7414 J. Phys. Chem.

B 2001, 105, 7414-7423

Synthesis and Properties of γ-Fe2O3 Nanoclusters within Mesoporous Aluminosilicate

Matrices

Lei Zhang,† Georgia C. Papaefthymiou,‡,§ and Jackie Y. Ying*,†

Department of Chemical Engineering and Francis Bitter Magnet Laboratory,
Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
ReceiVed: January 17, 2001

Iron oxide nanoclusters were synthesized within mesoporous MCM-41 aluminosilicate matrices via evaporation-
condensation of volatile Fe(CO)5. The well-defined hexagonally packed cylindrical pore structure of MCM-
41 led to the derivation of γ-Fe2O3 particles with spherical and elongated morphologies. Magnetization studies
and Mössbauer spectroscopy indicated that the γ-Fe2O3/MCM-41 nanocomposites exhibited interesting
superparamagnetic behavior. A blue shift in the absorption edge relative to bulk iron oxide was noted in the
UV-Vis spectra. Strain at the particle-support interface and quantum confinement effects played a critical
role in determining the overall magnetic and optical behavior of the γ-Fe2O3/MCM-41 nanocomposites. The
iron oxide nanoclusters within the MCM-41 matrix showed high thermal stability and increased magnetization
when calcined at high temperatures.

