Bergaya F. Handbook of Clay Science

Подождите немного. Документ загружается.

12.4.1. PRINCIPLES OF X-RAY PHOTOELECTRON SPECTROSCOPY

XPS measures the kinetic-energy distribution of electrons (photoelectrons) emit-

ted from core levels of the elements constituting a solid when the sample is irradiated

by X-rays (Fig. 12.4.1). The measured core electron binding energy (BE

), deﬁ ned as

the difference between the energy of the primary photon (hn) and the kinetic energy

of the photoelectron (KE

), allows the element in question to be identiﬁed:

¼ hn  KE

ð1Þ

In practice, Eq. (1) must be corrected for some experimental factors, as described

below.

When an atom emits a photoelectron, it is left with a vacancy in its inner orbital,

and becomes unstable. Auger transition is one of the processes by which the un-

stable, excited atom undergoes relaxation (Fig. 12.4.1). Here, an outer-orbital elec-

tron is transferred to the vacanc y, and the excess energy is released through the

emission of another outer-orbital electron (XYZ Auger electron). The Auger electron

kinetic energy (KE

), given approximately by Eq. (2), is a characteristic of an Auger

transition for a speciﬁc atom:

¼ BE

 BE

ð2Þ

XPS measures both photoelectrons and Auger electrons.

The X-ray beam penetrates the sample to a depth of 1–100 mm, causing the ejec-

tion of photoelectrons and Auger electrons. Electrons that escape to the surface from

deeper parts of the sample may lose kinetic energy through inelastic scattering. It is

such inelastically scattered electrons that give rise to the background in an XPS

spectrum. The inelastic mean free-path of an electron in the solid depends on its

Fig. 12.4.1. Photoelectron (e

) and Auger electron (e

) emission processes induced by X-ray.

Chapter 12.4: X-ray Photoelectron Spectroscopy866

kinetic energy. Both photoelectron and Auger electron measured by conventional

XPS have kinetic energy less than 1500 eV and their inelastic free-path is very short,

typically of the order of nm. Therefore XPS is a surface-sensitive technique. All

elements, other than H and He, are detectable by XPS when their concentrations

exceed 0.1–1%. For samples with a ﬂat surface, the spectral background can be

reduced by using a total-reﬂection X-ray or a grazing-incidence X-ray because of its

shallow depth of penetration (Kawai et al., 1995). Since XPS is non-destructive, the

characteristic features of photoelectron and Auger electron spectra also provide

information on the state of chemical bonding of the elements concerned.

Recent developments in XPS inst rumentation enable photoelectron images with a

lateral resolution in the micrometer to submicrometer range to be obtained through

scanning or direct imaging techniques (Garwood, 1995). In particular, photoelectron

emission microscopy (PEEM) is a promising technique for observing direct photo-

electron images of surfa ce chemical composition (Kordesch, 1995; De Stasio et al.,

1998). A lateral resolution down to 20 nm was achieved using X-ray PEEM

(X-PEEM) with a synchrotron-radiation photon source (De Stasio et al., 1999). The

application of X-PEEM to the analysis of geological samples was reported by De

Stasio et al. (2001).

12.4.2. EXPERIMENTAL TECHNIQUES

A. Instrumentation and Sample Handling

XPS measurements are conducted under ultra-high vacuum (o10

–8

Torr) so as to

avoid collision between photoelectrons and gas molecules in the spectrometer, while

surface contamination from residual gases is minimized. Fig. 12.4.2 shows a sche-

matic diagram of a typic al X-ray photoelectron spectrometer, consisting of an X-ray

source, an electron energy analyzer, and a photoelectron detector. Common X-ray

sources are Al Ka (1486.6 eV) and Mg Ka (1253.6 eV). By using monochromatized

X-rays with a narrow line width, satellite spectra exited by Ka

3,4

and Kb lines can be

eliminated, and the energy resolution of photoelectrons improved. The ejected pho-

toelectrons are transferred to an electron energy analyzer, and separat ed according

to their kinetic energy. Among the various types of analyzers (Briggs and Seah,

1990), the concentric hemispherical analyzer (CHA) and cylindrical mirror analyzer

(CMA) are the most commonly employed. Following energy analysis, the photo-

electrons are detected by electron multipliers.

