Mechanochemical reduction of magnesium ferrite

Article abstract of DOI:10.1021/jp020270z

The changes in magnesium ferrite, MgFe₂O₄, due to high-energy milling in a stainless steel vial have been investigated by Moì?ssbauer spectroscopy, X-ray diffraction, electron microscopy, and thermal analysis. The milling process reduces the average crystallite size of MgFe₂O₄ to the nanometer range. Prolonged mechanical treatment leads to the chemical reduction of MgFe₂O₄ and to the formation of a solid solution of FeO and MgO. In addition to the solid solution, metallic iron has been found as a byproduct. The fraction of the reaction products increases with increasing milling time. The range of thermal stability of the metastable milled reduction products has been determined by studying their response to changes in temperature.

Full text of DOI:10.1021/jp020270z

6672
J. Phys. Chem. B 2002, 106, 6672-6678
Mechanochemical Reduction of Magnesium Ferrite
Vladim´ır Sˇepela´k†
Institute of Geotechnics, SloVak Academy of Sciences, WatsonoVa 45, SK-04353 Kosˇice, SloVakia
Marcus Menzel and Klaus Dieter Becker*
Institute of Physical and Theoretical Chemistry, Technical UniVersity of Braunschweig,
Hans-Sommer-Str. 10, D-38106 Braunschweig, Germany
Frank Krumeich
Laboratory of Inorganic Chemistry, Swiss Federal Institute of Technology Zurich,
Ho¨nggerberg, CH-8093 Zurich, Switzerland
ReceiVed: January 29, 2002
The changes in magnesium ferrite, MgFe2O4, due to high-energy milling in a stainless steel vial have been
investigated by Mo¨ssbauer spectroscopy, X-ray diffraction, electron microscopy, and thermal analysis. The
milling process reduces the average crystallite size of MgFe2O4 to the nanometer range. Prolonged mechanical
treatment leads to the chemical reduction of MgFe2O4 and to the formation of a solid solution of FeO and
MgO. In addition to the solid solution, metallic iron has been found as a byproduct. The fraction of the
reaction products increases with increasing milling time. The range of thermal stability of the metastable
milled reduction products has been determined by studying their response to changes in temperature.
Introduction
has been reported in only a few papers.4,20-26 There have been
reports on the reduction of Fe3O4 during high-energy
milling.4,20-22 NiFe2O4 and CuFe2O4 appear to be the only
ternary spinels that have been investigated.24-26 The high-energy
milling of NiFe2O4 in a steel vial using steel balls was found to
induce a mechanochemical reduction process of the material
leading to the formation of a disordered solid solution of FeO
and NiO with wu¨stite structure.24,25 Goya et al.26 followed the
phase evolution of CuFe2O4 during high-energy milling. They
showed that the final product of the mechanical treatment of
copper ferrite is a two-phase mixture consisting of Fe3O4 and a
spinel solid solution (CuxFe3-xO4). This indicates that the high-
energy milling of the ferrite spinel generates a complex series
of solid-state transformations, including mechanochemical
decomposition and reduction.
Mechanically activated processes continue to attract consider-
able attention in the field of solid-state chemistry, physics, and
materials science. Mechanical treatment (mechanical activation,
high-energy milling) of crystalline compounds has been em-
ployed to prepare solids far from equilibrium, such as amorphous
alloys, nanostructured materials, and quasicrystalline phases.1
Reduction processes induced by mechanical treatment rep-
resent an important class of solid-state reactions. Many examples
have been reported where reduction of simple binary oxides by
a more reactive metal occurs during milling.2-10 In ball-milling
experiments involving R-Fe2O3 in air, reduction of R-Fe2O3 to
11-14
Fe3O4
and subsequently, to the Fe1-xO phase15 has been
observed to occur in a closed stainless steel container after
prolonged milling. Depending upon the value of the adiabatic
temperature of the redox reaction and milling conditions, the
reaction may occur either in a steady-state manner during milling
or by unstable thermal combustion of the reactants.16 In addition
to simple oxides, there have also been reports on the reduction
of chlorides during high-energy milling.17
While the mechanically induced redox reactions involving
simple binary oxides are discussed extensively in the literature,
the mechanochemical reduction of higher complex oxides is
treated only to a limited extent. In this context, the mechanically
induced reduction of ilmenite (FeTiO3) and scheelite (CaWO4)
may be mentioned.18,19 To our best knowledge, the application
of high-energy milling to the chemical reduction of oxide spinels
In this work, we present a detailed study of the phase
evolution of MgFe2O4 during high-energy milling in a stainless
steel vial. Magnesium ferrite is a soft magnetic material,27 which
finds a number of applications in heterogeneous catalysis and
adsorption, in sensor and in magnetic technologies. To empha-
size the site occupancy on the atomic level, MgFe2O4 may be
written as (Mg1-λFeλ)[MgλFe2-λ]O4, where parentheses and
square brackets denote cation sites of tetrahedral (A) and
octahedral [B] coordination, respectively. λ represents the so-
called degree of inversion (defined as the fraction of the (A)
sites occupied by Fe3+ cations) characterizing the distribution
of cations among the two nonequivalent sites. In recent
work,28-30 we have demonstrated that high-energy milling of
MgFe2O4 in ceramic-covered vials using ceramic R-Al2O3 balls
does not lead to chemical reduction: no other phases except
the nanoscale MgFe2O4 were detected in the milled powders.