Introduction
Nanoclusters and nanoarrays have attracted a great deal of
research attention due to their potential use as structural,
catalytic, magnetic, electronic, and optical materials.1,2,3,4,5
Ferrimagnetic γ-Fe2O3 is of particular interest due to its
dominant role in magnetic storage media.6 Highly dispersed
γ-Fe2O3 nanoclusters possess unique magnetic and electronic
characteristics, such as superparamagnetism and macroscopic
quantum tunneling associated with size quantization and elec-
tronic quantum confinement effects.7 By packing magnetic
nanoclusters into a host matrix, tailored interfacial confinement
environment, intercluster interaction, and effective charge carrier
transport may be obtained. Such composite structures have been
achieved by using cross-linked ion-exchange polymers, zeolitic
molecular sieves, and sol-gel derived materials.8,9,10
Crystalline zeolitic matrices can host highly ordered super-
structures of monodispersed magnetic particles but are limited
by their microporous dimensions (typically 0.5 to 1.3 nm). Sol-
gel derived silica materials can support particles of larger sizes
within their interconnected porous framework but suffer from
a low degree of particle packing order within the amorphous
matrices of broad pore size distributions.
We have recently reported on two novel approaches for the
synthesis of γ-Fe2O3 nanoclusters via (i) controlled chemical
precipitation of iron oxide followed by silica coating11,12 and
(ii) ion-exchange and oxidation of ferrous cations within a
porous sulfonated silica matrix.11 Here, we describe the use of
MCM-41 with hexagonally-packed mesopores of uniform
diameter13 as a matrix for hosting guest nanoclusters for
fundamental studies and potential nanostructured device ap-
plications. The one-dimensional cylindrical pore channels of
MCM-41 can be ideally suited for simultaneous control of
cluster size and morphology. Much effort has been devoted
* To whom correspondence should be addressed.
†Department of Chemical Engineering, M. I. T.
‡ Francis Bitter Magnet Laboratory, M. I. T.
§ Present address: Department of Physics, Villanova University, Vill-
anova, PA 19085.
10.1021/jp010174i CCC: $20.00 © 2001 American Chemical Society
Published on Web 07/18/2001
toward filling the mesopores with ultrafine clusters and nano-
wires. Nanoclusters of Fe2O3, polyaniline, Ge, GaAs, InP, Pt,
Pd, Ru have been deposited within the MCM-41 inorganic
matrix.14-21
In this paper, mesoporous MCM-41 matrices were used to
host the synthesis of magnetic γ-Fe2O3 nanoclusters. The
resulting nanocomposite materials demonstrated unique mag-
netic and optical properties associated with quantum confine-
ment effects and particle-support interactions. Various experi-
mental techniques were employed to characterize the MCM-41
matrix structure, the iron oxide phase, and the nanocluster
dispersion within the porous support. Size quantization effects
on the magnetic and optical properties of the γ-Fe2O3/MCM-
41 nanocomposites were investigated. Finally, these new
findings were compared to our previously reported studies on
γ-Fe2O3 nanoparticles embedded within amorphous silica
matrices.11,12
Synthesis of Fe2O3 Clusters within Hexagonally-Packed
Mesoporous Hosts
Synthesis and Characterization of Mesoporous Alumino-
silicate. Al-doped mesoporous silica was synthesized from an
inorganic siliceous precursor using organic cationic trimethyl-
ammonium surfactants (C16H33(CH3)3NBr or CTAB) as a
supramolecular templating agent. A 3.65 g portion of CTAB
surfactants was completely dissolved in H2O. A 22.2 g portion
of 27% sodium silicate solution dispersed in 50 g of water was
slowly added to the surfactant solution at room temperature with
vigorous stirring. After being stirred for 5 min, the pH of the
mixture was adjusted to 11.5 with diluted H2SO4. A desired
amount of aqueous aluminum sulfate solution was then gradually
introduced to the loosely bonded silica gel and stirred for 3 h
at room temperature before the sample was hydrothermally
treated at 100-150 °C for a desired period. The molar
composition of the wet gel could be represented as 1 SiO2: 0.1
CTAB: 120 H2O: x Al2O3 (where x ) 0 to 0.1). The solid
powder obtained was washed with ethanol and water, filtered,
and then calcined at 540 °C in air for 6 h to remove the organics.
Synthesis and Properties of γ-Fe2O3 Nanoclusters
Figure 1. XRD patterns of calcined mesoporous AlSi-R samples of
different Al dopant concentrations aged at 100 °C: (a) R ) 5, (b) R )
25, and (c) R ) 50.
Figure 2. N2 adsorption-desorption isotherm of calcined AlSi-25.
The samples were designated as AlSi-R, where R was given
by the precursor molar ratio of Si and Al.
The long-range pore packing order of the mesoporous Al-
doped silica was characterized by X-ray diffractometry (XRD)
(Siemens D5000 θ-θ diffractometer operated at 45 kV and 40
mA using Ni-filtered Cu KR, radiation) and transmission electron
microscopy (TEM) (JEOL 200CX, 200 kV). The XRD patterns
of calcined AlSi-R shown in Figure 1 corresponded to a typical
hexagonal P6m space group with the intense (100) peak as well
as minor peaks corresponding to (110), (200) and (210)
diffractions.
N2 adsorption analysis was performed on a Micromeritics
ASAP 2010 gas sorption instrument. The N2 adsorption-
desorption isotherm of the calcined AlSi-25 sample showed
negligible hysteresis (Figure 2). The sample has a narrow
Barrett-Joyner-Halenda (BJH) pore size distribution (Figure
3), with a Brunauer-Emmett-Teller (BET) surface area of 1217
m2/g.
Introduction of Fe2O3 Nanoclusters by Evaporation-
Condensation. The microstructural characteristics of porous
materials, such as surface area, pore diameter, and pore size
distribution, would affect the local environment for hosting
quantum-sized metal oxide clusters. MCM-41 materials with
hexagonally packed mesopores of 2-10 nm diameter may
promote the formation of ordered superstructures of the magnetic
phase within their well-defined 1-D cylindrical pore channels.
Both the pore morphology and the dielectric properties of the
mesoporous aluminosilicate materials are attractive matrix
characteristics.
An evaporation-condensation apparatus was used in our
synthesis (Figure 4). Fe(CO)5, a highly volatile precursor, was
carefully stored under argon and kept in the dark, while
J. Phys. Chem. B, Vol. 105, No. 31, 2001 7415
Figure 3. BJH desorption pore size distribution of calcined AlSi-25
calculated from the N2 desorption isotherm.
Figure 4. A schematic of the evaporation-condensation apparatus.
mesoporous aluminosilicate (AlSi-25) was degassed at 350 °C
overnight. Typically, 3 mL of Fe(CO)5 and 0.2 g of mesoporous
AlSi-25 were introduced to an evaporation flask and a receiving
tube, respectively. After assembling the two parts, the meso-
porous aluminosilicate was further evacuated for 3 h at 250 °C,
which was maintained by a heating tape around the receiving
tube. The receiving tube was then cooled to the desired
temperature. Upon opening valves 1 and 2, Fe(CO)5 was
evaporated at room temperature and condensed within the
mesopores of the aluminosilicate support. Excess Fe(CO)5 was
then pumped out of the receiving tube to the cold trap by
opening valve 3. The mesoporous silica containing Fe(CO)5 was
next subjected to a H2 stream at 100-200 °C for several hours
to decompose and reduce the iron complex. The as-prepared
Fe2O3/AlSi-25 nanocomposite was further calcined in a tube
furnace under flowing O2 for 3 h at the desired temperature.
The iron loading depended on the synthesis parameters, such
as the amount of Fe(CO)5, evaporation rate, saturation time,
reaction temperature, and decomposition conditions. The evapo-
ration-condensation procedure could be repeated several times
to achieve a high iron loading.
Results and Discussions
Microstructure of Fe2O3/Mesoporous Aluminosilicate Nano-
composites. Three Fe2O3/AlSi-25 nanocomposite samples were
synthesized with different Fe loadings. Elemental analysis results
from direct-current plasma emission spectroscopy (Luvak Inc.)
are summarized in Table 1. BET surface area and pore volume
7416 J. Phys. Chem. B, Vol. 105, No. 31, 2001 Zhang et al.

Figure 5. XRD patterns (i: 2θ ) 1.5-20° and ii: 2θ ) 15-75°) of Fe2O3/AlSi-25 nanocomposite #2: (a) as-prepared, (b) calcined at 300 °C,
(c) calcined at 500 °C, and (d) calcined at 800 °C.

