You are on page 1of 20

Geochimical Study of Hydrothermal Sources in Analavory, Itasy Region, Madagascar

Gérard SARAZIN 1* and Nelly Ghislaine RAKOTO 2

1 Water Geochemistry Team, Institute of Earth Physics of Paris & Paris-Diderot University, 1 rue de
Jussieu 75238 Paris, France

2 Department of Mineral Chemistry and Physical Chemistry, Environmental Chemistry Laboratory,


Faculty of Sciences, University of Antananarivo, Antananarivo 101, Madagascar

 Correspondence, email : gsarazin@gmail.com

Abstract

Springs from Analavory are geothermal waters which have undergone a complete cooling during their
way towards the surface. Chemical analysis at emergence show that they are, at depth, anoxic,
oversaturated with respect to CO2 and calcium carbonates. Outgassing and oxygen input at the
surface are the main parameters which allow to explain the observed composition. Appling classical
geothermometers lead to an estimation of 150°C in the deep geothermal reservoir.

Keywords : geothermal waters, geothermometers, geysers, carbonates geochemistry, Madagascar.

1. Introduction

The geothermal field of Analavory (Itasy, Miarinarivo) is a well-known site in Madagascar. Unlike the
hot geysers found in Iceland or North America, the carbonated springs of Analavory are « cold
geysers » [1], where the water temperature is nearly equal to the ambient temperature. The
pulsation of these springs is caused by irregular releases of CO2, which can propel the water to
varying heights ranging from 0.3 to 3 meters above the ground.

The aim of this study is :

- To understand the geochemical processes that explain the chemical composition of these waters.

- To assess the potential for geothermal exploitation.

2. Site Presentation

The geographic coordinates of the study area are : 18° 55.562’ S, 46° 38.410’ E, Altitude : 950 m
(Figure 1).

Figure 1 : Study area


The site is located on the Precambrian basement, consisting of gneiss and granite, locally overlain by
basaltic and trachytic rocks that are evidence of volcanic activity in the region between the Cenozoic
and Quaternary periods.

3. Materials and Methods

3-1. Field Measurements and Sample Collection

Due to limited measurement equipment availability, only the following parameters were measured in
situ :

- Temperature (Pt-100 probe) connected to the pH meter.


- The conductivity was measured using a digital « pen » conductometer with automatic
calibration. It provides values within the range of 0 to 1999 µS/cm. Since the measured
conductivities exceeded this range, samples G2 (Figure 2a) and G3 (Figure 2b) were precisely
diluted 5 times with deionized water (conductivity of 1 µS/cm) to obtain measurements
within the device’s scale.
- The pH was measured with a pH meter (VWR-pH110™) and a combined electrode, with a
precision of ± 0.02 units. The pH meter was calibrated using two buffer solutions (pH 4.01
and pH 7.00).

Figure 2 : Samples G2 and G3 collected in the field.


The G2 sample is from a source located in a natural funnel (Figures 2a and 2b), while G3 was
collected from a runoff channel about ten meters from the emergence of G2. The collected samples,
G2 and G3, were each stored in two completely filled Falcon™ tubes of 50 mL without any gas
bubbles. Additionally, groundwater was sampled from a well located at coordinates 18° 55.836’ S, 46°
38.746’ E, on the edge of the Andranomandroatra trail, approximately 780 meters southeast of the
geothermal springs. The temperature, conductivity, and pH were measured immediately after sample
collection, and the samples were stored in two Falcon tubes of 50 mL.

3-2. Laboratory Measurements

Upon returning to the university, it was observed that the tubes containing the samples had formed a
light orange precipitate. The « well » sample showed no visual changes. Subsequent analyses
conducted at the Laboratory of Water Geochemistry (LGE) at the Institute of Earth Physics of Paris
(IPGP) were performed as follows :

- For total concentrations of dissolved elements : a known volume of the two samples was acidified
with a known micro-volume of suprapure 60% HNO3 to completely dissolve the formed precipitate.

- For the concentrations of elements remaining in solution, the non-acidified samples were filtered
using Minisart™ membranes with a porosity of 0.2 µm, resulting in a filtrate for each sample : G’2 and
G’3.

3-2-1. Major Cation Analysis

The analysis of major cations : Na, K, Fe, Li, and boron concentration (B[III]) was performed using ICP-
AES spectroscopy on a Thermo Optek™ instrument. The sensitivity is 0.1 µM/L, and the precision is ±
5%. The concentrations of dissolved calcium and magnesium were measured using complexometric
titration in a basic medium (EDTA 10-2 mol/L with the colored indicators NET and HHSNN).

3-2-2. Major Anion Analysis

The major anions were measured using ion chromatography on a Dionex™ instrument. The sensitivity
is on the order of 1 µmol/L, and the precision is ± 5%. The chloride analysis was performed using both
ion chromatography and direct potentiometry (except for the « well » sample). The precision is ± 1%,
and the sensitivity is on the order of 200 µmol/L.

3-2-3. Alkalinity Measurement

Alkalinity was measured using the Gran titration method [3]. The measurements were conducted on
samples diluted 20 times, from which 20 mL were taken and titrated with 0.01 M HCl using a
Metrohm™ electronic burette. The precision is ± 0.5%.