Fig. 12.4.3 shows a wide-scan spectrum of Na

-montmorillonite excited by Al

Ka, demonstrating the multi-element detection capability of XPS. In this respect, it

is comparable with X-ray ﬂuorescence (XRF) spectrometry for bulk elemental anal-

ysis. The wide-scan XPS spectrum shows the photoelectron and Auger electron

lines of the constituent elemen ts, notably Si (2s, 2p), Al (2s, 2p), Mg (KLL Auger),

Na (1s, KLL Auger), and O (2s, 1s, KLL Auger). The set of core photoelectron lines,

12.4.2. Experimental Techniques 867

together with the X-ray-induced Auger electron lines, are useful for element iden-

tiﬁcation. At the same time, however, the probability of overlap with lines of dif-

ferent atoms increases. This can sometimes cause difﬁculties in identifying and

quantifying minor constituents.

Because clays are electrical insulators, positive electrostatic charges can build up at

the sample surface due to electron emission during measurement. In practice, the

measured kinetic energy (KE

) differs from the ideal value (KE

), identiﬁable with

Fig. 12.4.2. Schematic diagram of XPS measurement system equipped with a concentric

hemispherical analyzer (CHA).

Fig. 12.4.3. Wide-scan X-ray photoelectron spectrum of Na-montmorillonite excited by Al

Ka radiation.

Chapter 12.4: X-ray Photoelectron Spectroscopy868

in Eq. (1) or KE

in Eq. (2). The relationship between KE

and KE

is given by:

¼ KE

 F

 E

ð3Þ

where F

is the work function of the spectrometer; that is, the energy required to

bring the electron from its zero binding energy (or Fermi) level to that of the spec-

trometer. E

is the additional retarding energy due to sample charging. In order to

determine the photoelectron binding energy and Auger electron kinetic energy, the

measured electron energy should be corrected for F

and E

The C 1s or Au 4f

7/2

line is widely used as a primary standard for the calibration of

electron energies. It is often convenient to use the C 1s line (of hydrocarbons) for this

purpose since this peak usually arises from adventitious surface contamination by

vacuum pump oil. However, the C 1s binding energy depends on the nature of the

carbon source, and hence is not an accurate binding energy standard. Further, dif-

ferential charging of the sample and hydrocarbons (due to differences in electric non-

conductance) may cause an unexpected error in binding energy determination. On the

other hand, the Au 4f

7/2

line of a gold ﬁlm, evaporated onto the sample, is a reliable

standard since its binding energy (84.0 eV) was precisely measured (Ebel et al., 1983;

Anthony and Seah, 1984). Thus, the electron binding energy of the constituent elements

in a given sample is frequently determined relative to that of Au 4f

7/2

on the assumption

that the Fermi level of the sample coincides with that of the evaporated gold.

Mineral samples for analysis by XPS usually come in the form of a powder or a

thin section. Before being inserted into the apparatus, the sample is dried and at-

tached to a metal sample holder by a double-sided sticky tape, an electrically con-

ductive paste, or by other means. Samples may also be mounted by placing a

suspension of the powder on the sample holder, and by allowing the dispersant to

evaporate. Alternat ively, the sample may be pressed into a soft metal, such as in-

dium. Whatever method is adopted, due care should be taken to avoid or, at least,

minimize surface contamination.

