Study of glasses containing Zn- and Fe- contaminated Electric Arc Furnace
Dust by means of μ-XRF mapping and μ-XAFS
F. Pinakidou, M. Katsikini, E. C. Paloura, P. Kavouras, Th. Kehagias, Ph. Komninou, Th. Karakostas
Aristotle Univ. of Thessaloniki, Dept. of Physics, GR54124 Thessaloniki, Greece.
A. Erko
BESSY GmbH, Albert Einstein Str. 15, 12489 Berlin, Germany.
Abstract
We apply x-ray fluorescence mapping (XRF), micro- and conventional x-ray absorption
fine structure (μ-XAFS and XAFS) spectroscopies in order to study the microstructure
around Fe and Zn in as-casted and annealed vitrified samples that contain Electric Arc
Furnace dust from metal processing industries. It is demonstrated that annealing does
not induce any changes in the distribution of either Fe or Zn, which are homogeneously
distributed in both the as-casted and annealed samples. The μ-XAFS spectra recorded at
the Fe-K and Zn-K edges, at different positions of the annealed and as-casted samples,
reveal that Fe forms both FeO6 and FeO4 polyhedra, i.e. acts as an intermediate oxide.
On the other hand, Zn occupies tetrahedral sites in both the as-casted and annealed
sample, i.e. the environment around Zn is not affected by the annealing procedure.
Keywords: X-Ray Fluorescence Spectroscopy, EXAFS, NEXAFS, μ-EXAFS, Electric Arc Furnace Dust.
1. Introduction
Electric arc furnace dust (EAFD) is one of the largest solid waste streams produced by steel mills, and is
classified as a waste under the Resource Conservation and Recovery Act (RCRA) of the U.S. Environmental
Protection Agency (EPA). Steel production by electric-arc furnace (EAF) technology has been of increasing
importance over the past 20 years, and it is expected that, in the coming years, it will dominate the steel
production. During the EAF production of steel, about 15 - 20 kg of dust is formed per tone of steel. This dust
contains heavy metals and thus is considered as a toxic waste. These metals are found both as free oxides (e.g.
PbO, ZnO) as well as in the form of composite structures with iron oxides (e.g. ZnFe2O4). More specifically,
EAF dusts contain iron, zinc, calcium and silicon in the form of simple or mixed oxides, as well as copper,
manganese, chromium, cadmium and lead which either originate from the scrap iron raw material or are
introduced as additives. Successful recycling of the valuable metals (iron, zinc, and lead) reduces the disposal
problems and results in resource conservation. Hydrometallurgical or pyrometallurgical processes can recover
only a part of heavy metals (such as Zn) from the EAFD and hence it is of major importance to develop an
additional process that stabilizes metallic Zn and the other heavy metals found in the residues of Zn recovery
processes. One of the most promising methods for safe disposal of the EAF dust is vitrification (Pisciella et al.,
2001, Pellino et al., 2002) that leads to the production of chemically stable materials, vitreous or glass-ceramic,
that can be safely disposed or used for construction and decorative applications (Barbieri et al., 2000, Sheng,
2002).
The aim of this work is to determine the distribution, the local environment and the oxidation state of Fe and Zn
atoms in the glass matrix of vitrified EAFD-rich industrial wastes. It is known that the structural role of both Fe
and Zn depends on their valence state (Pinakidou et al., 2005, Lusvardi et al., 2002): Fe+2 acts as a glass modifier
and participates in the formation of FeO6 polyhedra while Fe+3 and Zn+2 are intermediate oxides, i.e. can also act
as glass network formers by constituting FeO4 and ZnO4 tetrahedra. Therefore it is of great importance to study

E. C. Paloura, paloura@auth.gr, tel.:+302310998036, fax: +302310998036
the structural role of both Fe and Zn oxides, since the structural integrity of the glass matrix depends strongly on
the type of polyhedra that the Fe and Zn atoms form. The coordination and the oxidation state of Fe and Zn in
solidified-stabilized EAFD samples were determined using X-ray absorption fine structure (XAFS)
measurements at the Fe-K and Zn-K-edges. The XAFS spectrum, which shows the dependence of the x-ray
absorption coefficient on the impinging photon’s energy above the absorption edge of a specific element,
consists of two regions: the extended X-ray Absorption Fine Structure (EXAFS) region and Near-edge Extended
X-ray Absorption Fine Structure (NEXAFS) region. The former reflects the microstructure (as that is determined
by the nearest neighbor distances and coordination numbers) of the absorbing atom while the latter reflects the
density of empty states of the absorbing atom and it is affected by the symmetry, the local coordination geometry
and the valence of the absorbing atom, which can be derived by proper analysis of the NEXAFS spectra. The
distribution of Fe and Zn, in the as-casted and the annealed samples, was determined using X-ray fluorescence
(XRF) mapping while the combination of XRF with micro-XAFS (μ-XAFS) permitted the identification of local
changes in the microstructure around Fe and Zn.
