Micron 37 (2006) 473477

www.elsevier.com/locate/micron

Determination of the oxidation state for iron oxide minerals

by energy-filtering TEM
U. Golla-Schindler a,*, R. Hinrichs b, O. Bomati-Miguel c, A. Putnis a
a
Institut fur Mineralogie, University of Muenster, 48149 Muenster, Germany
b
Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, Brazil
c
Insituto de Ciencia de Materiales de Madrid, Cantoblanco, 28049 Madrid, Spain

Abstract
The oxidation state of iron oxide nanoparticles was determined using the two principally different technical realisations of energy filtering
TEM, in one case using the JEOL 3010 equipped with a LaB6 cathode and a post-column GIF and in the second, the newly designed LIBRA
200FE equipped with an corrected in-column 908 energy filter and a field emission gun (Schottky emitter). The samples studied were oxide-coated
iron nanoparticles, and iron oxide inclusions in feldspars in granites.
Five possible candidates exist for the iron-oxide phases: FeO, a-Fe2O3 (hematite), g-Fe2O3 (maghemite), Fe3O4 (magnetite) or a-FeO(OH)
(goethite). Fingerprinting the O K-edge ELNES allows to distinguish between oxide phases with the same stochiometry and enables to make a first
selection of possible candidates. The additional determination of the chemical composition allows unique identification of the phase present. For
the oxide coated iron nanoparticles the most probable iron oxide phase of the shell is maghemite, which was additionally confirmed by HRTEM
studies. The second studied system were iron oxide needles in alkali feldspar, where we obtained hematite as the most probable phase. There we
additionally demonstrated the drastic changes of the ELNES of the O K-edge for the alkali feldspar and iron oxide needle by spatially resolved
EELS.
q 2005 Elsevier Ltd. All rights reserved.

Keywords: Electron energy loss spectroscopy (EELS); Electron loss near edge structures (ELNES); Energy-filtering transmission electron microscope;
Fingerprinting; Spatially resolved EELS; Oxides, granites, nanoparticles