3-2-4. Additional Analyses

Ammonium, dissolved inorganic phosphorus (SRP), and dissolved silica were measured using
conventional colorimetric methods [4] on a spectrophotometer (Thermo Optek Evolution 600™). The
sensitivities and precisions are provided in Table 1 below :

Table 1 : Sensitivity and precision of colorimetric measurements


3-3. Solid Analysis

The precipitation of solids (travertine) from the cold geysers’ waters in Analavory is clearly visible in
the field and covers the local rocks near the springs. A few tens of grams were brought to the LGE,
dried in an oven at 80 °C, and then ground. Observations were made using a scanning electron
microscope (SEM) coupled with a secondary X-ray detector (EDAX), and analyses were performed
using X-ray diffraction (XRD) and X-ray fluorescence (XRF). The solid analysis was also conducted
through wet chemistry by dissolving a known mass of the sample in a known volume of concentrated
nitric acid (Suprapur™ Merck at 60%). The same procedure as for water analysis was then followed.

3-3-1. Quantification of precipitated solids

The quantity of solids precipitated during the evolution of samples G2 and G3 into G’2 and G’3 was
measured using a known volume of suspension in G’2 and G’3. After homogenizing the suspension, a
2 mL aliquot was quickly taken and filtered through a 0.2 µm pore size membrane. The membrane,
whose initial mass is known, was then dried in an oven at 60°C for 24 hours and weighed to obtain
the solid mass. Weighings were performed using a Mettler™ MX-5 microbalance with a precision of ±
2 µg.

4. Results

4-1. Study of solids

4-1-1. Direct analysis of solids

The analyses were conducted at the Laboratory of Materials Chemistry and Catalysis (University of
Paris-Diderot). XRD : A carbonate phase consisting of 54% magnesium calcite and 42% aragonite was
identified. Iron and silicon detected by XRF did not show identifiable mineral phases in XRD,
indicating that these elements are present in amorphous phases. XRF : The semi-quantitative analysis
yielded the following mass percentages : Ca = 81 ; Mg = 0.5 ; Fe = 14 ; Si = 3 ; along with trace
amounts of Na, Sr, and Cl. Considering only the carbonate phase, excluding iron and silicon which
total 17%, calcium represents 81 x 1.17 = 94.8% of the solid. Therefore, the magnesium calcite has
the formula : Ca0.95Mg0.05CO3.

4-1-2. Wet chemical analysis of solids

A known quantity of solid, 0.5029 g, was dissolved in concentrated acid (60% suprapur Merck HNO3).
The dissolution in a few mL of acid left no solid residue, confirming a composition corresponding to a
mixture of carbonates (magnesium calcite and aragonite) and iron oxy-hydroxides. These components
are soluble in acid and give the solid precipitated by the thermal water a red-orange color. The
absence of residue after the dissolution indicates the absence of silicates (no detection by XRD). The
detected silica by XRF is amorphous. The resulting solution was diluted to an exact volume of 50 mL
with ultrapure water (MilliQ™). It was used for the analysis of calcium and magnesium
(complexometry) as well as iron (colorimetry). The results allowed establishing the formula for
magnesium calcite as : Ca0.89Mg0.11CO3.

The observed difference in the two formulas can be explained mainly for two reasons :

1. XRF has a precision of ± 2%, and this precision decreases for lighter elements.

2. The wet chemical analysis method cannot distinguish between calcium and magnesium elements
originating from the dissolution of magnesium calcite and aragonite.

The wet chemical measurement of iron (colorimetry) performed in duplicate gives a concentration in
the solid ranging from 12.5% to 13.6% by weight, a value comparable to that obtained by XRF
analysis.

4-2. Scanning Electron Microscope (SEM) – Energy Dispersive X-ray Analysis (EDAX) Observation

The observations were carried out at the Cryptogamy laboratory of the National Museum of Natural
History (MNHN). Image (1) is a photograph of the particles after grinding the solid, providing a scale
for observation. The following images (2 to 7), in false colors, visualize the distribution of elements :
Ca, Mg, O, Si, Fe, and K (Figures 3 to 6). It is difficult to detect positional correlations of the elements
in the solid due to the dominance of calcium. However, the association of calcium with oxygen and
magnesium is undeniable, confirming the presence of magnesium carbonate(s).

Figure 3

Figure 4
Figure 5

Figure 6 : Images MEB-EDAX

However, iron appears to be concentrated in a few micrometer-sized grains, which would confirm the
presence of a distinct phase of oxy-hydroxides. Silicon does not show any positional correlation with
the other elements. It is likely localized in amorphous silica precipitated by the thermal water at the
spring emergence.

4-2-1. Quantity of precipitated solids from samples G2 and G3

 Sample G2

From 2 mL of homogeneous suspension, 1.641 mg of solid is obtained after filtration. Assuming a


molar mass equal to that of calcite (CaCO3 ; M = 100 g/mol), the sample can precipitate 8.20 mmol/L
of carbonates.

 Sample G3

For an identical volume of 2 mL of suspension, a quantity of 1.675 mg of solid is obtained. Using the
same calculation as before, it can be deduced that G3 can precipitate 8.37 mmol/L of carbonates.

4-3. Analyses of G2, G’2, G3, and G’3

The analytical results are shown in Table 2. Conductivity is in µS/cm and concentrations are in µmol/L.
Table 2 : Temperature (T) is in °C, conductivity (Cond.) is in µS/cm. Δ% is the deviation from zero (in
percentage) of the ion balance. Concentrations are in µmol/L. The « well » sample contains fluoride
(F-) at a concentration of 14.4 µmol/L and dissolved organic carbon (DOC) at a concentration of 303
µmol/L. These two species were not detected in samples G and G’. Ammonium was not detected in
any of the samples.

