Bergaya F. Handbook of Clay Science

Подождите немного. Документ загружается.

(co-precipitation) as uranate. Uranium-bearing goethites can also be reduced by

bacteria to form insoluble U(IV) complexes (Frederickson et al., 2000, 2002).

The general consensus is that oxyanions on ox ides dominantly form IS complexes

with the bidentate type being the most common (Hayes et al., 1987; Manceau et al.,

1992a, 1992b, 1992c; M anceau and Charlet, 1992, 1994; Waychunas et al., 1993,

1995; Manceau, 1995; Foster et al., 1998a; Furhman et al., 1998; Reich et al., 1998;

Moyes et al., 2000; Manning et al., 2002). However, the IR study by Peak et al.

(1999) shows that sulphate sorption by goethite is OS when the ionic strength is low

and the pH is above 6, but both OS and IS occur at higher ionic strengths and lower

pH. The proportion of IS to OS complexes formed increases as pH falls and ionic

strength rise s. Peak et al. (1999) also noted that the spectra of sorbed SO

2–

goethite resemble those of schwertmannite, a ferric hydroxysulphate species, indi-

cating that IS complexes probably predominate.

B. Mineral Neoformation Reactions: Smectites as Nucleati on Templates

In studying the kinetics of sorption and precipitation reactions of Co

on hectorite,

Schlegel et al. (1998, 1999a, 1999b) found that at near neutral pH and low surface

coverage, sorbed Co forms both OS and IS complexes at the edges of the mineral.

The proportion of IS complexes increased with reaction time, suggesting that Co

migrated from exchange sites to edge sites during precipitation (Schlegel et al.,

1999b). Precipitation of Co

as part of a layer double hydroxide epitaxial on to the

edges was conﬁrmed by P-EXAFS (Schlegel et al., 1999a, 1999b). Initially, sorbed IS

Co had a greater proportion of Si and Mg neighbours in the ﬁrst cation (CoyM

)

shell but with time, the number of Co atoms in the CoyM

shell increased. Sorption

of Co was also accompanie d by Mg

release. Since the reaction ceased when sufﬁ-

cient Mg

had been released, Schlegel et al. (1998, 1999b) proposed incongruent

dissolution of the hectorite, at least initially, and that Co

replaced only the Mg

exposed at the edge surface.

A recent Ni K powder and P-EXAFS study of Ni sorption by montmorillonite

indicated the neoformation of an Ni phyllosilicate without the addition of soluble Si

(Da

hn et al., 2002b, 2003). Other workers (d’Espinosa de la Caillerie et al., 1995a,

1995b; Scheidegger et al., 1996a, 1996b, 1997, 1998; Scheckel and Sparks, 2000) ob-

served the formation of an Ni–Al layer double hydroxide during the sorption of Ni

by pyrophyllite, gibbsite and alumina. Recently, Galoisy et al. (2003) observed that Si

was randomly grafted on to trioctahedral Ni clusters. The Ni K EXAFS spectra of

montmorillonte, reacted with 660 mMNi

at pH 8 and ionic strength of 0.2 M

Ca(NO

)

, showed an Ni-rich precipitate in which Ni (on average) was surrounded by

edge-sharing Ni (2.6Ni, d ¼ 308 pm) and corner-sharing Si (4.2Si, d ¼ 326 pm). This is

consitent with Ni sorbing near edge-exposed Al octahedra (Da

hn et al., 2002b). The

kinetic study showed that Ni-enriched clusters formed within 1 day, with the high Ca

content ensuring negligible Ni

OS complex formation. The angular dependence of

Chapter 12.3: X-ray Absorption Spectroscopy836

the P-EXAFS amplitude indicated that the NiyAl interatomic dimension was

in-plane, and those of the NiySi were out-of-plane with respect to the smectite.

Subsequent reactions with about 20 mMNi

at pH 7.2 and ionic strength of 0.3 M

Ca(NO

)

, indicated structural parameters of 2 edge-sharing Al (d ¼ 300 pm), 2 cor-

ner-sharing Si (d ¼ 312 pm) and 4 other corner-sharing Si (d ¼ 326 pm). The low

values of N

NiyAl

may indicate the absence of octahedral substitution, even after 1 year.

