Why Arsenic Dissolves into Groundwater — Redox in Kumamoto’s Aquifer, Read with PHREEQC

Introduction: beyond the metallic smell

In #11 we noted that the metallic smell of well water is dissolved iron (Fe²⁺) in reducing groundwater — iron that oxidizes to reddish-brown hydroxide once it meets air. We promised that “such trace constituents are governed by redox (oxidation–reduction).” This article keeps that promise.

The lead is arsenic (As). Arsenic is highly toxic, and the drinking-water limit is a strict 10 µg/L (both WHO and Japan). In the Kumamoto area, this arsenic exceeds the limit naturally — not from pollution. Hossain et al. (2016) found As ranging 0.1–60.6 µg/L, with 36% of wells in the stagnant Kumamoto Plain exceeding 10 µg/L.

Why would arsenic dissolve naturally? As with the fluoride of #12, the answer is slow water–rock reaction over time. But this time, what moves arsenic is not pH — it is electrons. We reproduce the mechanism in PHREEQC in three stages: ① the redox sequence, ② arsenic speciation, and ③ release of arsenic from iron oxide.

Same stage as #12, a different lead

Strikingly, the signature of high-arsenic groundwater resembles the high-fluoride water of #12. For the high-As area (39 samples), Hossain et al. (2016) report:

High pH (mean 7.97, range 7.14–9.45)
Reducing (ORP mean +42 mV, down to −283 mV; low dissolved oxygen)
Low NO₃ (often below detection) and high dissolved iron (Fe mean 511, max 5283 µg/L)
Clustered in long-residence stagnant zones (central Kumamoto Plain)

Sediment-core analysis shows that arsenic is adsorbed on iron and aluminum oxides/hydroxides (As correlates with Fe and Al, not Mn) and is geogenic in origin.

So the stage is the same as #12 — “high pH, stagnant.” But whereas #12 saw high pH desorb fluoride, in #13 reduction (electron transfer) dissolves the iron oxide and takes arsenic with it. The lead role passes to redox.

The redox sequence — electron acceptors are used in order

What does it actually mean for groundwater to “turn reducing”? The key is that when microbes decompose organic matter, they use up electron acceptors in order of energy yield. This order is the redox sequence.

In PHREEQC, adding organic matter (CH₂O) step by step to an oxic recharge water reproduces the sequence directly.

SOLUTION 1  Oxic recharge water        # water with O2, NO3, SO4
    pe 14;  O(0) 0.5;  N(5) 0.5;  S(6) 1.0 ...
EQUILIBRIUM_PHASES 1
    Pyrolusite  0.0  5e-5              # MnO2 (electron acceptor)
    Fe(OH)3(a)  0.0  2e-3              # ferrihydrite (acceptor & Fe source)
REACTION 1
    CH2O 1.0;  0 0.1 ... 3.0 mmol      # add organic matter (reductant)

The result is Figure 1.

Figure 1: The redox sequence. Top: as organic matter (CH₂O) is added, pe (electron availability) falls in steps. Bottom: electron acceptors are consumed in the order O₂ → NO₃⁻ → MnO₂ → Fe(OH)₃ → SO₄²⁻, with Fe²⁺ and H₂S appearing. The red band is the iron-reduction zone — where arsenic is released.

The order is clear:

First O₂ (aerobic respiration) is consumed.
Then NO₃⁻ (denitrification).
Once O₂ and NO₃⁻ are gone, pe crashes and MnO₂ → Fe(OH)₃ reduction begins (Fe²⁺ rises sharply).
Further on, SO₄²⁻ is reduced, producing H₂S.

This sequence explains the signature of high-arsenic water. Arsenic is released in the red band — where iron oxide Fe(OH)₃ dissolves. And crucially, NO₃⁻ is consumed before iron. So where NO₃⁻ still remains, iron reduction has not been reached, and arsenic is not released. The paper’s “high As = low NO₃” is exactly this ordering of the sequence.

