Download PDF

Analytical and Bioanalytical Chemistry

, Volume 384, Issue 3, pp 817–826 | Cite as

Multiple-scattering extended X-ray absorption fine structure analysis of nanostructured iron(III) oxide in the pore system of mesoporous carbon CMK-1

Original Paper
First Online:
Received:
Revised:
Accepted:

Abstract

This work is devoted to the EXAFS analysis of nanostructured iron(III) oxide synthesized inside the pore system of mesoporous carbon CMK-1. A detailed study of the recording, preparation and evaluation of data recorded in fluorescence mode at the iron K-edge with and without multiple scattering is shown. The results obtained show that the local structure of Fe3+ inside nanostructured iron(III) oxide is different to that of the bulk material. Due to the small particle size, data analysis is much more difficult and data preparation more complex. Incorporating multiple scattering paths in the Fourier transforms and back-transforms during data evaluation gives structural insights that cannot be obtained using other spectroscopic methods, and this technique was used to draw conclusions about the first four coordination spheres of the nanostructured iron(III) oxide.

Keywords

Fe-K XAFS Nanostructured host/guest compounds Mesoporous carbon Nanostructured iron(III) oxide 

Introduction

In recent years a variety of nanostructured host/guest compounds have been studied by several research groups. The mesoporous silica phases MCM-48 and SBA-15 silica are in the focus of much of this research. Numerous reports have been published on the incorporation of metal atoms into mesoporous MCM-48 (metal atoms such as Ag [1], Co [2, 3], Cu [3, 4], Fe [3, 4, 5, 6, 7, 8], La [9], Mn [10], Ni [3], Pb [5], Pd [11],Ti [12, 13, 14, 15], V [16, 17, 18], Zn [5]) and into mesoporous SBA-15 (metal atoms such as Ag [19], Au [19], Cd [20], Fe [21] Ge [22], Mn [20, 23], Pd [24], Pt [19], Ti [25]). These nanostructured host/guest compounds are of interest for a wide number of applications in materials science, such as in catalysis [3, 4, 5, 9, 16, 26, 27], magnetic and diluted magnetic semiconductors [20, 23], and magnetic studies [8].

A new family of host structures, mesoporous carbons, was introduced to host/guest chemistry in 2001 [28, 29, 30, 31, 32, 33, 34]. The advantages of mesoporous carbons over mesoporous silica phases include their stability in strong acids and bases. They are also stable under high pressure and have a reducible matrix. Joo et al. [35] first reported on platinum nanoparticle incorporation into the mesoporous carbon CMK-3. The electrocatalytic activity of his host/guest compound for oxygen reduction was tested, which may be relevant to fuel-cell devices. In 2003, the authors reported on the incorporation of iron(III) oxide inside the pore system of mesoporous carbon CMK-1 [36]. The focus then turned to the intrapore synthesis of iron(III) oxide. EXAFS (extended X-ray absorption fine structure spectroscopy) experiments were carried out to obtain information on the local structure of the iron(III) oxide. This new nanostructured host/guest compound Fe2O3@CMK-1 was found to be highly desirable for catalysis (of methanol decomposition for instance [37, 38]) and advanced magnetic material preparation after intrapore reduction of the iron(III) oxide.

Gathering information on the local structure and oxidation state of the guest species is a very important activity in host/guest chemistry. EXAFS is often used to characterize nanostructured metal oxides in host materials [3, 6, 38, 39, 40, 41, 42]. These EXAFS evaluations are mostly restricted to the first three metal–metal coordination shells. In this case, the radial distribution function (and therefore the potential structural insights that this could provide) is not fitted, and results are not refined by incorporating multiple scattering paths.

The crystal structure of alpha iron(III) oxide was first solved by Pauling and Hendricks in 1925 [44]; it was then refined in 1966 by Blake [45], the results of which are still valid today. Submicron iron oxide particles can be synthesized with average crystallite dimensions of up to 5 nm using wet chemical techniques with stabilization procedures [43, 44, 45, 46, 47]. Nevertheless, these procedures reveal a broad size distribution, with only a few nanosized particles. In contrast, synthesis within mesoporous host structures leads to particles of a defined size (two dimensions are restricted by the pore diameter and the third dimension by the pore length). The absence of chemical boundaries between the host and the guest compound, as demonstrated in leach experiments, is another important point in relation to nanocomposites [48].

With the introduction of synchrotron radiation as an X-ray absorption spectroscopy source in the 1970s, the structures of haematite and other iron oxides were investigated once more. Several reviews of XAFS spectroscopy have been published [43, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66]. Very recently, a comprehensive review was published by Waychunas et al., who carried out XAFS investigations of nanoparticulate iron oxide minerals in soils and sediments [67].

However, XAFS investigations of nanostructured iron oxides in mesoporous materials have only been performed since 1999. In spite of the great potential of XAFS in this area, only a few papers have investigated it [3, 6, 36]. In these papers, the EXAFS analyses are mostly restricted to three coordination spheres about the iron atom, neglecting multiple scattering contributions to the experimental data.

Experimental

Synthesis and characterization

The synthesis of the host structure CMK-1 was carried out according to the procedures described in the literature [36]. Three impregnation, drying and calcination cycles were applied to form iron(III) oxide inside the host structure.