2. Experimental
The under study samples are vitrified products of EAF dust which mainly consists of zinc (ZnO at 34.9% w.t.)
and ferric oxides (in the form of ZnFe2O4 at 33.9%w.t.). The vitrification procedure involves mixing of the EAF
dust with SiO2, Na2CO3 and CaCO3 powders and co-melting at 1400C for 2h, followed by quenching (Kavouras
et al, 2003a, Kavouras et al, 2003b). The under study sample consists of 20% EAFD, 55% SiO2, 15% CaO and
10% Na2O and was subsequently heated at a temperature near glass transition temperature (T g=583.8oC), namely
20C higher than Tg, for residual stress relaxation. Annealing in two stages was applied to all starting products,
i.e. thermal treatment process that was composed of two isothermal stages. In the first isothermal stage,
nucleation takes place and in the second crystal growth occurs. The duration of these two stages were fixed at 20
min for the nucleation and 30 min for the crystal growth process. The nucleation stage was placed 200C below
exothermic peaks maxima, while crystal growth stage was placed a few degrees above them. Previous work
conducted in the under study sample disclosed that upon annealing, surface crystallization of wollastonite
(CaSiO3) initiates from the edges towards the center (Kavouras et al., 2005). A powder hematite sample (αFe2O3) is used as reference for the XAFS measurements.
The XRF, EXAFS and μ-XAFS measurements were conducted at the KMC2 beamline at the synchrotron
radiation (SR) facility BESSY-II in Berlin. The beamline is equipped with a double-crystal monochromator and
capillary optics that reduce the beam diameter to 5μm. The XRF maps were recorded using excitation photons of
7200eV, i.e. higher than the Fe-K absorption. The angle of incidence of the SR beam was 45 o and the energy
dispersive (Röntec) fluorescence detector was positioned normal to the beam. The Fe-K and Zn-K edge
fluorescence photons emitted from the sample were discriminated by setting electronically the energy range of
interest from 6097 to 7181eV and 8037 to 9092eV respectively. The μ-XAFS spectra were recorded at the Fe-K
and Zn-K edges, in the fluorescence yield mode, at two spots: one at the edge (position E) and the other at the
centre (position C) of the annealed sample. The spectrum of the Fe 2O3 reference sample was recorded in the
transmission mode using ionization chambers. The EXAFS spectra for the as-casted and annealed samples were
recorded at the Fe-K and Zn-K edges in the fluorescence yield mode using a Si-PIN photodiode. Energy
calibration was performed using the absorption edge position of the spectrum of a Fe foil.
3. Results and discussion.
3.1 XRF results
XRF spectroscopy was used for the qualitative compositional analysis of the under study samples. The XRF
spectra, shown in Fig.1, were recorded on several different spots on the as-casted and annealed samples with an
excitation energy equal to 9400eV, i.e. high enough to excite both Fe and Zn. As shown in the spectra, the under
study samples contain Fe, Zn, Ca and Mn, while the presence of Si and Na, that also exist in the glass, could not
be detected due to the high excitation energy and the resulting low absorption cross section for low Z elements.
The concentration of either Fe or Zn does not vary among the various measured spots and is the same in both the
as-casted and annealed samples.
Figure 1: Fluorescence spectra normalized at the Zn Kα peak recorded at the edge and the centre of the annealed sample
and at random positions of the as-casted and annealed sample.
In order to detect with higher accuracy any local inhomogeneities in the distribution of Zn and Fe, we resorted to
XRF mapping. The XRF map recorded from the annealed sample, shown in Fig.2, reveals that the distribution of
Fe and Zn is homogeneous and formation of Fe and Zn clusters has been avoided (as shown in Fig.2 the
variation in the concentration of either Fe or Zn in the annealed sample is less than 15%). This observation,
along with the EXAFS results discussed in the following, indicates that both Fe and Zn have been successfully
incorporated into the vitreous matrix.