The PHREEQC how-to is here

Computing the redox sequence in PHREEQC is walked through hands-on in PHREEQC from scratch #13: the redox sequence, and denitrification diagnosis in #14: diagnosing nitrate. Here we apply those results to the field case of Kumamoto’s arsenic.

Why arsenic moves — reductive dissolution of iron oxide

Arsenic in the sediment is adsorbed on the surface of iron oxide (hydrous ferric oxide, Hfo) — effectively “pinned.” In oxic conditions the iron oxide is stable and arsenic stays put.

But when the redox sequence reaches the iron-reduction rung, this changes. As Fe(OH)₃ accepts electrons and dissolves (reductive dissolution), the arsenic on its surface loses its host and is released to the water. In PHREEQC, reducing an Hfo surface that carries adsorbed arsenic (SURFACE) reproduces this release.

Figure 2: Release of arsenic by reductive dissolution of iron oxide (concept demo). Initially As is adsorbed on Hfo and dissolved As is below the limit (blue dashed = remaining Fe(OH)₃, red = dissolved As). As organic matter drives reduction, Fe(OH)₃ dissolves and the adsorbed As is released — dissolved As rises past the 10 µg/L limit.

Figure 2 shows exactly the picture: not “poison mixed in,” but arsenic that was pinned on iron oxide dissolving along with the iron. As Fe(OH)₃ decreases (blue), dissolved arsenic (red) rises in its place — this mirror relationship is the heart of arsenic mobilization.

An honest limit: this is a “concept demo”

This sorption/desorption calculation (the Dzombak–Morel Hfo model) uses default values for specific surface area and site density — not values specific to Kumamoto’s sediment. So we make no claim about exact concentrations; it is a visualization of the principle that reduction releases arsenic. Hossain et al. (2016) themselves treat this desorption qualitatively.

Arsenic’s “face” — from arsenate to arsenite

Arsenic does not merely dissolve or not. It changes chemical species (its face) with redox. In oxic water, arsenate As(V) (HAsO₄²⁻, etc.) dominates; in reducing water, arsenite As(III) (H₃AsO₃). Arsenite is generally more mobile and more toxic.

Here it helps to connect the ORP (redox potential Eh) a field meter reads with the pe PHREEQC uses.

pe and ORP (Eh) — a ruler for electrons

pe is a dimensionless measure of how readily electrons are exchanged (pe = −log{e⁻}). High pe = oxidizing, low pe = reducing. If pH is “a ruler for protons (H⁺),” pe is “a ruler for electrons (e⁻).”
The field ORP (Eh, mV) corresponds one-to-one with pe. At 25 °C, \(E_h(\mathrm{mV}) \approx 59.2 \times \mathrm{pe}\) (pe = 1 ≈ +59 mV). PHREEQC computes in pe, the field measures Eh — this equation bridges them.
But a Pt-electrode ORP is a mixed potential of many redox couples that are not necessarily in mutual equilibrium. Read a measured ORP as a guide, not an exact pe.

Fixing pH = 8 and sweeping pe from oxidizing to reducing gives Figure 3.

Figure 3: Fractions of reduced arsenic and iron species (y-axis = reduced fraction) versus pe (bottom axis) and Eh (top axis). Fe(II) becomes dominant at pe ≈ +1.7, and As(III) (arsenite) at pe ≈ −1 (Eh ≈ −59 mV). Iron is reduced first, then arsenic shifts to arsenite.

Two points stand out.

Iron first, arsenic second. Fe(III)→Fe(II) crosses over at pe ≈ +1.7, As(V)→As(III) at pe ≈ −1 — consistent with the sequence order (iron reduction before sulfate reduction) in Figure 1.
The more reducing, the more arsenite As(III). This reproduces, from public data, the paper’s Eh–pH picture of “arsenite under reducing conditions.”