These results were consistent with the neoformation of a dioctahedral Ni phyllosilicate

at the montmoillonite edges (Da

hn et al., 2002b, 2003).

hn et al. (2002b, 2003) were unable to account for sufﬁcient dissolved Si for

precipitation of phyllosilicates. However, the STx-1 montmorillonite used was likely

to contain opal CT (cristobalite–tridamite) as an impurity. Even in theo0.1 mm

fraction of STx-1 sufﬁcient opal CT remained as indicated by moderately broad

XRD reﬂection at about 0.409 nm (Gates, 2004, unpublished data). Opal CT could

well be the source of sufﬁcient dissolved Si for the precipitation of Ni complexes on

STx-1 just as it could be for ThySi signals in the complexes described above. If so,

the addition of Si would not be necessary as it was for hectorite (Schlegel et al., 1998)

which is much more solubl e than STx-1 under the conditions used. Similarly,

Manceau et al. (1999b) showed that a Co

phyllosilicate is neo formed in the pres-

ence of quartz.

Research on the neoformation of minerals raises many questions. How does a

stable mineral phase (e.g., pyrophyllite) generate sufﬁcient soluble Si to react with

sorbed metals? What takes the place of OS-sorbed metals on the exchange complex

of smectite when they ‘switch’ to IS? Does the layer charge change? Why should

sorption of divalent transition metals at the edges of phyllosilicate s result in mineral

dissolution? Montmorillonite is generally far more stable than hectorite at pHp8.

The ability of metals such as Cu

,Ni

,Al

, to undergo hydrolysis (Bargar et al.,

1997a, 1997b, 1998, 2004) and release of H

is well known. These reactions ap-

parently occur even in well-buffered aq ueous systems at pH 7–8 (Hayes and Katz,

1996; Da

hn et al., 2002b, 2003; Galoisy et al., 2003). Protons (H

aqueous

) can create

an unstable H

-smectite. That is, acid attack on sites of isomorphous substitution in

the tetrahedral and octahedral sheets leads to the release of tetrahedral Si

,Al

and octahedral Fe

,Fe

and Mg

into solution (Janek and Komadel, 1993,

1998). These metal cations can then participate in precipitation reactions at nucle-

ation sites, probably at the mineral edges (Scheidegger et al., 1996a, 1996b, 1997,

1998; Scheinost et al., 1999; Schlegel et al., 1999a, 1999b, 2001a, 2001b; Da

hn et al.,

2003). Eventually non-hydrolysable cations (Mg

) released will replace H

,Al

or other metals on the exchange complex , and ‘auto-transformation’ ceases (Janek

and Komadel, 1993). Thus, auto-transformation is initiated by sorbed hydrolysable

metals such as Ni

,Zn

or Co

. Through mineral dissol ution, this reaction can

supply both the cations necessary to satisfy the exchange complex as well as the

soluble Si and other elements needed for the precipitation of the metals into new

phases. Whether auto-transformation reactions are responsible for the precipitation

of metal (hydr)oxide or phyllosilicate phases is another question altogether. EXAFS

12.3.5. XAFS Studies of Reactivity of Clays and Clay Minerals 837

studies would suggest that conventional concepts of low-temperature transformation

of clays and clay minerals require rethinking.

C. Redox Reactions (Abiotic and Biotic) of Clays and Clay Minerals

We already mentioned (Section 12.3.2) that the energies at which XAFS absorptions

occur are sensitive to the ox idation state of the target element. A numb er of re-

searchers (Bajt et a l., 1994, 1995; Delaney et al., 1996a, 1996b, 1998; Berry et al.,

2003; Berry and O’Neill, 2004) used this relationship to estimate the oxidation state

of metals in a variety of mineral samples, including primary aluminosilicates and

(secondary) clays and clay minerals in soils and sediments. For example, the XA NES

energies of the Cr K-edge are very sensitive to Cr oxidation state. Sutton et al. (1993)

differentiated ﬁve valence states of Cr in lunar basalts: Cr[0], Cr[II], Cr[III], Cr[IV]

and Cr[VI], using a microprobe. Olivine minerals were dominated by Cr[II] and

pyroxenes by Cr[III]. The dominance of Cr(II) indicated that the basalt was crys-

tallised under highly reducing conditions. For a wide range of primary minerals the