5. Discussion

5-1. Geochemical evolution of the waters

The examination of the analysis table (Table 2) clearly shows that samples G2 and G3 differ mainly in
the following measured parameters : pH, conductivity, and alkalinity. When comparing samples G to
the « re-equilibrated » samples G’, significant differences can be observed in :

- pH

- Conductivity

- Alkalinity

- Concentrations of dissolved calcium, magnesium, and iron

5-2. Variations in conductivity

Sample G2 was directly collected from a spring (Figure 2), while G3 was collected from a natural
channel formed from the G2 source, located about ten meters downstream. Despite the short
distance traveled on the surface, the water has enough time to lose CO2, which can explain both the
decrease in conductivity and the increase in pH. In fact, dissolved CO2 converts to the gaseous state
following the reaction chain :

HCO3- + H+ ⇌ H2CO3 ⇌ CO2 (gas) + H2O (1)

The theoretical conductivity can be calculated using Kohlrausch’s law :

Σ = ∑ λ0i ⋅ zi ⋅ Ci (2)

Where :

Λ0i : ionic molar conductivity in Siemens per meter squared per equivalent : S.m2.eq-1 (1 eq = 103
mol/m3 x charge of the ion)

Ci : concentration in mol.m-3

Zi : absolute value of the ionic charge (unitless)


Σ is expressed in S.m-1. Kohlrausch’s relation applies only to very dilute solutions. For more
concentrated solutions, as is the case here, an empirical correction is applied to account for the ionic
strength. The modified Kohlrausch’s relation becomes :

Σ = ∑ ci ⋅ zi ⋅ (1 – I) (3)

Where I is the ionic strength, and σ x 1000 is the ionic conductivity at 25 °C expressed in µS/cm.

The measured and calculated conductivities are compared in the table below :

Table 3 : Measured and Calculated Conductivities

The conductivities calculated at 25°C have been corrected for temperature using the equation :

Σ(t) = σ(25)⋅[1 + 0.02⋅(t – 25)] [6] (4a)

Where t is the temperature in °C.

It can be observed that the measured and calculated conductivities deviate within a very small range,
even for water G2 which remains within the margin of uncertainty for this type of calculation with a
difference below 10%. We can now question the processes that could explain these variations in
conductivity. If degassing of the water is the only process, Equation (1) shows that we must verify the
equality :

Δ[H+] = Δ[HCO3-] (4b)

The relative variation of the H+ ion concentration can be calculated using the equation :

Δ(H+) = (H+) (ln 10 ⋅ ΔpH) (5)

For G2/G’2, the above relation allows us to calculate Δ(H+) = 1.8 µmol/L. Equation 3, with λi
(corrected for ionic strength) = 254.3 µS.cm<sup>2</sup>, gives a conductivity variation of 4.4 x
10<sup>-4</sup> µS/cm. Equality 4b, with the same calculation, allows us to evaluate the
conductivity variation related to the neutralization of HCO3-. Using λi = 29.5 µS.cm<sup>2</sup>, we
obtain 5.3 x 10<sup>-5</sup> µS/cm. The total decrease in conductivity due to CO2 degassing is
therefore 5 x 10<sup>-4</sup> µS/cm. A similar calculation for G3/G’3 yields a conductivity decrease
of 2 x 10<sup>-4</sup> µS/cm. From these calculations, we can conclude that water degassing
cannot explain the observed conductivity variations, which are 2170 and 1345 µS/cm for G2 and G3,
respectively. It is therefore much more likely that the precipitation of carbonates at the G2
emergence and during runoff (G3), removing Ca2+, Mg2+, Fe2+, and CO32- ions from the water, is
responsible for these conductivity variations.

5-3. Alkalinity Variations and Precipitation of Carbonate Phases

The precipitation of carbonates and ferric oxyhydroxides is evident when observing the newly formed
mineral deposits (travertine) on rocks (granite and basalt) in the immediate vicinity of the
carbonated-gaseous springs.

The attack of travertine with concentrated strong acid (30% HCl) leaves no visible solid residue.
Therefore, it can be affirmed that it consists of a mixture of calcium and magnesium carbonates, as
well as ferric oxyhydroxides, as identified in the solid study. The precipitation processes of these
solids can be simply described by the following reactions :

CO2 (dissolved) + H2O + Ca2+ ⇌ CaCO3 (solid) + 2 H+ (6)

The calcium carbonate formed can be calcite and/or aragonite, while the magnesium present in the
water does not form its own mineral phase but instead partially substitutes for the Ca2+ ion in the
rhombohedral lattice of calcite, forming magnesian calcite. The other process involves dissolved iron.
The hydrothermal waters, which we will demonstrate later to be undoubtedly hydrothermal in
nature, contain dissolved iron in the form of ferrous ion Fe2+ or soluble complexes of this species
(Fe(OH)+, Fe(OH)2, FeSO4). Indeed, hydrothermal waters, due to their deep origin, are anoxic. Under
these conditions, the leaching of host rocks can release soluble Fe(II) from ferromagnesian minerals
(olivine, pyroxenes, amphiboles, micas, etc.). This dissolution is faster at higher temperatures. At the
emergence point, upon contact with oxygen from the air, the oxidation of soluble Fe(II) to insoluble
Fe(III) is immediate :

Fe2+ + ¼ O2 (gas) + 5/2 H2O ⇌ Fe(OH)3 (solid) + 2 H+ (7)

This oxidation leads to the appearance of orange-colored amorphous phases, which are undetectable
by X-ray diffraction. Reactions (6) and (7) are responsible, a priori, for significant variations in the
chemical composition of the water from its initial state (G2 and G3) to its final state (G’2 and G’3)
after CO2 degassing and solid precipitation. The major modifications involve :

- pH

- Conductivity (5.1.2)

- Alkalinity variation and concentrations of Ca2+, Mg2+, and Fe2+.

