Bergaya F. Handbook of Clay Science

Подождите немного. Документ загружается.

these studies is within error for coordination number determinations, the interatomic

distances are different . Contributions from wave vectors at higher k inﬂuence the ﬁt

to wave vectors at lower k, resulting in different R

eff

. Interestingly, incorporating

more shells at higher R*, and ignoring any contributions of multiple-scattering

events, improve the ﬁt on R

eff

. This can be seen by comparing the results in Table

12.3.1 with those of Vantelon et al. (2003).

F. Polarised EXAFS Spectroscopy

Application of P-EXAFS to layer silicat es is unique in that it allows all out-of-plane

contributions to the EXAFS spectrum to be systematically removed, and accurate

structural information about the in-plane contributions only (and vice versa) to be

obtained. A backscattering phase shif t between a heavy (e.g., Fe, Mn, Ni, etc.) and a

light (e.g., Al, Mg) atom enhances quantiﬁcation of the heterogeneous nature of

octahedral and tetrahedral cation occupancies in smectites and other layered min-

erals. P-EXAFS also has the inherent ability to increase spectral resolution enabling

differentiation of atomic shells separated by only 10–20 pm (Manceau et al., 1998).

In powder EXAFS, the resolution power in the R* range is limited by low signal-to-

noise ratios in the high k region. On the other hand, the P-EXAFS method increases

the signal-to-noise ratio at high k, thereby improving the resolving power in R*.

For powdered systems (random disorder of the particles) P-EXAFS spectra taken

at several angles will show little, if any, variation in amplitude (Manceau, 1990). For

self-supporting clay mineral ﬁlms where the layers are aligned, changes in the ﬁlm

angle (a) with respect to the X-ray beam will alter the amplitude (pleochroism).

Backscattering atoms situated out-of-plane (tetrahedral Si, Al or Fe, structural ox-

ygen) from the target absorber (e.g., octahedral Fe) will contribute positive

pleochroism (amplitude enhancement) and those atoms within the ﬁlm plane (e.g.,

Al, Mg, Fe) will contribute negative pleochroism to the EXAFS amplitude with

increasing a.

Exactly how the P-EXAFS amplitude varies with a for Garﬁeld nontronite

(Manceau et al., 1998) is shown in Table 12.3.2 and Fig. 12.3.7A .InTable 12.3.2, the

term b is the average angle between the absorber and the backscattering atoms of a

particular shell, with respect to the ﬁlm plane normal, and N

cryst

the number of

Fig. 12.3.6. Analysis of ligand (FeyO

) and nearest cation (FeyM

) contributions to the k

weighted Fe K EXAFS spectra of ferruginous smectite SWa-1, collected on an oriented ﬁlm at

a ¼ 351 (equivalent to a powder EXAFS). The FeyO

and FeyM

shells were isolated from

the Fourier transform at R* ranges of 0.8–2.05 A

and 2.05–3.2 A

, respectively, and back-

transformed (top panel). The FeyO

(upper middle panel) and FeyM

(lower middle panel)

shells were ﬁt to the structural parameters displayed. A ﬁt of the total contribution of the

FeyO

and FeyM

shells, as well as higher R* shells, to the EXAFS is shown in the bottom

panel. See Table 12.3.1 for ﬁtting parameters of this latter ﬁt.

Chapter 12.3: X-ray Absorption Spectroscopy806

4 6 8 10 12

Wavevector k (Å

-1

)

χ(k)

Fe K EXAFS

Fe-O

shell

Fe-M

shell

Fe K EXAFS

FEFF Fit

χ(k)

Fe-O

shell

FEFF Fit

χ(q)

2 O d = 201 pm

4 O d = 202 pm

SWa-1

Thin film

α = 35˚

Fe-M

shell

FEFF Fit

χ(q)

