You are on page 1of 6

Journal of Colloid and Interface Science 286 (2005) 596–601

www.elsevier.com/locate/jcis

Adsorption of several metal ions onto a model soil sample:


Equilibrium and EPR studies
Karine Flogeac, Emmanuel Guillon ∗ , Michel Aplincourt
GRECI (Groupe de Recherche en Chimie Inorganique), Université de Reims Champagne-Ardenne, B.P. 1039, F-51687 Reims cedex 2, France
Received 27 August 2004; accepted 21 January 2005
Available online 9 March 2005

Abstract
Soils play an important role in the control of metallic cations in the environment. Therefore, knowledge of the adsorption properties of soil
is crucial in understanding and solving pollution problems. Adsorption isotherms provide a macroscopic view of the retention phenomena.
The aim of this paper is to study iron, manganese, and chromium adsorption onto a soil sample as a function of the reaction time, pH, and
metal concentration. The adsorption isotherms allow the determination of the affinity order of metals for the surface of the soil sample as
such: Fe3+ > Cr3+ > Mn2+ . The equilibrium data fit well with the Langmuir and Freundlich models and confirm the affinity order of the
soil sample for these metals. These adsorption data are combined with EPR spectroscopy to obtain structural information about the surface
complexes formed. Iron is held in inner-sphere complexes. Manganese is simultaneously held in outer- and inner-sphere complexes. Due to
poor resolution, chromium was not detected by EPR and thus it is impossible to infer coordination sphere and coordination number. Iron and
manganese are in an octahedral environment.
 2005 Elsevier Inc. All rights reserved.

Keywords: Soil; Iron; Manganese; Chromium; Adsorption isotherms; EPR

1. Introduction plays a significant role in the type, strength, and reversibility


of the metal complexes formed and, thus, determines if the
Metals, especially transition metals, complexed or not, metal ion is immobilized or leached. This is why many stud-
present a particular risk because, in contrast to many pol- ies established the importance of independent surfaces in the
lutants (organic compounds), they are stable in the en- retention of metals [3–8], such as organic matter (lignocel-
vironment (i.e., no degradation). Consequently, retention lulosic substrate, humic substances), goethite, gibbsite, and
processes at liquid–solid interfaces play a major role in envi- silica.
ronmental studies, especially in the control of metallic cation The metal ions investigated in this study are iron, man-
transport toward surface and ground waters [1,2]. Therefore, ganese, and chromium. In small quantities, iron and man-
a complete understanding of the availability of long-term ganese are necessary for living organisms. However, in areas
pollutants depends on our knowledge of the adsorption re- where the level of metals is important because of ore smelt-
actions taking place at the soil surfaces. ing, land application of biosolids, or other anthropogenic
processes, metals are often a detriment to the environment.
In natural environments, many sorbent phases are able
Chromium is a metal used in various industrial processes:
to adsorb metal ions such as metal oxides and oxyhydrox-
tanneries, cement industries, plating and alloying industries,
ides, clay minerals, and organic matter. The sorbent phase
and corrosive paints [9]. Chromium, in its trivalent form,
is an essential trace element for plants and animals: it is
* Corresponding author. Fax: +33(0)3-26-91-32-43. involved in glucose metabolism and nucleic acid synthe-
E-mail address: emmanuel.guillon@univ-reims.fr (E. Guillon). sis [10]. However, chromium(III) has also been shown to
0021-9797/$ – see front matter  2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcis.2005.01.027
K. Flogeac et al. / Journal of Colloid and Interface Science 286 (2005) 596–601 597

be a potential hazard, especially in the aquatic environment. 2.3. Adsorption experiments