Figure 2: 7500x15μm XRF maps of the distribution of Fe and Zn in the annealed sample. The regions depicted as E and C,
refer to the positions at the edge and at the centre of the annealed sample, respectively.
3.2 EXAFS and μ-EXAFS results
The XAFS and μ-EXAFS spectra were recorded at the Fe- and Zn-K-edges at room temperature. In order to
avoid self-absorption effects the EXAFS spectra were recorded at grazing incidence (θ=85o to the sample
surface). However, due to geometrical limitations set by the capillary optics the μ-EXAFS spectra were recorded
at θ=45o and were corrected for self-absorption phenomena (Tröger, 1992). Prior to analysis, the EXAFS and μEXAFS spectra were subjected to subtraction of the atomic absorption using the AUTOBK program (Mustre de
Leon, 1991) and corrected for monochromator-induced energy shifts using the spectrum of a Fe foil. The model
for the fitting was constructed using the FEFF8.0 (Mustre de Leon, 1991) program. The Fourier transforms (FT)
of the k3 - weighted EXAFS spectra at the Fe- and Zn-K edges are shown in Fig. 3(a) and (b), respectively. The
FT was calculated in the k-range 2.5-9.3 Å-1 for both the Fe- and Zn-K edges. As shown in the figure, the FTs of
the under-study samples have well-resolved structure up to a distance of about 3.5Å from the absorbing atom.
Therefore, mid-range order exists, possibly in the form of nanocrystallites, around both the Fe and Zn atoms.
Figure 3: Fourier transforms of the EXAFS spectra recorded at the a) Fe-K and b) Zn-K edge, respectively. The raw data
and the fitting are shown in thin and thick solid lines, respectively.
The EXAFS spectra recorded at the Fe-K-edge were fitted using a mixed model according to which X% of the
Fe atoms occupy octahedral sites that belong to the ZnFe2O3 phase, while (1-X)% form tetrahedra, and
participate in the formation of the glass. In both the as-casted and annealed samples, the fitting procedure was
performed using four shells: the first nearest neighbor (nn) shell consists of oxygen atoms that are either
octahedrally or tetrahedrally coordinated to the Fe atom. The second, third and forth nn shells comprise of Fe, Zn
and O atoms, according to the ZnFe2O4 crystalline model. In the ZnFe2O4 crystalline phase, each Zn atom is
tetrahedrally coordinated with 4 oxygen atoms, that link to 12 FeO6 octahedra. The Zn-K EXAFS spectra were
fitted using the ZnFe2O4 model and the fitting procedure was performed assuming four shells: in the 1 st nn Zn is
tetrahedrally coordinated with four oxygen atoms at a distance 2.02Å, while the 2nd, 3rd and 4th comprise of Fe, O
and Zn atoms, respectively. The analysis of the spectra from the hematite reference sample reveals the expected
trigonal distortion of the FeO 6 octahedron which causes a splitting of the first nearest neighboring shell (three
oxygen atoms at 1.93Å and three oxygen atoms at 2.09Å, respectively) (R. W. G. Wyckoff, 1964). The results of
the EXAFS analysis are listed in Table I.
Table I: EXAFS results for the Fe-K and Zn-K edges. The symbols N and R indicate the coordination
numbers and interatomic distances in the first nearest-neighbor shell. The asterisk indicates
parameters that were kept fixed during the analysis.
Fe-K edge
Samples
As-casted
Annealed at 850oC
Annealed at the edge
Annealed at the center
Zn-K edge
Percentage (%)
of FeOx
RFe-O
(0.01Å)
DW (X10-3)
(Å-2)
FeO6=51 (4%)
FeO4=49
FeO6=47 (6%)
FeO4=53
FeO6=69 (10%)
FeO4=31
FeO6=57 (7%)
FeO4=43
1.86
1.95
1.86*
1.92
1.86*
1.92
1.86*
1.94
2
3.1
2*
3.1*
2*
3.1*
2*
3.1*
N (O)
(10%)
RZn-o
(0.01Å)
DW (X10-3)
(Å-2)
4.3
1.94
4.4
4.1
1.94
3.7
4.2
1.94
5.4
4.1
1.94
3.9
As shown in Table I, in the as-casted sample, about 50% of the Fe is tetrahedrally coordinated with oxygen
atoms while the rest 50% is octahedrally coordinated with oxygen atoms. The corresponding Fe-O distances,
which are characteristic of the tetrahedral and octahedral coordinations, are 1.86 and 1.93Å, respectively.