2.1 Fe d = 305 pm

0.9 Al d = 303 pm

4 Si d = 237 pm

See Table 1 for Fit

12.3.1. Synchrotron-Based Techniques 807

atoms of the shell of the idealised structure. The relationship between N

cryst

and the

effective number of neighbouring atoms (N

eff

) actually seen by P-EXAFS at a given

ﬁlm angle, is given by (Manceau et al., 1999)

k; aðÞ¼

eff

cryst

iso

kðÞ ð9Þ

In Eq. (9), w

(k,a) is the amplitude determined by P-EXAFS for an atomic pair as

a function of reciprocal k-space and ﬁlm angle a and w

iso

(k) the amplitude deter-

mined on perfectly random systems. N

eff

for each ﬁlm angle is given by

eff

¼ 3N

cryst

cos

bsin

a þ

sin

bcos



ð10Þ

Note that at a ¼ 01

eff

cryst

sin

ð11Þ

and at a ¼ 901

eff

¼ 3N

cyrst

cos

b ð12Þ

From knowledge of the crystallographic angle, b, between the absorbing and the

backscattering atoms, as well as the idealised number of neighbours, N

cryst

,itis

possible to calculate the effective number of neighbours, N

eff

, observed at any angle

in a P-EXAFS experiment. Alternatively, b canbecalculatedfromknowledgeofN

eff

determined from P-EXAFS spectra or from simulations. The ability to predict N

eff

Table 12.3.2. Angular dependence of some atomic shell contributions to P-EXAFS spectra of

Garﬁeld nontronite. Modiﬁed after Manceau et al. (1999a), with R* calculated using FEFF.

The b angle is measured with respect to ﬁlm normal. See text for deﬁnitions

Shell R* (pm) ob>(1) N

cryst

eff

FeyO

197, 204 57 6 6.3 6.2 6.0 5.6 5.4 5.3

FeyFe

304 90 3 4.5 4.0 3.0 1.1 0.1 0

Fey(Si,Al)

326 32 4 1.7 2.5 4.0 7.3 8.4 8.6

FeyO

345 11 2 0.1 0.8 2.0 4.4 5.6 5.8

FeyO

374, 382 73 6 8.2 7.4 6.0 3.2 1.7 1.5

FeyO

402, 419 37 4 2.2 2.8 4.0 6.3 7.5 7.6

Fey(Si,Al)

449 52 4 3.7 3.8 4.0 4.3 4.5 4.5

FeyNa 510 15 2 0.2 0.8 2.0 4.2 5.4 5.6

FeyFe

528 90 6 9 7.9 6.0 2.3 0.3 0

FeyFe

609 90 3 4.5 4.0 3.0 1.1 0.1 0

Chapter 12.3: X-ray Absorption Spectroscopy808

0°

20°

35°

60°

90°

12.5105 7.52.5

Wavevector k (Å

-1

)

χ(k)

0 153045607590

eff

Film angle α

Fig. 12.3.7. (A) Variation of N

eff

as a function of ﬁlm angle a for Fe K P-EXAFS spectra. (B)

Polarised k

weighted Fe K EXAFS spectra of ferruginous smectite SWa-1. The a ¼ 901

spectrum was calculated from a linear extension of the experimental amplitude as a function of

cos

a. Isobestic points occur where the EXAFS amplitude (k

w(k)) is invariant regardless of a,

and separate in-plane contributions, where k

w(k) increases with increasing a, from out-of-

plane contributions, where k

w(k) decreases with increasing a.

12.3.1. Synchrotron-Based Techniques 809

from structural information provides better ﬁts of the spectra, as it decreases the number

of ﬁtting variables, or at least puts realistic limits upon them.

Table 12.3.2 shows that for shells whose atoms lie at shallow angles with respect to

theﬁlmnormal(e.g.,FeyO

,FeyO

and Fey(Si,Al)

), large increases in N

eff

occur

with increasing a. For those shells having atoms residing at steep angles with respect to

theﬁlmnormal(e.g.,FeyFe

,FeyFe

and FeyO

), N

eff

decreases by

differing amounts: N

eff

diminishes to 0 for the FeyFe

,FeyFe

and FeyFe

shells at

a ¼ 901, but to 1.5 for O

. For shells with atoms lying near b ¼ 551,littlechangeoccurs

in N

eff

as a function of a.Thus,b  551 is a critical angle where N

eff

is invariant for all

values of a.Ataﬁlmangleofa  351, the effective number of neighbours is approx-

imately equal to the crystallographic number (N

eff

E N

cryst

), a fact demonstrated by

Manceau et al. (1988) for single crystal and powdered mica. Thus, a  351 is another

critical angle in that P-EXAFS spectra collected at a  351 will be identical to powder

EXAFS spectra.

