A soil is identified as an hydromorphic or gleyey soil when some of its characteristics are due to an excess of water (Avery, 1973; Baize and Girard, 1995). However, different conditions with regard to hydric variations and biological and biogeochemical processes must be simultaneously fulfilled:
an excess of water;
the restriction of the oxygen sources;
the presence of bioavailable substrates;
temperature conditions favorable to microflora activity;
the presence of elements the oxidation state of which can change and thus record more or less irreversibly the variations in aerobic/anaerobic con- ditions, named geochemical markers of hydromorphic soils (Trolard et al., 1998).
Among those elements the oxidation state of which can change under earth surface conditions, iron is the most abundant, and the switching between the di- and trivalent redox states of iron initiates a number of significant geochemical reactions. Moreover, the large difference in mobil- ity between Fe2þ, soluble, and Fe3þ, insoluble, results in the segregation of iron in the horizon and/or in soil sequences (Segalen, 1971; Marshall, 1977). These variations are ordered in the landscapes and are considered as responsible for concretioning processes (Marshall, 1977) and for color variations (Taylor, 1981).
In all soil classification systems, soil characteristics depending on iron dynamics have been used to differentiate well-drained from poorly drained soils (Soil Survey Staff, 1975, 1976; IUSS Working Group WRB, 1994, 2006; Baize and Girard, 1995; Driessenet al., 2001): soil color, field tests, and rH measurements.
2.1. Soil color
Soil color is closely related to the nature of iron oxides, more specifically to their degree of hydration and their amount (Vyssotskii, 1905 [1999];
Taylor, 1981; Cornell and Schwertmann, 2003; Photo 1). Classically,
Geochemistry of Green Rusts and Fougerite 229
brown, red, or yellow colors have been associated with the occurrence of Fe (III) oxides, while gray, green, blue, or black colors have been associated with the occurrence of Fe(II) in the solids (oxides or sulfides), the white color being interpreted as indicating the absence of any Fe-bearing mineral.
Among these colors, moderately reduced waterlogged soils are characterized by the blue-green one (Photo 2 ) which turns into ochre when the soils are open to the outer atmosphere (Vyssotskii, 1905 [1999];
Ponnamperuma et al ., 1967; Photo 3). It has been often ascribed to the occurrence in the milieu of mixed Fe(II)–Fe(III) compounds with a likely structure of GRs. This assumption has been formulated since 1960 and largely discussed in the literature (see the reviews by Taylor, 1981, and Lewis, 1997). GRs were first observed in corrosion products of steel (Stampfl, 1969; Bigham and Tuovinen, 1985; Geninet al., 1994; Al-Agha et al., 1995) then in a waste sludge (Koch and Mrup, 1991), recognized as a
Photo 1 Soil profile developed in a loamy cover on granite at Quintin, in Brittany, showing all colors from blue (bottom) to yellow, orange and red (top). Photo INRA.
230 Fabienne Trolard and Guilhem Bourrie´
mineral in soils (Trolardet al., 1996, 1997), and finally homologated in 2004 by the International Mineralogical Association (Trolard, 2006; Trolard et al., 2007).
2.2. Field tests
Field tests that give a semiquantitative appreciation of the redox status of soil have been proposed by Childs (1981) and by Bartlett and James (1995).
Childs’ test gives evidence for the presence of Fe(II) in nonsilicated soil Fe, by the appearance of a strong red color on a freshly broken surface of a field-wet soil sample after spraying it with a 0.2%a, a0-dipyridyl (DIPY) s o lu ti on i n 1 M ( p H ẳ 7) ammonium acetate (Childs, 1981). After 1–2 min the intensity of the color is evaluated as follows: positive (red color, Photo 4), weakly positive (pink color or redder patches on clods), and negative or
Photo 2 Gleysol in Fouge`res, showing the characteristic green-blue color.
Photo INRA.
Photo 3 Change of soil color in Fouge`res after 30 mn exposure to the air.
Geochemistry of Green Rusts and Fougerite 231
barely detectable. Bartlett and James’ test gives evidence for the presence of easily reducible iron oxide fraction: 0.1 M oxalic acid is poured to saturate the soil on a spot plate that is placed in sunlight for 10 min, then five drops of DIPY indicator (10 mM DIPY in pHẳ4.8, 1.25 M ammonium acetate) are added. Similarly, distinctions are made: very positive (dark purple color), positive (purple color), weakly positive (pale purple color), and negative or barely detectable.