It has been found, however, that these MgFe2O4 nanoparticles
* Author to whom correspondence should be addressed. Fax: +49-531-
3917305. E-mail: k-d.becker@tu-bs.de.
† Present address: Institute of Physical and Theoretical Chemistry,
Technical University of Braunschweig, Hans-Sommer-Str. 10, D-38106
Braunschweig, Germany. Fax: +49-531-3917305. E-mail: v.sepelak@
tu-bs.de.
10.1021/jp020270z CCC: $22.00 © 2002 American Chemical Society
Published on Web 06/06/2002

Mechanochemical Reduction of Magnesium Ferrite
J. Phys. Chem. B, Vol. 106, No. 26, 2002 6673
are structurally and magnetically disordered due to mechanically
induced changes in the cation distribution and spin canting.
Experimental Section
Polycrystalline magnesium ferrite was prepared by conven-
tional solid-state reaction (further referred to as the nonactivated
material). Stoichiometric mixtures of the powdered reactants
R-Fe2O3 and MgO (Merck, Darmstadt) were homogenized in a
ball mill. The powdered mixtures were pressed into tablets under
a pressure of 30 MPa in a steel die in order to obtain a high
degree of compaction of the reactants. The tablets were 20 mm
in diameter and 4 mm thick. They were prefired twice at 1200
K for 24 h, ground, pressed, and finally sintered at 1300 K for
24 h (all processes in air). The single-phase nature of the as-
prepared samples was confirmed by X-ray diffraction (XRD)
and Mo¨ssbauer spectroscopy.
The high-energy milling process (dry milling) was carried
out in an AGO 2 planetary ball mill (Institute of Solid State
Chemistry and Mechanochemistry, Novosibirsk) at room tem-
perature. The nonactivated sample (10 g) was ground for various
times in a closed, air-filled vial (150 cm3 in volume) made of
hardened stainless steel together with hardened stainless steel
balls. The ball-to-powder weight ratio was 20:1. Milling
experiments were performed in air at 750 rpm.
Figure 1. Mo¨ssbauer spectrum of nonactivated MgFe2O4 taken at 300
K in an applied magnetic field of 5.5 T. The field is applied
perpendicular to the direction of the propagation of the Mo¨ssbauer
radiation. (A) and [B] denote cation sites of tetrahedral and octahedral
coordination, respectively.
Mo¨ssbauer measurements were performed in transmission
geometry using a microcomputer-controlled spectrometer in
constant acceleration mode using a 57Co/Rh γ-ray source. The
velocity scale was calibrated relative to 57Fe in Rh. The
transmitted γ-radiation was detected by means of a proportional
counter. “Recoil” spectral analysis software31 was used for the
quantitative evaluation of the Mo¨ssbauer spectra.
XRD patterns were collected using a URD 6 powder
diffractometer (Seifert-FPM, Freiberg) with Cu KR or Fe KR
radiation. XRD patterns were analyzed using the STADI P
software (Stoe, Darmstadt). The JCPDS database was utilized
for phase identification. An XRK-A Paar camera (Paar GmbH,
Graz) was employed for in situ X-ray diffraction in the
temperature range between 300 and 1100 K. Samples were
heated in a flowing nitrogen atmosphere at a heating rate of 2
K min-1. Typically, the partial pressure of residual oxygen in
the nitrogen gas used in the present experiments was of the
order of 10-4 atm.
The morphology of the powder and the sizes of individual
crystallites were studied by means of transmission electron
microscopy (TEM) (model CM30ST, Philips, Eindhoven). The
particle size distribution was measured by laser radiation
scattering (Laser-Particle-Sizer Analyzette 22 Granulometer,
Fritsch, Idar-Oberstein). Mean particle diameters and specific
surface areas were calculated from the granulometric data.
Thermoanalytical measurements were performed using a
SETARAM TAG 24 thermobalance with simultaneous dif-
ferential thermal analysis. Samples of 130 mg were heated in
open platinum crucibles up to 1100 K at a heating rate of 10 K
min-1 in flowing nitrogen gas.