Indeed, in vitro tests showed that trivalent chromium is a po-
tential toxin because it is a competitive inhibitor of many All experiments were carried out at a fixed ionic strength
cellular processes [11]. of 0.1 mol L−1 KNO3 . Two different types of adsorption ex-
The present paper aims at studying the adsorption of periments were conducted: variable pH and variable metal
iron, manganese, and chromium onto a soil sample from concentration. Both studies were conducted using batch ex-
the Champagne–Ardenne region (France) as a function of periments at room temperature (293 K).
the reaction time, pH, and metal concentration. This study For the adsorption studies as a function of the pH, the
is combined with electronic paramagnetic resonance (EPR) soil sample (2 g L−1 ) was immersed in 15 ml of background
spectroscopy to characterize the geometry of the surface electrolyte solution and stirred using a magnetic stir bar for
complexes formed. 4 h, which corresponds to the hydration time of the solid. Af-
ter this pre-equilibration step, the metallic cation was added
at a concentration of 2 × 10−4 mol L−1 , and the pH was in-
2. Materials and methods crementally adjusted to a fixed value by dropwise addition
of 0.1 mol L−1 HNO3 or 0.1 mol L−1 KOH. The final vol-
2.1. Reagents ume was adjusted to 25 ml. The flasks were then shaken
with an automatic shaker at 293 K for the time necessary
KNO3 , HNO3 , and KOH were obtained in the purest to ensure complete adsorption. After filtration through a
commercially available grade (Prolabo for KNO3 , Fluka 0.20-µm polyamide membrane, the supernatants were acidi-
for HNO3 and KOH) and used without further purifica- fied and the unsorbed metal ion concentration was measured
tion. Ferric nitrate (Fe(NO3 )3 ·9H2 O), manganese nitrate by a Varian ICP-AES spectrometer. The amount of metal ad-
(Mn(NO3 )2 ·4H2 O), and chromium nitrate (Cr(NO3 )3 ·6H2 O) sorbed was deduced from the initial concentration.
were purchased from Fluka. The adsorption equilibrium was predetermined by kinetic
experiments in which different contact times between the
2.2. Soil sample extraction metal and the solid were applied at a constant pH value. The
experimental pH was taken to be the pH measured after the
reaction time.
The soil sample was extracted from the Champagne–
The adsorption isotherms as function of metal concen-
Ardenne region (France), which is known to be a calcare-
tration experiments were carried out in the same way at a
ous region. The sample was dried at 293 K and sieved at
fixed pH value (3.2 for iron, 7.7 for manganese, and 5.0 for
<100 µm. The crude solid was suspended at 293 K for 24 h
chromium) with a metal concentration ranging from 10 to
in chloroform (10 ml per gram), essentially to extract free
1500 µmol L−1 .
lipids. After filtration, the residue was mixed for 24 h with
a 1 mol L−1 HCl solution (10 ml per gram) to eliminate
2.4. Adsorption isotherms
carbonates, sulfates, adsorbed metals, and hydroxides. Af-
ter a new filtration and a water wash, necessary to obtain
a neutral pH, an alkali treatment was implemented (NaOH The equilibrium data were analyzed in accordance with
0.1 mol L−1 , 10 ml per gram) for 6 h to make humic and the Langmuir and Freundlich sorption isotherms. These
fulvic acids soluble. Then, the insoluble solid obtained was isotherms allow to describe adsorption phenomena of met-
stirred in a nitric acid solution (5 × 10−2 mol L−1 , 10 ml als from aqueous solution onto soil sample.
per gram) to saturate the proton surface sites and vacuum- The Langmuir isotherm is expressed by
dried. These treatments allowed a quasi-insoluble solid ad- Ce Ce 1
sorbent in the pH range studied (2–10), which is a neces- = max + max with Cs = Kd Ce , (1)
Cs Cs Cs b
sary condition for further metallic cation adsorption experi-
ments. The solid was air-dried and ground to pass a 100-µm where Kd is the distribution coefficient that characterizes
sieve. the affinity of the metal for the sorbent, Csmax is the maxi-
The chemical composition and structure of the soil sam- mum adsorption capacity of the solid corresponding to com-
ple, after treatment, are detailed elsewhere [12]. Briefly, the plete monolayer coverage, Ce is the measured concentra-
site density was estimated by potentiometric titrations to be tion (µmol L−1 ) of metal in solution when the equilibrium
equal to 3.2 sites/nm2 , and the specific surface area deter- is reached, Cs is the measured adsorption per unit weight
mined by the BET method (N2 , 77 K) was 33.5 m2 g−1 . An of solid (µmol g−1 ), and b represents the Langmuir bond-
X-ray diffraction (XRD) analysis identified the solid as sil- ing term (L µmol−1 ) related to the adsorption energy. The
ica (SiO2 , 60%), illite and smectite (27%), kaolinite (6%), Freundlich isotherm, which assumes that different sites with
goethite (α-FeOOH, 5%), and rutile (TiO2 , 2%) for the min- several adsorption energies are involved, is expressed by
eral phase. The organic fraction represented 5% of the total
composition of the soil particles. log Cs = log KF + n log Ce , (2)
598 K. Flogeac et al. / Journal of Colloid and Interface Science 286 (2005) 596–601