Hence, Fe participates in the formation of the vitreous matrix by constituting FeO 4 and FeO6 polyhedra, i.e. the
intermediate role of the Fe oxide is revealed. As far as the 2nd and 3rd nn shells are concerned, the Fe and Zn
atoms occupy sites at distances 2.99 and 3.35Å, from the central Fe atom, respectively. This finding indicates
that annealing does not cause any major difference in the microstructure around Fe.
The EXAFS analysis at the Zn-K edge reveals that the bonding environment around the Zn atom is not affected
by annealing. In both the as-casted and annealed samples, the Zn atom is coordinated with 4 oxygen atoms at a
distance 1.94Å and thus acts as a glass former. The 2nd nn shell is found to consist of approximately 12 Fe atoms
at a distance of 3.35Å, i.e. in agreement with the crystalline model of ZnFe 2O4. Finally, the Zn-O distance in the
3rd nn shell ranges from 3.53 to 3.46Å (the coordination number is not discussed since the determined changes
are within the 30% error bar).
In order to investigate possible changes in the microstructure due to local compositional variations as well as due
to differences in the cooling rate between the center and at the edge of the samples, we resorted to μ-EXAFS
spectra that we recorded at the Fe-K and Zn-K edges, from two spots of the sample surface: spot E corresponds
to the edge of the sample annealed at 850oC while spot C corresponds to the center of the same sample (hereafter
denoted as E and C, respectively). The Fourier transforms (FTs) of the k 3 - weighted μ-EXAFS spectra recorded
at the Fe- and Zn-K edges are shown in Fig. 4(a) and (b), respectively. The FTs were calculated in the k-range
2.7-9.3 Å-1 and 2.5-9.3 Å-1 for the Fe- and Zn-K edges respectively. In accordance with the EXAFS analysis, the
existence of mid-range order around both the Fe and Zn atoms is again evident in the FTs.
Figure.4: Fourier Transforms of the a) Fe-K and b) Zn-K edge EXAFS spectra recorded at the edge and at the centre of the
annealed sample. The raw data and the fitting are shown in thin and thick line, respectively.
The Fe-K μ-EXAFS spectra were fitted using the same mixed model used for the analysis of the EXAFS spectra.
The Zn-K μ-EXAFS spectra were fitted using the ZnFe3O4 model and the fitting procedure was performed
assuming three shells: in the 1st nn Zn participates in the formation of ZnO4 at a distance 2.02Å, while the 2nd
and 3rd are comprised of Fe and Zn atoms.
The analysis of the μ-EXAFS spectra from the annealed sample (Table I) reveals that the percentage of the Fe
atoms that constitute octahedra is, within the error bar, equal at the center and at the edge of the sample. More
specifically, 55% of the Fe atoms form octahedra at the edges of the sample, while 65% form octahedra at the
center of the sample. The Fe-O bond length in the FeO6 polyhedron is equal to 1.93Å whilst the respective value
in the FeO4 tetrahedron is 1.86Å. Therefore, the intermediate role of Fe is verified by the μ-EXAFS results: Fe
participates in the formation of the vitreous matrix by forming FeO 4 tetrahedra and acts as a glass modifier when
bonded in FeO6 octahedra. As far as the 2nd and 3rd nn shell are concerned, we detect a small distortion around Fe
at the center of the sample, i.e. the Fe and Zn atoms appear to be in shorter distances than those at the edge of the
sample (2.92 and 3.16Å at the center and 2.99 and 3.60Å at the edge, respectively. Thus, Fe partially occupies
octahedral sites that belong to ZnFe2O4 nanocrystallites, which are not resolved in the XRF maps, due to the
limited spatial resolution of the experimental setup.
On the other hand, the μ-EXAFS analysis at the Zn-K edge reveals no alteration in the bonding environment of
the Zn atom. In both regions, the Zn atom participates in the formation of ZnO 4 tetrahedra, i.e. Zn is coordinated
with 4 oxygen atoms at a distance 1.94Å and thus acts as a glass former. In the 2 nd nn shell, the Zn-Fe distance is
equal to 3.35Å at both the edge and the center of the sample, indicating the invariability of the Zn microstructure
in different regions of the sample.