TABLE 1: Characterization of Calcined Mesoporous those on the dark field image. Elemental mapping with the point
AlSi-25 and As-Prepared Fe2O3/A1Si-25 Nanocomposites probe (2 nm probe size) was thus employed to verify the Fe
Fe2O3/ Fe2O3/ Fe2O3/ locations. When the X-ray probe was pointed at the bright
sample AlSi-25 AlSi-25 #1 AlSi-25 #2 A1Si-25 #3 spherical spots shown in the dark field image, strong Fe
Fe loading 0 1.27 6.82 57.5 absorption was detected, indicating the presence of Fe. When
(wt %) the probe was moved to the adjacent dark regions, no Fe was
receiving tube - 25 50 80
temperature (°C) detected. In certain areas, a lamellar-like structure was noted
reduction 100 °C, 1 h; 200 °C, 5 h 125 °C, 5hr in the bright field image, corresponding to hexagonally packed
temperature then 150 °C, 5 h pore arrays oriented perpendicular to the electron beam. This
and time
surface area 1217 929 763 362 was viewed as elongated bright lines on the dark field image.
(m2/g) The point mapping along the bright lines corresponded to a
pore volume (cm2/g) 1.10 0.93 0.66 0.29 continuous detection of Fe absorption, indicating the presence
of elongated iron oxide particles. For the sample calcined at
of the Fe2O3/AlSi-25 samples as determined by N2 adsorption
500 °C, a homogeneous Fe mapping with dark field image was
analysis decreased with increasing Fe loading, as shown in Table
obtained, reflecting both spherical and elongated morphologies
1. At a high Fe loading, substantial pore plugging by large Fe2O3
for the iron oxide particles (Figure 8). No particle agglomeration
nanoclusters might have occurred. The particle agglomeration
or sintering was observed at 500 °C, suggesting a high thermal
at the higher loadings led to a significant reduction of the
stability for Fe2O3/AlSi-25 nanocomposite #1.
magnetization due to strong antiferromagnetic coupling among
the closely interacting particles. Magnetic and Optical Properties of γ-Fe2O3/Mesoporous
The XRD pattern of the as-prepared Fe2O3/AlSi-25 nano- Aluminosilicate Nanocomposites. For nanometer-sized single
composite #2 showed that the hexagonally packed mesostructure domain magnetic clusters, magnetic behavior is governed by
of AlSi-25 was preserved (Figure 5i(a)) and no additional peaks the magnetic energy barrier defined as the product of the cluster
associated with Fe2O3 crystalline phase was found (Figure 5ii volume and its magnetic anisotropy density.22 At a temperature
(a)). As the calcination temperature increased, the d(100) spacing T g TB (TB ) critical magnetic blocking temperature), the
of the hexagonal phase decreased slightly due to pore shrinkage thermal energy is comparable to the magnetic energy barrier.
from condensation. A broad band at 2θ ∼ 22° associated with The cluster magnetization vector is thermally excited to fluctuate
amorphous SiO2 was noted in all XRD patterns in Figure 5ii. between the easy directions of magnetization in a manner
No crystalline peaks were observed at high 2θ angles (15°- analogous to that of paramagnetic atoms in an applied field,
75°), indicating the preservation of ultrafine grain sizes for iron except with a much larger magnetic moment. This is the well-
oxide clusters upon calcination. known phenomenon of superparamagnetism. For superpara-
The morphology of the Fe2O3 nanoclusters within the MCM- magnetic particles, the total magnetic moment displays a
41 matrix was studied using both TEM (Akashi 002B, 200 kV), paramagnetic behavior obeying a classical Langevin function
and scanning transmission electron microscopy (STEM) (Vacuum above the TB.
Generators HB603) with elemental mapping. For the as-prepared The macroscopic magnetic properties of our nanocomposites
and 500 °C-calcined Fe2O3/AlSi-25 nanocomposite #1, TEM were investigated by superconducting quantum interference
bright field images showed a hexagonally packed array with device (SQUID). magnetometry (Quantum Design, MPMS,
dark phase contrast inside the pores when viewing normal to characteristic measurement time (τm) ) 102 s) over a wide range
the axis of the hexagonally packed pores (Figure 6). The STEM of temperatures (5 K < T < 300 K) and applied magnetic fields
dark field image showed highly dispersed bright spherical spots (0 T < H < 3 T). Zero field cooling (ZFC) and field cooling
of ∼2-3 nm in diameter (Figure 7). The Fe elemental mapping (FC) magnetization curves were recorded, and magnetic hys-
illustrated that Fe was highly dispersed throughout the image teresis loops at low temperatures were studied.
area, with higher concentrations indicated by the bright spots. The internal magnetization and dynamic spin relaxation
Because of image drifting during scanning, it was very difficult processes of the nanocomposites were also examined with
to precisely match the Fe locations on the elemental map with Mössbauer spectroscopy. The fast 10-8 sec characteristic
Synthesis and Properties of γ-Fe2O3 Nanoclusters
Figure 6. TEM micrographs of Fe2O3/AlSi-25 nanocomposite #1: (a)
as-prepared and (b) calcined at 500 °C.
measurement time for Mössbauer spectroscopy yielded informa-
tion on blocking temperature, below which spin relaxation was
slowed. This enabled the investigation of dynamic and static
magnetic properties of the nanocomposites over a convenient
temperature range and provided the capability of zero external
field microscopic magnetic study. Mössbauer spectra for the
nanocomposites were recorded as a function of temperature for
4.2 K < T < 300 K using a liquid helium cryostat. They were
computer-fitted to theoretical curves using a nonlinear least-
squares curve fitting procedure.
For nanometer-sized particles, the crystallites could be too
small or poorly developed to be amenable to XRD structural
determination, rendering phase identification of the iron oxide
difficult. No crystalline peaks associated with the iron oxide
phase were observed in the XRD studies of the nanocomposites
in Figure 5, as discussed earlier. We have made use of the
Mössbauer electronic parameters, the isomer shift (δ), the
quadrupole splitting (∆Eo), and the internal hyperfine field (Hhf)-
values in conjunction with the magnetization strength of the
J. Phys. Chem. B, Vol. 105, No. 31, 2001 7417
nanocomposites to determine the iron oxide phase as γ-Fe2O3
and to establish its thermal stability against the γ f R phase
transformation.
Mössbauer Studies. In the cluster size range of 2 to 4 nm, a
large fraction of the iron atoms lies on the cluster surface
(assuming a surface layer comprised of two atomic layers), and
the Mössbauer spectra would reflect the electronic and magnetic
properties of surface iron ions to a large extent. Thus, variations
in surface and strain anisotropy would be reflected in (i) the
quadrupole splitting, (ii) the magnetic hyperfine field, (iii) the
magnetic anisotropy energy, blocking temperature, and magnetic
coercivity, and (iv) the collective magnetic excitations detected
below the blocking temperature. A simple idealized core-shell
model is proposed to analyze the electronic and magnetic
contributions from surface and interior iron cations. We assume
that the highly crystalline core is uniformly magnetized and
could undergo a coherent rotation when an external field (H) is
applied. However, the surface spins are canted relative to the
interior lattice magnetization direction due to strong pinning
forces at the surface, which would induce noncoherent spin
reversal modes.22,23
The Mössbauer spectra of the as-prepared Fe2O3/AlSi-25
nanocomposite #1 at various temperatures are shown in Figure
9. The overall temperature profile is consistent with the
observation of magnetically ordered particles of Fe2O3 exhibiting
dynamic superparamagnetic relaxation, due to thermally acti-
vated reversals of the particle magnetization.24 The dependence
of the superparamagnetic relaxation time on temperature (T),
magnetic anisotropy energy density (K), and particle volume
(V), is given by τ ) τo exp (KV/kT), where k is Boltzmann’s
constant and τo is a constant characteristic of the material.24 At
high temperatures, a superparamagnetic relaxation time shorter
than the Mössbauer measuring time (τ < 10-8 s) would obscure
observation of the internal magnetic order of the particles, giving
rise to a quadrupolar spectrum. At lower temperatures, super-
paramagnetic relaxation would be slowed, revealing the internal
magnetic order of the particles in a magnetically split Mössbauer
spectrum.
At T ) 300 K, the broad absorption lines (Figure 9a) called
for spectral fits with two superimposed quadrupole doublets of
similar isomer shift but different quadrupole splitting values,
as tabulated in Table 2. These two subcomponents were
associated with interior and surface iron sites according to our
proposed core-shell model for the analysis of electronic and
magnetic contributions from interior and surface atoms. The
larger quadrupole splitting (1.25 mm/s) was associated with
surface iron atoms due to ligand coordination distortion at the
surface. A ratio of Asurface:Acore ) 39:61 was obtained, as
measured by the absorption area under each subspectrum.
The spectra at lower temperatures were similarly fitted using
the core-shell model assuming a distribution of hyperfine fields.
At 4.2 K, where only magnetically split spectra were observed
(Figure 9f), two magnetic subsites were poorly resolved. The
smaller hyperfine field (see Table 2) compared to the corre-
sponding values of bulk γ-Fe2O3 (496 kOe at room tempera-
ture25 and 527 kOe saturation field at 0 K26) was associated
with the surface atoms. It has been known that the sharp
discontinuity of the Fe-O-Fe superexchange bonds and canted
spin structure on the surface would result in diminished magnetic
exchange. The contribution from surface spins was greatly
suppressed at 4.2 K as Asurface:Acore ) 21:79. We attribute this
to the frozen core spin structure and persistent surface spin
fluctuations. At intermediate temperatures, the spectra exhibit
greater complexity due to the superposition of quadrupolar and
7418 J. Phys. Chem. B, Vol. 105, No. 31, 2001 Zhang et al.