1. Introduction composition (Egerton, 1996; Reimer, 1995), and the fine

Several analytical techniques are available to provide
chemical and structural information on the macroscopic and
microscopic scale, but there is often a lack of information
regarding the nanoscale processes associated with mineral
mechanisms and phase transitions. The new generation of field
emission electron microscopes equipped with an energy filter
enables excellent spatial as well as energy resolution, allowing
the acquisition of detailed information about the atomic
structure, the chemical composition and the local electronic
states of the object and opens new avenues for advanced
applications, like establishing the correlation of macroscopic
with microscopic and nanoscopic properties of materials.
Electron energy loss spectroscopy (EELS) offers the
possibility of obtaining point information on the chemical
* Corresponding author.
E-mail address: golla@nwz.uni-muenster.de (U. Golla-Schindler).
0968-4328/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.micron.2005.11.002
structure of the electron energy loss spectrum (ELNES)
provides information about the atomic environment and state
(Mitterbauer et al., 2003; Lazar et al., 2003). In this study we
used spatially resolved EELS (Reimer, 1995), which involves
orienting the exsolution precipitate, interface etc. of interest
perpendicular to the energy dispersive direction and taking
image spectra (Golla-Schindler et al., 2003; Kimoto et al.,
1997; Reimer et al., 1992; Walther and Mader, 1999). This
allows a correlation of the ELNES to the features in the
specimen structure and provides a precise calibration of the
energy scale. The combination of ESI and spatially resolved
EELS allows to distinguish between different mineral phases
with the same stoichiometry.
In the past several investigations were carried out to
characterize different iron oxide minerals. Especially ELNES
of the O K-edge enables to distinguish between several iron
oxide phases. The quality of the results depends critically on
the attainable energy resolution, which limits the available
resolution of the ELNES structure. The new generation of
TEMs equipped with a field emission gun and a corrected
energy filter enable an energy resolution below 1 eV in
474
standard applications. With this tool it is possible to obtain
spectra in nanosized samples, that allow the identification of
phases through the comparison with results of O K-edge
ELNES investigations (Colliex et al., 1991; Paterson and
Krivanek, 1990) of known iron oxide phases.
The systems investigated in this work were artificial oxide-
coated iron nanoparticles (NPs) and natural iron-bearing
precipitates in granitic rocks. Both systems dealt with iron
oxide minerals, where the near edge structure of the O K-edge
and the Fe L-edge could be used to determine the oxidation
state, the iron oxide phase, and the quantitative chemical
composition (Golla-Schindler et al., 2003). The possible
candidates for the iron-oxide phases were FeO, a-Fe2O3
(hematite), g-Fe 2O 3 (maghemite), Fe3O 4 (magnetite),
a-FeO(OH) (goethite).
Nanoparticles have been increasingly studied due to their
variety of applications for many different purposes, especially
in life sciences. Although they are used in several medical
applications (like UV-sun screening, magnetic resonance
contrasting, drug delivery into the human body, etc.), there is
a lack of information about their shape and chemical
composition on the nanoscale. This may induce complications
in understanding their toxicological effects for wrong doses,
e.g. allergic reactions, inflammation and cell death. The NPs
investigated in this work were colloidal iron oxide NPs used for
medical purposes (Bomati-Miguel et al., 2005), consisting of
an oxidized rim and iron core. The major aim of the
investigation is to determine the iron oxide phase of the rim.
The second studied system consisted of iron-bearing
inclusions in red clouded feldpars in granitic rocks. The
mineralogy of the investigated granites mainly consisted of
feldspars (albite, plagioclase and K-feldspar), quartz and
biotite. The reddish coloration of the rock is supposed to be
due to the existence of small hematite inclusions imbedded in
(a)
(c)
1000
CCD counts x 103
800
600 O K-edge
400
200
0
500 550 600 650 700 750 800
Fig. 1. (a) Conventional TEM image of the iron based nanoparticles obtained with the JEOL 3010, (b) corresponding ELNES of the O K-edge and (c) exemplary
EELS spectrum used for the determination of the Fe/O ratio (d) reference EELS spectra for the most important oxides (Colliex et al., 1991).
U. Golla-Schindler et al. / Micron 37 (2006) 473477
the alkali feldspar (Boone, 1969), but up to now optical
petrography and microprobe analysis could not confirm this
assumption.
2. Experiments, results and discussions
The investigations were performed using two different
TEMs. A conventional JEOL 3010 operating at 297 kV
equipped with a LaB6 cathode, postcolumn Gatan Imaging
Filter and a 1 K slow-scan CCD camera (Gatan) and a LIBRA
200FE operating at 200 kV, equipped with a field emission
gun, a 4 K slow-scan CCD camera (Gatan) and a corrected 908
in-column Omega energy filter.
To optimize the signal to noise ratio (SNR) the experimental
conditions were chosen following the recommendations of
Berger and Kohl (1993). All images and spectra were gain and
dark count-corrected. Image processing and analysis routines
were written in the script language within Digital Micrograph
and the quantitative analysis of the resulting spectra were
performed using Gatan EL/P program packages and the
software IGOR (WaveMetrics).
For the TEM specimen preparation we used different
techniques for the two investigated systems. The TEM
specimens of the iron oxide particles were prepared by using
two techniques a dry preparation powdering the NPS on a lacey
carbon grid or a wet preparation ultrasonically dispersing the
NPs in toluol and drop them onto a lacey carbon grid. The TEM
specimens of the granitic rocks were obtained from thin section
by ion milling.
Fig. 1(a) shows a conventional TEM image of an iron oxide
nanoparticles agglomerate. Most of the particles show an
almost spherical shape and are around 1520 nm wide. Rim
and core are clearly distinguishable, because of the differences
(b) 100 O K-edge
CCD counts x 103
80
60
40
20
0 A B C D
500 520 540 560 580
Energy Loss (eV)
(d)
Fe L2,3-edge
Counts
Energy Loss (eV) Energy loss (eV)
 U. Golla-Schindler et al. / Micron 37 (2006) 473477 475

Fig. 2. (a) High resolution image obtained with the LIBRA 200FE (b) magnified cutout of the interesting region of (a) and (c) the corresponding linescans for the
determination of the lattice distances.