In a typical procedure, 1 g of dried CMK-1 was stirred for 30 minutes in 50 mL of an ethanolic solution of iron nitrate (4.6 mol/L) at room temperature. The carbon iron nitrate composite was separated from the solution by centrifugation and dried under vacuum at room temperature. Afterwards the host/guest compound was calcined at 573 K for five hours, leading to the transformation of the iron nitrate into iron(III) oxide. By performing this impregnation cycle three times, an iron content of ~65 weight percent [wt%] was obtained. Fe2O3@CMK-1 is also known as a host/guest compound. Bulk haematite, which served as the reference compound in this work, was synthesized in the same way as described above.

The host/guest compound was characterized using powder X-ray diffraction, infrared spectroscopy, nitrogen physisorption, transmission electron microscopy and Raman spectroscopy, in order to demonstrate the preservation of the host structure during the impregnation/calcination process and the formation of the iron(III) oxide almost exclusively inside the pore system of the mesoporous carbon. Further information on the characterization is given in [36].

XAFS: sample preparation

The bulk haematite and the host/guest compound were measured as freshly prepared polyethylene pellets (1 cm in diameter). After drying in vacuum at room temperature, 20–35 mg of the host/guest compound or 5–7 mg of the bulk haematite respectively were crushed and mixed together with 45 mg dried polyethylene to give a homogeneous mixture. The transition metal content of this mixture lead to an absorption jump of Δμ d=0.5–1.0.

Recording the spectra

Fe K-edge spectra were recorded at the EXAFSII beam line at HASYLAB@DESY in Hamburg, Germany. The positron storage ring DORIS III gave a positron energy of 4.45 GeV, an initial beam current of 140 mA and injection at 85 mA in five-bunch mode. The runtime was approximately eight hours.Nonfocused synchrotron radiation was obtained for the EXAFS experiments by bending magnets. The resolution of the Fe K-edge (7112 eV) was between 0.8 and 1.0 eV at the beam line used.

The experimental set-up is shown in Fig. 1. The desired wavelength of the white beam is filtered using a Si(111) double crystal (monochromator). The beam size at the host/guest compound was about 0.8×6 mm. The host/guest compounds were cooled to nearly liquid nitrogen temperature (85 K) using a liquid nitrogen cryostat. The fluorescence signal was detected with a five-element germanium fluorescence detector. The intensity of the beam was controlled using digital monochromator stabilization (DMOSTAB), which was set to 50% on the left side of the rocking curve in order to avoid higher harmonic oscillations.
Fig. 1

Schematic representation of the set-up at the EXAFSII beamline at HASYLAB

The EXAFS spectra were collected for 6950–8200 eV (ΔE=1250 eV). Due to the small amount of iron(III) oxide inside the host, the spectra were recorded in fluorescence mode. All spectra were recorded up to four times to improve the signal-to-noise ratio after summation.

Data reduction

The program WinXAS version 2.3 [68] was used for data reduction.

The formulae, theorems, menu items and so on used are described in detail in the program manual and in references therein [69]. Spectral preparation and reduction consists of

  • Summation of four spectra

  • Energy calibration

  • Background correction

  • Setting E 0

  • Determination of the atomic absorption (μ 0−fit)

  • Fourier transformation and back transformation

Data reduction was performed in order to convert the recorded data into a fittable form. The fitting and refinement were carried out on the Fourier transforms and their back-transforms.

To improve the signal-to-noise ratio, four energy-calibrated and normalized fluorescence spectra were averaged. Energy calibration was performed using a reference spectrum which was recorded in transmission mode for every host/guest compound spectrum. Background correction was achieved by using one polynomial at the pre-edge region and another at the post-edge region. The maximum order of each polynomials was restricted to two in order to keep the atomic adsorption function μ 0 oscillation-resistant. An advanced spline fit was carried out to determine the atomic absorption μ 0 in the EXAFS region. In contrast to the cubic spline fit, the range of k was divided up here into to four sections with its own number of knots. This spline function gives better results for the host/guest system. The extraction of the atomic absorption function is the trickiest part and must be carried out very carefully to prevent data manipulation. In the advanced spline fit, the number of knots for the spline and the selected spectral range were varied; the weighting of k was always three. The range setting is sensitive to the number of independent parameters N ind (Eq. 1). In general the range was defined to be from the first minimum after the white line of the spectrum up to k=18 Å−1, depending on the quality of the recorded spectra. This was used to obtain maximum information while minimizing the number of independent parameters. The EXAFS oscillations were separated using a Fourier transform with a Bessel window with a beta of 4.
$$N_{{{\text{ind}}}} = \frac{{2\Delta R\Delta k}}{\pi } + 2$$
(1)
Here N ind is the number of independent parameters, ∆R is the fit range, and ∆k is the range in k-space.
The Fourier transformation gives a radial distribution function (RDF). The positions r of the emerged peak maxima are phase-shifted distances between the central and backscattering atom. Figure 2 shows the non-phase corrected RDF of bulk haematite.
Fig. 2

Radial distribution function (RDF, not phase-corrected) of the Π(k).k 3 function from the EXAFS spectrum of bulk haematite

The first peak corresponds to the iron–oxygen distance in haematite. The second peak corresponds to iron–iron distances between face- and edge-shared octahedra. Peak three corresponds mainly to the iron–iron distance between corner-shared octahedra. Peak four is determined through multiple scattering effects and cannot be assigned to coordination spheres.

Back transformation was performed on the RDF with a k-weight of 3 in the range where the iron–iron shells appear; typically \(\widetilde{R} = 2.2 - 4.2{\mathop {\text{A}}\limits^ \circ }\). The distances obtained correspond to the distances between iron and its neighbouring atoms in the respective spheres in haematite.