3.3 NEXAFS results
The Fe-K spectra of the under study samples and the reference Fe 2O3 compound are shown in Fig. 5(a). The
NEXAFS spectra of Fe-containing compounds are characterized by a pre-edge peak that is attributed to 1s  3d
transitions (Callas et al., 1983). In a centrosymmetric environment (e.g. octahedral) the intensity of the pre-edge
peak is weak due to the quadrupole character of the transition. Contrary to that, in a non-centrosymmetric
environment (e.g. tetrahedral) the pre-edge peak gains intensity due to the d-p mixing of the final state that gives
dipole character to the transition. Thus, the intensity of the pre-edge peak is inversely proportional to the
coordination number of the absorbing atom due to the loss of inversion symmetry of the iron site (Wu 2001 et
al., Randall et al., 1995). Not only the intensity but also the shape of the pre-edge peak depends on the local
coordination around Fe. According to Westre (1997), fitting of the pre-edge peak with one Lorentzian is a
fingerprint of the tetrahedral coordination of the Fe atoms. Contrary to that, two Lorentzians are necessary to fit
the pre-edge peak when Fe occupies octahedral sites. Finally, it should be mentioned that the position of the preedge peaks is a measure of the portion of various oxidations states of Fe in a mixed valence sample, i.e. the
Fe+3/ΣFe ratio, and depends on the bonding environment of the Fe atom (Bajt et al., 1994).
Figure 5: The a) Fe-K edge and b) Zn-K edge NEXAFS spectra from the as-casted and annealed sample. The spectra from
the annealed sample were recorded at the edge, at the centre and at a random position. The spectrum of the reference Fe 2O3
is also included in Fig. 5(a).
Prior to analysis the spectra were normalized for the transmission function of the monochromator and were
subjected to linear background subtraction (by fitting the spectrum far before the absorption edge) and
normalization to the edge jump. The characteristics of the Fe-K and Zn-K NEXAFS spectra were determined via
fitting using a sigmoidal (Boltzman function) to simulate transitions to the continuum and Gaussians to simulate
transitions to bound states (Stöhr, 1992). The pre-edge peak in the Fe-K edge NEXAFS spectra was fitted using
Lorentzians.
The presence of the pre-edge peak in the NEXAFS spectra of all the under study samples indicates that the
polyhedron around Fe is either a tetrahedron and/or an asymmetric octahedron. In the reference hematite (Fe 2O3)
sample, where Fe belongs to a distorted octahedral environment, two Lorentzians (centered at 7113.7 and
7116.7eV, respectively) were necessary to fit the pre-edge region. Contrary to that the pre-edge peak in the
spectra of the under-study samples was fitted using one Lorentzian. It should be pointed out that, in the case of a
mixed bonding environment (both octahedral and tetrahedral), due to the weak contribution of the
centrosymmetric environment, one Lorentzian would be adequate to fit the pre-edge peak. Therefore, in the case
of the under study samples, the use of one Lorentzian is an indication that the Fe either belongs to a tetrahedral
environment, a distorted octahedral or a mixed bonding environment. However, if we take into account the
EXAFS results, we can propose that the use of one Lorentzian is an indication that the Fe atom is coordinated
both octahedrally and tetrahedrally with oxygen atoms. Finally, the energy position of the pre-edge peak
(centered at 7113.8eV) is the same in all the under study samples indicating that the Fe +3/ΣFe ratio remains
constant in all vitrified products, i.e. is unaffected by annealing and is invariant in different positions of the
annealed sample.
The energy position of the absorption edge (Eabs), in the under study samples is equal to that in hematite
(7122.8eV). According to Salem (1978), the E abs is a fingerprint of the valence of the absorbing atom and can be
affected by modifications in the ionicity of the Fe-O bond and the symmetry (bondlength and coordination)
around the Fe atom. Since we did not detect any energy shifts it can be proposed that any alteration in the
parameters mentioned above is very weak and smaller than the resolution of the NEXAFS spectra ( 0.5eV).
The NEXAFS spectra recorded at the Zn-K edges are shown in Fig 5(b). The spectra recorded from the as-casted
sample, as well as at the edge and at the center have structure at energies above the absorption edge (E abs),
which indicates that the material is nanocrysytalline around the Zn atom. The weaker structure in the spectrum
recorded at a random position of the annealed sample could indicate that locally the sample is amorphous due to the
inhomogeneous distribution of the ZnFe2O4 nanocrystallites.