B. Quantitative Analysis

The area under a given peak in a photoelectron spectrum is directly proportional to

the conc entration of the element in question, assuming that the element is homo-

geneously distribut ed in the analyzed volume. Since peak intensity increases as the

effective surface area of the sample increases, it is difﬁcult to estimate the absolute

concentration of a given element. In practice, one derives the concentration of el-

ement A relative to that of a reference element B (atomic ratio, N

) from the

measured photoelectron intensities (I

for element A and I

for element B):







ð4Þ

12.4.2. Experimental Techniques 869

where S

is the relative atomic sensitivity factor. Included in this factor are such

parameters as the photoionization cross-section, anisotropy of photoelectron emission,

detection efﬁciency of the spectrometer, and mean free-paths of photoelectrons. Rela-

tive atomic sensitivity factors are derived either theoretically (from calculated

photoionization cross-sections and other parameters), or experimentally (from XPS

measurements of reference compounds of known homogeneous chemical composi-

tion). Commercial XPS instruments are equipped with a microcomputer for instrument

control, data acquisition, and data handling. Relative surface concentrations (elemen-

tal compositions) can be calculated by the computer using appropriate software

(smoothing, background subtraction, peak deconvolution, peak area calculation, etc.).

The uncertainty in elemental quantiﬁcation by XPS mainly arises from the

application of the atomic sensitivity factor to samples with different matrices (Seah,

1980). In our experience, the relative surface concentration of an element can be

obtained within an error of 10–20%, using experimentally determined relative

atomic sensitivity factors. However, for photoelectrons with a low kinetic energy

(i.e., a high binding energy) the values derived are less certain. This is because the

mean free-path of the corresponding photoelectrons is shorter, and their intensity is

more susceptible to surface contamination, than their high-energy counterparts.

The reproducibility in deriving relative surface concentrations from XPS spectra

was checked by Seyama and Soma (1984, 1988). From measurement of the area

intensity ratios of Si 2s to Al 2p of montmorillonite samples having different ex-

changeable cations, these workers concluded that the values are reproducible within

710%. On the surfaces of freshly prepared silicates (clay and related minerals), the

relative abundances of major cati ons (Si, Al, Fe, Mg, Na, K, etc.) usually correspond

to their bulk chemical compositions. However, the surface concentration of oxygen

tends to exceed the bulk value because of the presence of OH



ions and/or water

molecules at the surface (Seyama and Soma, 1985, 1988).

C. Chemical Information

Characteristic features of photoelectron and Auger electron spectra, such as peak

position (electron bind ing or kinetic energy), satellite structure, and multiplet split-

ting, reﬂect the bonding state of elements. For some elements, it is possible to

measure both photoelectron and Auger electron spectra. In such cases, it is especially

informative to obtain a two-dimensional plot of both electron energies (Wagner

et al., 1979). Such a ‘chemical-state plot’ was used to characterize the bonding state

of some elements in clay minerals (Seyama and Soma, 1984, 1988; Dutta et al., 1999).

Wagner et al. (1979) also measured the modiﬁed Auger pa rameter (a

), deﬁned as

the sum of the photoelectron binding energy and Auger electron kinetic energy of an

element, and proposed that the value of a

is unique to each chemical-bonding state.

For bonding state characterization it is customary to compare the photoelectron

binding energy and/or Auger electron kinetic energy of an element in the sample

Chapter 12.4: X-ray Photoelectron Spectroscopy870

with that of a reference compound. The shift in peak (electron energy) position

relative to that of the reference compound is called the chemical shift. The chemical

shift in the photoelectron binding energy (DBE

) is dependent on the state (chemical

environment) of the atom, in particular its electron density (oxidation number).

The lower the electron density, the higher the photoelectron binding energy. The

Fe 2p photoelectron spectrum is a case in point. The Fe 2p

3/2

binding energy of

in silicate minerals (about 712 eV) is higher than that of Fe

(710–711 eV).

Further, the Fe 2p spectra of silicate minerals have satellite peaks characteristic of

the oxidation state of the iron (Seyama and Soma, 1987). The nature of atoms

surrounding the host atom, that is of nearest neighbours (and, in some cases, non-

nearest neighbours), also inﬂuences the chemical shift. For example, for silicate

minerals there is a positive correlation between the photoelectron binding energies of

Si 2s (2p) and O 1s (Seyama and Soma, 1985, 1988). These values refer to the silicate

framework (structure), and decrease as the negative charge on the silicate structure

increases.