Results and Discussion
The mechanically induced evolution of MgFe2O4 submitted
to the high-energy milling process has been followed by 57Fe
Mo¨ssbauer spectroscopy. Figure 1 shows the Mo¨ssbauer spec-
trum of nonactivated MgFe2O4 taken at room temperature in
an applied magnetic field of 5.5 T. The subspectra of both (A)-
and [B]-site ferric ions are indicated. The Mo¨ssbauer spectrum
of nonactivated MgFe2O4 taken in zero applied magnetic field,
Figure 2a, exhibits a considerable overlap of the (A) and [B]
Figure 2. Room-temperature Mo¨ssbauer spectrum of nonactivated
MgFe2O4 (a) and of material mechanically treated for 10 min (b) and
20 min (c) using steel milling tools.
subspectra due to the similar values of the hyperfine fields at
the iron nuclei on the two sublattices of the structure. In the
presence of an external magnetic field, the effective magnetiza-

6674 J. Phys. Chem. B, Vol. 106, No. 26, 2002
Sˇepela´k et al.
tion of the individual particles is aligned along the field and
the resulting magnetic field at the nuclei is either enhanced or
decreased by the applied field.32 In the case of nonactivated
MgFe2O4 with a collinear ferrimagnetic ordering,30 the presence
of an external magnetic field results in an almost complete
resolution of the overlapping patterns due to Fe3+ in octahedral
and tetrahedral coordination (Figure 1). The degree of inversion
calculated from the subspectral area ratio of the Mo¨ssbauer
spectrum without and with an applied magnetic field was found
to be λ ) 0.913(2) and λ ) 0.917(2), respectively; for details,
see Sˇepela´k et al.29 The values of the hyperfine magnetic fields
(HA ) 48.16(9) T, HB ) 50.01(1) T), of the isomer shifts (ISA
) 0.13(3) mm s-1, ISB ) 0.24(8) mm s-1), and of the degree
of inversion are in agreement with previously published data32
showing that MgFe2O4 is a partly inverse spinel.
of spectra with different relaxation times, since τ is sensitive to
the particle volume. Consequently, the spectral components
assigned to the Fe3+ ions (Figures 2b and 2c) can be understood
to arise from 57Fe in ferrite particles with relaxation times τ <
τL (superparamagnetic doublet) and τ > τL (slowly collapsing
broadened sextets due to Fe3+(A) and Fe3+[B]). The broad shape
of the collapsing sextets provides clear evidence of a wide
distribution of magnetic fields at the Fe3+ nuclei in disordered
MgFe2O4. This is particularly true for the octahedrally coordi-
nated ferric ions, Fe3+[B], which are very sensitive to the
number of their (A)-site nearest magnetic neighbors, Fe3+(A).29
The isomer shift of the superparamagnetic signal component
(IS ) 0.24(6) mm s-1) is characteristic of octahedrally
coordinated ferric ions.33 The slightly varying splitting of this
doublet (0.78(6) mm s-1 < QS < 0.81(7) mm s-1) indicates
changes in the relaxation time τ that may be due to changing
particle volumes in the course of the milling process.
Mechanical activation brings about significant changes in the
spectra. With increasing milling time, the hyperfine magnetic
sextets of the tetrahedrally and octahedrally coordinated iron
cations become asymmetric toward the inside of each line,
slowly collapse, and are gradually replaced by a central doublet
with broad lines (Figure 2). It should be mentioned in this
context that the new doublet structure in the center of the spectra
is already visible after only 3 min of milling. Further milling
leads to a gradual increase of its relative weight. The important
observation is that the newly created central doublet is asym-
metric and consist of two components that we attribute to Fe3+
and Fe2+. These assignments are based on the isomer shifts of
the subspectra: IS ) 0.24(6) mm s-1 for Fe3+ and IS )
0.81(8) mm s-1 for the Fe2+ state.33 The quadrupolar splitting
of the ferrous signal component is about 1.16 mm s-1, in good
agreement with splittings observed in Mg1-xFexO mixed crys-
tals.34,35 Here, composition-dependent splittings have been
reported with maximum values of about 1.2 mm s-1 for x )
0.6. Under the assumption that the Debye-Waller factors of
the iron species are identical, the Fe2+/(Fe3+ + Fe2+) ratio was
calculated from the areas of the two doublets. It was found to
increase with increasing milling time from zero for the non-
activated sample to about 0.32 and 0.4 for the sample milled
for 10 and 20 min, respectively (Figure 2). The fraction of
divalent iron ions relative to the total iron content amounts to
about 12.8% and 21.9% for the sample milled for 10 and 20
min, respectively. The appearance of two central doublets is
characteristic of the mechanical treatment of spinel ferrites by
means of steel milling tools.23,24 In the case of high-energy
milling using ceramic R-Al2O3 vial and balls, only a single
doublet due to Fe3+ is observed.28-30,36,37 Recently, this observa-
tion has also been reported for milling processes using tungsten
carbide (WC) balls and vials.38-40
In addition to the intensity decrease of the initial magnetic
sextets and the formation of both Fe3+ and Fe2+ doublet
components, the mechanical treatment leads to the formation
of a new hyperfine magnetic sextet with an isomer shift of IS
) -0.11(2) mm s-1 and an magnetic hyperfine field of H )
33.9(9) T, see Fe0 in Figures 2b and 2c. This magnetic field is
characteristic for metallic iron, which is given by H(R-Fe) )
33.04 T at room temperature.42 To conclude, the Mo¨ssbauer
data clearly show that high-energy milling of MgFe2O4 is
accompanied by the formation of reduced phases containing
Fe2+ and magnetically ordered metallic iron.
