Jackson S.D., Hargreaves J.S.J. Metal Oxide Catalysis

Подождите немного. Документ загружается.

304 7 X-ray Absorption Spectroscopy of Oxides and Oxidation Catalysts

goes through pathways that are either evacuated or ﬁ lled with He to reduce absorp-

tion of the less hard X - rays. For softer X - rays, such as the transition metal L - edges,

Al, Si, S, P or As, a vacuum system has to be used to reduce the non - speciﬁ c

absorption of the X - rays. However, even in these cases gas atmospheres can be

used as long as pathlengths are kept short. One of the major advantages of X - ray

absorption spectroscopy is the comparably undemanding experimental setup. It

has been used for in situ measurements of heterogeneous catalytic reactions,

electrochemistry and fast reaction kinetics.

Figure 7.4 shows the experimental set - up of a typical XAS experiment. For most

EXAFS experiments the source is a synchrotron. Laboratory sources are rare and

usually do not exhibit enough intensity to get usable EXAFS spectra.

The photon sources from synchrotrons range from bending magnets, to wig-

glers and undulators. The speciﬁ c device used depends on the frequency range

and brightness required for the speciﬁ c application.

Often the white beam is reﬂ ected by a mirror, which can be used for harmonic

rejection and focusing on the sample.

The mirrors normally have to be cooled (often using gallium) since they are

exposed to more or less the full intensity of the white beam. The monochromators

for harder X - rays are often silicon single crystals exposing (220) or (111) planes.

For lower energy photons such as the Mg, Si and Al edges, Y66B single crystals

can be used. Below 1000 eV (e.g. ﬁ rst row transition metal L - edges, the O K - edge

and the C K - edge) the technology changes signiﬁ cantly and gratings are used.

In transmission methods, after the monochromator, the optical path traverses

through the I

detector, the sample chamber, the I

detector and, for energy calibra-

tion, through a foil and the I

detector.

7.1.3

Detection Methods

Absorption is the simplest method for measuring an X - ray spectrum. The incom-

ing radiation, I

is measured and related it to the transmitted radiation ( I

Figure 7.4 A typical XAS experimental set - up.

Absorption is then given by:

µt

⎛

⎝

⎞

⎠

(7.4)

where t is the sample thickness. and µ is the linear mass absorption coefﬁ cient.

This method makes use of all the photons that are incident on the sample.

To measure the absorption signal, a monochromator is stepped through the

energy range and the variation of the mass absorption coefﬁ cient measured as a

function of the Bragg angles (i.e. energy.) For signiﬁ cantly faster scanning, the

full EXAFS energy range of radiation is shone on the sample and the emitted

radiation is detected spatially using X - ray detector plates [5 – 9] . The detectors most

commonly used for absorption are ion chambers, which are ﬁ lled with gases with

compositions dependent on the X - ray energy. The absorption of the gas should be

in the range of 20% for the I

detector and 80% for the I

detector.

Sometimes it is of advantage to measure absorption not directly, but by associ-

ated processes. This is mainly used when the X - ray absorption is only a small

fraction of the total absorption process. Problems with absorption are often

observed with highly diluted samples and also for low - energy EXAFS. In these

cases the transmission results from the difference of either two nearly identical

signals or two vastly different signals, both needing very accurate data.

Two of the techniques applied in this category are Auger electron process and

X - ray ﬂ uorescence process. A schematic in Figure 7.5 shows the processes for

these two techniques. These two processes compete with one another, and their

relative abundance is dependent on the atomic number of the absorber. In general

ﬂ uorescence is more pronounced for heavier elements, whereas the Auger process

is more prominent for lighter elements.

In addition to the removal of one core electron by the incident X - rays (photo-

electron process), the Auger process is taking place approximately 10

− 14

seconds

after the photoelectron event. In this process an outer electron falls into the inner

orbital vacancy from the photoelectron process and a second electron is emitted

Figure 7.5 Processes resulting from the removal of a photoelectron by X - rays.

7.1 Principles of EXAFS 305

306 7 X-ray Absorption Spectroscopy of Oxides and Oxidation Catalysts

carrying the excess energy. The Auger electron carries and energy equal to the

difference between the energy of the initial ion and the double charged ﬁ nal state.