Data analysis

Fit simulation and refinement was carried out on the Fourier transforms as well as on the back-transforms using the ab initio theoretical XAS code FEFF8.2 [70].

XAS simulation and refinement using the ab initio theoretical XAS code FEFF8.2 was carried out on the Fourier transforms and on the back-transforms in the same way. First the FEFF input file was generated.The input file for haematite (available on the ATOMS website [71]) was converted to an FEFF input file using the program ATOMS [72]. Single and multiple scattering paths were obtained by executing FEFF with the respective FEFF input file [68, 70].

To extract structural information the fast curve wave theory [73] was used. It can be described with Eq. (2). The parameters F j, σ’ and λ were calculated using the von Barth potential for ground states, the Hedin-Lundquist exchange potential for excited states [74] and the relaxed approximation for the core-hole [75].
$$\chi {\left( k \right)} = {\sum\limits_j {\frac{{N_{j} S^{2}_{0} F_{j} {\left( k \right)}}}{{kR^{2}_{j} }}*e^{{{\left( { - 2k^{2} \sigma ^{2}_{j} } \right)}}} *e^{{{\left( {\frac{{ - 2R_{j} }}{\lambda }} \right)}}} *} }e^{{{\left( {\raise0.7ex\hbox{$2$} \!\mathord{\left/ {\vphantom {2 3}}\right.\kern-\nulldelimiterspace}\!\lower0.7ex\hbox{$3$}\sigma ^{}_{j} jk^{4} } \right)}}} *\sin {\left[ {2kR_{j} + \delta _{j} {\left( k \right)} - \raise0.7ex\hbox{$4$} \!\mathord{\left/ {\vphantom {4 3}}\right.\kern-\nulldelimiterspace}\!\lower0.7ex\hbox{$3$}\sigma ^{\prime }_{j} k^{3} } \right]}$$
(2)
where F(k) is the backscattering amplitude, δ(k) is the phase, λ is the photoelectron mean free path, j is the j-th scattering path, N is the coordination number, R is the distance, σ 2 is the disorder parameter, \(S_{0} ^{2} \) is the amplitude reduction factor, σ’ is the third cumulant, and σ’ is the fourth cumulant.
All of the single scattering paths and the multiple scattering paths with an amplitude intensity larger than 2% in comparison to the first shell up to the fifth angular path were used for bulk haematite. Normally this set of reduced FEFF input files is used for XAFS simulation and refinement. However, the problem with this is that these paths calculated from single-crystal data are often too large for nanosized clusters. Therefore, these FEFF input files were refined in two steps; see Fig. 3. Firstly, a FEFF simulation was carried out with the reduced FEFF input file. During this step, the current relative weights and angles for each path were calculated anew with respect to the adapted structure of bulk haematite. Secondly, these new FEFF files were filtered (relative weight >3%) and stored as optimized FEFF files.
Fig. 3

Usual and new refinement procedures for the FEFF file

The optimized FEFF file contains only significant FEFF paths and amplitudes for the host/guest compound and so is more suitable for the refinement procedures.

The distance R is chosen based on the appearance of the shells in the respective XAS function. To keep the number of free running parameters N frp significantly smaller than the number of independent parameters N idp during the refinements, some restrictions were placed on the variables:

  • The amplitude reduction factor for bulk haematite was constrained to 0.5–0.95 and the result was used as a constant in the host/guest compound fit.

  • The Debye-Waller factors were constrained to 0.001–0.01 Å2

  • The Debye-Waller factors for the multiple scattering paths were calculated from those of the single scattering paths

  • The coordination numbers and distances of the multiple scattering paths were correlated with the respective single scattering paths

  • The energy shift was set to be the same for each element in the system

The simulation was carried out with these restrictions and was it refined up to an acceptable level. These parameters were then used as the initial parameters in the next fit refinement procedure. The most restrictive refinement was done first of all, and then more and more parameters were set free. The refinement was continued as long as the following requirements were achieved:

  • Low deviation parameter χ 2

  • Residual is smaller than 10, calculated via Eq. 3

  • Correlations in the correlation matrix are between 0.96 and 1

  • F-Test result is >0.6 after refinement with more free variable running parameters, proving that a global minimum has been reached.

$$R{\left[ \% \right]} = \frac{{{\sum\limits_{i = 1}^N {{\left| {y_{{\exp }} {\left( i \right)} - y_{{{\text{theo}}}} {\left( i \right)}} \right|}} }}}{{{\sum\limits_{i = 1}^N {{\left| {y_{{\exp }} {\left( i \right)}} \right|}} }}} \times 100,$$
(3)
Here N is the number of data points, y exp are the experimental data points, and y theo are the theoretical data points.

In many cases the errors were estimated at 1% for the distance, 5% for the Debye-Waller factor, 10% for the inner potential correction and 5% for the coordination number [76]. Due to the (nano)sizes of the particles and the consequent distortion in the crystal structure of the iron(III) oxide, these error limits need more consideration.

Results and discussion

XAFS measurements were carried out in fluorescence mode in order to investigate the nature of the iron coordination within the iron oxide nanoparticle.

Due to the conditions used during the synthesis of the host/guest compound, it can be assumed that iron(III) oxide with a haematite-like structure is formed inside the pore system of the mesoporous carbon.