5-3-1. pH Variation

The table below summarizes the measured pH variations :


Table 4 : (1) Samples equilibrated with the atmosphere and in equilibrium with the precipitate of
carbonates and Fe(III) oxyhydroxides formed upon contact with air within a few hours

The observed pH increases correspond to a decrease in the concentration of the H+ ion by 3.4 x 10-7
and 1.4 x 10-7 mol/L for G2 and G3, respectively. This apparent contradiction with Reactions (6) and
(7), which indicate an increase in H+ concentration and a decrease in pH, is only apparent. This is
because Reactions (6) and (7) occur simultaneously with Reaction (1), which partially removes H2CO3
from the solution, logically leading to an increase in pH.

Therefore, we must assume that out of these two opposing processes, the loss of CO2 is the faster
process compared to the precipitation of solids. However, the precipitation is facilitated by the
degassing of CO2, as the consumption of protons by Reaction (1) shifts Reactions (6) and (7) to the
right. Using Equation (3) and Table 3, we can calculate the conductivity variations attributed to pH
changes when transitioning from the compositions of G2 and G3 to G’2 and G’3. These variations,
considering the always very low concentration of the H+ ion and despite its high limiting conductivity,
are negligible (see 5.1.1).

5-3-2. Variations in Calcium and Magnesium Ion Concentrations and their Impact on Alkalinity

It is worth revisiting the definitions of water alkalinity [7-9]. Alkalinity can be defined either by the
« inactive » species, which are the conjugate acids and bases of strong acids and bases, as follows :

Alc = [Na+] + [K+] + 2[Ca2+] + 2[Mg2+] – ([Cl-] + [NO3-] + 2[SO42-]) (8)

Or by the « active » acid-base species present :

Alc = [HCO3-] + 2[CO32-] + [OH-] – [H+] (9)

This is often written as :

Alc + [H+] = [HCO3-] + 2[CO32-] + [OH-] (10)

For waters where the dominant acid-base system is the « carbonate system, » which includes the
species H2CO3, HCO3-, and CO32-, the precipitation of carbonates (magnesian calcite and aragonite)
will result in a decrease in alkalinity given by :

ΔAlc = 2Δ([Ca2+] + [Mg2+]) (11)

Similarly, it can be shown that for the precipitation of Fe(III) oxyhydroxide, the variation in alkalinity is
given by Reaction (7) :
ΔAlc = 2Δ[Fe2+] (12)

However, the removal (or input) of CO2 does not change the alkalinity of water. In fact, Reaction (1)
shows that the same amount of HCO3- (counted positively) is consumed as H+ (counted negatively) in
Equation (8). For most natural waters, which typically have a pH ranging from 6 to 8.5, the inequality
[HCO3-] >> [CO32-] holds true, and alkalinity is often simplified to the expression : Alc = [HCO3-].
Considering the previous remarks, we can use Table 2 to verify if the relationships in Equations (11)
and (12) are experimentally satisfied. This validation will confirm the previous hypotheses regarding
the evolution of waters G2 and G3 into G’2 and G’3. We can quantify the processes of solid
precipitation by performing balances of dissolved calcium, magnesium, and iron in the acidified
samples after sampling. In these samples, the total concentrations of these three elements are
measured. The same analysis is conducted on samples where the water is allowed to precipitate
solids after oxidation and degassing. Alkalinity is measured under both conditions. Equations (12) and
(13) can be verified through the following calculation :

- Calculate ΔAlc using alkalinity measurements.

- Determine the variation : Δ([Ca2+]+[Mg2+]) that should be observed.

On the other hand, from these measurements, we can deduce the quantity of [Ca,Mg] carbonates
precipitated and compare the calculated result to the measurements performed using a
microbalance. It is known that for G’2, 1.641 mg of precipitate was filtered in 2 mL of suspension, and
for G’3, 1.675 mg of precipitate was filtered in 2 mL of suspension. Assuming an average molar mass
of 100 g/mol for carbonate, which can be approximated as calcite considering the presence of
aragonite and the low magnesium content of magnesium calcite, we can calculate :

- For G2, precipitation of 8.20 mmol/L of CaCO3

- For G3, precipitation of 8.37 mmol/L of CaCO3

Calculation of the quantity of precipitated carbonates :

 Source G2 :

Alc1 = 59857 µmol/L (total alkalinity measurement before precipitation)

Alc2 = 44053 µmol/L (alkalinity measurement at equilibrium with the precipitate)

The measured ΔAlc = 15804 µmol/L

This decrease in alkalinity should result in a decrease in the concentrations of dissolved Ca and Mg,
theoretically equal to :