In addition to the well-defined spectral contributions described
above, the spectra of the milled material (Figures 2b and 2c)
also contain a broad structureless background. We had to
introduce this “sagging” background absorption to account for
unresolved magnetic relaxation processes in a qualitative way,
see, e.g., ref 43. The structureless background contains about
15% of the total spectral intensity. According to the respective
spectral intensity ratio, the fraction of magnetically ordered iron
increases from 0% for the nonactivated sample to about 6.3%
of total iron for the sample milled for 20 min. The total atomic
fraction of iron contained in the reduced phases reaches a value
of about 28.2% after this time of milling. These percentages,
however, may represent only lower-bound values because part
of the signal intensity of the reduced phases, notably that due
to metallic iron, may be contained in the sagging background
of the spectrum.
To determine the phase evolution of MgFe2O4 during high-
energy milling in greater detail and to support the findings of
Mo¨ssbauer spectroscopy, the mechanochemical reduction was
also followed by X-ray powder diffraction. The XRD pattern
of the nonactivated sample (Figure 3a) is characterized by the
sharp crystalline peaks corresponding to MgFe2O4 (JCPDS
17-0464). During the early stages of milling, XRD merely
reveals a decrease of the intensity and an associated broadening
of the Bragg peaks of the spinel. This reflects the formation of
the disordered state with a small crystallite size and with internal
strain introduced during the mechanical treatment. With increas-
ing milling time, qualitative changes are observed in diffraction
patterns of the samples. This can be attributed to the formation
of a solid solution of Mg1-xFexO which possesses the wu¨stite
structure (JCPDS 35-1393) and of crystalline iron (JCPDS
06-0696). This observation is similar to the behavior encountered
in the case of NiFe2O4 under the same milling conditions where
the mechanically induced reduction led to the formation of
metallic iron and of solid solution Ni1-xFexO.24 In both
mechanochemically reduced MgFe2O4 and NiFe2O4 spinels, the
To elucidate the origin of the central doublet component
assigned to the Fe3+ ions, it should be noted that the high-energy
milling of MgFe2O4 is accompanied by the reduction of the
crystallite size to the nanometer range (∼10 nm). It is well-
known that the most striking effect of crystallite size reduction
on Mo¨ssbauer spectra is due to superparamagnetic relaxation.
The latter arises if crystallite sizes are so small that thermally
induced energy fluctuations can overcome the anisotropy energy
and change the direction of the magnetization of a particle from
one easy axis to another.41 A material can be observed as
superparamagnetic in the Mo¨ssbauer spectrum if the relaxation
time τ for the reversal of magnetization in a particle is smaller
than the Larmor precession time τL of the nuclear magnetic
moment in the local field.41 In an assembly of particles with
different volumes (nonuniformity of particle sizes), the experi-
mental Mo¨ssbauer spectra will be given by the superposition

Mechanochemical Reduction of Magnesium Ferrite
J. Phys. Chem. B, Vol. 106, No. 26, 2002 6675
Figure 3. X-ray diffraction pattern of nonactivated MgFe2O4 (a) and
of material mechanically treated for 20 min (b). Diffraction lines of
MgFe2O4 are denoted by Miller indices (Figure 3a). The asterisk
indicates a diffraction peak of the sample holder (Cu KR radiation).
multiphase nature of the samples and the broadness of the
diffraction lines of the nanosize milled material make it
impossible to determine quantitative structural data or the exact
chemical composition of the Me1-xFexO or of a possible
Me-Fe solid solution (Me ) Mg or Ni).
Figure 4. TEM images of the products of mechanochemical reduction
of MgFe2O4.
An extensive study44,45 of the influence of the milling
conditions on the contamination of oxide (SiO2) powders by
metallic iron during high-energy milling has shown that the Fe
content originating from the abrasion of the metallic milling
tools increases with increasing milling time and specific mill
power. However, under conditions of dry milling, abrasion
exhibits a saturation behavior limiting iron wear to about 0.7
at. %, which has been attributed to an encrustation of the metallic
surfaces.44,45 Under the assumption that the material under study
in the present work behaves similarly to the oxide studied in
refs 44 and 45, and because pre-used balls and vials were utilized
in our study, we expect that the contamination by iron abrasion
is less than 1 at. %. The relatively high content (∼28.2%) of
the reduced phases containing Fe2+ and Fe, therefore, provides
evidence that the abrasion cannot be the main source of
crystalline metallic iron detected in our study. Thus, on the basis
of the available information on iron abrasion in dry milling
experiments, we conclude that both the solid solution of
Mg1-xFexO and the ferromagnetic Fe are predominantly obtained
as the result of the mechanochemical reduction of MgFe2O4.