Fe oxidation state could be reasonably calibrated using the centroid of the Fe pre-

edge signal (Bajt et al., 1994). Variations of this technique were used to calibrate

XANES spectra for biotite micas (Dyar et al., 2001), amphiboles (Delaney et al.,

1996a, 1996b), and silicate glasses (Berry et al., 2003). Inter estingly, the calibration

curves developed by Dyar et al. (2001) were indistinguishable from those of Bajt

et al. (1994) and Delaney et al. (1996a, 1996b), indicating the insensitivity of this

technique to gross mineral differences. Gates et al. (1997) found that features of the

edge crest were more useful for assessing Fe oxidation states of reduced smectites.

Recently, Berry et al. (2003) and Berry and O’Neill (2004) used the derivative of the

absorption edge to determine Cr(II)/total Cr ratios in silicate glasses. They indicate

that local disorder effects should be considered when analysing a speciﬁc class of

minerals.

The oxidation states of structural constituents in clays and clay minerals, or the

reactions of clays and clay minerals with the surrounding environment, were rarely

investigated by XAFS, primarily because of sample handling difﬁculties. For

smectites, such studies were limited to Fe-rich species (Gates et al., 1997; Hunter

et al., 1999; Manceau et al., 2000a) although Cr(VI) redox reactions at smectite

surfaces were reported (Fendorf, 1995; Brigati et al., 2000; Taylor et al., 2000).

Similar reactions of transition metals or metaloids at the surfaces of phyllomangan-

ates (Charlet and Manceau, 1992; Silvester et al., 1995) and (hydr)oxides (Peterson et

al., 1996, 1997; Drits et al., 1997, 1998a; Fitts et al., 2000; Kendelewicz et al., 2000;

Foster et al., 2002; Tournassat et al., 2002) wer e also investigated (Table 12.3.4).

Aspects of some of these studies will be detailed below.

P-EXAFS Study of Reduced Nontronite Structure

P-EXAFS analysis of the structures of reduced nontronite showed that in the R*

300–600 pm region the octahedral sheet differs dramatically from its oxidised or

Chapter 12.3: X-ray Absorption Spectroscopy838

unaltered counterpart (Manceau et al., 2000b). There was evidence to suggest that on

reduction from Fe(III) to Fe(II), Fe atoms migrate from M2 to M1 sites, thereby

forming clusters of vac ancies and domains of trioctahedral character. The increase to

0.154 nm of the XRD d(06-33) reﬂection, and the appearance of an OH stretching

vibration near 3623 cm

–1

on reduction of >90% of the octahedral Fe, were con-

sistent with the formation of trioctahedral domains. The migration of Fe(II) from cis

to trans sites was probably accompanied by dehydroxylation and re-protonation/

hydration (see Chapter 8), in a fashion similar to transitions during thermal de-

hydroxylation of smectite (Drits et al., 1998b).

Reduction of more than 90% of the octahedral Fe(III) resul ted in decreased

amplitude of the ﬁrst two shells in the FTs of the Fe K P-EXAFS spectra. Fitting

inferred a decrease in the number of oxygen ligands and an increase in the FeyO

interatomic distances (6O at d ¼ 201 pm for the oxidised Garﬁeld nontronite as

against 5.3O at d ¼ 210 pm for the form). Likewise the FeyFe

parameters changed

from 4.5Fe

at d ¼ 303 pm for the oxidised nontronite to 4.9Fe

at d ¼ 312 pm for

the reduced form. The FT of the reduced nontronite further showed a loss of the

FeyFe

shell near R*E500 pm, but an increase in the FeyFe

shell near

R*E600 pm, compared to the oxidised nontronite. Recall that a focussing effect

of co-linear multiple-scattering events can increase amplitudes in this region of the

EXAFS spectra of trioctahedral structures. Manceau et al. (2002b) considered that

these features were consistent with such an effect.

The reoxidation of reduced nontronites produced materials that were similar to,

but not identical with, the unaltered nontronite, indicating that the reaction was not

fully reversible, at least following complete reduction of the nontronite (Manceau

et al., 2000b). Infrared studies of highly reduced nontronites and ferruginous

smectites (Komadel et al., 1995; Fialips et al., 2002a, 2002b) support this conclusion,

but other studies with slightly reduced montmorillonite, ferruginous smectite and

nontronite (Gates et al., 1996, 1997; Hunter et al., 1999) indica ted that the reduc-

tion–reoxidation reaction was reversible within the errors of the measur ements. Both

the total Fe content of the smectite and the absolute level of reduction are apparently

important factors affecting the reversibility of the redox reaction (see Chapter 8).

