mud lake and dust control duties, and we would rather tell you where the boundary is than sell you a grade that will not perform on hot liquor.

Why does high-goethite bauxite need so much more polymer?

Goethite is fine, poorly crystalline and very high in surface area, so there is far more area for polymer to adsorb onto before bridging can begin. It also holds water in the settled bed, so underflow density suffers at the same time as dose rises. A refinery shifting to a higher-goethite ore blend usually finds washer capacity becomes the plant constraint, and polymer optimisation recovers only part of what the mineralogy cost.

Does dry stacking change which polyacrylamide grade to use?

Yes, and usually downward in molecular weight. Wet impoundment rewards fast free settling, so very high MW and large open flocs are ideal. Filtration for dry stacking rewards filterability and manageable yield stress, and large open flocs hold water and blind the cloth. Plants moving from lake disposal to filtered residue commonly drop MW and reduce dose, which surprises operators used to pushing dose up for settling rate.

Can polyacrylamide help with red mud dust on a residue stack?

Yes, applied as a dilute surface spray it binds a crust much as it does on erodible soil. The complication is residual alkalinity: on an un-neutralised stack an anionic polymer degrades faster at the surface, so re-application intervals are shorter. Where the residue has been carbonated or seawater-neutralised, a nonionic grade lasts longer. Dose and application rate should be set on a test plot rather than from a general figure.

How should we bench test polymer for a Bayer circuit?

Hot and undiluted. Run tests at process temperature in a heated or jacketed vessel and at plant liquor strength, and match polymer contact time to the plant's dosing residence time. Testing cold or on diluted liquor removes the caustic hydrolysis and charge-screening mechanisms and overstates performance substantially. Prepare stock in clean water at 0.1–0.3% with full hydration time, never in process liquor.