Δ([Ca2+]+[Mg]) = ΔAlc/2

Thus, Δ([Ca2+]+[Mg]) calculated = 7902 µmol/L

Therefore, according to this calculation, the water from G2 can precipitate 7.9 mmol/L of solid
carbonates, which corresponds to 0.790 g/L of solid (with a molar mass of 100 g/mol). For a volume
of 2 mL, this corresponds to a solid mass of 1.580 mg. The calculated value differs from the measured
value by less than 4%. Considering experimental uncertainties, this calculation demonstrates that the
measured variation in alkalinity is indeed due to the precipitation of alkaline earth carbonates when
the water comes into contact with the atmosphere. A similar calculation can be done regarding the
alkalinity variation related to Reaction (6), which leads to Equation (12). The total concentration of
dissolved iron is 227 µmol/L. After oxidation and precipitation of the carbonate-iron hydroxide
mixture, the concentration is below the detection limit (<1 µmol/L). The calculated alkalinity variation
using Equation (12) is therefore 552 µmol/L, representing only 3.5% of the ΔAlc value.

 Sample G3 :

A similar calculation yields :

ΔAlc = 14830 µmol/L and Δ([Ca2+]+[Mg]) calculated = 7415 µmol/L

Thus, with a molar mass of 100 g/mol, the theoretically precipitated solid is 0.742 g/L, which
corresponds to 7.42 mmol/L or 1.484 mg for 2 mL. The measured quantity of solid differs from the
calculated quantity by 13%, which is an acceptable deviation and does not contradict the initial
hypothesis attributing the variations in water alkalinity to the precipitation of solids. Regarding the
ΔAlc variation related to the precipitation of iron (III) oxyhydroxides, it is 524 µmol/L, representing
3.5% of the ΔAlc value for this sample.

5-4. Thermodynamic Equilibrium or Non-Equilibrium of the System ?

It is now legitimate to ask whether the thermodynamic system represented by the Analavory geysers
is in equilibrium with the atmosphere and the solids it precipitates. To answer this question, we used
the Visualminteq™ software. This program calculates the speciation of all chemical species, i.e., the
concentration of all individual ions as well as the complexes formed by the individual ions and
available ligands, based on the analytical data entered into the program. The temperature can be set,
and the pH is entered as data. It is necessary to ensure an ion balance such that the difference
between the anionic and cationic charges is less than 15%. This condition is met for G2 and G3, as the
deviations from electro-neutrality are respectively 0.3% and 0.0% (Table 2). For the two samples that
were allowed to equilibrate with the atmosphere, G’2 and G’3, the ion balances are 12% and 1.5%
respectively. The program also calculates the saturation coefficients of water with respect to
atmospheric gases that can be imposed, as well as the saturation coefficients of water with respect to
the solids that can potentially precipitate based on the dissolved species entered into the calculation
program. For gases, the saturation coefficient with respect to atmospheric CO2 is calculated as
follows :

Ω CO =

P CO 2 ( eau) / p CO 2 ( atm)

Where pCO2(atm) is the current « normal » partial pressure of carbon dioxide in the atmosphere,
which is 4 x 10-4 atm (equivalent to exactly 401.52 ppm in March 2015-[10]). pCO2(eau) is calculated
using the equation :

P CO 2 ( eau) =

(H 2 CO 3 ) / K H

Where (H2CO3) is the activity of carbonic acid (calculated by the program) and KH is the Henry’s
constant.

Where (H2CO3) is the activity of carbonic acid (calculated by the program) and KH is the Henry’s
constant value at the specified temperature. For a mineral of type An Bm, the saturation coefficient is
given by :

( A )n ⋅ (B )m

Ω =
N m K

Where (A) and (B) are the activities of species A and B in solution, and KS is the solubility product of
the solid calculated at the user-entered temperature.

The iterative calculation stops when convergence is achieved for the ionic strength value. Activity
coefficients are calculated using the empirical Davies formula. The input data consists of the
analytical results from Table 2. The chosen temperature is 28 °C. Table 4 is a selection of calculations
performed by the Visualminteq software. The calculations for lines G*2 and G*3 were obtained by
forcing the atmosphere/water/mineral system to reach equilibrium with calcite (Ω(CaCO3) = 1), which
allows determining the quantity of this mineral that the water must precipitate to reach this state.
These quantities will be compared with our experimental results when the initial G2 and G3 waters
naturally precipitate the mixture of carbonates/iron(III) oxyhydroxides to form G’2 and G’3.

5-5. Comparison of G2 and G3

At the emergence of the G2 spring, it is observed that CO2 degasses with boiling, which is confirmed
by the calculation as the partial pressure of this gas in water is 1.12 atm (Table 6), a value higher than
atmospheric pressure, and a very high saturation coefficient (ΩCO2 = 2800). The water is
supersaturated with respect to calcite (and also aragonite) with ΩCaCO3 = 12.6.

In the runoff channel (G3), the water has already lost a significant amount of CO2 as the pH has
increased from 6.43 to 6.70, and the partial pressure of CO2 in the water has decreased by a factor of
2.6, resulting in ΩCO2 = 1077. As expected, there is an increase in ΩCaCO3, which changes from 12.6
to 19.5. This increase is due to degassing, which leads to a rise in pH and a simultaneous increase in
the activity of the CO32- ion. Consequently, the precipitation of different calcium carbonates is
thermodynamically possible. These calculations correctly interpret what is observed in the natural
environment and demonstrate that the atmosphere/water/mineral system is far from its equilibrium
state.