4. Conclusions
XRF mapping of vitrified EAF dust revealed that the distribution of Fe and Zn is homogeneous in both the ascasted and the annealed samples. According to the EXAFS results the environment around the Fe atom is mixed,
i.e. Fe occupies both tetrahedral and octahedral sites and thus acts both as a glass former and as a modifier. The
Fe-O distance is found equal to 1.93Å and 1.86 Å for tetrahedrally and octahedrally coordinated Fe, respectively.
Annealing does not affect the concentration of FeO6 and thus does not alter the structural role of Fe. Therefore,
Fe participates in the formation of the glassy network acting as an intermediate oxide. It forms FeO4 tetrahedra
that link to the vitreous matrix, while occupies octahedral sites in ZnFeO 4 nanocrystallites. On the other hand, Zn
occupies tetrahedral sites in both the as-casted and annealed sample, i.e. the environment around Zn is not affected by
the annealing procedure. Finally, analysis of μ-EXAFS recorded from different spots of the under study samples
reveals that there are no detectable local changes in the microstructure of either Fe or Zn.
References
Bajt S., S. R. Ruton and J. S. Delabey, 1994, Geoch. et Cosmoch. Acta 58, 5209.
Barbieri, L., A. Corradi, I. Lancellotti, 2000, J. Eur. Ceram. Soc. 20, 2477.
Callas G. and J. Paetiau, 1983 Solid State Commun. 48, 625.
Kavouras P., G. Kaimakamis, T. A. Ioannidis, T. Kehagias, P. Komninou, S. Kokkou, E. Pavlidou, I. Antonopoulos, M.
Sofoniou, A. Zouboulis, C. P. Hadjiantoniou, G. Nouet, A. Prakouras and T. Karakostas, 2003a, Waste Management 23, 361.
Kavouras P., P. Komninou, K. Chrissafis, G. Kaimakamis, S. Kokkou, K. Paraskevopoulos, and T. Karakostas, 2003b, J. Eur.
Ceram. Soc. 23, 1305.
Kavouras P., I. Tsilika, Th. Kehagias, K. Chrissafis, G. Kaimakamis, D. Papadopoulos, S. Kokkou, Ph. Komninou and Th.
Karakostas, ibid 2005, in press.
Lusvardi G., Malavasi G., Menabue L.and M. C. Menziani, 2002, J. Phys. Chem. B 106, 9753.
Mustre de Leon J., J. J. Rehr, S. I. Zabinsky, R. C. Albers, 1991, Phys. Rev. B 44, 4146.
Pellino M., A. Karamanov, P. Pisciella, S. Criscussi and D. Zonetti, 2002, Waste Management 22, 945.
Pinakidou F., M. Katsikini, E.C. Paloura, P. Kavouras, Ph. Komninou, T. Karakostas and A. Erko, 2005, Phys. Scripta (in
press).
Pisciella P., S. Crisucci, A. Karamanov and M. Pelino, 2001, Waste Management 21, 1.
Randall C. R., L. J. Shu, Y. M. Chiou, K.S. Hagen, M. Ito, N. Kitajima, R. J. Lachicotte, Y. Zang and L. Que, 1995, Inorg.
Chem. 34, 1036.
Salem S. I., C. N. Chang, P. L. Lee and V. Severson, 1978, J. Phys. C: Solid State Phys. 11, 4085.
Sheng, J., 2002, Fuel 81, 253.
Stöhr J., 1992, NEXAFS Spectroscopy. Springer - Verlag, Berlin.
Tröger L., D. Arvanintis, K. Baberschke, H. Michaelis, U. Grimm and E. Zschech, 1992, Phys. Rev. B 46, 3283.
Westre T., P. Kennepohl, J. G. DeWitt, B. Hedman, K. O. Hodgson and E. I. Solomon, 1997, J. Am. Chem. Soc. 119, 6297.
Wu Z., D. C. Xian, C. R. Natoli, A. Marcelli, E. Paris and A. Mottana, 2001, Apl. Phys. Lett. 79, 1918.
Wyckoff R. W. G., 1964, Crystal Structures, vol. 2. John Wiley & Sons, New York, London, Sydney.
Acknowledgements
The experimental work was realized with financial support from the HPRI-1999-CT-00028 program. One of the
authors, F. Pinakidou, acknowledges financial support from the HERAKLEITOS program.