Redox Coupling between Octahedral Fe in Smectites and Sorbed TPB

Gates et al. (1997) and Hunter et al. (1999) used Fe K XANES and Fe K EXAFS to

follow the changes in oxidation state of octahedral Fe in montmorillonite, ferru-

ginous smectite and nontronite during reaction with Na-tetraphenylboron (TPB).

Abiotic oxidative degradation of TPB, occurring at Lewis sites within the interlayers,

resulted in measurable shifts (1–2 eV) and changes to features in the X-ray absorp-

tion edge, indicative of electron transfer to octahedral Fe(III). Analysis of Fe K

EXAFS spectra of the TPB-montmorillonite system revealed an increased spread in

the FeyO bond lengths, consistent with reduction of Fe(III) to Fe(II). Further, the

kinetics of the 2e



transfer from TPB to octahedral Fe were closely similar to those

determined using wet-chemical techniques to quantify TPB degradation (Hunter and

12.3.5. XAFS Studies of Reactivity of Clays and Clay Minerals 839

Bertsch, 1994; Gates et al., 1997; Hunter et al., 1999). Although the Fe(III) content

of nontronite is an order of magnitude greater than that of montmorillonite, the

total amount of Fe(III) reduced by reaction with TPB was nearly identical. It would

therefore appear that the reaction was more efﬁcient in montmorillonite than in

nontronite, but this might be because the nontronite particles were dispersed within

an Fe-free smectite medium (hectorite). The ordering of Fe(III) within the structure

of Wyoming montorillonite does not seem to limit electron transfer from the in-

terlayer.

Cr(VI) and Cr(III) Redox Reactions at Mineral Surfaces

Brigatti et al. (2000) used Cr K XANES to show that Cr(VI) was reduced to Cr(III)

by a natural Fe(II)-bearing chlorite. Another Cr K XANES study (Taylor et al.,

2000) showed that when reduced with dithionite ferruginous smectite mediated the

reduction of sorbed Cr(VI) to Cr( III) with a ratio of 3:1 of structural Fe(II) ox-

idation to sorbed Cr(VI) reduction. Furthermore, their evidence suggested that

Cr(VI) probably sorbed within the interlayer of reduced smectite, as the reaction was

greater for Na-exchanged than for K-exchanged red uced smectite. The oxidation of

Cr(VI) at magnetite surfaces (Peterson et al., 1996, 1997; Kendelewicz et al., 2000)

and iron sulphides (Patterson et al., 1997) results in the formation of immobile

hydrous Cr(III)Fe oxide phases. An interesting reaction occurs for Cr(III) sorption

onto birnessite (Charlet and Manceau, 1993; Silvester et al., 1995). The Cr(III)

initially either migrates into the phyllomanganate structure (ﬁlling vacancies), or

precipitates at the surface, but eventually electron transfer occurs, and the Cr(III) is

oxidised to Cr(VI) by Mn(IV) reduction to Mn(II). The Cr(VI) forms CrO

2

at the

surfaces, which ultimately is an undesirable product as Cr(VI) is more soluble, mo-

bile and toxic than Cr(III).

Redox Reactions at Oxide and Organic Surface s

Abiotic redox reactions of a variety of metals and metaloids on a variety of minerals,

e.g., Se(VI) oxidation on Fe oxyhydroxides (Myneni et al., 1997), As(V) oxida-

tion on birnessite (Foster et al., 2002; Manning et al., 2002; Tournessat et al., 2002)

and anatase (Foster et al., 1998a), Tc(VII) oxidation on FeS and green rust

(Wharton et al., 1999; Bunker et al., 2002) and U(VI) oxidation on FeS (Moyes et al.,

2000) and goethite (Duff et al., 2002) were studied since synchrotron radiation be-

came readily available to the geosciences. In nearly all these cases, the redox reaction

is associated with the formation of a surface precipitate. The sole exception is U(VI)

oxidation at goethite surfaces giving rise to layer substitution or co-precipitation

(Duff et al., 2002).