Bulk haematite is isostructural with corundum. It can be thought of as consisting of hcp arrays of oxygen ions where two thirds of the octahedral sites filled with iron(III) ions [77]. The arrangement of the iron ions leads to FeO6 octahedra. As shown in Fig. 4, three different kinds of octahedra linkages exist. Iron–iron distances of about 2.90 Å are observed between face-shared and distances of about 2.97 Å between edge-shared octahedra. The two other iron–iron distances (3.36 and 3.71 Å) are due to a corner-shared linkages of the octahedra [45, 54].
Fig. 4

Polyhedral model of haematite. Dark: iron, light: oxygen atoms

The Fourier transforms of bulk haematite and the host/guest compound are shown in Fig. 5. The iron–oxygen shells of the host/guest compound are mostly unchanged from bulk haematite. The radial distribution function of bulk haematite shows peaks of up to 7.5 Å, while peaks of up to 4 Å are distinguishable for the host/guest compound. The strong reduction in backscattering amplitude intensity of the iron–iron shells of the host/guest compound is due to the small particle size (and therefore the limited vertical extension) and structural disorder; these effects are well known for systems lacking long-range order. Therefore either an increase in the Debye-Waller factors and/or a reduction in the coordination numbers of outer shells must be assumed. It should be noted that the Debye-Waller factors and the coordination numbers are strongly correlated and should not be neglected in a multiple shell fit. A significant reduction in the coordination numbers of the outer shells only occurs for particles smaller than 5 nm.
Fig. 5

Radial distribution functions (RDF, not phase-corrected) of the Fe K-edge oscillations χ(k). k 3 of the host/guest compound Fe2O3@CMK-1 and of bulk haematite. The RDF of bulk haematite has been divided by two

Refined structure parameters were obtained for bulk haematite (host/guest compound) by firstly fitting the back-transforms with four (three) iron–iron shells) (see Fig. 6), and then including the iron–oxygen shells (two shells in both cases, see Fig. 7).
Fig. 6

Back-transforms of the Fe–Fe coordination shells and the results from least squares refinements of bulk haematite and the host/guest compound

Fig. 7

Back-transforms of the Fe–O and Fe–Fe coordination shells and the results from least squares refinements of bulk haematite and the host/guest compound

Figures 6 and 7 show the back-transforms and the results of the refinements. The refined data are listed in Table 1, the refinement parameters in Table 2. The differences between experimental and fit data are marginal for both tables. Deviations at low k are probably due to the restricted energy shift, the influence of which is most prominent at low k values.
Table 1

Refined structural parameters of the iron–iron shells of the host/guest compound and bulk haematite as well as corresponding results extracted from single-crystal data for haematite, obtained with.without refining the single scattering paths for the iron–oxygen coordination shells

Refinement

Shell

Host/guest compound

Bulk haematite

Haematite single-crystal [46]

With Fe–O

Without Fe–O

With Fe–O

Without Fe–O

N

Fe–O

2.90

-

3.01

-

3

R [Å]

1.91

-

1.95

-

1.94

∆σ 2 [Å2]

0.007

-

0.003

-

-

N

Fe–O

2.85

-

3.01

-

3

R [Å]

2.13

-

2.10

-

2.11

∆σ 2 [Å2]

0.008

-

0.004

-

-

N

Fe–Fe

0.74

0.73

0.90

0.91

1

R [Å]

3.00

3.01

2.88

2.88

2.89

∆σ 2 [Å2]

0.003

0.004

0.004

0.003

-

N

Fe–Fe

1.35

1.31

2.98

3

3

R [Å]

3.04

3.03

2.94

2.95

2.96

∆σ 2 [Å2]

0.004

0.005

0.003

0.004

-

N

Fe–Fe

0.53

0.62

3

2.87

3

R [Å]

3.39

3.40

3.34

3.36

3.35

∆σ 2 [Å2]

0.003

0.003

0.004

0.003

-

N

Fe–O

2.66

fixed

2.89

fixed

3

R [Å]

3.40

fixed

3.38

fixed

3.39

∆σ 2 [Å2]

0.008

fixed

0.006

fixed

-

N

Fe–O

2.84

fixed

2.93

fixed

3

R [Å]

3.60

fixed

3.61

fixed

3.59

∆σ 2 [Å2]

0.009

fixed

0.006

fixed

-

N

Fe–Fe

-

-

2.90

2.89

3

R [Å]

-

-

3.72

3.71

3.69

∆σ 2 [Å2]

-

-

0.003

0.002

-

Refinements were carried out on the back-transforms

N: coordination number; R: bond length; ∆σ 2: Debye-Waller factor. Even though the two higher Fe-O shells make no significant contribution to the backscattering amplitude, they were included (as fixed values) in order to be able to compare them with the results gained from the refined Fourier transforms

Table 2

Refinement parameters for the back-transform (BT) fitted host/guest compound and bulk haematite

 

BT data without Fe–O

BT data with Fe–O

Bulk haematite

Host/guest compound

Bulk haematite

Host/guest compound

BTR

2.1–4.1 Å

2.1–4.1 Å

1.1–4.1 Å

1.0–4.1 Å

\(S_{0} ^{2} \)

0.90

0.90

0.93

0.93

∆E 0 [eV]

−3.03

−4.37

Fe: −4.11

Fe: −5.55

   

O: −6.71

O: −6.91

R [%]

4.6

5.0

6.2

7.0

N idp

15

15

21

18

N frp

10

9

11

10

SS

4+2 fixed

3+2 fixed

6

5

MS

0

0

0

0

BTR: Back-transform range from the Fourier transforms; \(S_{0} ^{2} \): amplitude reduction factor; ∆E 0: energy shift; R: residual; N frp: number of free running parameters; N idp: max number of free running parameters

Comparing the fit results for the refinement with and without the iron–oxygen shells, it is remarkable that there are no significant differences between the distances and Debye-Waller factors. Due to this, and the structural information from the iron–iron shells, the iron–oxygen shell can be neglected in further data analyses and discussions.