where KF is the Freundlich distribution coefficient related to


the total adsorption capacity of the solid, and n is a nonlinear
constant.

2.5. EPR experiments

The solid state EPR spectra were obtained with a Bruker


ELEXYS 500 spectrometer operating at an X-band fre-
quency with a 100-kHz modulation frequency. The spectra
were recorded in the following conditions: microwave power
1 mW for chromium and 15 mW for iron and manganese;
Fig. 1. Chromium, manganese, and iron adsorption on the soil sample
modulation amplitude 0.4 mT. The spectra were carried out (2 g L−1 ) as a function of pH at 293 K in KNO3 0.1 mol L−1 medium,
at 10, 77, and 298 K for the iron, chromium, and manganese with [Mn+ ] = 2 × 10−4 mol L−1 . Points on the continuous lines represent
complexes, respectively. The soil sample was also recorded, the metal cation precipitation curves (without solids).
for comparison, at each temperature. Simulations of the EPR
spectrum were performed using WINSYMPHONIA Bruker
software package.

3. Results and discussion

3.1. Adsorption experiments

Kinetic experiments were first carried out to determine


the equilibrium time of the complexation reaction between
iron, manganese or chromium, and the soil sample. The ki-
netic of each metal adsorption was studied by varying the
contact time from 5 min to 48 h. The adsorption equilibrium Fig. 2. Chromium, manganese, and iron adsorption isotherms on the soil
was reached in 10 h for iron and manganese and in 24 h for sample (2 g L−1 ) at 293 K.
chromium. These results agree with several previous studies
on metal adsorption [13–15]. Several mechanisms have been only surface complexation occurred. The fact that the ad-
postulated to explain this slow uptake [13,15]. The most sorption edge systematically occurred at a pH value lower
likely is that in the first hours of the adsorption process, an than that of the metal cation precipitation, confirms the sur-
equilibrium is reached between the adsorbed metal amount face complexation and rules out the metal hydroxide surface
and the metal cation amount in solution (fast process). Then, precipitation. Consequently, in order to avoid metal hydroly-
formation of polynuclear complexes, diffusion into the soil sis processes at the surface, pH values equal to 3.2, 5.0, and
sample, and surface precipitation could occur. In order to 7.7 were chosen for isotherm experiments as a function of
obtain a complete adsorption of the metals, and for practical the metal concentration for iron, chromium, and manganese,
reasons, the experiments were carried out with an equilib- respectively.
rium time equal to one night for iron and manganese, and The adsorption isotherms of metal ions onto soil sam-
24 h for chromium. ples are reported in Fig. 2. These isotherms can be subdi-
The effect of the solution pH on the adsorption of met- vided in three parts. In the low-concentration part, the metal
als is reported in Fig. 1. Adsorption of metals is strongly introduced is quantitatively adsorbed. In the intermediate
dependent on the pH, increasing when the pH of the solu- part, there is a plateau or a pseudo-plateau, which prob-
tion increases. The adsorption edge is different from each ably corresponds to the saturation of the surface or to a
metal. Indeed, the adsorption begins around pH 2.0 for iron, co-precipitation phenomenon (i.e., sorption and precipita-
2.5 for chromium, and 6.0 for manganese. The adsorption of tion simultaneously). In the larger concentration part, the
the metallic cations by the soil sample occurs in a relatively amount of metal adsorbed increases again with the concen-
narrow pH range (around 2 pH units). The affinity of each tration of the metal ion introduced, corresponding to a sur-
metal ion is reflected in the percentage adsorption obtained face precipitation. Equilibrium adsorption experiments were
at a fixed pH. Thus, in all the pH range, the order of affinity performed to provide the maximum metal adsorption capaci-
was Fe3+ > Cr3+ > Mn2+ . More generally, with increasing ties of the soil sample and the presence of a plateau indicates
pH, the adsorption edges were reached for Fe3+ , Cr3+ , and that this maximum adsorption capacity has been reached.
Mn2+ in this sequence. These results were compared with This capacity can be estimated at 8 mg g−1 (155 µmol g−1 )
the precipitation of metals to ensure that no hydrolysis and for chromium, 4 mg g−1 (70 µmol g−1 ) for manganese, and
K. Flogeac et al. / Journal of Colloid and Interface Science 286 (2005) 596–601 599