The in-plane contributions can be completely removed from an a ¼ 901 EXAFS

spectrum, whi ch can then be used to estimate the magnitude of out-of-plane con-

tributions. This estimate is then used to remove any out-of-plane contributions from

the a ¼ 01 spectrum to obtain an estimate of in-plane contributions only. Trans-

mission measurements at a ¼ 901 are not easily obtained experimentally—it should

be possible to use glancing angle or wide angle X-ray scattering (GAXS or WAXS)

analyses—and P-EXAFS data are usually collected at several angles of the self-

supporting ﬁlm with respect to the X-ray beam (Figs. 12.3.7B and 12.3.8A).

Processing P-EXAFS data requires careful ampli tude normalisation because the

amount of sample probed increases with increasing a (it doubles from a ¼ 01 to 601).

The normalised a ¼ 01,201,351, y ,601 spectra for a given sample are then used to

calculate an a ¼ 901 spectrum by linear regression of the k

w(k) (or k

w(k)) amplitude

against cos

a, for all points in reciprocal (k) space (Manceau et al., 1998). The

resulting calculated a ¼ 901 spectrum is then processed in the same way as for other

EXAFS spectra, the difference being that the a ¼ 901 contributions are isolated from

the forward FTs and subtracted from those isolated from the a ¼ 01 spectrum to

obtain a partial EXAFS spectrum containing only in-plane contributions.

Recall that the FeyM

shell in a powder EXAFS spectrum of a smectite is a

composite of in-plane (octahedral cations) and out-of-plane (ligand and tetrahedral

cations) contributions. The difference of the resulting in-plane FeyM

partial

P-EXAFS spectra from that obtained on powder samples where the out-of-plane

scattering events contribute, is striking (Fig. 12.3.8B) and, as was shown abov e,

inﬂuences the resulting analysis (Table 12.3.1 and discussion in Section 12.3.2D). As

the subtraction process removes the out-of-plane contributions, a phase shift to

lower wave vectors is observed in the partial FeyM

P-EXAFS spectrum relative to

the partial FeyM

powder EXAFS spectrum.

The a ¼ 901 contributions are normally only calculated for a limited k(q) range,

but there is no reason, other than increased complexity, that the entire spectrum

could not be processed and analysed in a similar fashion. Detailed analysis of highly

Chapter 12.3: X-ray Absorption Spectroscopy810

Wavevector k (Å

-1

)

Powder EXAFS

SWa-1

Fe-M

shell

P - EXAFS

(k)

3 5 7 9 11 13

046

FT (k

χ(k)

R* (Å)

0°

20°

35°

60°

90°

Fe-O

Fe-Fe

Fe-Si

Fe-O

Fe-Na

Fig. 12.3.8. (A) The resulting Fourier transforms of k

weighted Fe K P-EXAFS in Fig.

12.3.7B. Assignments to shells based on Manceau et al. (2000a). (B) Comparison of the

FeyM

shell isolated from the a ¼ 351 spectrum of ferruginous smectite SWa-1 before

(powder EXAFS) and after (P-EXAFS) subtracting the tetrahedral contributions using the

procedures described in Section 12.3.3.

12.3.1. Synchrotron-Based Techniques 811

textured self-supporting nontronite ﬁlms (Manceau et al., 1998, 2000a, 2000b) in-

dicates that P-EXAFS spectra returns the appropriate values calculated from theory.

Interestingly, Schlegel et al. (1999a, 1999b, 2001a) and Da

hn et al. (2002a, 2002b,

2003) recently collected P-EXAFS data at angles as high as a ¼ 801. Given the

uncertainties associated with measurement and extraction of EXAFS amplitudes, as

well as the dispersion inherent in self-supporting ﬁlms (Manceau et al., 1998, 1999;

Schlegel et al., 1999a, 1999b; Manceau et al., 2000a, 2000b; Schlegel et al., 2001a;

hn et al., 2002a, 2002b, 2003) this measuremen t is essentially equivalent to one

calculated from an a ¼ 901 spectrum (compare the N

eff

values at a ¼ 801 and a ¼ 901

in Fig. 12.3.7A).

