Introduction: tap water has a limit for a reason
Drinking water carries quality standards that define what is safe to drink. Fluoride (F⁻) is one of them: Japan’s limit is 0.8 mg/L (the WHO guideline is 1.5 mg/L). A little fluoride prevents tooth decay, but chronic excess causes dental and skeletal fluorosis.
What is a little surprising is that this fluoride can exceed the limit naturally — not from industrial pollution. Groundwater in the western part of the Kumamoto area is one example: inside a volcanic aquifer, the water itself concentrates fluoride as it reacts with rock.
Following Hossain et al. (2016), this article reads “why fluoride builds up in natural groundwater” in the language of geochemistry, and reproduces the key reactions in PHREEQC. It is the second chapter of a story begun in #11, where rainwater dissolved calcite into a Ca-HCO₃ recharge water. What happens next, as that water travels on?
This article is about the inorganic fluoride ion F⁻. It is chemically distinct from PFAS (per- and polyfluoroalkyl substances), which have entirely different sources and behavior. PFAS deserves — and will get — its own separate article.
Continuing from #11: as Ca-HCO₃ recharge water travels on
In #11, rainwater took up soil CO₂, became a weak acid, and dissolved minerals into a Ca-HCO₃ groundwater — the “entry water” that almost all groundwater passes through first.
Kumamoto’s aquifer is a volcanic one, built mainly of Aso pyroclastic-flow deposits: highly permeable, with active flow. The Ca-HCO₃ water born in the recharge zone undergoes two more major changes as it flows down-gradient:
- Cation exchange on clay minerals gradually replaces Ca with Na.
- Silicate weathering of volcanic minerals raises pH and HCO₃⁻.
Where these two overlap, a setting forms in which fluoride becomes mobile. Let us take them in turn.
Hossain et al. (2016) analyzed 50 groundwater samples here. Fluoride ranged from 0.1 to 1.57 mg/L, and 58% of the first (shallow) aquifer and 26% of the second (deep) aquifer exceeded the 0.8 mg/L limit. The high-fluoride waters shared a clear signature — Na-HCO₃ type, high pH (7.05–9.45), high HCO₃, low Ca, high Na/Ca — and clustered in long-residence stagnant zones (groundwater age over 55 years).
Process ① — cation exchange toward Na-HCO₃ water
At the heart of the high-fluoride signature is low Ca, high Na (a high Na/Ca ratio), and this is produced by cation exchange.
Clay-mineral surfaces are negatively charged and loosely hold cations. When Ca-HCO₃ groundwater contacts them, Ca²⁺ in the water is captured by the clay and Na⁺ is released in its place.
\[\mathrm{Ca^{2+} + 2Na\text{-}X \;\rightarrow\; Ca\text{-}X_2 + 2Na^{+}}\]
(X is an exchange site.) The water loses Ca and gains Na, shifting from Ca-HCO₃ to Na-HCO₃ type. Hossain et al. (2016) found the Na/Ca ratio rising from 0.4 to 27 along this path; Schoeller’s chloro-alkaline indices (CAI) confirm the exchange.
Following it in PHREEQC
Starting from the Ca-HCO₃ recharge water of #11, we react it with a Na-form exchanger (clay) and vary the exchanger amount to advance the reaction.
SOLUTION 1 Recharge Ca-HCO3 water (≈ #11 calcite-equilibrium water)
temp 20
pH 7.3
units mmol/kgw
Ca 1.4
Alkalinity 2.8 as HCO3
Na 0.3
Cl 0.3
F 0.02
END
USE solution 1
EXCHANGE 1
NaX 0.0010 # pure Na-exchanger; vary 0.0005–0.0025 mol to advance reaction
ENDThe result is Figure 1. The horizontal axis is the amount of Ca²⁺ removed by exchange (i.e., the reaction progress).
Two things stand out.