The Auger process is competing with the ﬂ uorescence process.

Fluorescence radiation is a result of the ﬁ lling of a core hole (created by the

absorption process) by outer electrons. For K - edges K

α 1

radiation is emitted, which

is characteristic of the absorbing element. There is signiﬁ cant background created

by Compton radiation and elastic scattering. This radiation is at higher energy

than the ﬂ uorescence line. Thus, detectors need to be energy discriminative. The

scattering contributions can be removed using as a ﬁ lter a metal foil with an atomic

number ( Z ) one lower than the element of interest. Compton and elastic scattering

frequencies are often above the absorption edge of the Z − 1 element and so they

are removed.

Modern solid - state detectors have windowing functions that can ﬁ lter the ﬂ uo-

rescence radiation.

7.1.4

Data Reduction

As mentioned, EXAFS is deﬁ ned by the variation of the mass absorption

coefﬁ cient:

µµ

()

−

()

(7.5)

Equation 7.5 describes the energy - dependent EXAFS oscillations that can be

interpreted using the EXAFS equation (as given in Equation 7.1 or Equation 7.2 ).

χ ( k ) EXAFS as a function of electron wave vector.

The mass absorption coefﬁ cient is determined experimentally from the curves

as illustrated in Figure 7.2 µ

( E ) is the smooth atomic background, whereas µ

( E

)

is the edge step. The edge step is the mass absorption coefﬁ cient at the edge posi-

tion. This can be determined by extrapolating the post - edge baseline or from the

edge position. The edge position can be deﬁ ned as the ﬁ rst point of inﬂ ection at

the absorption edge.

EXAFS analysis is performed by a number of data treatment steps:

1. Deglitch the individual spectra.

2. Add the experimental spectra and individual contributions of the detectors

(for, e.g., ﬂ uorescence).

3. Convert intensities to µ .

4. Subtract a background function pre - edge.

5. Normalize the spectra to an absorption of 1.

6. Subtract a background function post - edge to obtain χ ( E ).

7. Obtain E

and convert to χ ( k ), that is, convert from energy to k - space.

8. Deﬁ ne a model and calculate the sum of the model function from the EXAFS

equation as, for example, from Equation 7.3 .

9. Compare the calculated function with experimental χ ( k ) and minimize the

differences.

Steps 8 and 9 are speciﬁ c to the SRS Daresbury Laboratory suite of programs,

which will be the focus of this discussion. Other programs use Fourier ﬁ ltering

techniques to isolate the contribution of the various shells. A number of other

software programs use this technique. Fourier ﬁ ltering can be problematic as

outlined in refs [10, 11] .

Deglitching is often necessary depending on the type and spectral region of the

monochromator. The easiest way of deglitching is removal of the datapoints. A more

detailed guide to deglitching can be found on: http://srs.dl.ac.uk/xrs/index.html

The Daresbury suite of programs used to add and convert data are Excalib and

Exspline, and the EXAFS ﬁ tting is called Excurv98.

There are a number of other programs available (often free of charge or with

nominal costs) such as FEFF and viper (which uses FEFF code).

Exspline is an X - window - based program that uses spline ﬁ rst for pre - and post -

edge ﬁ tting baselines. It has the advantage that the Fourier transform, the EXAFS

and the pre - and post - baseline ﬁ t can be observed interactively. This is a great

improvement on Exback, which used non - interactive baseline correction.

Owing to the various processes detailed above, the EXAFS oscillations decay

quickly with the photoelectron wavevector ( k ). The EXAFS is often weighed as

k , k

or k

to increase the intensity of the oscillations at higher k values.

Figure 7.6 shows the EXAFS oscillation together with the amplitude of the

Fourier transform of the Al K - edge of the zeolite Na - FAU. The spectrum and its

Fourier transform ( FT ) are shown together with ﬁ tted data. The ﬁ rst peak in the

Fourier transform can be found at 1.67 Å . However the real position of the peak

is at 1.74 Å .