For further information on experimental set-up, sample preparation and spectral

analysis of P-EXAFS data, the reader is referred to the papers by Manceau et al.

(1998, 1999, 2000a) and Manceau and Schlegel (2001). Poor alignmen t of individual

particles within self-supporting ﬁlms remains a potential limitation of P-EXAFS as

this will result in dispersion, amplitude dampening for out-of-pl ane features, am-

plitude augmentation of in-plane features and loss of linearity in the pleochroic

relation. However, for typical smectite self-supporting ﬁlms, the error arising from

this condition is either small enough to be neglected or can be corrected using charts

prepared by Manceau and Schlegel (2001).

12.3.2. XAFS STUDIES ON SMECTITE STRUCTURE

A. Structural Reﬁnement by P-EXAFS

It is only recently that P-EXAFS was applied to investigate the structures of ﬁne-

grained and poorly crystalline layer silicates (Manceau et al., 1998, 1999a; Schlegel

et al., 1999a; Manceau et al., 2000a, 2000b; Schlegel et al., 2001a; Da

hn et al., 2002a,

2002b, 2003), although it was earlier used to study the location of Fe in micas and

chlorites (Manceau et al., 1988, 1990a; Manceau, 1990; Waychunas and Brown,

1990; Dyar et al., 2002) and to determine bond angles in graphite intercalates

(Bonnin et al., 1986). Heald and Stern (1977) were the ﬁrst to use anisot ropic X-ray

absorption to study single crystals. Manceau et al. (1988) applied P-EXAFS to

examine Fe distribution in chlorite and biotite, observing that the second peak

(FeyM

) in the FT occurred at shorter R* distance when the electric vector was in-

plane with the octahedral sheet than when it was out-of-plane. Direct comparison of

the back-transformed partial EXAFS spectra with that of biotite showed that about

25% of the total Fe was located in interlayer sites of the chlorite.

P-EXAFS was extended to structural reﬁnements of nontronite (Manceau et al.,

1998, 1999a, 2000a, 2000b) and phyllomanganates (Manceau et al., 1999a) and more

recently to determine the location of adsorbed metals on hectorite (Schlegel et al.,

1999a, 1999b, 2000) and montmorillonite (Da

hn et al., 2002a, 2002b, 2003) surfaces.

P-EXAFS improved our understanding of sorption sites (Hazemann et al., 1992;

Chapter 12.3: X-ray Absorption Spectroscopy812

O’Day et al., 1994a; Schlegel et al., 1999a, 1999b, 2001a, 2001b; Da

hn et al., 2002a,

2002b, 2003). The technique is also potentially capable of assessing site occupancies

of Al and Mg in Fe-poor beidellites, montmorillonites and saponites, where

NMR fails (Muller et al., 1997; Manceau et al., 2000a; Vantelon et al., 2003).

P-EXAFS has yet to be applied to obtain structural reﬁnements of other layered

minerals such as palygorskites, sepiolites and layered double hydroxides. In addition,

P-EXAFS would be suitable for assessing trace metal substitution (e.g., Zn or Mn

for Al) in smectites formed under various geochemical conditions or exposed to

different weathering regimes.

The structures of some nontronites and ferruginous smectite were reﬁned by

Manceau et al. (1998, 1999a, 2000a) using a combination of P-EXAFS and mod-

elling. Distance-valence least-squares (DVLS) modelling of the c2/m symmetry of

nontronite enabled the contribution of atomic scattering paths (from FEFF) to

partial EXAFS spectra to be determinded (Manceau et al., 1998). The DVLS and

FEFF method was found to produce calculated spectra in excellent agreement with

experimental spectra. In addition, Manceau et al. (1999a) provided a method to

determine crystallographic angles associated with the layered structure. Manceau

et al. (2000a) found that Fe(III) predominantly occupies M2 (cis) octahedra in

agreement with earlier electron diffraction studies (Tsipursky and Drits, 1984). For

the Fe-poor smectites, Vantelon et al. (2003) recently observed that modelling pow-

der EXAFS spectra in which Fe(III) cations occupied both M1 (trans) and M2

octahedra provided good agreement with experimental spectra, but attempts at fur-

ther reﬁnement of the montmorillonite structure using P-EXAFS have yet to be

published.

