Arsenic Removal From Water: Speciation, Oxidation and Adsorption

Why does arsenic speciation control removability?

Arsenic occurs in natural water in two oxidation states, and their acid–base chemistry is what makes one easy to remove and the other difficult. Arsenite, As(III), exists as arsenious acid H₃AsO₃, which has a first dissociation constant of pK_a1 ≈ 9.2. Across the normal treatment window of pH 6–8.5 it is therefore an uncharged, fully protonated molecule that is barely attracted to a charged sorbent or a growing hydroxide floc.

Arsenate, As(V), exists as arsenic acid H₃AsO₄ with pK_a1 ≈ 2.2, pK_a2 ≈ 7.0 and pK_a3 ≈ 11.5. At pH 6–9 it is present as the anions H₂AsO₄⁻ and HAsO₄²⁻, which form strong inner-sphere complexes with positively charged iron-oxide surfaces. The practical consequence is unavoidable: unless the raw water is already fully oxidised, the first unit process must convert As(III) to As(V).

How do you oxidise As(III) to As(V)?

Pre-oxidation is the critical first step. Dissolved oxygen alone is far too slow, so a chemical oxidant is dosed ahead of the removal stage. Practical options and their characteristics:

• Free chlorine (hypochlorite) — fast and cheap; oxidises As(III) within seconds to minutes, but forms trihalomethanes if natural organic matter is present.
• Potassium permanganate (KMnO₄) — rapid and robust; also oxidises Fe(II)/Mn(II), which is convenient when iron removal is co-located.
• Chlorine dioxide, ozone and solid-phase (Fe/Mn) media — effective alternatives where disinfection by-products must be minimised.

The stoichiometry is modest — roughly one mole of a two-electron oxidant per mole of As(III) — so oxidant demand is normally dominated by competing reductants (Fe(II), Mn(II), sulphide, NOM), not by arsenic itself. A short contact tank ahead of coagulation or the adsorbers is usually sufficient. Once oxidation is complete, the downstream chemistry described below does the work.

Coagulation and co-precipitation on iron (hydr)oxides

Dosing a ferric salt — ferric chloride (FeCl₃) or ferric sulphate — hydrolyses to form amorphous ferric hydroxide (hydrous ferric oxide, HFO). Arsenate is removed by two coupled mechanisms: surface complexation (adsorption of HAsO₄²⁻ onto Fe–OH surface sites, forming Fe–O–As bonds) and, at higher loadings, surface precipitation as the surface saturates and a mixed ferric–arsenate solid grows outward. This continuum between adsorption and precipitation on iron oxyhydroxide is exactly the behaviour described by Benjamin (1983) and formalised in the surface-precipitation model of Farley, Dzombak and Morel (1985).

Because removal is driven by the surface, the governing operational variable is the Fe:As mass ratio, not simply the ferric dose. Meeting a 10 µg/L target from tens of µg/L of raw arsenic typically needs a Fe:As mass ratio of about 20:1 to 40:1 (a molar ratio near 15–30). The resulting arsenic-bearing HFO is then separated by clarification, membrane microfiltration of the coagulated solids, or media filtration. The chemistry is a close cousin of standard metal-salt coagulation covered in our coagulation and flocculation guide.

Adsorption onto GFH and iron-oxide media

Where flow is small, sludge handling is unwelcome, or arsenic is the only target, fixed-bed adsorption is often preferred to coagulation. Granular ferric hydroxide (GFH), granular ferric oxide and iron-oxide-coated media present a high density of Fe–OH surface sites in a packed column. Water passes through at a fixed empty-bed contact time (EBCT) of typically 3–5 minutes, arsenate loads onto the media by the same surface-complexation mechanism, and the bed is run to a breakthrough set by the discharge limit, then replaced or regenerated.

This simple mass balance — total arsenic the bed can hold divided by the arsenic added per litre of water — predicts media life to within the accuracy of the isotherm, and is the basis of the worked example below.

Worked example: GFH adsorber bed life

Estimate the media life of a granular ferric hydroxide adsorber for a groundwater already oxidised to As(V).

Inputs: influent C₀ = 40 µg/L = 0.040 mg/L; target (breakthrough) C_b = 10 µg/L = 0.010 mg/L; GFH Freundlich K_F = 5 (mg/g)(L/mg)^0.25, 1/n = 0.25; bulk density ρ_b = 1,250 g/L; EBCT = 5 min.

• Loading at breakthrough (evaluate the isotherm at C_b): q_b = K_F · C_b^1/n = 5 × (0.010)^0.25 = 5 × 0.316 = 1.58 mg As/g.
• Arsenic the bed can hold per litre of media: q_b × ρ_b = 1.58 mg/g × 1,250 g/L = 1,975 mg As per litre of media.
• Arsenic added per litre of water: C₀ − C_b = 0.040 − 0.010 = 0.030 mg/L.
• Bed volumes to breakthrough: BV = 1,975 / 0.030 ≈ 65,800 BV.
• Run time: at a 5-minute EBCT the bed treats 12 BV/h, so 65,800 ÷ 12 ≈ 5,480 hours ≈ 7.5 months of continuous operation before media change-out.

The result (roughly 30,000–70,000 BV is typical for GFH) is sensitive to competing anions and pH, so confirm the isotherm on the actual water. Halving the influent to 20 µg/L would roughly double the run to breakthrough, because the same media capacity is consumed at half the rate.

How do pH and competing anions affect performance?

Adsorption of arsenate onto iron (hydr)oxides is strongly pH-dependent. The oxide surface is positively charged below its point of zero charge (roughly pH 7–8 for HFO/GFH), so anion uptake is strongest in mildly acidic to neutral water and falls off sharply above pH 8. Lowering pH into the 6.5–7.5 band before the adsorbers is one of the most effective ways to extend bed life.

Competing anions occupy the same surface sites. In order of nuisance: phosphate (a direct chemical analogue of arsenate and the most aggressive competitor), silicate (significant above ~20 mg/L SiO₂, especially at higher pH), and to a lesser extent fluoride, bicarbonate and sulphate. High phosphate or silicate can cut GFH bed life several-fold, which is why site-specific isotherm and column testing is indispensable before sizing.

When do anion exchange and RO make sense?

Anion exchange on a strong-base resin removes arsenate (again, only after oxidation of As(III), which the resin cannot hold). Its weakness is competition: sulphate is preferred over arsenate, so exchange is only economic in low-sulphate water, and chromatographic peaking can briefly push effluent arsenic above influent near exhaustion — a real compliance hazard that demands conservative run lengths.

Reverse osmosis rejects both species physically, but rejection of neutral As(III) is markedly poorer than of charged As(V), so pre-oxidation still pays. RO reaches the lowest residuals (<2 µg/L) and suits high-value or multi-contaminant duties, at the cost of energy and a concentrate stream. Whichever route is chosen, the driver is the same statutory limit — the US EPA and WHO 10 µg/L value — and demonstrating compliance is easier with the monitoring and reporting framework we help clients build. Arsenic sits within the wider duty covered in our heavy-metals removal and compliance guide.