Other electron transfers coupled with lattice substitutions of transition metals in

oxides were observed. Sorbed Co(II) migrates into the octahedral vacancies within

the structure of phyllomanganates and forms a redox couple with Mn(IV) (Charlet

and Manceau, 1993; Manceau et al., 1997). When sufﬁcient Mn(IV) was reduced to

Mn(II), the latter is expelled from the crystal lattice. Similar reactions were reported

Chapter 12.3: X-ray Absorption Spectroscopy840

for Ce(III) and Tl(I) (Bidoglio et al., 1993) on phyllomanganate surfaces. In common

with their Co(III) counterparts, complexes of Ce(IV) and Tl(III) are much less sol-

uble and remain as part of the manganate structure.

Several studies were recently conducted on biological systems. Schulze et al.

(1995a, 1995b) showed that reactions taking place in the rhizosphere of wheat roots,

infected by the take-all fungus, result in the oxidation and precipitation of Mn(IV) at

the plant root surfaces. Studies by Templeton et al. (1999, 2003a, 2003b) showed that

Pb redox reactions at mineral surfaces can be mediated by microbial bioﬁlms.

Tokunaga et al. (1997, 1998) studied Se and Cr redox reactions in soil aggregates

mediated by soil microorganisms.

12.3.6. XAFS STUDIES OF OTHER CLAY MINERALS

In order to gain a better understanding of the XAFS spectral features of smectites

and oxyhydroxides, a host of reference minerals were studied in detail. The accu-

mulated information was thoroughly reviewed elsewhere (Hodgson et al., 1984;

Koningsberger and Prins, 1988; Hawthorne, 1988; Decarreau, 1990; Mottana and

Burragato, 1990; Schulze et al., 1999; Bertsch and Hunter, 2001; Fenter et al., 2002).

A. Other Layer Silicates

As shown in Table 12.3.4 , most XAFS studies of layer silicates (e.g., kaolin, illite,

sepiolite, pyrophyllite, talc, vermiculite/hydrobioite) wer e conducted in conjunction

with sorption studies, either as sorbents or reaction products. However, Bugaev et al.

(1998) showed that, for model Al-containing compounds having from 4-fold to

12-fold coordination, Al K XANES spectra could be reproduced considering single-,

double- and triple-scattering processes on multi-atom chains. They reported that the

Al K XANES spectrum of pyrophyllite could be reproduced with excellent agree-

ment with single and linear multiple-scattering processes occurring between atoms of

the ﬁrst ligand shell: d(AlyO) ¼ 2O at 188 pm and 4O at 192 pm. The maximum

divergence angle for the multiple-scattering paths was 141 and the maximum total

path length was 1800 pm. Spectral ﬁtting of diaspore (a-AlOOH), gave a similar

total path length, but a much smaller divergence angle from linear scattering paths

(o41) with a greater degree of distortion about the Al octahedra (d(AlyO) ¼ 2O at

185, 1O at 186 and 3O at 198 pm) (Bugaev et al., 1998). For kaolinite Bugaev et al.

(1998) found an inversion of the amplitude of two peaks associated with octahedral

Al compared with other Al-bearing layer silicates (Fig. 12.3.3). This was ascribed to

the presence of two distorted Al octahedral sites: site 1 d(AlyO) ¼ 2O at 188, 4O at

194 pm and site 2 d(AlyO) ¼ 2O at 189 and 4O at 194 pm. The parameters for site 2

octahedra increased the amplitude of the second octahedral Al peak. This illustrates

the potential usefulness of XANES (NEXAFS) in identifying and quantifying local

structural information in a similar manner as EXAFS.