B. Dioctahedral vs. Trioctahedral Structural Types

Trioctahedral smectites may readily be distinguished from dioctahedral smectites by

XRD (see Chapter 12.1) and IR spectr oscopy (see Chapter 12.6). Since the b unit cell

parameter is longer in trioctahedral smectites, the d(060) (or d(06-33)) line in the

XRD pattern is close to 0.153 nm as compared with E0.149 nm for dioctahedral

smectites (Brindley and Brown, 1980). The OH stretchi ng region in the IR spectrum

of trioctahedral smectites is shifted by about 100 cm

–1

to higher energies relative to

dioctahedral smectites of similar composition, and also shows considerable

pleochroism under plane polarisation (Farmer, 1974).

P-EXAFS is arguably the best XAFS technique for distinguishing tri- from di-

octahedral smectites because the out-of-plane contributions from the tetrahedral

sheet can be isolated and removed (Manceau et al., 1988, 1998; Manceau, 1990). In

fact, if the tetrahedral Si and Al contributions are not accounted for in powder

EXAFS analysis of smectites, the contribution from octahedral Al and Mg would be

overestimated. This is due to overlap of the backscattering phases for these atoms

at the two different interatom ic distances typically associated with octahedra and

tetrahedra (Fig. 12.3.5)(Manceau et al., 1988). As a result, the in-phase difference

12.3.2. XAFS Studies on Smectite Structure 813

between MySi

(tet)

and MyAl

(oct)

shells due to differences in interatomic distances

(MySi

(tet)

E316–328 pm; MyAl

(oct)

E300–310 pm) is cancelled. If the octahedral

sheet contai ns heavy atoms, these atoms would be underestimated if the tetrahedral

contributions were not fully accounted for. The same would also occur if the con-

tributions from lighter octahedral atoms are not taken into account, or the lighter

atoms themselves are underestimated. As will be shown in Sections 12.3.1C and

12.3.1D, much progress was made in dealing with complex chemistries by combining

P-EXAFS with other methods.

In powder EXAFS, trioctahedral character is indicated by the amplitude and

position of the ﬁrst cation shell (Muller et al., 1997). This shell could be used a priori

(Manceau et al., 1998) to distinguish between the two groups of minerals since an

absorber would be surrounded by six other octahedral cations in a trioctahedral

structure, but only by three other cations in a dioctahedral structure. However, when

Al, Mg and different oxidation states of Fe are present, misinterpretations can easily

be made. Manceau (1990) therefore recommended that EXAFS be used in a sup-

portive role for such studies. The ﬁrst cation shell in the FT of the Fe K P-EXAFS

spectra is shifted to slightly lower R*(30 pm) and is only 50–60% as intense for

biotite (trioctahedral mica) compared with nontronite (dioctahedral smectite), de-

spite there being only 3Fe

nearest octahedral neighbours in nontronite compared

with 4Fe

and 2Al or Mg nearest octahedral neighbours in biotite (Manceau,

1990). The magnitude of amplitude change, as a function of ﬁlm angle, is greater in

biotite than in nontronite. Two strongly pleochroic shells, in the 380–620 pm R*

range are observed for biotite, but not for nontronite. This indicates increased co-

herence in both interatomic distances and crystallographic angles in the trioctahedral

structure as compared with its dioctahedral counterpart. A third cation shell

(FeyFe

) at about 580 pm is observed for biotite (Manceau et al., 1998), the am-

plitude of which is increased by constructive inter ference between outgoing and

incoming scattered waves in co-linear multiple-scattering events (O’Day et al.,

1994c). These co-linear scattering paths do not exist to any great extent in dioc-

tahedral structures (Manceau et al., 1998). In a dioctahedral structure, tetrahedral

sheet rotation and tilting results in a splitting of FeyO

and Fe yO

interatomic

distances into two dist inct ranges (Table 12.3.2), thus diminishing the overall signal

of these ligand shells due to interference. These and add itional ligand shells at R*as

high as 600 pm also interfere with octahedral cation backscattering contributions

(Fig. 12.3. 5 ).