Figure 7. Elemental mappings and STEM images of as-prepared Fe2O3/AlSi-25 nanocomposite #1.

Figure 8. Elemental mappings and STEM images of Fe2O3/AlSi-25 nanocomposite #1 calcined at 500 °C.

magnetic subspectra arising from the presence of a distribution
of particle sizes.
The observed temperature dependence of the hyperfine fields
(Hhfl and Hhf2) in Table 2 indicated that below the blocking
temperature, where spin reversals were suppressed, the mag-
netization vectors of the particles might still undergo collective
magnetic excitations around the easy axis of magnetization.27
These low-energy excitations could also prevent the observed
hyperfine fields from reaching their saturation values above 0
K. In addition, the surface fields (Hhf2) exhibited stronger
temperature dependence compared to the core fields (Hhfl)
presumably due to surface-spin fluctuation modes.28
The cluster size distribution can be obtained from the fraction
f(T) of the clusters with volume greater than the critical volume
(Vc) defined by superparamagnetism.29 The plot of the fraction
of particles contributing to the magnetic subspectrum as a
function of temperature exhibited two inflection points (see
Figure 10). This suggested that there were two distinct blocking
temperatures as determined by the inflection points, TB1 ) 60
K and TB2 ) 110 K, corresponding to two particle size ranges
in the as-prepared Fe2O3/AlSi-25 nanocomposite #1. This was
in agreement with our STEM and point X-ray probe mapping
results, which indicated the presence of spherical and elongated
particles. We could associate the lower blocking temperature
Synthesis and Properties of γ-Fe2O3 Nanoclusters J. Phys. Chem. B, Vol. 105, No. 31, 2001 7419

Figure 9. Mössbauer spectra of as-prepared Fe2O3/AlSi-25 nanocomposite #1 collected at (a) 300 K, (b) 75 K, (c) 60 K, (d) 50 K, (e) 25 K, and
(f) 4.2 K. The solid line is a least-squares fit to the experimental data, assuming a superposition of sites as shown above the experimental points
(see text). The magnetic subcomponent was fitted to a distribution of magnetic hyperfine fields in order to reproduce the observed spectral broadening.