Csmax values obtained by the Freundlich isotherms are in


accordance with the experimental data. These values indi-
cate that metal ions are significantly uptaken by the soil
sample in these conditions. Therefore, it implicates low mo-
bility of metal ions in natural environment. As expected
with the experimental isotherms, the greatest equilibrium
fixation capacity Csmax was obtained for Cr3+ ions, that is,
151 µmol g−1 , which decreased to 104 µmol g−1 for Fe3+
and 77.5 µmol g−1 for Mn2+ . This affinity sequence is dif-
(a) ferent from the one deduced for the pH influence. The in-
version of chromium and iron values is probably due to the
co-precipitation phenomenon, which cannot be unambigu-
ously distinguished from the sorption one [14]. It has been
shown, in the case of chromium, the formation of polynu-
clear species since low pH values, which can be sorbed at
the same time than the chromium(III) ion. The values of the
Langmuir bonding term b give an idea of the adsorption in-
tensity. Indeed, the energy of a metallic cation adsorption on
a solid surface increases with b. These values are close to
the ones obtained for the adsorption of metal cations onto
(b) biosorbent [16]. The Freundlich adsorption intensity con-
stants n are similar to constants previously reported [17].
Fig. 3. Langmuir adsorption (a) and Freundlich adsorption (b) isotherms for In natural environment, systems are more complicated
Cr3+ adsorption on soil sample (2 g L−1 ) at 293 K.
because several other parameters govern the adsorption phe-
nomena, and the natural solids are much more complex sys-
Table 1
tems. But our experiments give some useful pieces of infor-
Langmuir and Freundlich parameters for chromium, manganese, and iron
adsorption onto soil sample (2 g L−1 ) at 293 K mation on the retention properties of a model soil sample.
This can be the starting point for the study of a natural sam-
Langmuir Freundlich
ple without pretreatment.
Csmax (µmol/g) b (L/µmol) r2 KF (g/L) n r2
After the macroscopic studies of metal adsorption onto
Chromium 151.0 8.4 × 10−3 0.967 3.05 0.67 0.966 the soil sample, we carried out molecular-scale studies of
Manganese 77.5 8.8 × 10−3 0.993 1.15 0.46 0.943 surface complexes using EPR spectroscopy, to obtain infor-
Iron 104.0 15.0 × 10−3 0.840 6.03 0.51 0.779
mation on the geometry of the surface complexes formed.