Clearly, such a mechanically induced chemical reduction in an
oxygen-containing atmosphere represents an unusual and un-
expected observation.
due to the scarcity of these observations. This is compatible
with the above-mentioned information on the small extent of
iron abrasion to be expected in dry high-energy milling.44,45
Results from electron microscopy and particle size analysis
reveal that the sizes of the powder particles of the nonactivated
sample range between 10 µm to 50 µm with a specific surface
of about 1.06 m2 cm-3. In the course of milling, the powder
particles are subjected to continuous fragmentation. As a result,
the powders become much finer and more uniform in shape.
While the nonactivated MgFe2O4 consisted predominantly of
individual particles, the milled samples consist of aggregates
of fine particles. The specific surface area of the powders
obtained was 3.2 m2 cm-3. Dark-field TEM images of the
mechanochemically reduced MgFe2O4 (Figure 4a) show many
bright spots indicating many small crystals. Their sizes were
estimated from TEM to be about 10 nm (Figure 4b).
The spinel ferrites are known to be stable and chemically
inert refractory compounds. According to the phase diagram of
the Mg-Fe-O system,46,47 MgFe2O4 can be reduced at very
low oxygen activities. In a study performed at relatively low
temperatures (in the range from 720 to 970 K) and using flowing
hydrogen gas to ensure low oxygen activities, it has been shown
that MgFe2O4 can be reduced leading to the formation of the
solid solution Mg1-xFexO and of metallic iron.35,48 Hence, under
these conditions, the same reaction products are obtained that
have also been observed in the case of the mechanochemical
reduction of MgFe2O4 performed in air at room temperature.
Thus, the mechanically induced reduction of MgFe2O4 offers
As discussed above, abrasion of the milling tools is unavoid-
able to some extent. Accordingly, by close inspection of the
milled samples by electron microscopy and EDX analysis
chromium-containing particles have been detected, which
originate from the wear of the hardened steel vial and balls (Fe18-
Cr8Ni). However, a quantitative EDX analysis was not possible

6676 J. Phys. Chem. B, Vol. 106, No. 26, 2002
Sˇepela´k et al.
Figure 5. (a) The XRD patterns of the mechanochemically reduced MgFe2O4 taken during the thermal treatment in the high-temperature X-ray
camera (Fe KR radiation). A gradually disappearing diffraction line (due to oxidation) of metallic iron is indicated by an arrow. The sample was
measured on a platinum support (*). (b) Detail of Figure 5a: the XRD pattern at 1000 K showing the asymmetry of the (311) diffraction line.
an alternative route to the conventional thermally activated
solid-gas reduction process.
The products of the mechanochemical reduction reaction are
metastable with respect to structural and compositional changes
under temperature and environmental conditions. In situ high-
temperature XRD in a flowing nitrogen atmosphere in the range
from 293 to 600 K showed no change in the diffraction patterns
of the milled powders. Thus, the range of thermal stability of
the reduction products extends up to 600 K under these
conditions. However, at temperatures above 600 K a gradual
recrystallization of the nanoscale powders takes place. This is
manifested by the gradual narrowing of diffraction lines and
by an increase of their intensities, as shown in Figure 5a. The
growth of the crystallites is accompanied by compositional
changes of the reduction products. The metallic iron is gradually
oxidized to Fe3O4 (JCPDS 19-0629) by the residual oxygen
Figure 6. Mo¨ssbauer spectrum of the mechanochemically reduced
MgFe2O4 after annealing at 1000 K in nitrogen atmosphere. Cation
present in the nitrogen gas. This process is manifested by the
gradual disappearance of the diffraction line of metallic iron
and by the increase of the diffraction line intensities of the spinel
phase (Figure 5a). More direct evidence for the formation of
magnetite, however, is provided by Mo¨ssbauer spectroscopy.
The second product of the mechanochemical reduction, the solid
solution Mg1-xFexO, is relatively stable with respect to tem-
perature. The presence of this phase in the thermally treated
samples is reflected by the asymmetry of the (311) diffraction
line of the spinel phase (at 2Θ ≈ 45°) on its right-hand side
(Figure 5b).
After the annealing of the multiphase mixture, the formation
of magnetite and the partial oxidation of the solid magnesio-
wu¨stite solution was confirmed by Mo¨ssbauer spectroscopy. The
Mo¨ssbauer spectrum of the sample milled for 20 min after
annealing at 1000 K in nitrogen atmosphere (Figure 6) shows
the sextets corresponding to magnetically ordered MgFe2O4 and
Fe3O4 as well as the paramagnetic doublet characteristic of Fe2+
cations in the solid solution. No indication of Fe and Fe2O3
was observed. As expected, the superparamagnetic doublet
component assigned to the Fe3+ ions vanished due to recrys-
tallization and the growth of spinel ferrite particles. The
disappearance of the broad, structureless absorption used to
sites of tetrahedral (A) and octahedral [B] coordination in the spinels
MgFe2O4 (MF) and Fe3O4 (M) are indicated by (A)MF, [B]MF, and (A)M,
[B]M, respectively. The doublet in the center of the spectrum corre-
sponds to ferrous ions in magnesiowu¨stite, Mg1-xFexO.
describe the “sagging” of the Mo¨ssbauer spectra (Figure 2)
indicates that this spectral component is associated with the
presence of small particles exhibiting magnetic relaxation
phenomena (superparamagnetism and/or collective magnetic
excitations). The quantitative evaluation of the Mo¨ssbauer
spectra revealed that the thermal treatment of the milled samples
results in a decrease of the Fe2+ content in the solid solution
by about 5% when compared with the sample before annealing.