There are three differences between the bulk material and the host/guest compound. Firstly, it is not possible to use four iron–iron shells for refinement in the case of the host/guest compound. The first or the fourth iron–iron shell cannot be fitted, as shown earlier in our group [3, 6]. The loss of important multiple scattering paths during the back-transform procedure might be one reason for this. The importance of the first iron–iron shell is apparent from the Fourier transformed fit results. Secondly, the increase in the coordination number is different, and thirdly the distances for the iron–iron shells are different too. The last two points will be discussed later on for the Fourier transform refinement results.

Figures 8 and 9 show the experimental RDFs of bulk haematite and the host/guest compound, as well as their fit results. Two iron–oxygen shells were used to fit the first peak around 1.5 Å and four iron–iron shells to fit the second FT peak. The fitted peaks at higher R-values are due to multiple scattering effects. The refined data are listed in Table 3, the refinement parameters in Table 4. The coordination numbers and distances are similar to the single crystal data for both. Normally the Debye-Waller factors increase for higher shells, as can be seen for the Fe–O shells, and tend to become smaller from Fe–Fe to Fe–O interactions, as observed in Fe–SiO2 and NiO–SiO2 nanocomposites [43]. This effect is not observed for the iron–iron shells here, and this can largely be explained by the smaller particle size.
Fig. 8

Fourier transforms (FT) of bulk haematite and the host/guest compound Fe2O3@CMK-1, and the results of corresponding least square fits carried out on the respective iron–iron and iron–oxygen shells (fixed) with the aid of multiple scattering paths

Fig. 9

Fourier transforms (FT) of bulk haematite and the host/guest compound Fe2O3@CMK-1, and the results of corresponding least square fits carried out on the respective iron–iron and iron–oxygen shells with the aid of multiple scattering paths

Table 3

Refined structural parameters of the iron–iron shells of the host/guest compound and bulk haematite as well as the corresponding results from single-crystal haematite data, obtained with/without refining the single scattering paths of the iron–oxygen coordination shells

Refinement

Shell

Host/guest compound

Bulk haematite

Haematite single-crystal [46]

With Fe–O

Without Fe–O

With Fe–O

Without Fe–O

 

N

Fe–O

2.92

-

3.01

-

3

R [Å]

1.92

-

1.95

-

1.94

∆σ 2 [Å2]

0.008

-

0.006

-

-

N

Fe–O

2.88

-

2.99

-

3

R [Å]

2.12

-

2.10

-

2.11

∆σ 2 [Å2]

0.008

-

0.007

-

-

N

Fe–Fe

0.52

0.46

0.87

0.85

1

R [Å]

2.95

2.96

2.90

2.90

2.89

∆σ 2 [Å2]

0.004

0.003

0.004

0.004

-

N

Fe–Fe

1.05

0.90

2.99

2.98

3

R [Å]

2.99

2.98

2.96

2.97

2.96

∆σ 2 [Å2]

0.003

0.004

0.003

0.004

-

N

Fe–Fe

0.76

0.38

3.00

2.99

3

R [Å]

3.26

3.23

3.36

3.37

3.35

∆σ2 [Å2]

0.003

0.004

0.004

0.003

-

N

Fe–O

2.79

fixed

3.03

fixed

3

R [Å]

3.31

fixed

3.40

fixed

3.39

∆σ 2 [Å2]

0.007

fixed

0.006

fixed

-

N

Fe–O

2.81

fixed

2.96

fixed

3

R [Å]

3.50

fixed

3.61

fixed

3.59

∆σ 2 [Å2]

0.009

fixed

0.007

fixed

-

N

Fe–Fe

1.39

0.73

4.22

3.87

3

R [Å]

3.53

3.54

3.71

3.70

3.69

∆σ 2 [Å2]

0.004

0.004

0.004

0.003

-

Refinements were carried out on the Fourier transforms

N: coordination number; R: bond length; ∆σ 2: Debye-Waller factor

Table 4

Refinement parameters for the Fourier transform-fitted host/guest compound and bulk haematite. The spectral range in k space is 3.0–15.0 Å−1 for all

 

FT data without Fe–O

FT data with Fe–O

Bulk haematite

Host/guest compound

Bulk haematite

Host/guest compound

Fit range

2.1–5.3 Å

2.1–5.3 Å

1.1–5.3 Å

1.0–4.1 Å

S 0 2

0.94

0.94

0.92

0.92

∆E 0 [eV]

−3.98

−6.07

Fe: −5.03

Fe: −6.11

   

O: −6.42

O: −4.38

R [%]

5.1

7.7

4.3

6.2

N idp

27

27

34

26

N frp

14

13

22

15

SS

4+2 fixed

4+2 fixed

8

6

MS

7

7

7

7

\(S_{0} ^{2} \): amplitude reduction factor; ∆E 0: energy shift; R: residual; N frp: number of free running parameters; N idp: max number of free running parameters

The results and fit parameters gained from the refined backscattering and the Fourier transform calculations, performed with and without the iron–iron shells and the multiple scattering paths, are given in the respective tables and figures. The differences between the experimental and the refined fit data are marginal for all four bulk haematite Fe–Fe shells. Also, the refined fit results barely changed whether the iron–oxygen shells were included or not, which means that these shells can be neglected for further data analysis and discussion.

A good fit result was obtained for bulk haematite. The coordination numbers and distances agree well with the corresponding single-crystal data.