12.3.6. XAFS Studies of Other Clay Minerals 841

B. Oxides and Oxyhydroxides

The structures and reactivity of Fe and Mn (hydr)oxides were studied in detail since

the availability of synchrotron X-ray sources. Such studies provided tremendous

insight into many low-temperature geochemical reactions that take place in soils and

sediments. Synchrotron-based XAFS analyses, applied to the condensation of iron

oxyhydroxide precipitates from gels (Combes et al., 1986, 1989a, 1990; Hazemann et

al., 1992; Manceau and Drits, 1993; Bottero et al., 1994; Ford et al., 1999a), showed

that the rate of hydrolysis controls the local structural order beyond the ﬁrst ligand

shell (Charlet and Manceau, 1993). The polymerisation of Fe into clusters is only

observed by EXAFS as peaks in the FT near 305 and 345 pm, when the hydrolysis

rate exceeds 1OH per 1Fe. The polymeric gels tend to be locally ordered with struc-

tures similar to a-FeOOH (goethite) or b-FeOOH (akaganeite), respectively, if the

products are prepared from Fe(III)-nitrate or Fe(III)-chlo rite salts. Such local struc-

tural order is preserved during the crystallisation and dehydration of the gels

(Combes et al., 1989b; Drits et al., 1993; Manceau and Drits, 1993). Synchrotron-

based SAXS studies (Bottero et al., 1994) showed that the hydrolysis rates inﬂuence

crystallite formation and aggregation in (hydr) oxide gels. When the fractal dimen-

sion exceeds 2, aggregate ﬂocculation leads to the formation of HFO or d-FeOOH

(ferrioxyhite or 2-line ferrihydrite) precipitates. On further ageing, these precipitates

convert into hematite. Similar reactions occur with the M n (hydr)oxides, although

these appear to be more complex (Charlet and Manceau, 1993) because of the ex-

istence of a wide variety of Mn oxide polymorphs. The precipitation of Fe gels in the

presence of SiO

was studied by Waychunas et al. (1999), Doelsch et al. (2000, 2002)

and Masion et al. (2001). The studies of Masion et al. (2001) showed that at low

hydrolysis rates, Si can hinder the precipitation of Fe(II) gels, but not that of Fe(III)

gels, indicating that hydrolysis reactions control both sho rt- and long-range order

formation.

XAFS studies of the sorption of metals and metaloids on surfaces of oxides, or

their coprecipitation within the structure, provide information on the nature of the

reactive sites (e.g., Com bes et al., 1992; Bajt et al., 1993; Spadini et al., 1994; Waite

et al., 1994; Bargar et al., 1996; Bochatay et al., 1997; Axe et al., 1998, 2000; Collins

et al., 1998, 1999; Randal et al., 1998, 1999; Ford et al., 1999b ; Manceau et al.,

2000c, 2000d; Arai et al., 2001). It is commonly assumed that speciﬁc crystallo-

graphic faces have speciﬁc reactivities and are differentially responsive to sorbing

metals/metaloids due to the dynamics of surface potential, metal hydrolysis, surface

hydration and steric effects (e.g., Manceau, 1995; Ford et al., 1997a; Ford and

Bertsch, 1999; Criscenti and Sverjensky, 2002).

The crystal dimensions of Fe(III) oxides formed in the presence of other metals

are affected by the concentration and type of meta l impurity (Ford et al., 1997b).

However, Manceau et al. (2000c) showed that the bulk structures of two different

FeOOH polymorphs strongly inﬂuence the types and amounts of sorption complexes

formed, at least for weakly hydrolysing metals. The goethite (a-FeOOH) structure is

Chapter 12.3: X-ray Absorption Spectroscopy842

composed of chains of double rows of Fe(III) octahedra, in which the octahedra

share edges within the chains and corners between chains. The lepidocrocite

(g-FeOOH) structure is composed of chains of double edge-sharing octahedra, but

the octahedra share edges between chains rather than corners, as in goethite. Cd K

EXAFS clearly shows that at low surface coverage (11–13%) isolated Cd octahedra

(CdyO d ¼ 224–225 pm) share corners (CdyFe d ¼ 380–390 pm) and edges

(CdyFe d ¼ 329 pm) with Fe octahedra in both goethite and lepidocrocite. At

high surface coverage, only isolated edge-sharing Cd octahedra (bidentate) were

observed in lepidocrocite, and only isolated corner-sharing Cd monodentate octa-

hedra were detected in goethite. Manceau et al. (2000c) suggested that in addition to

differences in bonding between chains in these structures, there are differences in the

stacking sequences of the anion layers (hexagonal in goethite and cubic in lepid-

ocrocite). The resultant morphological forms (prismatic to acicular in goethite and

lath-like in lepidocrocite) inﬂuence the sorption complexes formed.