in electron-density. The image shows that not all particles are Fe/O 0.75G0.05. It has to be taken into account that only the
ideally round and that there is a fluctuation in size. endmember maghemite has a defect spinel structure with an
Fig. 1(b) shows the ELNES fingerprinting of one exemplary average of 21/3 vacant sites per unit cell and that there exist a
NP of Fig. 1(a). The spectrum was taken when the beam was solid solution between magnetite and maghemite with
focused with the brightness knob to a spot and positioned as chemical compositions in between magnetite and maghemite.
good as possible at the rim of the particle. HRTEM images were taken to compare the lattice distances
There are four distinct positions for the O K-edge where with expected values for the two possible phases. Fig. 2(a)
peaks can appear or be absent, labelled A, B, C, D (Colliex shows a HRTEM image of the core-shell nanoparticles, where
et al., 1991; Paterson and Krivanek, 1990). The ELNES of the the continuous lattice fringes demonstrate that the shell is
iron oxide nanoparticles shows peaks at all of these positions. topotactical with the core. Fig. 2(b) shows the corresponding
Considering that the phases FeO, a-Fe2O3 (hematite) do not linescans. The lattice distances agree with expected values for
show a peak in position C, both phases can be excluded by this the oxide phase of g-Fe2O3 (maghemite) shown in Table 1,
fingerprinting technique. Additional information can be which is consistent with the quantitative result of the Fe/O
obtained from a quantification of the Fe/O ratio, which is ratio.
different for each remaining possible oxide phase. For
Table 1
this calculation we used experimentally and theoretically Comparison of available lattice fringes for different iron oxides
determined k factors (Golla and Putnis, 2001) and obtained
Profile Experimental Hematite Maghematite Magnetite
Fe/OZ0.74G0.12. This value eliminates a-FeO(OH)
d (A) d (A) d (A) d (A)
(goethite), leaving the possibilities of g-Fe2O3 and Fe3O4
1 2.24G0.04 2.29 (006) 2.23 (213) 2.10 (004)
with the by Colliex et al. (1991) experimentally determinated
2 2.41G0.02 2.52 (K210) 2.41 (222) 2.42 (222)
Fe/O ratios for maghemite, Fe/O 0.68G0.04 and magnetite,
476 U. Golla-Schindler et al. / Micron 37 (2006) 473477

(c)
(a) 80
O K-edge

3
CCD counts x 10
60

40

20
A B C D
0
500 520 540 560 580
Energy loss (eV)

(d)
(b) 400

3
CCD counts x 10
350 O K-edge
300
Fe L2,3-edge
250
200
150
500 550 600 650 700 750
Energy Loss (eV)

Fig. 3. (a) Image of the granitic rock, (b) conventional TEM images of an iron oxide needle in the alkali feldspar matrix obtained with the JEOL 3010,
(c) corresponding ELNES of the O K-edge and (d) EELS spectrum used for the determination of the Fe/O ratio.

Fig. 3(a) shows an optical image of a granite from Finland,
where the pink coloured feldspar is clearly visible. TEM
samples of the feldspar examined with EDS showed small iron
oxide needles associated with pore-spaces in the feldspar. One
example of a typical shape is shown in Fig. 3(b). To determine
which iron phase is present the ELNES fingerprinting
technique of the O K-edge provided helpful information.
Fig. 3(c) show the ELNES of the O K-edge of the needle of
Fig. 3(b), where at positions A, B, and D there is a peak while at
position C the peak is absent. Due to this fact the iron oxide
phases g-Fe2O3 (maghemite) and Fe3O4 (magnetite) could be
excluded and only FeO, a-Fe2O3 (hematite) and a-FeO(OH)
(goethite) remained. The calculation of the Fe/O ratio resulted
in Fe/OZ0.63G0.11, which is in a good agreement with the
expected value for a-Fe2O3 (hematite)Z0.65G0.04. Selected
area electron diffraction analysis confirmed this result (Putnis
et al., 2005).
An ideal technique for the precise study of the chemical
shift and the ELNES peak positions at interfaces of different
mineral phases is spatially resolved EELS. To apply this
technique the interface of interest has to be aligned parallel to
the energy dispersive plane of the energy filter. This enables to