5-5-1. The reaction pathway from G2 to G’2 to G*2

The transition from G2 to G’2 can be described as the slow « equilibration » of the sample collected
in the field with an atmosphere where the partial pressure of CO2 is reduced to approximately 4.00 x
10-4 atm, and the temperature is close to 25°C, which is very similar to the field temperature.
Additionally, the water spontaneously deposits a solid phase consisting predominantly of calcium
carbonates (with iron oxy-hydroxides remaining as minor neoformed constituents). As previously
shown with Reaction (6), the precipitation of calcium carbonates should lower the pH by releasing H+
ions and consuming CO32- ions. However, contrary to this expectation, the pH increases from 6.43 to
8.56, while the water’s supersaturation with respect to CaCO3 increases from 12.6 to 495. These
results confirm the intuitive prediction that CO2 degassing is extremely rapid, and it primarily controls
the water’s pH. On the other hand, the precipitation of calcium carbonates is very slow, as indicated
by the sustained high supersaturation : the pH increase enhances the activity of CO32- ions, leading
to an increase in the Ω value from 12.6 to 495. By « forcing » water G’2 to equilibrate with calcite
(G*2) while maintaining the measured pH value (pH = 8.54), equilibrium is reached when 2.58 x 10-2
mol/L of calcite has precipitated. In our experimental measurements, we found that G’2 had
precipitated 8.2 x 10-3 mol/L of calcium carbonates. Considering Reaction (6) again :

CO2 (dissolved) + H2O + Ca2+ ⇌ CaCO3 (solid) + 2 H+


The extent of reaction ξ is equal to the moles of CaCO3 precipitated, which is 8.2 x 10-3 mol/L for G’2.
Knowing that at equilibrium ξeq = 2.58 x 10-2 mol/L, the Reaction (6) is only shifted to the right by
32%.

5-5-2. The reaction pathway from G3 to G’3 to G*3

The same approach applies to G3, although the numerical outcomes may slightly differ. Here as well,
it is the degassing of CO2 that controls the pH as the water transitions from G3 to G’3, resulting in a
significant supersaturation relative to calcite (Ω(CaCO3) = 415). As for the « forced » equilibration
(G*3), it allows for the calculation of a reaction extent of precipitation of 34%.

Table 5 : Calculations performed by Visualminteq using analytical data.

6. Potentialités géothermiques des geysers d’Analavory

6-1. The hydrothermal origin of the springs

The hydrothermal origin of the Analavory springs is a subject of controversy as they are considered
cold geysers, i.e., waters of « phreatic » type that have simply come into contact with carbon dioxide
gas during their underground journey, a relic of past volcanic activity [1] from the Cenozoic era (66
Ma) to the Quaternary period (8,000 years for the lava flow that formed Lake Itasy by damming the
river). However, an examination of the chemical composition of G2 leaves no doubt about the deep
origin of the Analavory springs. A simple comparison with the analysis of phreatic water sampled
from a well near the site shows that (Table 2) :

The ionic concentrations of all major elements are consistently higher in the « geysers. »

The difference is remarkable for dissolved silica, boron, and lithium ; these last two elements, present
in G2 and G3, are below the detection limit in the phreatic water.

This hydrothermal origin can be confirmed by comparing the composition of G2 with two famous
sources in the Antsirabe geothermal field : Ranovisy and Visy Gasy [11] :

Table 6 : Concentrations are in µmol/L. The Andronakely source is a surface water adjacent to the
thermal waters of Antsirabe.
The examination of Table 6 clearly shows that the chemical composition of the Analavory springs
cannot be explained by a simple dissolution of CO2 in phreatic water. It is therefore legitimate to
attempt to estimate the geothermal potential of the Analavory springs using « geothermometers, »
which provide fairly accurate information about the temperature of the deep reservoir.

6-2. Geothermometers

A geothermometer is a tool based solely on the measurement of the chemical composition of


hydrothermal waters. They were developed in the 1970s following the first oil crisis, which prompted
countries to explore alternative energy sources to fossil fuels such as coal, oil, and gas.
Geothermometers have been the subject of numerous articles [12-15]. There are two main
categories :

 Silica geothermometers
 Cation geothermometers

6-2-1. Silica geothermometers

They are all based on the fact that the solubility of silica (SiO2) increases rapidly with temperature.
The presence of silica in all deep rocks of the Earth’s crust, whether magmatic (granites) or volcanic
(basalts), gives these geothermometers universal applicability. Furthermore, the existence of the
geothermal gradient raises the temperature of rocks and water circulating in fractures, resulting in
exponentially increased rates of water-rock interactions. This leads to the hypothesis that at depth,
these interactions are in thermodynamic equilibrium.