An ongoing controversy concerns the presence of tetrahedral Fe(III) in the

structure of many iron (hydr)oxides. Manceau et al. (1990b), Manceau and

Drits (1993) and Drits et al. (1993) proposed that the (hydr)oxides of iron, and

presumably other metals, were structurally similar in that inter-conversion

and mixing of polyhedra could adequately describe the EXAFS spectra. Thus, fe-

rrihydrite was proposed to be a mixture of (i) defect-free goethite-like structure with

double chains of edge-sharing octahedra; (ii) defective goethite-like structure, with

the proportion of defect sites being higher for 2-line than 6-line ferrihydrite; and

(iii) ultra-di spersed hematite grains with scattering domain sizes o2nm (Drits et al.,

1993). Others related the various features in the XAFS spectra and XRD patterns

of ferrihydrite to tetrahedral Fe(III) (Eggleton and Fitzpatrick, 1988; Zhao et al.,

1993, 1994).

Whether or not tetrahedrally coordinated Fe(III) exists in ferrihydrite is not sim-

ply an academic question. Ferrihydrite readily converts into maghemite when

heated, and may constitute a major source of maghemite and goethite in soils (see the

review by Jambor and Dutrizac, 1998). Ferrihy drite contains much structural (co-

ordinated) water that is readily lost with common laboratory treatments. Such de-

watering is important in the conversion of ferrihydri te into goethite or maghemite,

and the formation of maghemite from ferrihydrite gives rise to tetrahedral Fe(III)

sites. Thus the stoichiomet ries of ferrihydrite and other poorly crystalline hydr(ox-

ides) are important, especially in determining the amount of ‘crystalline’ water

(Mavrocordatos and Fortin, 2002). The accumulated evidence from XAFS (Calas

and Petiau, 1983; Waychunas et al., 1983; Manceau and Combes, 1988; Combes

et al., 1989a, 1989b, 1990; Manceau et al., 1990b; Drits et al., 1993; Manceau and

Drits, 1993; Waychunas et al., 1993; Bottero et al., 1994; Galoisy et al., 2001a,

2001b) and photoelectron energy loss spectroscopy (TE M PEELS) (Mavrocordatos

and Fortin, 2002) strongly indicate that for oxyhydroxides of Fe in soils and se-

diments, as well as in mineral and glass phases, tetrahedral coordination of Fe

the exception rather than the rule.

12.3.6. XAFS Studies of Other Clay Minerals 843

In nano-size ferrihydrite particles, a signi ﬁcant proportion (up to 30%) of the

Fe(III) at the crystallite edges could be coordination under-saturated (CUS) with

respect to the oxygen and hydroxyl framework (Zhao et al., 1994) The occurance of

highly distorted octahedra would impact on the amplitude of the XAFS signal

(Manceau and Gates, 1997). The bulk Fe atoms are coordinated to oxo (O) and

hydroxyl (OH) bridging anions, whereas the edge Fe atoms have bridgi ng OH anions

to the bulk structure but non-bridging H

O ligands to satisfy the excess valence

charge. Distortion effects of Fe octahedra (Fig. 12.3.3A) at the surfaces of nano-

crystalline ferrihydrite (due to coordination with water molecules and hydroxyls)

were shown by Manceau and Gates (1997) to be partly responsible for the enhanced

Fe K pre-edge XAFS features observed by Zhao et al. (1993, 1994). Manceau and

Gates (1997) further proposed that two types of H

O molecules were present in

ferrihydrite nano-crystals: those directly coordinated to the surface Fe atoms, and

those hydrogen-bonded to the coordination water. The doubly hydrogen-bonded

water (forming two H-bonds with two other coordination waters) is more stable than

the singly hydrogen-bonded water. That is, the latter can be removed at relatively

moderate humidities or low temperatures. High-resolution thermogravimetry (Ford

and Bertsch, 1999) lends considerable support to this hypothesis.

Finely divided goethite particles with varied morphologies were shown to have

three reversible surface dehydration/dehydroxylation temperatures: near 195–215,

220–240 and 240–260 1C, probably representing H

O/OH bonded to 1 (S

-OH),

2(S

-OH) and 3 (S

-OH) surface atoms, respectively. These three dehydration/

dehydroxylation events preceded the bulk dehydroxylation temperature near

270–280 1C; their resolution from each other and from the bulk dehydroxylation

appears to correspond to the degree of crystallinity of the goethite which, in turn,

depends on the ferrihydrite-to-goethite transformation rate.