The discrepancy between the two values is a result of the phase - shift function,

which is included in Equation 7.3 . Phase shifts are dependent on the absorber

atom and can be calculated by ab initio methods. The model building for phase - shift

ﬁ les include a model of the most abundant atomic neighbor for a speciﬁ c atom.

7.1 Principles of EXAFS 307

Figure 7.6 EXAFS and FT transformed spectra of Na - FAU.

The dashed line represents ﬁ tted and the solid line

experimental data.

308 7 X-ray Absorption Spectroscopy of Oxides and Oxidation Catalysts

Fourier transformation of EXAFS makes χ ( k ) a complex number, but often the

FT is only represented by its amplitude, as shown in Figure 7.6 .

Often the FT (as seen in Figure 7.6 ) shows well separated peaks and is then

used to perform the ﬁ tting procedure in Fourier space or to ﬁ lter the data. Of

course, in this case a windowing function has to be applied with all its associated

problems. The ﬁ ltering is often performed at higher r values to remove some of

the noise.

The ﬁ ltering process itself introduces an arbitrary function, which can cause

artifacts. For instance, the statistical signiﬁ cance of shells is altered [10] . Unless

Eigen functions are used for the Fourier ﬁ ltering [11] the introduction of the arbi-

trary function to perform Fourier ﬁ ltering should be omitted. If absolutely neces-

sary (for example if there are problems with baseline correction) very wide ﬁ ltering

functions can help in reducing the problem.

The ﬁ tting procedure is normally carried out consecutively, that is, shells are

added to the ﬁ tting process as seems appropriate. Every shell should be checked

for statistical signiﬁ cance, that is, if the addition of, normally, three parameters

( σ , R,N ) is statistically viable. The check can be made by comparison of the change

in the Fit index with F - test values, as outlined in ref. [12] . The oscillations are often

very broad and with few features for oxides. In this case good statistical parameters

are especially important.

Interpretation of XANES is signiﬁ cantly more difﬁ cult than for EXAFS. The

EXAFS equation breaks down in the low - k XANES region and the mean free path

goes up. The XANES region is even more complicated by the signiﬁ cant multiple

scattering than EXAFS. However, the electronic transitions can be handled as a

ﬁ ngerprint and the XANES, the white line intensity and edge position are a

measure of the electron density. XANES can provide us with important informa-

tion of the chemistry of the probed solid. For instance for transition metals the

pre - edge peak can distinguish between octahedral, square planar, tetrahedral or

distorted tetrahedral coordination. The oxidation state of transition metal species

can also be determined from the XANES.

Figure 7.7 Differential XANES spectra of iron containing

compounds. Dotted line Fe metal, full line FeO dashed line

. Adapted from ref. [13] .

Figure 7.7 shows typical derivative XANES spectra of some iron oxides and iron

metal. The edge position is in the maximum of the second peak for the oxides. It

can be seen that the edge position changes signiﬁ cantly depending on the oxida-

tion state. It also shows that the variation of the XANES is signiﬁ cant, and can be

used as a ﬁ ngerprint to determine the species present in the oxide.

Pre - edges as observed in Figure 7.7 , which are found in transition metals with

partially ﬁ lled d - shells, can be indicative of the coordination environment of the

transition metal. The pre - edge can be interpreted as an s – d transition, which is

forbidden in octahedral symmetry. Only small pre - edges are found for transition

metal compounds having octahedral or distorted tetrahedral symmetry. On the

other hand, tetrahedral compounds can have big pre - edges which can be used as

an indicator for the coordination chemistry. For example tetrahedral FePO

and

Fe - MFI zeolites exhibit signiﬁ cant pre - edges, whereas most transition metal

exchanged zeolites, the iron oxides and iron salts only show smallish s – d

transitions.

It is interesting to note that oxygen XANES exhibits pre - edges if bound to a

transition metal (e.g. Co), as well.

XANES probes the transition of core electrons into unoccupied electronic states.

The ﬁ nal states may be mixed with other orbitals and can be used to determine

the coordination environment of an element in a compound, its electron density

and oxidation state. The intensity can be used to accurately determine relative

oxidation state ratios. Furthermore, XANES can be used to determine relative

amounts of species by linear combination of individual compounds. In recent

years, the codes for XANES calculation (especially John Rehr ’ s FEFF code), have

signiﬁ cantly improved and it can be expected that theoretical models of com-

pounds will be accurately determined by XANES measurement and calculation in

the future.