Silica geothermometers assume as a basic postulate that the reaction :

SiO2 (solid) + 2 H2O ⇌ H4SiO4 (dissolved) (13)

Is at equilibrium. The concentration of dissolved silica can therefore be directly related to


temperature since the equilibrium constant K(T) = (H4SiO4) varies with temperature according to the
Van ‘t Hoff equation, the integrated form of which is given by :

ΔH ⎛1 1⎞

Log K

= log K

+ T ⋅⎜ − ⎟ (14)

T T0

Ln 1 0 ⋅ R T T

⎝ 0 ⎠

The enthalpy change ΔHT itself being a function of temperature can be evaluated using the equation :

ΔH = ΔH0 + ΔC0 ⋅ (T – T0) (15)

T T0 P 0

All the necessary thermodynamic data for these calculations can be easily found in the tables [5]. The
equilibrium constant of Reaction (13) can be expressed as a function of temperature using an
expression of the form :
Log(H4SiO4) = a/T + b (16)

While quartz is the stable form of silica at high temperatures, there are many allotropes stable at
lower temperatures, the most common being chalcedony (crystallographically identical to quartz
[hexagonal – 1A3] but microcrystalline and therefore more soluble) and amorphous silica. For each
variety, the coefficients a and b in equation (16) have the following values [16] :

Table 7

The conditions that need to be met for Equation (16) to provide a realistic temperature are as
follows :

The ascent of hot water to the surface through a network of fractures must be rapid. If this is not the
case, the equilibrium (13) established at depth is not preserved because the water, on one hand,
cools down, and on the other hand, it has time to equilibrate with the rocks it encounters during its
ascent to the surface. The temperature of the deep reservoir will be underestimated.

The thermal water should not undergo significant mixing with subsurface phreatic waters. This would
dilute the dissolved silica acquired at depth and underestimate the temperature of the geothermal
reservoir.

Figure 7 shows the solubility curves of the three phases of silica as a function of temperature. The
points corresponding to G2 and G3, as well as the point from the phreatic well, are plotted on the
graph.

Figure 7
The G2/G3 waters are in near-equilibrium with amorphous silica at emergence, with saturation
indices of 1.14 and 1.02, respectively. This state is solely due to the cooling of the water at the
surface, as solubility decreases with temperature. Table 8 below provides the calculated deep
temperatures for G2/G3 using Equation (16) and Table 7 :

Table 8 : Estimation of deep temperatures

It is not possible to make assumptions about the mineral in equilibrium with the thermal waters in
the geothermal reservoir. Therefore, we will be satisfied with an estimated temperature range of 120
to 150 °C. These results can be compared to the calculated temperatures for the hottest sources in
the Antsirabe geothermal field [11], which fall within the range of 135-170 °C.

6-2-2. Cation geothermometers

They are based on the existence of water-mineral equilibria at high temperatures in deep geothermal
reservoirs. The cations involved in these equilibria are primarily sodium (Na+) and potassium (K+).
Other, less commonly used geothermometers involve calcium (Ca2+), lithium (Li+), and magnesium
(Mg2+) [16].

The Na-K geothermometer is the most commonly used geothermometer, and its principle is based on
the equilibrium between two minerals that are abundant in granitic rocks : potassium feldspar
KAlSi3O8 and sodium plagioclase (albite) NaAlSi3O8. The equilibrium involves an ion exchange
between the two solids :

NaAlSi3O8 (solid) + K+ (dissolved) ⇌ KAlSi3O8 (solid) + Na+ (dissolved) (17)

The equilibrium constant can be written as :

K(T) = (Na+)/(K+) (18)

The temperature dependence of this equilibrium can be described by the Van’t Hoff equation. The
advantage of this geothermometer compared to the silica geothermometer is that it is insensitive to
dilution because it involves an activity ratio. If the studied thermal water is not too concentrated, a
concentration ratio can be used, as we will do here, since the ionic strengths of G2 and G3 are
respectively 0.096 and 0.089 mol/L. The main limitations of the Na-K geothermometer can be
expressed as questions :

- Is the thermodynamic equilibrium reached at depth ?

- Are there secondary reactions or alteration of other minerals that could affect the Na/K ratio ?

The first question can be relatively easily answered by comparing the calculated temperature with
the temperature obtained using the silica geothermometer. For example, a « cold » water that has
never been in equilibrium with the two minerals would lead to an absurd result. For instance, the
calculated temperature for surface seawater would be 120°C ! The reactions of rock alteration during
the ascent of thermal water towards the surface are favored in waters rich in dissolved CO2, whose
acidity can selectively extract sodium or potassium from fractured rocks in the geothermal channels.
This can lead to overestimation or underestimation of the deep temperature depending on the
circumstances.

Many authors cited in the publication by Serra & Sanjuan have added « corrective » factors to the
fundamental Equation (18), often entirely empirical and calculated based on the composition of
geothermal waters for which the temperature of the deep reservoir is known through drilling. As a
result, there are no less than 13 relations of the type (Na+)/(K+) = f(T) [16]. The most « reliable »
relations are those based on the establishment of robust correlations between the composition of
waters and their deep temperature when it is known, allowing calibration of the geothermometer.
For example, these relations can be expressed as follows :

T (°C)

1190

= -273.15

Na

1.35

Log ()

1170

(19) and (20)

T (°C)

= -273.15

Na

1.42

Log ()

Published by Michard [17] adjusted with two correlations, each containing 60 data points. Each point
represents a thermal water from deep granitic rocks, where the emergence temperature is measured
and the deep temperature is known through prospecting or production drilling. Table 9 below
provides the calculated temperatures for G2/G3 using Relations (19) and (20) :

Table 9 : Na-K Temperatures


The results from Table 9 show a perfect agreement between the deep temperatures and the silica
geothermometer when Relation (18) is used. The temperature given by Relation (19) is probably
overestimated. The aberrant temperatures given for the phreatic water (Well) clearly indicate that
this water has no hydrothermal characteristics. In conclusion, it can be hypothesized that the deep
waters of the cold geysers in Analavory are in equilibrium with quartz, potassium feldspar, and albite
at a temperature of approximately 150 °C. At the surface temperature, the waters re-equilibrate with
amorphous silica, while the loss of CO2 and contact with oxygen lead to an increase in pH and the
precipitation of calcite and ferric hydroxide.