This indicates that in addition to the complete oxidation of
metallic iron, also a partial oxidation of the solid Mg1-xFexO
solution takes place during heating.
The information on the structural response of the mechano-
chemical reduction products to changes in temperature obtained
by in situ XRD and Mo¨ssbauer spectroscopy was complemented
by thermogravimetric measurements performed in a nitrogen
atmosphere. The thermal treatment of the nonactivated sample
in the range from 293 to 1000 K does not produce any mass
changes. In contrast, heating the milled samples (Figure 7) leads
to a mass increase due to oxidation of the reduced phases by

Mechanochemical Reduction of Magnesium Ferrite
J. Phys. Chem. B, Vol. 106, No. 26, 2002 6677
Another important factor to be considered in the context of
mechanochemical reaction is the mass of the balls in the high-
energy milling which strongly influences the impact energies.
It has been demonstrated that high-energy milling of spinel
ferrites using alumina balls, whose density is lower than that
of steel milling tools, does not lead to chemical reduction.28-30,36,37
Similarly, the use of WC milling tools, which are heavier than
steel ones, does not lead to chemical reduction of spinels.38-40
In view of these results, it appears that the mass of the balls
(and consequently the possible “hot spot” temperature) cannot
be the dominant factor responsible for the observed mechano-
chemical reduction process. Taking into account this fact
together with the present evidence and that obtained in the
analogous system NiFe2O4,24 it may be stated that the material
of the vial and balls plays an essential role in the mechano-
chemical transformation of spinel ferrites. As an alternative to
the above indicated views, the reduction process could be
understood as result of a short-time equilibration between spinel
and the encrusted metallic iron phase of the milling tools during
the impact period. The oxidation of the metallic ball, however,
may be kinetically hindered, leading instead to the liberation
of oxygen. Impact-induced local heating and high pressures may
provide further factors enhancing this reduction mechanism of
the spinel. In this context, the minute quantities of abraded iron
could play an initiating, catalytic role. To conclude, the
identification of the factors representing the main driving force
for mechanically induced redox processes, the elucidation of
the microscopic mechanism(s), and the determination of rate-
determining steps of mechanochemical reduction represent major
challenges for fundamental research and require further efforts.
Figure 7. Thermogravimetric curves of MgFe2O4 mechanically treated
for various times (tm).
the residual oxygen present in the flowing nitrogen gas. If the
atomic fraction of iron that reacts with oxygen during heating
amounted to only 11.3% (6.3% Fe0 + 5% Fe2+ in the solid
solution), a mass increase of about 2.8% could be expected upon
reoxidation of the metallic iron and partial reoxidation of the
solid solution. This is in reasonable agreement with the
experimental mass change of up to 3.5% observed for the milled
material (Figure 7).
In view of the fact that the milling process is carried out in
air, the appearance of reduced phases is a surprising experi-
mental result. Although several mechanisms have been proposed
for the reduction processes occurring during high-energy
milling,11-13,26,49 to our best knowledge, no conclusive explana-
tion for mechanically induced redox reactions has been given
yet. Kaczmarek et al.12,49 suggested that the mechanochemical
reduction process takes place at the surface of the milled oxide.
According to these authors, oxygen bonds on the cleaved oxide
surfaces are broken during the mechanical treatment leading to
the reduction of the sample and to a release of oxygen to the
vial’s internal volume. In this case, the rupture of oxide surface
layers and surface stress are assumed to provide the driving
force for the reduction.
It has also been proposed that the oxygen partial pressure in
the vial atmosphere is the critical parameter responsible for the
mechanochemical reduction. Zdujic et al.15,50 presumed that
adsorption of oxygen on atomically clean surfaces created by
particle fracture may cause a decrease of the oxygen partial
pressure to some critical value below which the given phase
becomes locally unstable and transforms to a stable one. The
transformation occurs at the moment of impact by the formation
of high-energy localized sites of short lifetime (sometimes called
“hot spots” or “thermal spikes”).
Also the role of milling tools has been considered in the
context of mechanically induced chemical reduction. It has been
assumed by numerous authors that the contamination originating
from the abrasion of the vial and the balls is insignificant for
the reduction reaction.12,13,26,49 In contrast, Kosmac et al.11
suggested on the basis of their experiments on milled R-Fe2O3
powders that contamination by Fe from the stainless steel vial
and balls can cause the reduction of Fe3+ to a mixture of Fe3+
and Fe2+, i.e.: 4R-Fe2O3 + Fe f 3Fe3O4. However, the
relatively small amount of iron contamination of up to 1%
typical for dry high-energy milling44,45 seems to imply that this
process cannot really account for the extensive reduction
observed in the present case. Thus, we conclude that an
interaction between a possible iron contamination and the milled
oxide is not the dominant process in the mechanochemical
reduction of MgFe2O4.