One exception can be seen, however, for the fourth iron–iron shell, where the coordination number is significantly smaller than for the single-crystal data. This reduction is probably due to powder particles in the bulk material, a well-known effect for higher shells. A significant shift in the Debye-Waller factors is not observed, which supports the presence of small particles.

A significant reduction in the coordination numbers for the host/guest compound can be observed. The reduction increases with distance. As for bulk haematite, the host/guest compound shows no significant shift in the Debye-Waller factors. These factors are very similar for both bulk haematite and the host/guest compound. The reduction in the coordination numbers combined with the minimal change in Debye-Waller factors shows that very small nanoparticles (<3 nm) are formed inside the pore system. The distances obtained for the second and fourth iron–iron shells differ from those obtained from single-crystal data. They correspond to edge- and corner-shared octahedra. The first iron–iron shell appears at 3.01 nm, a distance where only edge-shared octahedra can form. This very important structural fact, that no face-shared octahedra are found in the host/guest compound, can be lost when refining the back-transforms. On the other hand, no proof of the corner-shared linkage of the fourth shell was found when refining the back-transforms of the first iron–iron shell. It is therefore difficult to ascertain whether the host has haematite structure or whether it is a haematite-like iron(III) oxide. Full information about the coordination numbers and the distances are only obtained by fitting the Fourier transform over a wide range and using the relevant multiple scattering paths. Furthermore, the distances obtained from the refined FT are closer to the distances observed in the single-crystal data than to those gained by refining the back-transforms, due to the reason mentioned above.

As discussed in earlier, a specially refined FEFF file was used for XAFS simulation and refinement. This means that the weights and angles of the multiple scattering paths were adapted to the structure of this bulk haematite. Figures 10 and 11 show the single and the two strongest multiple scattering paths that contribute to the EXAFS fits of the host/guest compound and bulk haematite for 2–4 Å. The declared weights of about 18% (16% for bulk haematite, referring to the strongest amplitude) for the multiple scattering paths underlines once more the need to use them. The influence of the multiple scattering paths increases with increasing amplitude, as shown by the quality of the results for the host/guest compound. The refined FEFF file, and therefore the application of the multiple scattering paths, allows us to reach the conclusions made here.
Fig. 10

XAFS refinement of bulk haematite to experimental haematite FT[χ(k)*k 3]. The Fe–Fe coordination shells (weights 29, 87, 62 and 100%), the Fe–O coordination shells (weights 9 and 11%), as well as the four triple scattering paths with the highest contributions (weights estimated to be 6–18%) are shown

Fig. 11

XAFS refinement of the host/guest compound Fe2O3@CMK-1 as an overview (top) and at a higher magnification (bottom). The Fe–Fe coordination shells (weights 23, 89, 47 and 100%), the Fe–O coordination shells (weights 9 and 10%) as well as the four triple scattering paths with the highest contributions (weights estimated to be 3–14%)

Conclusions

Fluorescence EXAFS investigations were carried out in order to obtain information on the structure of nanosized iron(III) oxide inside the pore system of mesoporous carbon. Due to large number of structures possible (not only in terms of the host/guest chemistry), a detailed study of the recording, preparation and evaluation of the data is presented. The EXAFS results show that most of the structural information can be obtained by refining the radial distribution function with the relevant multiple scattering paths over a wide range.

It was shown that the use of multiple scattering paths in refined Fourier transforms performed over a wide range can reveal results that are difficult or impossible to probe with other spectroscopic methods. The authors think that the procedure described above can be applied to nanostructured particles of <4 nm in general, and it is not restricted to host/guest chemistry.

Notes

Acknowledgements

The authors would like to thank HASYLAB@DESY, Hamburg, Germany, for allocating beam time and financial support. Further financial support from the Justus Liebig University Giessen and the Fonds der Chemischen Industrie is gratefully acknowledged.

