Iron and Manganese Removal from Groundwater: The Oxidation-Filtration Approach

Why are iron and manganese a problem in water?

Iron and manganese are among the most common nuisance contaminants in anaerobic groundwater and some process waters. Under the reducing, low-oxygen conditions of a confined aquifer, both metals exist in their soluble, reduced forms — ferrous iron, Fe(II), and manganous manganese, Mn(II) — and pass invisibly through a borehole pump in true solution. The problem appears the moment that water meets air or any other oxidant: the metals oxidise and precipitate as coloured, insoluble oxides.

The consequences are almost entirely aesthetic and operational rather than acutely toxic, which is why the limits are set on the basis of consumer acceptability and asset protection:

• Iron produces red-brown staining of laundry and fixtures, a metallic taste, and rusty turbidity. It also feeds iron-oxidising bacteria such as Gallionella and Leptothrix, whose slimes foul pipes, boreholes and heat exchangers.
• Manganese is more insidious. It causes black staining and deposits at much lower concentrations, catalyses its own accumulation in distribution mains, and sloughs off as black water during flow transients. There is also growing evidence linking chronic manganese exposure to neurological effects, tightening the health rationale.

Raw groundwater iron is commonly 0.5–10 mg/L and manganese 0.1–2 mg/L, so the plant must often deliver one to two orders of magnitude of removal. For a broader look at the granular filters that do the capture, see our guide to media filtration design.

How does oxidation-filtration actually work?

The whole process rests on a single idea: change the oxidation state of the metal so that its solubility collapses. Fe(II) and Mn(II) salts are freely soluble; the oxidised species Fe(III) (as ferric hydroxide) and Mn(IV) (as manganese dioxide) have solubility products so small that they precipitate almost completely at normal water pH. Oxidise, then filter out the resulting particles.

The overall reactions with dissolved oxygen are:

From the stoichiometry, the theoretical oxygen demand is 0.14 mg O₂ per mg Fe(II) and 0.29 mg O₂ per mg Mn(II). These are tiny amounts — a few tenths of a mg/L of oxygen — which is why simple aeration (cascade, spray or draught aerators bringing dissolved oxygen to 8–11 mg/L) supplies far more oxidant than the reaction needs. The rate, not the quantity of oxidant, is what limits the design.

The two-step conceptual model — homogeneous oxidation in the reaction/contact tank, then removal of the oxide floc on a filter — is deliberately simplified. In practice a large part of real-world removal, especially for manganese, is heterogeneous: the oxide surfaces already present on the filter media adsorb the reduced metal and catalyse its oxidation, which is why the choice of media matters so much (covered below).

Why is manganese so much harder to remove than iron?

If iron and manganese both simply oxidise with oxygen, why does every operator find manganese the troublesome one? The answer is kinetics. The homogeneous oxidation of Fe(II) by dissolved oxygen is fast at circumneutral pH; the oxidation of Mn(II) by oxygen is agonisingly slow at the same pH and only becomes practical above about pH 9.

Both reactions are strongly pH-dependent because the rate law is second order in hydroxide concentration — the reactive species is the hydrolysed metal, not the bare aquo ion:

The difference between the two metals is the rate constant. At 20 °C and neutral pH the half-life of Fe(II) oxidation by oxygen is on the order of minutes, but the half-life of Mn(II) oxidation is measured in days to weeks — utterly impractical for a treatment plant. This is the central design fact: aeration alone will polish iron out but will barely touch manganese unless the pH is pushed up towards 9–9.5 or a stronger oxidant or a catalytic surface is provided.

A further complication: the MnO₂ that forms is autocatalytic. Freshly precipitated manganese dioxide adsorbs Mn(II) and accelerates its further oxidation, so a filter that has been "seasoned" with an oxide coating removes manganese far better than clean media. This is the basis of catalytic filtration.

Which oxidant should you use, and at what dose?

The oxidant choice trades off cost, kinetics, by-products and whether manganese has to be tackled. The four common options — oxygen (aeration), chlorine, potassium permanganate and ozone — each have a defined stoichiometric demand per mg of metal, derived from the electron balance of the redox half-reactions.

A few mechanistic points behind the numbers. Chlorine oxidises Fe(II) rapidly but the Mn(II) reaction is slow at neutral pH, so chlorine is mainly an iron oxidant unless a catalytic filter completes the manganese step. Permanganate is the classic manganese oxidant: Mn(VII) is reduced to MnO₂, which is itself the active catalytic solid, so a slight residual permanganate dose continuously regenerates the filter coating. Ozone is powerful but must be dosed carefully — excess ozone can oxidise Mn(IV) all the way to soluble permanganate, turning the water pink and defeating the purpose; our ozonation in water treatment guide covers ozone contacting and control in detail.