Conclusions
The changes in MgFe2O4 induced by high-energy milling in
a stainless steel vial have been investigated by Mo¨ssbauer
spectroscopy, X-ray diffraction, electron microscopy, and
thermal analysis. In addition to the formation of nanoscale ferrite
particles, the mechanical treatment results in the mechanochem-
ical reduction of MgFe2O4 leading to the formation of a solid
solution Mg1-xFexO and of a metallic iron phase. The fraction
of the reduced phases increases with increasing milling time.
Due to its local nature, Mo¨ssbauer spectroscopy has proven
its exceptional ability to discriminate different valence states,
magnetic states, and local coordination of iron ions in nano-
crystalline mechanically treated spinel ferrites. Both these
properties of local atomic and magnetic structure as well as
phase analysis of nanoscale materials are especially valuable
in view of the fact that diffraction techniques lose much of their
resolving power in such systems.
The nanocrystalline products of the mechanochemical reduc-
tion are metastable with respect to structural and compositional
changes at elevated temperatures. In a nitrogen atmosphere, the
range of the thermal stability of the nanoscale product extends
up to about 600 K. At temperatures above 600 K, the nanosize
product gradually begins to crystallize. This is accompanied by
the oxidation of the metallic iron to Fe3O4, while the solid
solution is relatively stable chemically during the annealing.
In the context of a mechanistic analysis, we conclude that
the metallic nature of the vial and of the balls plays an essential
role in the mechanochemical reduction process. However, the
details of this process and factors influencing it remain open
and require further study. It appears that the event of mechani-
cally induced redox reactions presents novel opportunities for
the nonthermal manipulation of materials and provides a
promising field for future fundamental as well as applied
research.

6678 J. Phys. Chem. B, Vol. 106, No. 26, 2002
Sˇepela´k et al.
(26) Goya, G. F.; Rechenberg, H. R.; Jiang, J. Z. J. Appl. Phys. 1998,
Acknowledgment. This work was supported by the Alex-
ander von Humboldt Foundation and Slovak Grant Agency for
Science (Grant 2/7040/20). Financial assistance by the Fonds
der Chemischen Industrie is gratefully acknowledged.
84, 1101.
(27) Willey, R. J.; Noirclerc, P.; Busca, G. Chem. Eng. Commun. 1993,
123, 1.
(28) Sˇepela´k, V.; Schultze, D.; Krumeich, F.; Steinike, U.; Becker, K.
D. Solid State Ionics 2001, 141-142, 677.
(29) Sˇepela´k, V.; Baabe, D.; Litterst, F. J.; Becker, K. D. J. Appl. Phys.
2000, 88, 5884.
References and Notes
(30) Sˇepela´k, V.; Baabe, D.; Litterst, F. J.; Becker, K. D. Hyperfine
(1) For example, Schumacher, P.; Warren, P.; Cantor, B. Metastable,
Mechanically Alloyed and Nanocrystalline Materials; Trans Tech Publica-
tions: Zu¨rich, 2001; Boldyrev, V. V. ReactiVity of Solids: Past, Present
and Future; Blackwell Science: Oxford, 1996.
(2) Avvakumov, E. G. Mechanical Methods of ActiVation of Chemical
Processes; Nauka: Novosibirsk, 1986 (in Russian).
(3) Schaffer, G. B.; McCormick, P. G. Appl. Phys. Lett. 1989, 55, 45.
(4) Takacs, L. Nanostruct. Mater. 1993, 2, 241.
(5) Basset, D.; Matteazzi, P.; Miani, F.; Le Cae¨r, G. Hyperfine
Interactions 1994, 95, 2235.
(6) Tokomitsu, K. Solid State Ionics 1997, 101-103, 25.
(7) Kerr, A.; Welham, N. J.; Willis, P. E. Nanostruct. Mater. 1999,
11, 233.
(8) Goya, G. F.; Rechenberg, H. R. J. Phys.: Condens. Matter 2000,
12, 10579.
(9) El-Eskandarany, M. S.; Omori, M.; Konno, T. J.; Sumiyama, K.;
Hirai, T.; Suzuki, K. Metall. Mater. Trans. A 2001, 32, 157.
(10) Oleszak, D.; Krasnowski, M. Mater. Sci. Forum 2001, 360-362,
235.
(11) Kosmac, T.; Courtney, T. H. J. Mater. Res. 1992, 7, 1519.
(12) Kaczmarek, W. A.; Ninham, B. W. IEEE Trans. Magn. 1994, 30,
732.
(13) Campbell, S. J.; Kaczmarek, W. A.; Wang, G. M. Nanostruct.