TABLE 2: Mo1 ssbauer Parameters for Fe2O3/AlSi-25 Nanocomposite #1

Fe2O3/AlSi-25 isomer shifta,b δ (mm/sec) quadrupole splittinga ∆Eo (mm/sec) hyperfine field Hhf (kOe)
nanocomposite #1 T (K) surface core surface core Hhf1 Hhf2
300 0.33 0.33 1.25 0.73
75 0.41 0.45 1.30 0.77 428.8 296.4
as-prepared 60 0.54 0.42 1.50 0.77 458.3 375.2
50 0.35 0.39 1.50 0.78 466.0 402.6
25 0.46 0.27 1.51 0.94 466.9 411.3
4.2 -0.026 487.9 458.2
calcination at 300 °C 300 0.36 0.34 1.24 0.74
4.2 -0.056 477.4 447.9
calcination at 500 °C 300 0.39 0.35 1.10 0.81
4.2 -0.132 487.5 442.2
bulk γ-Fe2O3 (A+B sites averaged) 298 0.35c 0 496d
a Uncertainties in δ and ∆E are ( 0.03 mm/s and ( 0.05 mm/s, respectively. b Isomer shift is relative to metallic iron at room temperature; 4.2
o
K values were fixed at those obtained at room temperature + 0.12 mm/s second-order Doppler shift. cReference (27). dReference (25).