- Na/Ca rises from 0.2 to 11 — Ca-HCO₃ turning into Na-HCO₃ (the same behavior as the paper’s 0.4 → 27; more exchanger reaches 27).
- pH stays nearly constant (≈ 7.3). This matters: pure cation exchange changes the water’s type but does not raise pH. So where does the high pH of high-fluoride water (7–9.4) come from? From the next lead actor — silicate weathering.
In other words, even with the same fluoride source, the more exchange proceeds in stagnant zones, the higher Na/Ca climbs and the lower Ca falls. The very setting that lets fluoride move is prepared on the water’s side.
Process ② — high pH and high HCO₃ mobilize fluoride
Alongside the Na-HCO₃ shift, weathering of volcanic silicates (plagioclase, pyroxene, biotite, …) proceeds. This consumes hydroxide, raising pH and supplying HCO₃⁻. The resulting high-pH, high-HCO₃, low-Ca environment moves fluoride by three routes.
(1) Release from mica (F–OH exchange). Sheet silicates like biotite hold fluoride at OH sites in their structure. At high pH (high OH⁻), this F swaps out with OH into the water.
\[\mathrm{Mica\text{-}F_2 + 2OH^- \;\rightarrow\; Mica\text{-}(OH)_2 + 2F^-}\]
(2) Dissolution of apatite. Fluorapatite Ca₅(PO₄)₃F dissolves in CO₂-bearing water, releasing fluoride and phosphate.
\[\mathrm{Ca_5(PO_4)_3F + 6CO_2 + 6H_2O \;\rightarrow\; 5Ca^{2+} + 3H_2PO_4^- + F^- + 6HCO_3^-}\]
(3) Desorption from metal oxides. Fluoride adsorbed on Fe/Al oxide–hydroxide surfaces is pushed off as high pH lowers the surface positive charge, high HCO₃ and PO₄ compete for sites, and high Na/Ca lowers the surface charge density.
Common to all three: the very properties of “Na-HCO₃-shifted” water — low Ca, high pH, high HCO₃ — are the key that mobilizes fluoride. Process ① sets the stage for process ②.
Route (3) is described only qualitatively even in Hossain et al. (2016). PHREEQC’s SURFACE (Dzombak–Morel model) can demonstrate the principle, but site-specific parameters (specific surface area, site density) are unavailable — so here we present the concept, without numerical claims.
The mineral brake — does fluorite hold fluoride back?
If fluoride only ever increased, water would become endlessly high in fluoride. In reality there is a ceiling, and the “brake” is fluorite CaF₂.
\[\mathrm{Ca^{2+} + 2F^- \;\rightleftharpoons\; CaF_2\ (\text{fluorite})}\]
If the water is supersaturated with fluorite, fluorite precipitates and fluoride is capped. If undersaturated, fluorite only dissolves and sets no ceiling. Here low Ca is decisive: with Ca lowered by the Na-HCO₃ shift, the product Ca × F² is small, fluorite stays undersaturated — the brake never engages, and fluoride is allowed to reach high levels.
Checking saturation indices against the paper
We put the Table 1 aquifer-average compositions into PHREEQC (WATEQ4F database) and compute saturation indices (SI) and fluoride speciation to compare with the reported values.
SOLUTION 1 Kumamoto 1st (shallow) aquifer average — Hossain et al. 2016, Table 1
temp 19.4; pH 7.83; pe 0.6
units mg/L
Na 95.1; K 5.8; Ca 12.4; Mg 3.9; Cl 101.2
S(6) 19.1 as SO4; N(5) 4.8 as NO3; Alkalinity 157.7 as HCO3
F 0.83; P 7.0 as PO4; Si 48.6 as SiO2
SELECTED_OUTPUT
-saturation_indices Calcite Fluorite Fluorapatite
-molalities F- MgF+ CaF+ NaF
ENDThe result is Figure 2.