5.5 mg g−1 (100 µmol g−1 ) for iron. These values indicate 3.2. EPR experiments
that the soil sample presents a significant affinity for metals.
Adsorption isotherms could be mathematically represented Before recording the spectra of the metal–soil sample
by the Langmuir and Freundlich adsorption isotherms. As systems, we have recorded the spectrum of the soil sample
an example, the Langmuir and the Freundlich adsorption alone [12] in order to compare it to the former ones. Fig. 4a
isotherms of chromium are represented in Fig. 3. The Lang- represents the spectrum of the soil sample at 10 K. Briefly,
muir isotherm parameters were deducted for each metal by this spectrum is dominated by high-spin Fe3+ contributions
performing a linear regression and are reported in Table 1. (d 5 , S = 5/2). We can distinguish at least two types of iron
The good agreement (r 2 ) with the experimental data sug- signal [18,19]. The first one corresponds to some rhombic
gests that the metal ions sorbed form a monolayer cov- iron(III) with an isotropic signal at g = 4.3, originating from
erage on the adsorbent surface. The sites involved in the the middle Kramers doublet, and a very weak feature at
metal sorption are probably the acid moieties from the or- g = 9 due to the highest Kramers doublet (Fig. 4b). The
ganic matter coated onto the mineral phases [12]. In the second one corresponds to some nearly tetrahedral or octa-
case of iron(III), the lower r 2 value may be due to weak hedral iron(III) sites giving a resonance at g = 2.1 (Fig. 4c).
co-precipitation of iron hydroxide. The Freundlich isotherm The first type of iron ions probably originates from trivalent
parameters are also added for comparison. Generally, to a iron in the illite/smectite or kaolinite fraction [20], and the
lesser extent, the equilibrium data were also well described second type originates from the goethite fraction [21]. This
by the Freundlich model, probably due to the real hetero- broad signal, at g = 2.1, is difficult to attribute because it
geneous nature of the soil sample surface sites. However, also resembles dipolar couplings. Moreover, another signal
adsorption data for each metal ion better fit the Langmuir appears around g = 2 (Figs. 4d and 4e). This signal seems
model, since the Langmuir isotherms present better cor- to correspond to a vanadyl porphyrin [22] with an axial g
relation coefficient values than the Freundlich model. The tensor from S = 1/2 with an hyperfine coupling to an ax-
600 K. Flogeac et al. / Journal of Colloid and Interface Science 286 (2005) 596–601

Fig. 6. EPR spectrum of manganese–soil system at 298 K.