The sulfur testimony — what δ³⁴S reveals

Near the bottom rung of the sequence is sulfate reduction. Hossain et al. (2016) found the sulfur isotope ratio δ³⁴S_SO₄ spanning 8.3–57.6‰. This wide range is evidence of isotopic fractionation: microbes reducing sulfate preferentially use the lighter sulfur (³²S), enriching the residual sulfate in heavy ³⁴S. In other words, sulfur is undergoing redox cycling in this groundwater.

Moreover, if the sulfide produced meets dissolved Fe(II), arsenic could be co-precipitated as a sulfide and thereby removed from the water. But the paper reports no direct evidence of this co-precipitation and leaves it as a suggestion. We take the same stance — no further than “it may occur.”

Sulfur redox in PHREEQC too

The oxidation of sulfide (pyrite → sulfate, acid water) and its reverse are covered in PHREEQC from scratch #6: pyrite oxidation and AMD — the backstage of the sulfur cycling that δ³⁴S records.

The three conditions for high arsenic

Pulling it together, the three conditions for high-arsenic groundwater concluded by Hossain et al. (2016) now make mechanistic sense.

High pH — lowers the positive charge on iron-oxide surfaces, easing desorption of arsenic (an anion) — shared with fluoride in #12.
Reducing (anoxic) — the redox sequence reaches the iron-reduction rung, dissolving arsenic along with the iron oxide.
Stagnation in recently deposited sediments — geologically young (recently deposited) layers such as the Ariake clay have not been well flushed by groundwater since burial, still holding their original arsenic adsorbed on iron oxide; there the groundwater stagnates and reacts over a long residence time.

Note that “young” here means the geological age of the sediment (recently deposited), which is distinct from the long residence time of the water within it. Conditions ① and ② are properties of water that has taken its time; ③ is the vessel — a young layer not yet flushed. All overlap where groundwater lingers. Like fluoride, arsenic concentration is a record of how long the groundwater has travelled — and how far reduction has gone.

Summary

In parts of Kumamoto, arsenic exceeds the drinking-water limit through natural reduction, not pollution.
The mechanism is the redox sequence: organic-matter decay consumes acceptors in the order O₂ → NO₃ → Mn → Fe → SO₄, and arsenic is released where iron oxide dissolves. Hence high As = low NO₃.
Arsenic changes face with pe (≈ Eh), from arsenate to arsenite, the more mobile arsenite dominating under reduction.
High-arsenic water clusters where the three conditions — high pH, reducing, and stagnation in poorly-flushed young sediments — overlap.
The order, the speciation, and the release can all be reproduced in PHREEQC.

Not poison mixed in, but arsenic moved by time and reduction. Like fluoride (#12), it is a phenomenon we can read with data and geochemistry.

Next — #14 Reading groundwater “age” with isotopes

The “long-residence stagnant zone” that recurred in #12 and #13 — how do we measure its age? Next, we read recharge sources with stable isotopes (δ¹⁸O, δD) and the meteoric water line, and groundwater age with tritium and ⁸⁵Kr, using data from Kumamoto and Kirishima.

References

Hossain, S., Hosono, T., Ide, K., Matsunaga, M., Shimada, J. (2016) Redox processes and occurrence of arsenic in a volcanic aquifer system of Kumamoto Area, Japan. Environmental Earth Sciences, 75:740.
Appelo, C.A.J. & Postma, D. (2005) Geochemistry, Groundwater and Pollution, 2nd ed. Balkema.
Smedley, P.L. & Kinniburgh, D.G. (2002) A review of the source, behaviour and distribution of arsenic in natural waters. Applied Geochemistry, 17, 517–568.
Dzombak, D.A. & Morel, F.M.M. (1990) Surface Complexation Modeling: Hydrous Ferric Oxide. Wiley.
Parkhurst, D.L. & Appelo, C.A.J. (2013) Description of input and examples for PHREEQC version 3. U.S. Geological Survey Techniques and Methods, book 6, chap. A43.