Worked example 1 — oxidant dosing. A borehole delivers 4.0 mg/L Fe(II) and 0.6 mg/L Mn(II). Compute the theoretical oxidant demand for each candidate:

• Oxygen: (0.14 × 4.0) + (0.29 × 0.6) = 0.56 + 0.17 = 0.73 mg/L O₂. Aeration to 9–10 mg/L dissolved oxygen supplies this many times over — oxidant mass is never the constraint, but Mn needs catalytic help.
• Chlorine: (0.64 × 4.0) + (1.30 × 0.6) = 2.56 + 0.78 = 3.34 mg/L Cl₂ theoretical; specify ~4–5 mg/L to cover ancillary demand and leave a residual.
• Permanganate: (0.94 × 4.0) + (1.92 × 0.6) = 3.76 + 1.15 = 4.91 mg/L KMnO₄. Dose slightly below stoichiometric on the manganese and let the greensand finish it, to avoid pink residual.

These figures set the chemical feed pump and day-tank sizing. Because permanganate is self-indicating (excess turns water pink), many plants trim the dose to the point of faint colour break-through and back off — a simple, robust control that needs no online analyser.

What is catalytic greensand and how does it work?

The workhorse of small and medium iron/manganese plants is catalytic filter media — manganese greensand, or one of the modern MnO₂-coated silica/anthracite equivalents (Birm, pyrolusite, and proprietary catalytic sands). These media do far more than physically strain out oxide floc; their manganese-dioxide-coated surfaces drive an adsorption-oxidation mechanism that removes manganese which homogeneous oxidation alone could never catch in the available contact time.

The mechanism has three coupled steps at the grain surface:

1. Adsorption. Dissolved Mn(II) (and Fe(II)) partitions onto the negatively charged, high-surface-area MnO₂ coating.
2. Surface oxidation. The adsorbed Mn(II) is oxidised by the higher-valence manganese oxide (and by any dissolved oxidant present), forming fresh MnO₂ in place. The reaction that is hopelessly slow in bulk solution is fast on the catalytic surface.
3. Regeneration. The oxidant (chlorine, or intermittent/continuous permanganate) re-oxidises the reduced surface back to MnO₂, restoring capacity.

Manganese greensand is operated in one of two modes. In intermittent regeneration, the bed is periodically taken offline and soaked/rinsed with a permanganate solution to rebuild the oxide coating; between regenerations it works on stored oxidising capacity. In continuous regeneration (CR), a small permanganate (or chlorine) dose is fed continuously ahead of the filter so the coating is regenerated on the fly and the bed never depletes. CR is the norm for steady groundwater duties because it decouples run length from adsorptive capacity.

The oxide-coated grain is essentially a self-renewing catalyst. Because a healthy MnO₂ coating is autocatalytic for manganese, a well-run bed becomes more effective over its first weeks of operation as the coating matures — the opposite of media that simply blind and foul. When selecting between greensand, catalytic anthracite and pyrolusite, weigh the trade-off between service rate, backwash demand and the oxidant regime; the underlying hydraulics follow the same rules as any rapid gravity or pressure filter in media filtration design. For plant hardware selection, our overview of process water treatment equipment sets out the pressure-vessel and skid options.

How do you size the filter and estimate sludge?

Two calculations dominate the physical design: the filter area (from the hydraulic loading) and the mass of oxide sludge produced (which sets backwash-water handling and disposal). Both are quick once the oxidation chemistry is settled.

Worked example 2 — filter sizing. Continue the borehole above at a design flow of Q = 40 m³/h. Choose a conservative service rate for combined Fe/Mn duty of v = 10 m/h:

That is met by, for example, two pressure vessels of ~1.6 m diameter (2.0 m² each) running in parallel — giving redundancy so one can backwash while the other stays in service. Check the empty-bed contact time (EBCT): at 0.85 m bed depth and 10 m/h, EBCT = 0.85 / 10 = 0.085 h = 5.1 minutes, comfortably within the 3–10 minute range typical for catalytic iron/manganese beds.