identification as γ-Fe2O3. In support of such a conclusion, we

Figure 10. The fraction of particles in Fe2O3/AlSi-25 nanocomposite
#1 contributing to the Mössbauer magnetic subspectrum as a function
of temperature. The arrows show the two distinct transitional regions.
with a distribution of spherical particles, and the higher blocking
temperature with a distribution of larger, elongated particles.
The Mössbauer parameters of the as-prepared Fe2O3/AlSi-
25 nanocomposite #1 were consistent with the iron oxide phase
observed (i) the absence or negligible quadrupolar perturbation
on the magnetic spectra and (ii) the reduced values of the
hyperfine fields. Hematite (R-Fe2O3) exhibits a large quadrupolar
perturbation of 0.12 mm/s at 300 K and -0.22 mm/s below the
Morin transition (T ) 260 K for bulk R-Fe2O3), and a saturation
hyperfine field (Hsat) of 544 kOe;30 as opposed to zero
quadrupolar perturbation and a saturation field of 527 kOe for
γ-Fe2O3.26 In addition, hematite has a canted anti-ferromagnetic
structure with very weak magnetization. The SQUID measure-
ments discussed below indicated that these nanocomposites
possessed considerable magnetization, consistent with the fer-
rimagnetic spinnel structure of γ-Fe2O3.
The geometrical confinement of the Fe2O3 clusters by the
regular pore structure of the inorganic AlSi-25 matrix could
potentially stabilize the nanoparticles at high temperatures. The
thermal stability of Fe2O3/AlSi-25 nanocomposite #1 samples
calcined at 300 and 500 °C was examined by Mössbauer
spectroscopy. The values for the isomer shift and quadrupole
splitting of both the surface and core components in the
Mössbauer spectrum at 300 K remained essentially unchanged
7420 J. Phys. Chem. B, Vol. 105, No. 31, 2001
Figure 11. Mössbauer spectra collected at 4.2 or 300 K for Fe2O3/
AlSi-25 nanocomposite #1 calcined at (a) 300 °C and (b) 500 °C under
O2. The solid line is a least-squares fit to the experimental data,
assuming a superposition of sites as shown above the experimental
points (see text). The magnetic subcomponent was fitted to a distribution
of magnetic hyperfine fields in order to reproduce the observed spectral
broadening.
for the nanocomposite calcined at 300 °C (Figure 11a, Table
2), indicating that no phase transition took place at this
calcination temperature. Different quadrupole splitting values
were obtained for both components for the sample calcined at
500 °C, suggesting an improved crystallinity due to atomic
restructuring at this calcination temperature (Figure 11b, Table
2). The Mössbauer spectrum at 4.2 K gave a small quadrupole
splitting perturbation value (-0.056 mm/s) for the sample
calcined at 300 °C, supporting the retention of the γ-Fe2O3
phase. The 4.2 K spectrum of the sample calcined at 500 °C,
however, showed a large quadrupole splitting perturbation value
of -0.132 mm/s, which may be interpreted as a partial phase
transformation from γ-Fe2O3 to R-Fe2O3 at this calcination
temperature (∆Eo of pure R-Fe2O3 ) -0.22 mm/s at T ) 80
K).30
SQUID Studies. The isothermal magnetization of as-prepared
Fe2O3/AlSi-25 nanocomposite #1 was obtained at two different
temperatures (Figure 12). The hysteresis loop at 300 K gave a
zero remanence (Mr) and coercivity (Hc), indicating superpara-
magnetic behavior. With increasing external magnetic field, the
magnetization increased rapidly without saturation up to 30 kG
Zhang et al.
Figure 12. Plots of magnetization vs applied field for as-prepared
Fe2O3/AlSi-25 nanocomposite #1 obtained at (a) 300 K and (b) 5 K.
Figure 13. Temperature dependence of the magnetization curves under
(a) field cooling and (b) zero field cooling for as-prepared Fe2O3/A1Si-
25 nanocomposite #1.
due to both spin canting in the ultrafine particle assembly31 and
distribution in the particle sizes. The lack of magnetic saturation
became more significant at low temperatures. The onset of
hysteresis occurred at ∼50 K. At 5 K, a symmetric hysteresis
loop was observed with an unsaturated magnetization of 57
emu/g at 30 kG applied field, a remanence of 1.5 emu/g and a
coercivity of 680 G, demonstrating the ferrimagnetic nature of
the nanocomposite at a low temperature. The spin canting
resulting from the discontinuity of the superexchange bonds
between Fe cations was responsible for the lower magnetization
(compared to bulk γ-Fe2O3: Ms ) 76 emu/g) and the lack of
saturation for the nanocomposite.32 Compared to bulk γ-Fe2O3
that has a remanence of 63 emu/g33 and a coercivity of 250-
400 G,34 the smaller remanence and larger coercivity of the
nanocomposite could be attributed to the single magnetic domain
clusters. For single-domain particles, coercivities are associated
with the rotation of the magnetization away from the easy axis;
whereas in the bulk system, coercivities reflect magnetic domain
wall motion in the presence of a magnetic field.22
The temperature-dependent magnetization in an applied field
of 200 G was investigated under field cooling (FC) and zero
field cooling (ZFC) (Figure 13). The ZFC curve (Figure 13b)
exhibited a maximum at 20 K and a superimposed contribution
to the magnetization with a maximum at 50 K. The sharp Tmax
at 20 K could be related to small spherical single-domain
magnetic particles. Because the Tmax is closely related to the
blocking temperature and increases with increasing cluster size
Synthesis and Properties of γ-Fe2O3 Nanoclusters
Figure 14. Temperature dependence of the magnetization curves under
(a) field cooling and (b) zero field cooling for Fe2O3/AlSi-25 nano-
composite #1: as-prepared, or calcined to 300 °C and 500 °C.
Figure 15. UV-vis diffuse reflectance spectra of Fe2O3/AlSi-25
nanocomposite #1: (a) as-prepared, (b) calcined at 300 °C, and (c)