Similar surface-induced effects were shown to occur on other mineral surfaces,

such as freshly cleaved g-Al

in water (England et al., 1999; Eng et al., 2000) and

when divalent heavy metals are substituted for Ca

in calcite (Reeder et al., 1999).

Changes in bond strength and residual charge on ligands, with changes in bond

length, are the origin of the high capacity of ultra small particle surfaces to take up

metals and metaloids from solution (Reeder et al., 1994, 2000, 2001; Lamble et al.,

1995, 1997; Rakovan et al., 2002). Distorton of surface coordination environments

undoubtedly plays a pivotal role in the formation of surface complexes and copre-

cipitates (Scheinost et al., 1999, 200 1 ; Manceau et al., 2000d; Mavrocordatos and

Fortin, 2002; Waychunas et al., 2002a, 2002b).

12.3.7. FUTURE OUTLOOK AND CONCLUDING REMA RKS

Many questions often emerge from impr oved understanding of nature. Synchro-

tron-based studies on clays and clay minerals certainly raised their fair share of

questions. For example, why is the M1 site more energetically favourable than the

Chapter 12.3: X-ray Absorption Spectroscopy844

M2 site for Fe(II)? Why are sorption sites at clay surfaces apparently more selective

or site-speciﬁc at low c oncentrations of a particular electrolyte than at high con-

centrations? Is this really selectivity or simply a function of the number of ‘‘ideal’’

sites available at a surface? How exactly do dioc tahedral smectites, or even minerals

such as quartz, act as templates for neoformation of trioctahedral phyllosilicate

species? How does the inherently ﬂexible structure of the 2:1 layer maintain its

crystalline integrity after such apparently destructive processes as redox or heat-

induced de hydroxylation? Questions such as these still face researchers attempting to

understand the nature and properties of smectites and related poorly crystalline

minerals.

In recent years we entered a truly new age of scientiﬁc investigation, one where

dynamic molecular modelling will progress to the forefront of improving our un-

derstanding of clay mineral structures and reactivity. But all models must be val-

idated by experiment. Without experimental fact a model is nothing more than an

opinion, and often an expensive one at that. In combination with other more

conventional and routine techniques, synchrotron-based studies will continue to play

an important part in gathering these experimental facts. As this review showed

the increased application of synchrotron-based analyses over the last decade greatly

improved our understanding of the structures and reactions of poorly crystal-

line clays and clay minerals. We should point out, however, that synchrotron-

based studies by themselves could generally only provide a limited amount of new

information. Only when combined with other spectroscopic and/or wet chemical

methods, can synchrotron-based techniques provide a ﬁrm foundation for models

to stand on.

REFERENCES

Allard, T., Ildefonse, P., Beaucaire, C., Calas, G., 1999. Structural chemistry of uranium

associated with Si, Al, Fe gels in a granitic uranium mine. Chemical Geology 158, 81–103.

Amonette, J.E., Zelazny, L.W. (Eds.), 1994. Quantitative methods in soil mineralogy. Pro-

ceedings of a Symposium Sponsored by Division S-9, San Antonia, TX, October, 1990.

SSSA Miscellaneous Publication. Soil Science Society of America, Madison, WI, 462pp.

Ankudinov, A.L., Ravel, B., Rehr, J.J., Conradson, S.D., 1998. Real-space multiple-scattering

calculation and interpretation of X-ray absorption near-edge structure. Physics Review B

58, 7565–7576.

Arai, Y., Elzinga, E.J., Sparks, D.L., 2001. X-ray absorption spectroscopic investigation of

arsenite and arsenate adsorption at the aluminium oxide–water interface. Journal of Col-

loid and Interface Science 235, 80–88.

Axe, L., Bunker, G.B., Anderson, P.R., Tyson, T.A., 1998. An XAFS analysis of strontium at

the hydrous ferric oxide surface. Journal of Colloid and Interface Science 199, 44–52.

Axe, L., Tyson, T.L., Trivedi, P., Morrison, T., 2000. Local structure analysis of strontium

sorption to hydrous manganese oxide. Journal of Colloid and Interface Science 224,

408–416.

References 845