Recent field applications of these two tests have been done in Brittany (Chaplot, 1998; Chaplotet al., 2000) and in Amazonia (Fritschet al., 2007), and performed on clods as well as on separate features in the case of heterogeneous soil samples, for example, spots, mottles, or nodules. In these two cases, the seasonal dynamics of the redox status of soil was studied.
In Brittany, seasonal variations in Fe(II) occurrence (Child’s test) were surveyed over 2 years along a loamy hillslope developed on granite or schist.
Results (Fig. 1) showed that in both situations, the depth of the limit between reduced and oxidized horizons varied with the seasons and that reduced horizons were located in the surface organo-mineral and deep saprolite horizons, which were separated by oxidized mineral horizons. In the granite system near the stream channel, an oxidized zone persisted in the saprolite despite the continuous waterlogging of soil, which implies that waterlogging is a necessary but not sufficient condition for hydromorphy.
In the Amazonian basin, similar observations were made during the dry and wet seasons (Fritsch et al., 2007). While during the dry season the Childs’ test gave negative results (Fig. 2), during the rainy season it showed a large positive response in different horizons of the sequence.
The presence of lepidocrocite in site 1 during the rainy season indicates that fougerite was formed, as fougerite is known as a precursor of lepido- crocite. Strongly positive results of the Bartlett and James’test (Fig. 3)
Photo 4 Red color obtained by Childs’ test on the gleysol in Fouge`res.
232 Fabienne Trolard and Guilhem Bourrie´
affected a larger thickness of the slightly clay-depleted topsoil in the transi- tion zone (site 3) than in the upslope or low-lying positions. In the clay loam subsoil, the test was commonly strongly positive on separated features such as the bright to dusky red mottles, weakly positive for dark red nodules, and barely detectable or negative for the yellow fringes and white masses.
All these observations show that Fe is subject to fast mineralogical transformations in gleysols.
2.3. rH measurements
Field tests and rH measurements are scarce, and evidences for Fe mobility and redox state of iron are mainly based upon morphological observations (color), total Fe measurements in soils, and mineralogy by X-ray diffraction (XRD). However, this gives only indirect evidences. No information is gained about the thermodynamic control and kinetics of the processes.
The diagnosis for gleyic properties (IUSS Working Group WRB, 2006) is an rH value of 19 or less, where rH( logpH2) is related to the redox potential (Eh) and pH by the relationship: rHẳEh/0.029ỵ2pH, whereEh
is expressed in V, and 0.029 stands for the approximate value of (ln 10)RT/
2Fat 298.15 K.
Stream channel
January
January
? ?
? ?
?
?
May
Point of field test Horizon
Topography
1m
2m 10 m Point of field test
Horizon Topography
1m
10 m
2m
March
A B C D
October October
December April
Stream channel
Figure 1 Dynamics of iron versus time, as evidenced with Childs’ test, in two sequences of soils on granite (left) and schists (right) in Brittany, France. (A) White:
negative or barely visually detectable; (B) dotted gray: weakly positive (pale purple 10R6/3); (C) gray: positive, purple 10R6/8; (D) black: strongly positive, dark purple 10R4/8. Adapted fromChaplot (1998) and Chaplotet al., (2000).
Geochemistry of Green Rusts and Fougerite 233
This implies measuringEh and pH in soil solution, as discussed below.
When computing rH, one must, however, use the exact temperature of the sample, and rH is more conveniently obtained as a function ofpe( log {e}, where {e} is the activity of the electron), and pH:
peẳ FEh
ðln10ịRT; ð1ị
rHẳ2ðpeỵpHị; ð2ị
where F ẳ96485.309 C mol1is the Faraday constant, R ẳ 8.314510 J mol1 K1 is the ideal gas constant, T ẳ 273.15 ỵ t is the absolute temperature (K), and Eh is the redox potential, in Volt, referred to the normal hydrogen electrode.
Usually, the potentialEis measured with a Pt electrode against either a calomel reference electrode or an Ag/AgCl reference electrode, and Eh
must be corrected for the standard potential of the reference electrode.
6 5 4
3
Wet season
Water table level
Water table level
2
1 0 m
1 m 2 m
6 5 4
3
Dry season 2
1
A B C 0
0 m 1 m 2 m
20 m
Figure 2 Dynamics of iron versus time, assessed with Childs’test, in the toposequence of Humaita, Bresil. (A) Black: positive (purple); (B) Gray: weakly positive (pale purple);
(C) White: negative or barely visually detectable. During the wet season (top), test is positive in large parts of the sequence, while during the dry season (bottom), the test is generally negative. Adapted fromFritschet al., (2007).