The value of the ratio E/D (D and E are the zero-field


splitting parameters) was found to be 1/3, in accordance
with rhombic symmetry. The spectrum feature is typical of a
FeO4 or FeO6 chromophore. Iron(III) in surface complexes
is held in inner-sphere complexes. A wide signal is also
observed around g = 2 (337.0 mT). It corresponds to iron
Fig. 4. EPR spectra at 10 K of soil sample (a), rhombic iron contribution (b),
nearly octahedral iron contribution (c), and vanadyl contribution (d) with an
hydroxide formed by co-precipitation of metal onto the soil
enlargement of the 350-mT region (e). sample surface [24]. Finally, a narrow band was observed
at g = 2.0045 (335.0 mT); it is assigned to semiquinonic
species from organic fraction of the soil sample. The pres-
ence of this radical signal, after subtraction of the soil sample
spectrum, indicates an increase of the semiquinonic species.
This increase is due to redox processes and electron trans-
fers between the organic radical and the metal center on the
soil sample surface [4]. The phenolic moieties are the main
functions involved in these redox processes [4] and such re-
dox phenomenon may be explained by the fact that iron is
probably adsorbed on the organic matter fraction of the soil
sample.
The EPR spectrum of the Mn–soil system, recorded at
room temperature, is represented in Fig. 6. It was also ob-
Fig. 5. EPR spectrum of iron–soil system at 10 K.
tained after subtraction of the soil sample spectrum at the
same temperature. It presents the six-line pattern, character-
ial A tensor. We can then determine g coupled with a A , istic of a manganese(II) ion with g = 1.979 (350.0 mT). This
and g⊥ coupled to A⊥ , arising from a S = 1/2 coupled to a spectrum is in accordance with a MnO6 chromophore and
I = 7/2 nucleus. The Hamiltonian parameters obtained were an octahedral environment around the metal cation. In ad-
g = 1.95, g⊥ = 1.96, A = 16.3 mT, and A⊥ = 5.0 mT. dition, it is similar to that of the Mn(H2 O)2+ 6 compound.
Superimposed to this spectrum we can observe a narrow This shows that the manganese(II) ion is predominantly
signal at g = 2, which corresponds to an organic free rad- bound to the surface acid sites by van der Waals interactions;
ical [23] originating from the organic fraction (Fig. 4e). i.e., the complex is an outer-sphere complex. The hyper-
The metal–soil sample systems for EPR analysis were ob- fine coupling constant of manganese(II) is equal to 8.2 mT.
tained by stirring a weighted amount (50 mg) of soil sample This value is particularly high compared to Mn(H2 O)2+ 6
with metal solution (2 × 10−4 mol L−1 ) at pH 3.2, 5.0, and one (2.2 mT). This relatively high value is due to an in-
7.7 for iron(III), chromium(III), and manganese(II), respec- teraction of Mn(H2 O)2+ 6 ions with the soil sample surface
tively. (outer-sphere complexes) and can also be explained by the
The EPR spectrum of the Fe-soil system (Fig. 5), recorded formation of hydroxo species such as Mn(OH)+ that can
at 10 K, was obtained after subtraction of the soil sample bind to the surface sites. So these two species are proba-
spectrum recorded at the same temperature. It consists of a bly sorbed on the surface. Moreover, a more precise analysis
strong band at g = 4.157 (157.0 mT) and a weak signal at of the spectrum shows the presence of several superimposed
g = 9.586 (64.0 mT). These peaks are characteristics of an hyperfine signals, which indicates the existence of several
Fe3+ ion in a high-spin state in a tetrahedral or octahedral sites of complexation for manganese onto the soil sample
site. The symmetry is given by D and E terms, which corre- surface. Hence, we cannot exclude the formation of some
spond to the tetragonal and rhombic distortion, respectively. inner-sphere manganese surface complexes.
K. Flogeac et al. / Journal of Colloid and Interface Science 286 (2005) 596–601 601