calcined at 500 °C.
and magnetic anisotropy constant, the broad Tmax at 50 K could
be attributed to elongated particles with larger average particle
volume and shape anisotropy. These blocking temperatures are
lower than those observed with Mössbauer spectroscopy due
to the longer measuring time associated with SQUID measure-
ments.
SQUID magnetometry was also applied to study the magnetic
properties of the calcined samples. The magnetization vs
temperature curves measured under FC and ZFC at 200 G
showed that the total magnetization increased with calcination
temperature (Figure 14). The splitting between FC and ZFC
curves occurred at a lower temperature for samples calcined at
a higher temperature. The improved crystallinity of the Fe2O3
core and the reduced contribution from disordered surface spins
gave rise to the increase in magnetization. The partial trans-
formation to R-Fe2O3, observed in the Mössbauer spectrum of
the 500 °C-calcined sample, was not significant enough to lead
to a reduction in the overall magnetization. This study indicated
that Fe2O3/AlSi-25 nanocomposite #1 would be an excellent
candidate for applications requiring high-temperature stability.
Optical Properties: The optical property of the Fe2O3/A1Si-
25 nanocomposites was studied using a Cary II UV-vis
spectrometer. The spectrum of the as-prepared Fe2O3/AlSi-25
nanocomposite #1 exhibited a blue shift of 70 nm relative to
bulk γ-Fe2O3 (Figure 15a). The widening of the band gap was
consistent with quantum confinement. Upon calcination, the
absorption edge shifted to slightly higher wavelengths (Figures
15b and c), corresponding to changes in (i) the core structure
of the Fe2O3 nanoclusters, and (ii) the interfacial effects from
the particle-support interaction.
Comparison of Fe2O3/SiO2-Based Nanocomposites Syn-
thesized via Different Routes. The electronic, magnetic, and
J. Phys. Chem. B, Vol. 105, No. 31, 2001 7421
TABLE 3: Summary of Magnetic Properties of the Fe2O3/
SiO2-based Nanocomposites
M Hc TB TB
as-prepared (emu/g Fe2O3) (G) (K) (K)
sample (at 3 T, 5 K) (at 5 K) (SQUID) (Mössbauer)
Fe2O3/ 30 270 10 25
Sulfonated-SiO2
nanocompositea
Fe2O3/ 20 635 15 23
SiO2-Coated
nanocompositeb
Fe2O3/AlSi-25 57 680 20, 50 60, 110
nanocomposite #1
a Reference (11). b Reference (12).
optical properties of the Fe2O3/AlSi-R nanocomposites reflected
the nanometer-scaled dimensions of the magnetic constituents,
the abundance of interfaces, and particle-support interactions.
These effects may be further brought to focus by a comparison
of the properties exhibited herewith and those for Fe2O3/
sulfonated-SiO2 nanocomposite and Fe2O3/SiO2-coated nano-
particles reported in our previous studies.11,12
The magnetic properties of the three differently prepared
Fe2O3/SiO2-based nanocomposites in our current and previous
studies are summarized in Table 3. A complex magnetic system
was obtained within the cylindrical mesopores of MCM-41
matrix using the evaporation-condensation approach described
in this study. Large magnetization and coercivity values were
obtained using the AlSi-25 supports (bulk γ-Fe2O3: 76 emu/g,
polymer supported γ-Fe2O3 nanoclusters: 46 emu/g,35 block
copolymer supported γ-Fe2O3 nanoclusters: 39 emu/g36). This
could be explained by the combination of magnetic contributions
from particles of different morphologies. For single magnetic
domain particles, the magnetization is significantly reduced due
to the quantum size effect. The magnetic dipole moment is
proportional to the particle volume. The combined effective
volume of the spherical and elongated Fe2O3 nanoclusters in
the cylindrical pores of AlSi-25 could be larger than that of the
spherical SiO2-coated Fe2O3 particles with diameters of ∼4 nm.
Therefore, a larger magnetization was obtained with the Fe2O3/
AlSi-25 nanocomposite. The coercivity of spinel-structured
nanocrystals greatly depends on morphology and surface
anisotropy. Compared to spherical particles, shape anisotropy
in acicular particles can produce a higher coercive field,
assuming an ellipsoid model with uniform magnetization. The
higher coercivity of Fe2O3/AlSi-25 nanocomposite #1 compared
with that of Fe2O3/SiO2-coated nanocomposite could be at-
tributed to the presence of elongated particles in the former.
On the other hand, the low coercivity in the Fe2O3/sulfonated-
SiO2 nanocomposite implied that (i) the internal spin structure
of the particles in this system significantly deviated from that
of uniform magnetization assumed in the Stoner and Wohlfarth
model37 and/or (ii) the anisotropy was dominated by strain
exerted by the sulfonated SiO2 matrix on the nanoclusters.
The optical properties of the three nanocomposites are
compared in Figure 16. The matrix-mediated synthesis of Fe2O3
nanoclusters within sulfonated SiO2 gel involved an intercon-
nected porous host with a relatively broad pore size distribution.
This might have resulted in relatively less stress on the magnetic
particles, and thus, the largest blue shift (110 nm) was obtained
with the Fe2O3/sulfonated-SiO2 nanocomposite (Figure 16a). In
the case of the Fe2O3/SiO2-coated sample, the presence of high
stress exerted by the SiO2 coating might have substantially
counteracted the quantum confinement effects, resulting in a
relatively small blue shift of 20 nm (Figure 16b). The strain
induced by particle-support interactions could significantly
7422 J. Phys. Chem. B, Vol. 105, No. 31, 2001
Figure 16. UV-vis diffuse reflectance spectra of (a) Fe2O3/sulfonated-
SiO2 nanocomposite, (b) Fe2O3/SiO2-coated nanocomposite, and (c)
Fe2O3/AlSi-25 nanocomposite #1.
distort the particle’s crystal lattice as to produce red-shifted
optical absorption as reported for a γ-Fe2O3/polymer nanocom-