234 Fabienne Trolard and Guilhem Bourrie´
For calomel reference electrode (Criaud and Fouillac, 1986)
Eh ẳEỵ0:24150:76103ðt25ị: ð3ị Although calomel electrode is more reliable, Ag/AgCl reference electrode is preferred due to the risk of pollution by mercury. However, Ag/AgCl electrodes are not highly reproducible (Bates and Macaskill, 1978), and, moreover, their standard potential depends on the nature and the concen- tration of the internal electrolyte solution. Usually, KCl is used, as transfer- ence numbers of Kþand Clare similar, which minimizes the error due to the liquid junction potential, but either at saturation or 3, 3.5, or 4 M. The standard potential of the Ag/AgCl electrode is given by
EAg=AgClẳE0@E0
@T ðT 298:15ị; ð4ị whereE0is the standard potential of the Ag/AgCl electrode at 298.15 K, 1 bar. Thermodynamic data (Table 1) lead to the practical equation:
6 5
Wet season
Dry season 4
2
Water table level
Water table level
1 0 m 3
1 m 2 m
6
A B C D 0 20 m
5 4
2 1
0 m 3 1 m 2 m
Figure 3 Dynamics of iron versus time, assessed with Bartlett and James’ test, in the toposequence of Humaita, Bresil. (A) Black: strongly positive, dark purple; (B) gray:
positive, purple; (C) pale gray: weakly positive, pale purple; (D) white: negative or barely visually detectable. Adapted fromFritschet al., (2007).
Geochemistry of Green Rusts and Fougerite 235
Table 1 Thermodynamic data for Ag/AgCl reference electrode
Compound DfG0/kcal mol1 DfH0/kcal mol1 S0/cal mol1K1 References
Ag(c) 0 0 10.206 a
AgCl(c) 26.224 30.362 22.97 a
Cl(aq.) 31.350 40.023 13.2 a
e (1/2 H2(g)) 0 0 15.6055 a
Equilibrium reaction:
AgClðcị ỵe ⇆Agðcị ỵCl
DRG0/kcal mol1 DRH0/kcal mol1 DRS0/cal mol1K1
5.126 9.661 15.1695 b
E/V @@ET0=mV K1
0.222 0.6578 a,b
0.22249 — c
a Latimer (1952).
b From (a), with@E0=@TẳDRS0=F(Bratsch, 1989), converted withFẳ23.060 cal V1mol1.
c Bates and Macaskill (1978).
E 0Ag= AgCl ẳ 0: 22249 0 :6578 10 3 ð t 25ị; ð5ị where E0 is from Bates and Macaskill (1978), while its derivative with temperature is obtained from Latimer (1952).
The half-cell chemical reaction for the Ag/AgCl electrode can be written as follows:
AgCl ð c ị ỵe ⇆ Ag ðc ị ỵ Cl
and according to Nernst’s law, the theoretical potential is obtained as follows:
EAg =AgCl ẳ E 0 @ E 0
@ T ð T 298 :15 ị ðln10 ịRT
F log Clf g; ð6ị where the curly braces stand for the activity of Cl in the internal electrolyte filling solution. For KCl 4.5 M, the mean activity coefficient is g ẳ0.583 at 25 C (Robinson and Stokes, 1970); hence, f Clg ẳ ffiffiffiffiffiffiffiffiffiffiffi
0 :583 p 4: 5 ẳ 3 :43, which leads to EAg /AgCl ẳ 191 mV. Practically, because of the difficulty to make reproducible Ag/AgCl electrodes, manufacturers give the corrections to apply. It must be kept in mind that electrodes should be checked periodically, and after replacing the internal filling solution. The correction applied is not always specified in the literature.
In our previous studies (Bourrie´ et al., 1999), standard calomel electrode was used and data were corrected according to Eq.(3). Later (Feder et al., 2005), we used Ag/AgCl with KCl in an agar gel manufactured by Ingold and the data were corrected according to
Eh ẳEỵ0:206710:7588103ðt25ị; ð7ị the electrode being checked against two reference solutions (230 mV and 470 mV5 mV). The precision obtained is 20 mV, and the sensitivity 0.1 mV.