Regarding the origin of dissolved iron (227 and 262 µmol/L for G2 and G3, respectively), it is
suggested that it comes from the dissolution of iron from pipelines used for aragonite mining [1]. This
hypothesis does not withstand an analysis of the physicochemical processes that can lead to the
formation of soluble Fe(II). In fact, to produce Fe2+ ions from Fe0 (metal of the pipes), an oxidant
capable of accepting electrons from the redox couple is required :

Fe0 → Fe2+ + 2 e-

However, the high concentration of dissolved iron in G2/G3 shows that it can only be dissolved iron in
the Fe(II) state, either as Fe2+ ions or complexes. The only oxidant available is dissolved oxygen, and
it is only present when the thermal water is in contact with the atmosphere. Therefore, it is much
more likely that Fe(II) originates from the deep alteration at high temperature of minerals such as
micas (biotite), pyroxenes, or amphiboles present in the granitic bedrock covered by basaltic flows.

7. Conclusion :

This geochemical study provides a better understanding of the physicochemical mechanisms that
account for the evolution of the « cold geysers » in Analavory when they come into contact with the
atmosphere. The specified and quantified degassing and solid precipitation reactions explain field
observations, particularly the encrusting precipitation of a mixture of calcium carbonates and oxy-
hydroxide ferric, whose mineralogical nature has been determined through in-depth solid analysis
using advanced analytical techniques (XRD, XRF, SEM-EDX). We have also clearly demonstrated the
hydrothermal nature of these waters. They originate from a deep geothermal reservoir and are not
simply surface waters that have dissolved « relic » CO2 from ancient volcanic activity, as previously
believed. Lastly, the use of geothermometers allows for estimating the deep reservoir temperature at
approximately 150 °C with a good level of confidence.
Acknowledgements :

I would like to express my sincere gratitude to Mrs. Eliana Maminomenjanahary for her assistance in
the field. I would also like to thank Mrs. Laure Cordier, Hassiba Lazar, Emmanuelle Raimbault, and
Sophie Nowak for their invaluable contribution to this extensive analytical work.

References :

[1] « Les geysers froids d’Analavory » (2015) www.earth-of-fire.com

[2] L. M. RAKOTOSON, « Physico-chemical studies of the geysers of Analavory. » DEA Thesis.


University of Antananarivo, Faculty of Sciences, August 2010.

[3] G. GRAN, « Determination of the equivalence point in potentiometric titrations, part II. » Analyst
(London). 77 (1952) 661-671.

[4] G. CHARLOT, « Chemical reactions in solution. » Masson publisher, Paris (1969).

[5] « Handbook of Chemistry and Physics » (2005).

[6] M. BAKALOWICZ, « Contribution of water geochemistry to the understanding of the karst aquifer
and karstification. » Doctoral thesis, Université P. et M. Curie Paris-6, Laboratory of the CNRS
Underground, 269 p. (1979).

[7] W. STUMM and J. J. MORGAN, « Aquatic Chemistry. » John Wiley and Sons, Editor (1981).

[8] G. Michard, « Chemical equilibria in natural waters. » Ellipses publisher (1989) 189 p.

[9] G. MICHARD, « Chemistry of natural waters. Principles of water geochemistry. » Publisud


publisher (2002) 445 p.

[10] NOAA National Oceanic & Atmospheric Administration – Earth System Research Laboratory
(ESRL), « Trends in Carbon Dioxide. »
http://en.wikipedia.org/wiki/Carbon_dioxide#In_the_Earth.27s_atmosphere (2015)

[11] G. SARAZIN, G. MICHARD, RAKOTO and L. PASTOR, « Geochemical study of the geothermal field
of Antsirabe (Madagascar). » Geochem. J. 20 (1986) 41-50.

[12] R. O. FOURNIER and A. H. TRUESDELL, « An empirical Na/K/Ca geothermometer for natural


waters. » Geochim. Cosmochim. Acta. 37. (1973) 1255-1275.

[13] S. ARNORSSON, « Application of the silica geothermometer in low temperature hydrothermal


areas in Iceland. » Am. J. of Sci. 275 (1975). 763-784.

[14] R. O. FOURNIER, « A revised equation for the Na/K geothermometer. » Geotherm. Res. Counc.
Trans. 3 (1979) 221-224.

[15] G. MICHARD, « Chemical geothermometers. » Bull. du BRGM. (2ème série). Section III. N°2 (1979)
183-189.

[16] H. SERRA and B. SANJUAN, « Literature review of chemical geothermometers applied to


geothermal waters. » Final report BRGM/RP-52430-FR. (2004).

[17] G. MICHARD, « Behaviour of major elements and some trace elements (Li, Rb, Cs, Sr, Fe, Mn, W,
F) in deep hot waters from granitic areas. » Chem. Geol. 89 (1990) 117-13.

You might also like