C. Tetrahedral Cation Distributions in Smectites

EXAFS Spectroscopy

Since Bonnin et al. (1985) ﬁrst applied Fe K pre-edge XAFS to determine the

amount of tetrahedral Fe in Garﬁeld nontronite, XAFS studies of site occupancy in

smectites become more common. Bonnin et al. (1985) found that the amount of

tetrahedral Fe in nontronites, if present, was less than about 5% of total Fe, in

Chapter 12.3: X-ray Absorption Spectroscopy814

accord with IR, Mo

ssbauer (Bonnin et al., 1985) and P-EXAFS (Manceau et al.,

1998) analyses. Using Fe K pre-edge and P-EXAFS analyses, Manceau et al. (2000a)

assumed that the Garﬁeld nontronite had nil tetrahedral Fe and used it as their

reference octahedral Fe(III) mineral. They found that the Panamint Valley nontro-

nite and the ferruginous smectite SWa-1 also contain ed nil tetrahedral Fe, but that

the German nontronite NG-1 contained about 17% Fe in tetrahedral sites.

Given the uncertainties associated with pre-edge analyses (Manceau and Gates,

1997), Gates et al. (2002) applied Fe K powder EXAFS, in conjunction with IR and

XRD methods, to two ferruginous smectites and 12 nontronites to determine the

distribution of Fe in tetrahedral and octahedral sites. The ﬁrst peak in the FT of Fe

K EXAFS spectra contains all the contributions of the ligan d ﬁrst shell. A nearly

p-difference in phase exists between the

FeyO(d ¼ 195 pm) and

FeyO

(d ¼ 202 pm) waveforms (Manceau et al., 2000a) in the reverse transform of this

peak. The FeyO contributions to the EXAFS spectra associated with tetrahedral

Fe interfere systemat ically with those contributions associated with octahedral Fe

(Fig. 12.3.9). This ﬁnding was used by Gates et al. (2002) to estimate tetrahedral Fe

in nontronites. For nontronites with signiﬁcant Al (Al

X5–12% ignited basis),

least-squares ﬁtting of linear combinations of the

FeyO(d ¼ 195 pm) and

FeyO(d ¼ 202 pm) waveforms to experimental data yielded tetrahedral Fe con-

tents within 3% of total Fe. However, the method was less sensitive (>5% of total

Fe) for nontronites with very low total Al contents. For example, the Spokane

nontronite with o3% Al

(ignited basis) was estimated to have 12% of total Fe

occupying tetrahedral sites (Fig. 12.3.9B) by this method. Although this value falls

within the error of the method, it is unrealistically low because the tetrahedral cation

composition woul d be unﬁlled. Near-IR or XRD analyses are more reliable over the

entire composition range studied (Gates et al., 2002), but for the majority of sampl es,

the XAS method would be highly useful. Obviously, reﬁnements in d(

FeyO) and

FeyO) for each individual nontronite would improve the ability of this powder

EXAFS technique in determining tetrahedral Fe.

Pre-Edge Spectroscopy

The effect of incorporating

Fe into the structure of nontronite is depicted in

Fig. 12.3.3B (Manceau et al., 2000a; Gates et al., 2002). The German nontronite,

NG-1, was found by other methods to have a s much as 17% of total Fe in tet-

rahedral sites. As such, it displays considerably enhanced pre-edge amplitude com-

pared to the Garﬁeld nontronite, despite being chemically similar. The pre-edge

amplitude is enhanced for the Spokane nontronite and the South Australian non-

tronite, NAu- 2, as well. For all three nontronites, the resolution of the t

-e

splitting of energy states is lost relative to Garﬁeld nontronite, suggesting that

with a smaller t

-e

splitting is present in these samples.

The pre-edge amplitude for Spokane nontronite suggests an unrealistically low

(Gates et al., 2002) estimate of

Fe (10%). The diminished amplitude of the

Spokane nontronite is likely related to a high symmetry about the octahedral sites.

12.3.2. XAFS Studies on Smectite Structure 815