In the case of chromium(III), the contributions of chrom- References


ium could unfortunately not be extracted from the spectrum
of the Cr–soil system because of the poor resolution of this [1] D.L. Sparks, Environmental Soil Chemistry, Academic Press, London,
spectrum. 1995.
These data clearly show a significant retention (adsorbed [2] M.F. Hochella, in: M.F. Hochella Jr., A.F. White (Eds.), Mineral–
Water Interface Geochemistry, vol. 23, Mineralogical Society of
amount) of iron, chromium, and manganese by the soil sam-
America, Washington, DC, 1990, pp. 87–132.
ple surface at different pH values on a case by case basis. [3] D.R. Roberts, R.G. Ford, D.L. Sparks, J. Colloid Interface Sci. 263
This implies that metal cations are immobilized in soils (2003) 364.
and/or are relatively slowly leached towards aquifers. Our [4] P. Merdy, E. Guillon, M. Aplincourt, New J. Chem. 26 (2002)
results give a starting point for the comprehension of the 1638.
bioavailability of metallic cations in soils and can be used [5] E. Guillon, P. Merdy, M. Aplincourt, Chem. Eur. J. 9 (2003) 4479.
in geochemical models. However, to fully understand these [6] K. Xia, A. Mehadi, R.W. Taylor, W.F. Bleam, J. Colloid Interface Sci.
185 (1997) 252.
processes, further studies examining the influence of pH,
[7] G.S. Pokrovski, J. Schott, F. Farges, J.L. Hazemann, Geochim. Cos-
time, and concentration directly onto natural soil samples mochim. Acta 67 (2003) 3559.
need to be assessed. [8] S.E. Fendorf, G.M. Lamble, M.G. Stapleton, M.J. Kelley, D.L. Sparks,
Environ. Sci. Technol. 28 (1994) 284.
[9] R.L. Ramos, A.J. Martinez, R.M.G. Coronado, Water Sci. Technol. 30
4. Conclusion (1994) 191.
[10] F.C. Richard, A.C.M. Bourg, Water Res. 25 (1991) 807.
This work allows a better understanding of the role of dif- [11] A.R. Walsh, J. O’Halloran, A.M. Gower, Ecotoxicol. Environ. Safety
ferent physicochemical parameters in metal retention, such 27 (1994) 168.
[12] K. Flogeac, E. Guillon, M. Aplincourt, E. Marceau, L. Stievano,
as reaction time, pH, or metal concentration. Indeed, we
P. Beaunier, Y.M. Frapart, Agronomy (2005), in press.
showed that the metal adsorption is relatively rapid (less than [13] A. Kokorevics, J. Gravitis, E. Chirkova, O. Bikovens, N. Druz, Cell.
24 h). The maximum adsorption is reached around pH 4 for Chem. Technol. 33 (1999) 251.
iron, 5 for chromium, and 8 for manganese. The maximum [14] K. Flogeac, E. Guillon, E. Marceau, M. Aplincourt, New J. Chem. 27
amount of metals bound to soil sample was found to be equal (2003) 714.
to 77.5 µmol g−1 for manganese, 104 µmol g−1 for iron, and [15] K. Csoban, M. Parkanyi-Berka, P. Joo, P. Behra, Colloids Surf. A 141
151 µmol g−1 for chromium. Thus, the sorption experiments (1998) 347.
[16] Z. Reddad, C. Gerente, Y. Andres, P. Le Cloirec, Environ. Sci. Tech-
indicated that the soil sample has a great affinity for iron, nol. 36 (2002) 2067.
manganese, and chromium cations. [17] J.C. Echeverría, M.T. Morera, C. Mazkiarán, J.J. Garrido, Environ.
We also characterized the surface complexes using EPR Pollut. 101 (1998) 275.
spectroscopy. This study showed that iron forms relatively [18] D. Goldfarb, M. Bernardo, K.G. Strohmaier, D.E.W. Vaughan,
stable complexes through inner-sphere complexes while H. Thomann, J. Am. Chem. Soc. 116 (1994) 6344.
manganese ions are involved in less stable complexes (mixed [19] R.S.T. Manhaes, L.T. Auler, M.S. Sthel, J. Alexandre, M.S.O. Mas-
sunaga, J.G. Carrio, D.R. Dos Santos, E.C. Da Silva, A. Garcia-
outer- and inner-sphere complexes).
Quiroz, H. Vargas, Appl. Clay Sci. 21 (2002) 303.
[20] A.U. Gehring, R. Karthein, Clay Miner. 25 (1990) 303.
[21] A.U. Gehring, A.M. Hofmeister, Clay Miner. 42 (1994) 409.
Acknowledgments [22] G.W. Hodgson, B. Hitchon, K. Taguchi, B.L. Baker, E. Peake, Geo-
chim. Cosmochim. Acta 32 (1968) 737.
We are grateful to the Région Champagne–Ardenne for a [23] P. Merdy, E. Guillon, J. Dumonceau, M. Aplincourt, Anal. Chim. Acta
grant to K.F. We also thank Y.M. Frapart (Université René 459 (2002) 133.
Descartes, Paris V, UMR 8601, France) for the EPR experi- [24] N. Senesi, G. Sposito, J.P. Martin, Sci. Total Environ. 62 (1987)
ments. 241.