Mater. 1995, 6, 735.
(14) Jiang, J. Z.; Zhou, Y. X.; Mørup, S.; Koch, C. B. Nanostruct. Mater.
1996, 7, 401.
Interactions 2000, 126, 143.
(31) Lagarec, K.; Rancourt, D. G. Recoil - Mo¨ssbauer Spectral Analysis
Software for Windows, Version 1.02; Department of Physics, University of
Ottawa, Ottawa, 1998.
(32) O’Neill, H. St. C.; Annersten, H.; Virgo, D. Am. Miner. 1992, 77,
725.
(33) Menil, F. J. Phys. Chem. Solids 1985, 46, 763.
(34) Becker, K. D.; Dreher, S. Ber. Bunsen-Ges. Phys. Chem. 1989,
93, 1382.
(35) Ge, X.; Li, M. S.; Shen, J. Y. J. Solid State Chem. 2001, 161, 38.
(36) Sˇepela´k, V.; Baabe, D.; Becker, K. D. J. Mater. Synth. Proces.
2000, 8, 333.
(37) Sˇepela´k, V.; Wissmann, S.; Becker, K. D. J. Mater. Sci. 1998, 33,
2845.
(38) Chinnasamy, C. N.; Narayanasamy, A.; Ponpandian, N.; Justin
Joseyphus, R.; Jeyadevan, B.; Tohji, K.; Chattopadhyay, K. J. Magn. Magn.
Mater. 2002, 238, 281.
(39) Chinnasamy, C. N.; Narayanasamy, A.; Ponpandian, N.; Chatto-
padhyay, K. Shinoda, K.; Jeyadevan, B.; Tohji, K.; Nakatsuka, K.;
Furubayashi, T.; Nakatani, I. Phys. ReV. B 2001, 63, 184108.
(40) Chinnasamy, C. N.; Narayanasamy, A.; Ponpandian, N.; Chatto-
padhyay, K. Mater. Sci. Eng. A 2001, 304-306, 983.
(41) Mørup, S. In Mo¨ssbauer Spectroscopy Applied to Inorganic
Chemistry. Vol. 2; Long, G. J., Ed.; Plenum: New York, 1987; p 89.
(42) Spikerman, J. J.; De Voe, J. R.; Travis, J. C. Standard Reference
Materials: Standards for Moessbauer Spectroscopy, NBS Special Report
260-20 1970.
(15) Zdujic´, M.; Jovalekic, CÄ .; Karanovic, Lj.; Mitric´, M. Mater. Sci.
Eng. A 1999, 262, 204.
(43) Hesse, J. Hyperfine Interactions 1989, 47, 357.
(44) Tka´cova´, K.; Sˇtevulova´, N.; Lipka, J.; Sˇepela´k, V. Powder Technol.
1995, 83, 163.
(45) Sˇtevulova´, N.; Ba´lintova´, M.; Tka´cova´, K. J. Mater. Synth. Process.
2000, 8, 265.
(46) Zalazinskii, A. G.; Balakirev, V. F.; Barkhatov, V. P.; Chufarov,
G. I. Russ. J. Phys. Chem. 1975, 49, 914.
(16) Yang, H.; McCormick, P. G. J. Solid State Chem. 1993, 107, 258.
(17) Ding, J.; Miao, W. F.; Tsuzuki, T.; McCormick, P. G.; Street, R.
In Synthesis and Processing of Nanocrystalline Powder; Bourell, D. L.,
Ed.; TMS: Warrendale, 1996; p 69.
(18) Welham, N. J. J. Alloy Compd. 1998, 274, 303.
(19) Welham, N. J. Mater. Sci. Eng. A 1998, 248, 230.
(20) Takacs, L. Mater. Lett. 1992, 13, 119.
(47) Paladino, A. E. J. Am. Ceram. Soc. 1960, 43, 183.
(48) Qian, Y. T.; Kershaw, R.; Dwight, K.; Wold, A. Mater. Res. Bull.
1983, 18, 543.
(49) Kaczmarek, W. A.; Onyszkiewicz, I.; Ninham, B. W. IEEE Trans.
Magn. 1994, 30, 4725.
(50) Zdujic, M.; Jovalekic, C.; Karanovic, Lj.; Mitric, M.; Poleti, D.;
Skala, D. Mater. Sci. Eng. A 1998, 245, 109.
(21) Sorescu, M. J. Mater. Sci. Lett. 1998, 17, 1059.
(22) Botta, P. M.; Aglietti, E. F.; Porto Lo´pez, J. M. J. Mater. Synth.
Process. 2000, 8, 345.
(23) Sˇepela´k, V.; Becker, K. D. J. Mater. Synth. Process. 2000, 8, 155.
(24) Menzel, M.; Sˇepela´k, V.; Becker, K. D. Solid State Ionics 2001,
141-142, 663.
(25) Shi, Y.; Ding, J. J. Appl. Phys. 2001, 90, 4078.