References

  1. 1.
    Wang L-Z, Shi J-L, Zhang W-H, Ruan M-L, Yu J, Yan D-S (1999) Chem Mater 11:3015CrossRefGoogle Scholar
  2. 2.
    Köhn R, Brieler F, Fröba M (2000) Stud Surf Sci Catal 129:341–348Google Scholar
  3. 3.
    Köhn R, Fröba M (2001) Catal Today 68:227–236CrossRefGoogle Scholar
  4. 4.
    Minchev C, Köhn R, Tsoncheva T, Dimitrov M, Mitov I, Paneva D, Huwe H, Fröba M (2002) Stud Surf Sci Catal 142:1245–1252CrossRefGoogle Scholar
  5. 5.
    Dapurkar SE, Badamali SK, Selvam P (2001) Catal Today 68:63–68CrossRefGoogle Scholar
  6. 6.
    Fröba M, Köhn R, Boufffaud G, Richard O, v. Tendeloo G (1999) Chem Mater 11:2858–2865CrossRefGoogle Scholar
  7. 7.
    Köhn R, Paneva D, Dimitrov M, Tsoncheva T, Mitov I, Minchev C, Fröba M (2003) Microporous Mesoporous Mater 63:125–137CrossRefGoogle Scholar
  8. 8.
    Köhn R, Fröba M (2003) Z Anorg Allg Chem 629:1673–1682CrossRefGoogle Scholar
  9. 9.
    Nguyen SV, Szabo V, Trong On D, Kaliaguine S (2002) Microporous Mesoporous Mater 54:51–61CrossRefGoogle Scholar
  10. 10.
    Zhao DY, Goldfarb DJ (1995) J Chem Soc Chem Commun 875–876Google Scholar
  11. 11.
    Yang C-M, Sheu H-S, Chao K-J (2002) Adv Funct Mater 12:143–148CrossRefGoogle Scholar
  12. 12.
    Morey MS, O’Brien S, Schwarz S, Stucky GD (2000) Chem Mater 12:898CrossRefGoogle Scholar
  13. 13.
    Walker JV, Morey M, Carlsson H, Davidson A, Stucky GD, Butler A (1997) J Am Chem Soc 119:6921CrossRefGoogle Scholar
  14. 14.
    Widenmeyer M, Grasser S, Köhler K, Anwander R (2001) Microporous Mesoporous Mater 44:327–336CrossRefGoogle Scholar
  15. 15.
    Schrijnemakers K, Vansant EF (2001) J Porous Mater 8:83–90CrossRefGoogle Scholar
  16. 16.
    Selvam P, Dapurkar SE (2004) Appl Catal A: General 276:257–265CrossRefGoogle Scholar
  17. 17.
    Van der Voort P, Baltes M, Vansant EF (1999) J Phys Chem B 103:10102–10108CrossRefGoogle Scholar
  18. 18.
    Van der Voort P, Vansant EF (2000) Microporous Mesoporous Mater 38:385–390CrossRefGoogle Scholar
  19. 19.
    Han,Y-J, Kim JM, Stucky GD (2000) Chem Mater 8(12):2068–2069CrossRefGoogle Scholar
  20. 20.
    Brieler F, Grundmann P, Fröba M, Chen L, Klar PJ, Heimbrodt W, Krug von Nidda H-A, Kurz T, Loidl A (2005) Chem Mater 17(4):795–803CrossRefGoogle Scholar
  21. 21.
    Martinez F, Han Y-J, Stucky G, Sotelo JL, Ovejero G, Melero JA (2002) Stud Surf Sci Catal 142B:1109–1116Google Scholar
  22. 22.
    Schlecht S, Yosef M, Fröba M (2004) Z Anorg Allg Chem 630:864–868CrossRefGoogle Scholar
  23. 23.
    Brieler F, Grundmann P, Fröba M, Chen L, Klar PJ, Heimbrodt W, Krug von Nidda H-A, Kurz T, Loidl A (2004) J Am Chem Soc 126:797–807CrossRefGoogle Scholar
  24. 24.
    Zhang Z, Dai S, Blom DA, Shen J (2002) Chem Mater 14:965–968CrossRefGoogle Scholar
  25. 25.
    Zhang WH, Lu J, Han B, Li M, Xiu J, Ying P, Li C (2002) Chem Mater 14:3413–3421CrossRefGoogle Scholar
  26. 26.
    Trong On D, Desplantier-Giscard D, Danumah C, Kaliaguine S (2001) Appl Catal A: General 222:299–357CrossRefGoogle Scholar
  27. 27.
    Weitkamp J, Glaeser R, Dittmeyer R (eds) (2004) Winnacker-Kuechler: Chemische Technik 1:645–718 Wiley-VCH Verlag, Weinheim, GermanyGoogle Scholar
  28. 28.
    Ryoo R, Joo SH (2004) Stud Surf Sci Catal 148:241–260Google Scholar
  29. 29.
    Ryoo R, Joo SH, Jun S (1999) J Phys Chem B 103(37):7743–7746CrossRefGoogle Scholar
  30. 30.
    Joo SH, Jun S, Ryoo R (2001) Microporous Mesoporous Mater 44(45):153–158CrossRefGoogle Scholar
  31. 31.
    Kruk M, Jaroniec M, Ryoo R, Joo SH (2000) J Phys Chem B 104:7960–7968CrossRefGoogle Scholar
  32. 32.
    Ryoo R, Joo SH, Krug M, Jaroniec M (2001) Adv Mater 13(9):677–681CrossRefGoogle Scholar
  33. 33.
    Kaneda M, Tsubakiyama T, Carlsson A, Sakamoto Y, Ohsuna T, Terasaki O (2002) J Phys Chem B 106:1256–1266CrossRefGoogle Scholar
  34. 34.
    Jun S, Joo SH, Ryoo R, Kruk M, Jaroniec M, Liu Z, Ohsuna T, Terasaki O (2000) J Am Chem Soc 122(43):10712–10713CrossRefGoogle Scholar
  35. 35.
    Joo SH, Choi SJ, Oh I, Kwak J, Liu Z, Terasaki O, Ryoo R (2001) Nature 412:169–172CrossRefGoogle Scholar
  36. 36.
    Huwe H, Fröba M (2003) Microporous Mesoporous Mater 60:151–158CrossRefGoogle Scholar
  37. 37.
    Minchev C, Huwe H, Tsoncheva T, Paneva D, Dimitrov M, Mitov I, Fröba M (2005) Microporous Mesoporous Mater 8:333–341CrossRefGoogle Scholar