| Mineral | PHREEQC (this study) | Paper | Meaning |
|---|---|---|---|
| Calcite CaCO₃ | −0.42 | −0.69 / −0.57 | near equilibrium (legacy of #11) |
| Fluorite CaF₂ | −1.81 / −2.01 | −2.19 / −2.26 | undersaturated = does not cap fluoride |
| Fluorapatite | +4.82 | +2.42 / +1.77 | supersaturated = viable source |
(first aquifer / second aquifer; calcite nearly identical)
Calcite and fluorite are undersaturated and match the paper well, reproducing the conclusion that fluorite (undersaturated) does not cap fluoride. Fluoride speciation is 98–99% F⁻, matching the paper’s “96–100%.” Fluorapatite is supersaturated in both (sign agrees), though our magnitude is larger — because the log K for fluorapatite differs substantially between database versions. The conclusion (“supersaturated, a viable source”) is unchanged.
When does fluorite actually brake?
Under what conditions does fluorite really cap fluoride? We test it by adding Ca step by step to a low-Ca Na-HCO₃ water.
Figure 3 makes the point plainly. In the low-Ca range where real high-fluoride waters live, fluorite is at SI ≈ −1.5 and puts no brake on fluoride at all. For fluorite to saturate and begin capping fluoride, Ca must climb toward 12 mmol/L (≈ 480 mg/L) — but Na-HCO₃-shifted water never gets there. So fluoride keeps accumulating.
\[\mathrm{CaF_2 + 2HCO_3^- \;\rightarrow\; CaCO_3 + 2F^- + H_2O + CO_2}\]
High HCO₃ in fact pushes the other way, favoring fluorite dissolution (fluoride release).
Why it clusters in stagnant zones
Every reaction above takes time. Cation exchange, silicate weathering, fluoride mobilization — all proceed the longer groundwater is in contact with rock. So high-fluoride water clusters not in fast recharge zones but in stagnant zones — the deep, sluggish areas of the plain, over 55 years old.
This is an extreme, concrete case of the principle from #11: water writes its history into its chemistry. Fluoride concentration is, in a sense, a record of how long the groundwater has travelled. This link between “time” and water quality is explored further in later articles (isotopes, residence time).
Summary
- In parts of Kumamoto, fluoride exceeds the drinking-water limit through natural water–rock reaction, not human pollution.
- The key is two-staged: ① cation exchange turns Ca-HCO₃ into low-Ca Na-HCO₃ water (without raising pH), and ② silicate weathering creates high pH and high HCO₃ that mobilize fluoride from mica, apatite, and metal oxides.
- Because Ca is low, fluorite stays undersaturated and never brakes fluoride.
- Since these reactions take time, high-fluoride water clusters in long-residence stagnant zones.
- And the whole sequence can be followed quantitatively in PHREEQC — SI comparison, the exchange path, and the fluorite control.
Even naturally, limits can be exceeded — but not chaotically. It is a necessity we can read with data and geochemistry.
Starting from the “metallic (iron) smell” mentioned in #11, the next article enters the world where redox (oxidation–reduction) governs groundwater chemistry. In the same Kumamoto area, Hossain et al. (2016, Environmental Earth Sciences) revealed arsenic mobilization — occurring, in fact, on a stage much like #12’s fluoride: high pH, reducing, stagnant.
References
- Hossain, S., Hosono, T., Yang, H., Shimada, J. (2016) Geochemical Processes Controlling Fluoride Enrichment in Groundwater at the Western Part of Kumamoto Area, Japan. Water, Air, & Soil Pollution, 227(10), 385.
- 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.
- Appelo, C.A.J. & Postma, D. (2005) Geochemistry, Groundwater and Pollution, 2nd ed. Balkema.
- Schoeller, H. (1967) Qualitative evaluation of groundwater resources. In Methods and Techniques of Groundwater Investigation and Development, UNESCO.
- Edmunds, W.M. & Smedley, P.L. (2013) Fluoride in natural waters. In Essentials of Medical Geology.