Worked example 3 — oxide sludge produced. The solids are the precipitated oxides. Per mg of metal removed, the dry oxide mass follows from the molar masses:

• Fe → Fe(OH)₃: molar mass ratio 106.9/55.8 = 1.92 mg oxide per mg Fe.
• Mn → MnO₂: molar mass ratio 86.9/54.9 = 1.58 mg oxide per mg Mn.

For our water (4.0 mg/L Fe, 0.6 mg/L Mn) at 40 m³/h = 960 m³/d:

• Iron oxide: 4.0 g/m³ × 1.92 × 960 m³/d = 7,373 g/d ≈ 7.4 kg/d.
• Manganese oxide: 0.6 g/m³ × 1.58 × 960 m³/d = 910 g/d ≈ 0.9 kg/d.
• Total dry solids ≈ 8.3 kg/d.

This oxide is carried out in the backwash water. If the backwash is, say, 3% of throughput and the settled sludge thickens to 2% dry solids, the wet sludge volume and its onward handling can be estimated directly — and it is worth designing the backwash recovery/thickening at the same time as the filters rather than as an afterthought. Where the residual stream also carries other regulated metals, the disposal route intersects with heavy metals removal compliance.

When should you use sequestration or biological removal?

Oxidation-filtration is the default, but two alternative strategies fit specific niches: sequestration for very low levels, and biological removal for the right water chemistry.

Sequestration does not remove the metals at all — it keeps them in solution and invisible. A sequestrant, usually a polyphosphate or silicate, complexes Fe(II)/Mn(II) and prevents them oxidising and precipitating in the distribution system. It is cheap and needs no filter, but it is strictly a low-concentration tactic: it is generally limited to total iron plus manganese below about 1 mg/L (some practitioners cap it nearer 0.5–1.0 mg/L Fe and 0.1 mg/L Mn), and it fails if the water is heated (polyphosphates hydrolyse) or if oxidation has already begun. Because the metal stays in the water, sequestration is unsuitable wherever the aesthetic limit must genuinely be met at the tap by removal — it only masks the problem.

Biological iron and manganese removal harnesses naturally occurring iron- and manganese-oxidising bacteria cultivated within a filter bed. Given controlled, slightly oxidising conditions (a small, carefully limited oxygen dose and a suitable pH/redox window), Gallionella and Leptothrix oxidise iron, and specialised manganese-oxidising organisms oxidise manganese biologically — crucially, at pH values (around 6.5–7.5) and rates that pure chemical oxidation could not achieve for manganese. The bacterial oxides are dense and filter well, backwash demand is often lower, and no strong chemical oxidant is needed. The catch is that biological plants must be acclimatised, the iron and manganese stages sometimes need to be staged separately (iron bacteria can inhibit the manganese stage), and the redox/oxygen control window is narrow. For high, stable groundwater loads a well-run biological plant can be the lowest-chemical-cost option; for variable or intermittent supplies, robust chemical oxidation-filtration remains easier to operate.

A pragmatic decision rule: use sequestration only for trace levels in cold, short distribution systems; use biological removal for high, steady loads where footprint and redox can be controlled; and use chemical oxidation-filtration — aeration or permanganate/chlorine ahead of catalytic media — for the great majority of duties, especially wherever manganese must be reliably driven below 0.05 mg/L. Where the treated water is destined for a membrane process, note that residual iron and manganese are potent foulants, so an upstream Fe/Mn stage protects downstream membrane filtration systems from irreversible oxide scaling.

Designing an iron and manganese oxidation-filtration plant

1. Characterise the raw water. Measure Fe(II), Mn(II), pH, alkalinity, dissolved oxygen, temperature and any competing reductants (sulphide, ammonia, organics) on fresh, unaerated samples.
2. Select the oxidant and dose. Compute the stoichiometric demand per mg Fe and Mn, add 10-50% for ancillary demand, and pick aeration, chlorine, permanganate or ozone to suit the manganese kinetics.
3. Set the pH. Raise pH if needed: iron oxidises above pH 7, manganese needs pH 9+ for oxygen oxidation or a catalytic surface at lower pH.
4. Choose the filter media. Select catalytic greensand or MnO2-coated media in continuous or intermittent regeneration mode; fix service rate (5-12 m/h) and bed depth (0.75-0.9 m).
5. Size the filters. Filter area A = Q / v; provide parallel vessels for redundancy during backwash, and confirm empty-bed contact time of 3-10 minutes.
6. Design residuals handling. Estimate oxide sludge (1.92 mg/mg Fe, 1.58 mg/mg Mn) and size backwash recovery, thickening and disposal accordingly.