Fig. 12.4.4, showing the Mg 1s and Mg KL

Auger electron spectra of

-montmorillonite excited by Al Ka radiation, gives another example of

Fig. 12.4.4. (A) Mg 1s and (B) KL

Auger spectra of Mg-montmorillonite (Reproduced,

by permission of The Royal Society of Chemistry, from Seyama and Soma, 1984).

12.4.2. Experimental Techniques 871

deriving chemical information from XPS analysis (Seyama and Soma, 1984). For

both electrons the spectra are broad, and each can be deconvoluted into two

components corresponding to: (i) non-exchangeable Mg

ions occupying octahe-

dral sites within the layer structure; and (ii) exchangeable Mg

ions occupying

interlayer sites.

Fig. 12.4.5 shows a chemical-state plot for magnesium, comparing the relation-

ship between Mg 1s binding energy and Mg KL

Auger kinetic energy for

-montmorillonite with that of magnesium halides and magnesium oxide. The

position of exchangeable (interlayer) Mg

ion falls between that of MgCl

.6H

and MgF

. In contrast, the point for non-exchangeable (structural) Mg

ion is

close to that for MgO. The Mg 1s binding energy for non-exchangeable Mg

ions is

1.5 eV lower, while the Mg KL

Auger kinetic energy is 2.0 eV higher, than the

corresponding values for exchangeable Mg

ions. These large differences may be

attributed to the effect of neighbouring atoms. Apparently, the ﬂow of electronic

charge to Mg

as well as the extra-atomic relaxation (a screening of the ﬁnal-state

ion in the Auger transition by electrons from neighbouring atoms) are larger for

non-exchangeable Mg

ions because of its stronger interaction with O

2–

and OH

–

ions within the aluminosilicate layers.

Although XPS is essentially non-destructive, irradiation by X-rays can occasion-

ally cause chemical damage (alteration) to the sample. For example, during XPS

measurement Cu

ions (structural, adsorbed, or interlayer) in clay minerals can be

converted into Cu

through photoreduction (Mosser et al., 1992).

Fig. 12.4.5. Chemical-state plot for Mg (Reproduced, by permission of The Royal Society of

Chemistry, from Seyama and Soma, 1984).

Chapter 12.4: X-ray Photoelectron Spectroscopy872

12.4.3. APPLICATIONS

Despite its limited sensitivity in terms of spatial and binding energy (chem ical

shift) resolution, XPS can provide valuabl e information on the surface composition

of clays and clay minerals, and the chemical bonding of the constituent elements.

The examples below illustrate this capability.

The particles of naturally occ urring clay minerals (phyllosilicates) are often coated

by ferric (hydr)oxides which, therefore, may significantly inﬂuence surface proper-

ties. Conventionally this (external) coating of iron can be extracted by treatment with

sodium citrate–dithionite–bicarbonate (CDB), leaving the structural iron more or

less intact. Using this approach, Soma et al. (1992) were able to characteriz e the

surface chemical composition of ﬁve New Zealand halloysites with different particle

morphologies by XPS. Except for the sample from Hamilton, most of the iron in the

halloysites were structural, and hence not extractable with CDB. The Hamilton

halloysite contained the highest amoun t of iron, of which about 80% was external

and extractable by CDB. The Fe 2p

3/2

photoelectron binding energy (711.6 eV) of

this sample was consistent with that of ferric (hydr)oxides, such as goethite (McIn-

tyre and Zetaruk, 1977). Its iron-to-silicon (Fe/Si) atomic ratio, determined by XPS,

was also significantly larger than the bulk value determined by X-ray ﬂuorescence

analysis. On the other hand, the Fe 2p

3/2

binding energy of the other halloysites

(about 712.8 eV) was characteristic of ferric ion in silicate structures (Seyama and

Soma, 1987). Their surface Fe/Si atomic ratio (by XPS) was close to, or less than, the

bulk value, indicating that the surface layers of some halloysites were depleted in

(structural) iron. This observation would have important implications for determin-

ing the surface charge characteristics and chemi cal reactivity. After CDB extraction,

the surface Fe/Si atomic ratio of Hamilton halloysite was comparable with the

corresponding bulk ratio, while the Fe 2p

3/2

binding energy shifted to 712.6 eV, close

to the value for structural iron. This example clearly demonstrates the capability of

XPS in providing infor mation on the radial distribution and chemical-bonding state

of an element (here iron) in minerals with a heterogeneous microstructure.