Fig. 4. (a) Conventional TEM image of an iron oxide needle in the alkali feldspar matrix with the LIBRA 200FE, (b) spectrum image of the O K-edge ELNES of the
selected area and (c) the corresponding linescans on the iron oxide needle and in the feldspar matrix.
 U. Golla-Schindler et al. / Micron 37 (2006) 473477
study accurately the changes of the ELNES at the interfaces,
because a whole series of spectra, a so called spectrum image,
is taken simultaneously. For this study we select a specimen
area, where an iron oxide needle is embedded in the alkali
feldspar matrix. Fig. 4(a) shows the selected area, Fig. 4(b) the
correlated spectrum image and Fig. 4(c) the corresponding
extracted EELS spectra (from the spectrum image on top of the
alkali feldspar and the iron oxide needle). The changes of the
ELNES in the spectrum image as well as in the extracted EELS
spectra are clearly visible.
3. Conclusion
The studies were performed with two different TEMs, the
JEOL 3010 equipped with a LaB6 cathode, a post column
energy filter and the newly designed LIBRA 200 FE equipped
with a field emission gun and an incolumn energy filter. The
comparison of the same investigations done with both
instruments showed consistent results.
In the artificial iron oxide NPs the evaluation of the shell
by fingerprinting the O K-edge identified two possible
candidates for the oxide phase, magnetite and maghemite.
The relative chemical composition Fe/OZ0.74G0.12 was
not precise enough in this case, to distinguish between
hematite and maghemite, but HRTEM studies showed that
the lattice distances are consistent with maghemite.
In the examination of the natural iron-oxide precipitates
from the red feldspars, the ELNES O K-edge fingerprinting
pointed to hematite as the most probable candidate for the iron
oxide phase. Chemical composition and SAED confirmed the
identification.
Spatially resolved EELS of the hematite needle in the
feldspar matrix showed accurate correlation between the
specimen topography and the ELNES features. In the image
spectra the energy shift and changes of the O K-edge ELNES
between the alkalifeldspar matrix and the iron oxide needle
can be clearly seen.
Concluding we have shown with two exemplary studies that
on a nanoscopic scale, energy filtering TEM is a powerful
technique to distinguish between mineral phases by analysing
the chemical composition and the ELNES to differentiate
between minerals with the same stoichiometry.
477
Acknowledgements
The authors acknowledge the support from the DFG and an
Alexander von Humboldt Fellowship and S. Veintmillas-
Verdaguer for helpful discussion.
References
Berger, A., Kohl, H., 1993. Optimum imaging parameters for element mapping
in an energy filtering transmission electron microscope. Optik 92, 175193.
Bomati-Miguel, O., Morales, M.P., Tartaj, P., Ruiz-Cabello, J., Bonville, P.,
Santos, M., Zhao, X., Veintmillas-Verdaguer, S., 2005. Fe-based
nanoparticulate metallic alloys as contrast agents for magnetic resonance
imaging. Biomaterials 26, 56955703.
Boone, G.M., 1969. Origin of clouded red feldspars: petrologic contrasts in a
granitic porphyry intrusion. Am. J. Sci. 267, 633668.
Colliex, C., Manoubi, T., Ortiz, C., 1991. Electron -energy-loss-spectroscopy
near-edge fine structures in tze iron -oxygen system. Phys. Rev. B 44,
1140211411.
Egerton, R.F., 1996. Electron Energy-Loss Spectroscopy in the Electron
Microscope. Plenum Press, New York.
Golla, U., Putnis, A., 2001. Valence state mapping and quantitative electron
spectroscopic imaging of exsolution in titanohematite by energy-filtered
TEM. Phys. Chem. Miner. 28, 119129.
Golla-Schindler, U., Benner, G., Putnis, A., 2003. Laterally resolved EELS for
ELNES mapping of the Fe L2,3 - and O K-edge. Ultramicroscopy 96, 573
582.
Kimoto, K., Sekiguchi, T., Aoyama, T., 1997. Element mapping using a field
emission transmission electron microscope with an imaging filter.
J. Electron Microsc. 46, 369374.
Lazar, S., Botton, G.A., Wu, M.-Y., Tichelaar, F.D., Zandbergen, H.W., 2003.
Materials science applications of HREELS in near edge structure analysis
and low-energy loss spectroscopy. Ultramicroscopy 96, 535546.
Mitterbauer, C., Kothleitner, G., Grogger, W., Zandbergen, H., Freitag, B.,
Tiemeijer, P., Hofer, F., 2003. Electron energy-loss near-edge structures of
3d transition metal oxides recorded at high-energy resolution. Ultramicro-
scopy 96, 469480.
Paterson, J.H., Krivanek, O.L., 1990. ELNES of 3d transitionmetal oxides.
Ultramicroscopy 32, 319325.
Putnis, A., Hinrichs, R., Putnis, C.V., Golla-Schindler, U., Collins, L.G., 2005.
Hematite in porous red-clouded feldspars: evidence of large-scale crustal
fluid-rock interaction. Submittede to Lithos.
Reimer, L., 1995. Energy-Filtering Transmission Electron Microscopy.
Springer, Berlin.
Reimer, L., Fromm, I., Hirsch, P., Plate, U., Rennekamp, R., 1992.
Combination of EELS modes and electron spectroscopic imaging and
diffraction in an energyfiltering electron microscope. Ultramicroscopy
46, 335347.
Walther, T., Mader, W., 1999. Application of spatially resolved electron
energy-loss spectroscopy to the quantitative analysis of semiconducting
layer structures. Inst Phys. Conf. Ser. 164, 121128.