XANES and EXAFS complement each other very well and are ideal to determine

the site structure of even mixtures of catalysts or oxides.

7.2

Applications of EXAFS

XAS is in general less accurate than XRD techniques. The accuracy depends on

the edge energy, the resolution of the monochromator and, last but not least,

the quality of model building. As a general rule distances can be determined up

to 5 × 10

− 3

Å and coordination numbers at maximum to 10%. This is clearly in -

ferior to XRD.

However, as mentioned in the introduction, the main advantage of XAS com-

pared to other structural techniques is its element speciﬁ city, its potential opera-

tion under more or less any experimental conditions, its applicability for amorphous

and X - ray amorphous materials and its straightforward sample preparation. In the

following, some applications of XAS in catalytic systems will be discussed.

7.2 Applications of EXAFS 309

310 7 X-ray Absorption Spectroscopy of Oxides and Oxidation Catalysts

XAS was originally used mainly for supported metal catalysts. XRD, despite

being applicable under reaction conditions of high temperatures and pressures,

is not normally able to detect small clusters of noble and other metals in supported

catalysts. A signiﬁ cant number of systems are too small to be detected by conven-

tional XRD techniques because they become X - ray amorphous due to line broaden-

ing. Group VIII noble metals were among the ﬁ rst catalysts investigated and the

work by Lytle and also Sinfelt is of particular note in this area [14, 15] . These classic

investigations, and especially in situ investigation [16] , paved the way for modern

XAS in catalysis.

7.2.1

Acid Zeolites

Zeolites are crystalline alumosilicates with a regular pore structure. These materi-

als have been used in major catalytic processes for a number of years. A signiﬁ cant

number of processes use zeolites as catalysts either as supports or as catalysts

themselves. The application using the largest quantities of zeolites is Fluid Cata-

lytic Cracking ( FCC ). The ﬁ rst zeolites used as catalysts for FCC were reported in

a 1961 patent by Exxon [17] and very quickly replaced many of the amorphous

silica alumina catalysts which dominated the process up to that time. The zeolites

with signiﬁ cant cracking activity are ultrastable Faujasite ( USY ), ZSM - 5 (alterna-

tively known as MFI), mordenite, offretite and erionite. By far the highest volume

of zeolite type used for this application are USY zeolites.

The catalytic activity of zeolites has its origin in the fact that some of the silicon

atoms in the crystalline framework of the solids are replaced by an aluminum

atom. Since aluminum is trivalent, the replacement of the tetravalent silicon

results in the introduction of a negative charge into the zeolite lattice. This negative

charge has to be compensated by a cation. When synthesized, most zeolite materi-

als have sodium as a counter - cation, resulting in very little catalytic activity for

acid - catalyzed reactions. Therefore zeolites are normally exchanged by rare earth

or ammonium cations to create catalytically active acidic sites. The ammonium

cations are normally decomposed at higher temperatures leaving residual H

species, forming the Br ø nsted acidic site. Mixed rare earth and ammonium zeo-

lites are among the most widely used zeolite formulations for FCC.

Despite the enormous commercial success and widespread use of acidic zeolite

catalysts, important questions on the local structure of zeolitic materials remain

unanswered. The local structure of aluminum in the zeolites has proven rather

elusive to investigation.

It has proved difﬁ cult to establish the local structure surrounding these alumi-

num sites by diffraction methods because the similarity of the backscattering

powers of aluminum and silicon result in the elements being virtually indistin-

guishable using these methods. Although aluminum is slightly larger than silicon,

most crystallographic data on zeolites does not report any distinction between

silicon and aluminum atoms in the lattice. Neutron diffraction [18] is one of

the few techniques able to provide direct structural data on the position of the

zeolitic proton. Neutron diffraction has been used both to characterize the average

proton site structure and to investigate the interaction of the sites with reactant

molecules.