XPS analysis of siliceous ferrihydrites provides insight into the bonding state and

localization of silicon (Soma et al., 1996). The surface Si/Fe atomic ratios of ﬁve

natural siliceous ferrihydrites (determined by XPS) were close to, or slightly smaller

than, their respective bulk values (0.18–0.43). This indicated that Si was well dis-

persed throughout the ferrihydrite matrix at the scale of the photoelectron mean

free-path (of the order of several nm), with a tendency for depletion in the outer

(surface) layers. The Si 2s binding energy of 152.9 7 0.2 eV for all ﬁve natural

samples was close to that for olivine (152.9–153.0 eV) (Seyama et al., 1996), a neso-

silicate with isolated SiO

tetrahedra. However, it was markedly lower than the value

for quartz (154.4 eV) (Seyama et al., 1996), indicating the absence of a three-

dimensional network of SiO

tetrahedra. Thus, both the surface Si/Fe atomic ratio

and the Si 2s photoelectron binding energy of natural siliceous ferrihydrites were

consistent with the model of Parﬁtt et al. (1992). These workers suggested that

12.4.3. Applications 873

silicate is bonded to, or forms a bridge between, the surfaces of ‘micro-crystalline

domains’ making up a primary ferrihydrite particle of several nm in diameter.

The XPS results for a synthetic ferrihydrite with an Si/Fe atomic ratio of about

0.1 and a low Si 2s binding energy also suggest that Si is well dispersed in the matrix

and, as such, would inhibit the conversion (of ferrihydrite) into hematite (Glasauer

et al., 2000). On the other hand, for a synthetic precipitate of ferrihydrite with a

relatively high-silicon content, the surface Si/Fe atomic ratio (0.48) was much larger

than the bulk value (0.18), while the Si 2s binding energy (154. 1 eV) was close to that

of quartz (Soma et al., 1996). These observations indicate that silica layers with

three-dimensionally polymerized SiO

units covered the outer particle surfaces of the

precipitated ferrihydrite. In line with this postulate, the O 1s photoelectron spectrum

showed two peaks of which the one with a higher binding energy was assigned to

oxygen in a three-dimensional SiO

network (Seyama and Soma, 1985).

An important aspect of the surface chemistry of phyllosilicates (clay minerals) is

related to their anisotropic layer structure, exposing basal-plane, and crystal-edge

surfaces. The spatial resolution of conventional XPS is not sufﬁcient to differentiate

between these two surfaces of micrometer-size clay mineral particles. However, an-

isotropic surface reactivity may be assessed using macrocrystalline micas with a

specimen surface area comparable to that of an X-ray beam. For example, Ilton

et al. (1997) used XPS to investigate the interaction of chromate ions (CrO

2

) with

mica (biotite and phlogopite) crystals in aqueous solutions containing an alkaline

salt. The study was undertaken because hexavalent chromium is toxic, and its fate in

the environment is of great concern. The photoelectron spectra of the treated biotite

showed the presence of chromi um on the edge surface but none on the basal plane.

In the case of phlogopite, however, no chromium was detected on either the edge or

basal surface. The Cr 2p

3/2

binding energy indica ted that chromium was bound as a

trivalent species to the edge surface of biotite. This ﬁnding was explained in terms of

the reduction of hexavalent chromate by ferrous cations, present in the layer struc-

ture, followed by adsorption (at the biotite edge surface) of the resultant trivalent

chromium cation. The overall process was promoted by sodium and lithium salts

whereas the presence of RbCl and CsCl in solution had an inhibiting effect. It is

suggested that hydrated Na

and Li

ions could replace interlayer K

ions in

biotite, and so facilitate the heterogeneous reduction of chromate at the edge/

solution interface.