  38. 38.
    Tsoncheva T, Paneva D, Mitov I, Huwe H, Fröba M, Dimitrov M, Minchev C (2004) React Kinet Catal Lett 83:299–305CrossRefGoogle Scholar
  39. 39.
    Tuel A, Arcon I, Millett JMM (1998) J Chem Soc Faraday Transactions 94(23):3501–3510CrossRefGoogle Scholar
  40. 40.
    Wong S-T, Lee J-F, Cheng S, Mou C-Y (2000) Appl Catal A: General 198:115–126CrossRefGoogle Scholar
  41. 41.
    Burch R, Cruise N, Gleeson D, Tsang SC (1998) J Mater Chem 8:227–231CrossRefGoogle Scholar
  42. 42.
    Henke BL, Gullikson EM, Davis JC (1993) Atomic Data Nucl Data Tables 54:181–342CrossRefGoogle Scholar
  43. 43.
    Corrias A, Ennas G, Mountjoy G, Paschina G (2000) Phys Chem Chem Phys 2:1045–1050CrossRefGoogle Scholar
  44. 44.
    Pauling L, Hendricks SB (1925) J Am Chem Soc 47:781–790CrossRefGoogle Scholar
  45. 45.
    Blake RL, Hessevick RE, Zoltai T, Finger LW (1966) Am Mineral 51:123–129Google Scholar
  46. 46.
    Ennas G, Marongiu G, Musino A, Falqui A, Ballirano P, Caminiti R (1999) J Mater Res 14:1570–1575Google Scholar
  47. 47.
    Onodera S, Kondo H, Kawana T (1996) MRS Bullettin 21:35–41Google Scholar
  48. 48.
    Lund CRF, Dumesic JA (1982) J Phys Chem 86:130–135CrossRefGoogle Scholar
  49. 49.
    Torok SB, Van Grieken RE (1994) Anal Chem 66:186–206CrossRefGoogle Scholar
  50. 50.
    Becker KD, Niemeier D, Wissmann S, Oversluizen M, Couves JW, Chadwick AV (1995) Nucl Instr Meth Phys Res B 97:111–114CrossRefGoogle Scholar
  51. 51.
    Lee PA, Citrin PH, Eisenberger P, Kincaid BM (1981) Rev Mod Phys 53:769–806CrossRefGoogle Scholar
  52. 52.
    De Groot FM, Fuggle JC, Thole BT, Sawatzky GA (1999) Phys Rev B 42:5459–5468CrossRefGoogle Scholar
  53. 53.
    Lee PA, Pendry JB (1975) Phys Rev B 11:2795–2811CrossRefGoogle Scholar
  54. 54.
    Manceau A, Drits VA (1993) Clay Minerals 28:165–184CrossRefGoogle Scholar
  55. 55.
    Moscovici J, Benzakour M, Decker S, Carnes C, Klabunde K, Michalowicz A (2001) J Synchrot Radiat 8:925–927CrossRefGoogle Scholar
  56. 56.
    Rehr JJ, Ankudinov A, Zabinsky SI (1998) Catal Today 39:263–269CrossRefGoogle Scholar
  57. 57.
    Meitzner G (1998) Catal Today 39:281–291CrossRefGoogle Scholar
  58. 58.
    Wu ZY, Gota S, Jollett F, Pollak M, Gautier-Soyer M, Natoli CR (1997) Phys Rev B 55:2570–2577CrossRefGoogle Scholar
  59. 59.
    Koningsberger DC, Mojet BI, Van Dorssen GE, Ramaker DE (2000) Topics Catal 10:143–155CrossRefGoogle Scholar
  60. 60.
    Kitamura M, Muramatsu S, Sugiura C (1986) Phys Rev B 33:5294–5300CrossRefGoogle Scholar
  61. 61.
    Crocombette JP, Pollak M, Jollett F, Thromat N, Gautier-Soyer M (1995) Phys Rev B 52:3143–3150CrossRefGoogle Scholar
  62. 62.
    Chou SH, Guo J, Ellis DE (1986) Phys Rev B 34:12–24CrossRefGoogle Scholar
  63. 63.
    Alberding N, Crozier ED (1983) Phys Rev B 27:3374–3382CrossRefGoogle Scholar
  64. 64.
    Sevillano E, Meuth H, Rehr JJ (1979) Phys Rev B 20:4908–4911CrossRefGoogle Scholar
  65. 65.
    Grunes LA (1983) Phys Rev B 27:2111–2131CrossRefGoogle Scholar
  66. 66.
    Caliebe WA, Kao CC, Hastings JB, Taguchi M, Kotani A, Uozumi T, de Groot FMF (1998) Phys Rev B 58:13452–13458CrossRefGoogle Scholar
  67. 67.
    Waychunas GA, Kim CS, Banfield JF (2005) J Nanoparticle Res 7:409–433CrossRefGoogle Scholar
  68. 68.
    Ressler T (1998) J Synchrot Radiat 5:118–122CrossRefGoogle Scholar
  69. 69.
  70. 70.
    Newville M, Ravel B, Haskel D, Rehr JJ, Stern EA, Yacoby Y (1995) Phys B 208:154–156CrossRefGoogle Scholar
  71. 71.
  72. 72.
    Ravel B (2001) J Synchrot Radiat 8:314–316CrossRefGoogle Scholar
  73. 73.
    Gurman SJ, Binsted N, Ross I (1984) J Phys C 17:143–151CrossRefGoogle Scholar
  74. 74.
    Van Barth U, Hedin L (1972) J Phys C 5:1629–1642CrossRefGoogle Scholar
  75. 75.
    Crozier ED (1997) Nucl Instr Meth Phys Res B 133:134–144CrossRefGoogle Scholar
  76. 76.
    Li GG, Bridges F, Booth CH (1995) 52:6332–6648Google Scholar
  77. 77.
    Greegor RB, Lytle FW (1980) J Catal 63:476–486CrossRefGoogle Scholar

Copyright information

© Springer-Verlag 2006

Authors and Affiliations

  1. 1.Institute of Inorganic and Analytical ChemistryJustus Liebig University GiessenGiessenGermany

Personalised recommendations