Magic Angle Spinning ( MAS ) Nuclear Magnetic Resonance ( NMR ) also had

some success in determining proton positions. Side - band analysis from solid - state

H NMR was used to determine proton positions [19, 20] in ZSM - 5 type zeolites.

In addition to the limited experimental techniques, some theoretical studies on

the local structure of aluminum sites in the zeolites were published in the early

1990s [21] . More recently, some systematic studies of the acid – base properties of

zeolites were determined using theoretical approaches. It was suggested that the

local structure of the zeolites has some inﬂ uence on the acid strength of the indi-

vidual sites [22, 23] . It has long been recognized, for example, that the concentration

of aluminum has signiﬁ cant inﬂ uence on the acid strength of individual sites.

Variations in the acid strength were suggested to depend on the structure type and

concentration of lattice aluminum, which result in a variation of the local site struc-

ture. In contrast to earlier studies [24] , the authors suggested no direct correlation

of acid strength with bond angles. The authors also reported that the Al

−

O bond

that carries the acidic proton is longer than the other Al

−

O bonds [22, 25, 26] .

The ﬁ rst EXAFS study of aluminum in zeolites was a study of Na - and H - FAU

(Faujasite) by Koningsberger and coworkers [27]. The authors observed slight

variations in the Al

−

O bond lengths when comparing Na - , H - and NH

- form zeo-

lites. However, they did not ﬁ nd any signiﬁ cant variations between the four dif-

ferent aluminum – oxygen bond distances. This is in contrast to the theoretical

calculations, which suggested an increased Al

−

O distance for one of the Al

−

bonds. Similarly, more recently Bell and coworkers [28] did not ﬁ nd an increased

distance Al

−

O distance for acidic and Cu - exchanged ZSM - 5 . Bell suggested that

his data does not support a distinction between Al

−

O bonds but did not give any

statistical support for this conclusion.

In line with the theoretical predictions [4, 29] an increased Al

−

O distance was

observed by some workers. Furthermore multiple scattering calculations enabled

calculation of the detailed local structure, including bond angles. The group that

initially reported very similar Al

−

O distances [27, 30] later observed a variation

under in situ conditions [31] . This is not too surprising, since other cations were

found to relax the zeolite local structure to a more regular tetrahedron [32] (Figure

7.8 ). Sorption of water and the presence of alkali metal or ammonium counter -

cations result in the more regular tetrahedral structure that had been reported in

the earlier papers [27, 30] .

Weaker bases such as toluene or isopropanol induced some relaxation of the

aluminum in acidic zeolites. Water also exhibited a similar effect [32] , R.W. Joyner,

A.D. Smith, and M. Stockenhuber 2007, unpublished].

One of the disadvantages of EXAFS in general is that it is an averaging tech-

nique. Thus it is very important to make sure the zeolites are very well deﬁ ned.

Other techniques such as

Al MAS NMR spectroscopy and in situ infrared spec-

troscopy are needed to ensure single aluminum sites. Depending on the data

quality it is also possible to deﬁ ne more than one type of site; however, in this case

7.2 Applications of EXAFS 311

312 7 X-ray Absorption Spectroscopy of Oxides and Oxidation Catalysts

statistical analysis is absolutely crucial for the determination of the number of

statistically relevant shells [12] . XAS can also be used to characterize site variations.

It has been suggested that XANES can be used to distinguish between tetrahedral,

octahedral, distorted octahedral and square planar geometry [30, 33] .

However, the accuracy of the ab initio calculations for the large number of mul-

tiple scattering pathways present in the XANES region is limited and thus difﬁ cul-

ties arise in the interpretation of the data (see above).

XANES, particularly the edge position and white line intensity, can, however,

give a good indication of the electron density of the aluminum. The electron

density on the aluminum atom then in turn inﬂ uences the electron density on the

oxygen, which can have an inﬂ uence on acid strength of the solid.