Both biotite and phlogopite were able to adsorb Cr

cations from solutions

containing NaCl and KCl (Ilton et al., 2000). However, adsorption was larger, and

the photoelectron binding ene rgies of Cr 2p

3/2

and 3p were higher for the NaCl than

the KCl system. The adsorption of Cr

ions from KCl solution was apparently

conﬁned to crystal-edge surfaces. In NaCl solution, on the other hand, Cr

was

adsorbed at both edge and interlayer sites since Cr

and Na

could replace in-

terlayer K

ions in mica.

The interacti ons of organic matter with clay minerals have important technical

and environmental applications . As already remarked, the C 1s photoelectron line is

Chapter 12.4: X-ray Photoelectron Spectroscopy874

unsuitable for characterizing organic compounds adsorbed on mineral surfaces since

this peak is affected by hydrocarbons from vacuum pump oil. On the other hand, the

N 1s photoelectron line can often serve as a convenient marker of nitrogen-

containing organic molecules, such as amines. The adsorption from aqueous solu-

tions of partially hy drolyzed polyacrylamide (HPAM) by kaolinite, feldspar, and

quartz illustrates this point (Graveling et al., 1997; Allen et al., 1998). Measurement

of the N 1s and Si 2p lines in the XPS spectrum following adsorption (of HPAM)

allowed the surface concentration of N to be normalized for surface Si sites. Among

the three minerals investigated, kaolinite showed the highest adsorption of HPAM

(pH 2–10), probably involving hydrogen bonding between basal oxygens of the clay

mineral and amide groups of the polymer. The high reactivity of kaolinite was also

demonstrated by imaging XPS of mixed kaolinite/quartz crystals after adsorption of

HPAM (Allen et al., 1998). Using Si 2p, Al 2p, and N 1s photoelectron images with

o10 mm spatial resolution it was possible to differentiate between domains in the

kaolinite matrix containing more HPAM and quartz grains with fewer polymers. We

would expect that further developments in XPS-associated microscopy techniques

will extend the application of XPS to complex systems containing clays and clay

minerals.

The magnitude and distribution of surface charge have a great inﬂuence on

the charge characteristics of clay minerals. The application of XPS to assessing

the surface charge of some micas (margarite, muscovite, and sericite) was reported by

Gier and Johns (2000) and Johns and Gier (2001). After the replacement of surface

cations by Ba

these workers determined the atomic ratios of surface Ba

interlayer Ca

(for margarite), K

(for muscovite and sericite), and Na

(for

margarite, muscovite, and sericite). These measured ratios were indicative of the

relative layer charge on external surface and the uppermost interlayer. Assuming that

the interlayer charge was equally divided between the two external cleavage surfaces,

the Ba

/interlayer cation atomic ratio would indicate the occurrence of layer charge

asymmetry. The O/Ba atomic ratio of mica, determined by XPS, could also be used to

assess the magnitude and distribution of surface charge (Johns and Gier, 2001).

Photoelectron intensity from a single crystal surface as a function of electron take-

off angle shows a characteristic ﬂuctuating pattern. This phenomenon, known as

‘photoelectron diffraction’, is due to the elastic scattering of photoelectron s by

neighbouring atoms. Since the X-ray photoelectron diffraction (XPD) pattern

depends on crystal structure, it can yield information about the location of atoms

within the crysta l. Evans and co-wo rkers (Evans and Raftery, 1980, 1982; Ash et al.,

1987) measured the XPD patterns of mica single crystals, rotating at an angle about

a speciﬁc crystallographic axis, and examined the location (sites) of minor elements

in the minerals. They were able to show that in biotite Ti preferentially occupied

octahedral sites whereas in lepidolite Rb and Mn, respectively, occupied sites that

were essent ially identical with those of K (interlayer) and Li (octahedral).

Aluminium in silicate minerals can occupy both tetrahedral and octahedral sites.

The distribution of Al

ions between the two sites is important in determining the

12.4.3. Applications 875