A more direct approach to assess this has been made very recently [R.W. Joyner,

J.A. Purten, C. France and M. Stockenhuber, 2007, unpublished]. Oxygen EXAFS

was recorded for the ﬁ rst time in zeolites and it was possible to get accurate data

for Co

−

O and K

−

O structures from oxygen K - edge data. The XANES for oxygen

also exhibited very rich features which can easily distinguish between the various

counter - cations. It was also possible to separate oxygen interacting with cations

from “ spectator oxygen ” atoms. Oxygen X - ray absorption is relatively difﬁ cult

experimentally because a good vacuum has to be achieved. Furthermore, detector

technology is difﬁ cult for the low energies involved. This initial study used electron

yield detection, which is rather noisy and needs mixing of the zeolite with graphite

to get better conductivity. However, solid - state and gas microstrip detectors exist

that can be used for oxygen K - edge detection at 530 eV.

7.2.2

Transition Metal Exchanged Zeolites

Exchanging the proton forms of zeolites with transition metals leads to redox active

catalysts. Great interest in this technology was triggered at the beginning of the

1990s, when Iwamoto [34 – 36] reported the use of copper exchanged ZSM - 5 material

for the selective catalytic reduction of NO

. The incentive for this reaction came from

the potential use of the NO reduction process in diesel and lean - burn engines.

Figure 7.8 The relaxation of the aluminum tetrahedron upon

adsorption of ammonia on H - ZSM - 5. Adapted from ref. [33] .

EXAFS of transition metal exchanged zeolites is the best technique for

direct structure analysis of these materials. Under ideal conditions, the transition

metal forms single atoms or very small clusters, which are X - ray amorphous.

EXAFS, on the other hand, in particular for the harder edges, can determine the

local structure of exchanged zeolites very well. It is not surprising that EXAFS

spectroscopy has been widely used to characterize Cu - ZSM - 5 materials. Among

the ﬁ rst to characterize copper species in zeolites was Hamada as early as 1990

[37] . It was suggested that there are signiﬁ cant differences in the copper species

present in the ZSM - 5 zeolite as compared to bulk copper oxides, and a number

of authors proposed that the copper species present in zeolites depend on the

loading. At lower loadings single copper ions were thought to be present in cat-

ionic zeolite positions, whereas at higher loadings small copper clusters were

thought to be the active species in the NO

reduction process [37, 38] . The nucle-

arity of the clusters can be determined from ﬁ tting the coordination numbers and

should generally be treated with some caution. However, a combination of the

use of good standards and cautious data treatment generally allows determination

of the cluster sizes involved, even for oxidic materials, which are more difﬁ cult

to analyze than metal clusters. The advantage of in situ determination of EXAFS

and XANES is very well demonstrated in a number of studies, both of variation

of the local structure and of the oxidation state under reactive conditions [38, 39] .

Deactivation of the catalyst under reaction conditions is one of the biggest prob-

lems of Cu - zeolite catalysts. It is therefore not surprising that a number of studies

have attempted to understand the origin of the problem using XAS. Sulfur and

water tolerance have been investigated in detail [40, 41] .

It was suggested by a number of authors that sintering similar to metal catalysts

leads to deactivation of copper - modiﬁ ed zeolites.

Sulfate formation and coke deposition were often excluded as reasons for deac-

tivation, using EXAFS spectroscopy. The formation of larger clusters in the zeolite,

which can also lead to destruction of the host structure, can be observed by an

increase in the coordination number of the Cu

−

Cu scattering contribution to the

EXAFS.

The formation of extended oxidic species was deemed detrimental to the cata-

lytic performance by most authors. XAS has been instrumental in the detection

of these species.

Deactivation of the copper zeolites under de - NO

conditions was one of the

major reasons why the catalyst was never used in a commercial application. Recent

environmental legislation intensiﬁ ed the hunt for a water - and sulfur - stable active

catalyst. One of the most successful preparative methods was reported by Hall and

Feng [42, 43] . They reported excellent de - NO

performance based on an iron

exchanged ZSM - 5 zeolite. The activity was reported to remain constant for extended

times, even under high water and sulfur content conditions. The initial catalytic

study initiated a whole raft of characterization studies by a number of groups. The

interest was signiﬁ cantly increased when it became obvious that there are issues

with catalyst preparation reproducibility [44, 45] . XAS was crucial in the discussion

of the structure of active sites for de - NO

and the site responsible for the high

7.2 Applications of EXAFS 313