posite system.35 The hexagonally packed array of cylindrical
mesopores (2.5 nm pore diameter) of the AlSi-25 support could
induce more stress on the surface of Fe2O3 clusters than the
irregular pore structure of the sulfonated SiO2 gel (4 nm average
pore size) but less stress than that imposed by the silica coating.
The combined quantum confinement and stress effects led to
the intermediate level of optical blue shift (70 nm) in Fe2O3/
AlSi-25 nanocomposite #1 (Figure 16c).
Conclusions
Evaporation-condensation of volatile Fe(CO)5 was used to
produce Fe2O3 nanoclusters within a hexagonally packed
mesoporous aluminosilicate matrix. Various Fe loadings could
be obtained through control of cluster synthesis parameters.
Superparamagnetism was demonstrated for the Fe2O3/AlSi-25
nanocomposites at low Fe loadings. TEM and STEM with Fe
elemental mapping indicated the coexistence of spherical and
elongated iron oxide particles within the cylindrical pores of
AlSi-25 supports. The SQUID and Mössbauer properties were
consistent with γ-Fe2O3. Two distinct relaxation phenomena
were observed in both SQUID and Mössbauer measurements,
associated with the presence of two sets of particle size
distributions. The large magnetization and coercivity of Fe2O3/
AlSi-25 nanocomposite can be attributed to the presence of
spherical and elongated magnetic particles. The Fe2O3 nano-
clusters in this nanocomposite exhibited high stability against
γ f R phase transition. Only a partial phase transition to
R-Fe2O3 was detected by Mössbauer spectroscopy for the sample
calcined at 500 °C. The high thermal stability and increased
magnetization from calcination make this material an excellent
candidate for high-temperature applications. A blue shift in the
absorption edge was noted in the optical spectrum of this
nanocomposite system.
Acknowledgment. This work was supported by the Camille
and Henry Dreyfus Foundation. The authors thank M. Frongillo
and A. G. Garratt-Reed (MIT CMSE) for their assistance in
the TEM and STEM studies. G.C.P. thanks the Centro Brasileiro-
de Pesquisas Fı́sicas for support during her Visiting Scientist
appointment in Brazil. The authors would also like to acknowl-
edge Drs. R. Scorzelli and A. Bustamante Dominguez of the
Centro Brasileiro de Pesquisas Fı́sicas for the low-temperature
Mössbauer measurements.
Zhang et al.
References and Notes
(1) (a) Gletier, H. Prog. Mater. Sci. 1989, 33, 233. (b) Tronc, E.; Jolivet,
J. P. In Nanophase Materials: Synthesis-Properties-Applications; Hadji-
panayis, G. C., Siegel, R. W., Eds.; Kluwer: Boston, 1994; p 21. (c) Gleiter,
H. Mater. Sci. Forum 1995, 189-190, 67. (d) Shi, J.; Gider, S.; Bakcock,
K.; Awschalom, D. D. Science 1996, 271, 937. (e) Greer, A. L. Mater. Sci.
Forum 1998, 269-272, 3.
(2) (a) Thomas, J. M. Pure Appl. Chem. 1988, 60, 1517. (b) Ying, J.
Y. AIChE J. 2000, 46, 902. (c) Gates, B. C. Chem. ReV. 1995, 95, 511.
(3) (a) Simon, A. Angew. Chem., Int. Ed. Engl. 1988, 27, 159. (b)
Chien, C. L. Annu. ReV. Mater. Sci. 1995, 25, 129. (c) Leslie-Pelecky, D.
L.; Rieke, R. D. Chem. Mater. 1996, 8, 1770. (d) Schmid, G.; Hornyak, G.
L. Curr. Opin. Solid State Mater. Sci. 1997, 2, 204.
(4) (a) Valokitin, Y.; Sinzig, J.; DeJongh, L. J.; Schmid, G.; Moiseev,
I. I. Nature 1997, 384, 621. (b) Luth, H. Appl. Surf. Sci. 1998, 130-132,
855. (c) Chen, W.; Ahmed, H. AdV. Imaging Electron Phys. 1998, 102, 87.
(5) (a) Simon, U.; Schön, G. Angew. Chem., Int. Ed. Engl. 1993, 32,
250. (b) Gehr, R. J.; Boyd, R. W. Chem. Mater. 1996, 8, 1807.
(6) Berkowitz, A. E. In Nanomaterials: Synthesis, Properties and
Applications; Edelstein, A. S., Cammarata, R. C., Eds.; Institute of Physics
Publishing: Bristol, 1996; p 569.
(7) (a) Morrish, A. H. The Principles of Magnetism; Wiley: New York,
1996; Chapter 7. (b) Stucky, G. D.; MacDougall, J. Science 1990, 247,
669. (c) Gunther, L. In Magnetic Properties of Fine Particles; Dormann, J.
L.; Fiorani, D., Eds.; Elsevier: Amsterdam, 1992; p 213. (d) Dormann, J.
L.; Fiorani, D.; Tronc, E. NATO ASI Series E 1994, 260, 635.
(8) (a) Golden, J. H.; Deng, H.; DiSalvo, F. J.; Fréchet, J. M. J.;
Thompson, P. M. Science 1995, 268, 1463. (b) Ziolo, R. F.; Giannelis, E.
P.; Shull, R. D. Nanostr. Mater. 1993, 3, 85. (c) Nguyen, M. T.; Diaz, A.
F. AdV. Mater. 1994, 6, 858.
(9) (a) Shull, R. D.; Ritter, J. J.; Shapiro, A. J.; Swartzendruber, L. J.;
Bennett, L. H. J. Appl. Phys. 1990, 67, 4490. (b) Shull, R. D.; Kerch, H.
M.; Ritter, J. J. J. Appl. Phys. 1994, 75, 6840. (c) Wang, J. P.; Luo, H. L.
J. Appl. Phys. 1994, 75, 7425.
(10) (a) Bein, T.; Schmiester, G.; Jacobs, P. A. J. Phys. Chem. 1986,
90, 4851. (b) Herron, N.; Wang, Y.; Eddy, M.; Stucky, G. D.; Cox, D. E.;
Moller, K.; Bein, T. J. Am. Chem. Soc. 1989, 111, 530. (c) Lazaro, F. J.;
Lopez, A.; Larrea, A.; Pankhurst, Q. A.; Nieto, J. M.; Corma, A. IEEE
Trans. Magn. 1998, 34, 1030.
(11) Zhang, L.; Papaefthymiou, G. C.; Ziolo, R. F.; Ying, J. Y. Nanostr.
Mater. 1997, 9, 185.
(12) Zhang, L.; Papaefthymiou, G. C.; Ying, J. Y. J. Appl. Phys. 1997,
81, 6892.
(13) (a) Kresge, C. T.; Leonowicz, M. E.; Roth, W. J.; Vartuli, J. C.;
Beck, J. S. Nature 1992, 359, 710. (b) Huo, Q.; Margolese, D. I.; Ciesla,
U.; Feng, P.; Gier, T. E.; Sieger, P.; Leon, R.; Petroff, P. M.; Schüth, F.;
Stucky, G. D. Nature 1994, 368, 317. (c) Tanev, P. T.; Chibwe, M.;
Pinnavaia, T. J. Nature 1994, 368, 321. (d) Yang, H.; Kuperman, A.;
Coombs, N.; Mamiche-Afara, S.; Ozin, G. A. Science 1996, 379, 703. (e)
Schulz-Ekloff, G.; Rthaisky, J.; Zukol, A. J. Inorg. Mater. 1999, 1, 97. (f)
Ying, J. Y.; Mehnert, C. P.; Wong, M. S. Angew. Chem., Int. Ed. 1999, 38,
56.
(14) MacLachlan, M. J.; Aroca, P.; Coombs, N.; Manners, I.; Ozin. G.
A. AdV. Mater. 1998, 10, 144.
(15) Wu, C.-G.; Bein, T. Chem. Mater. 1994, 6, 1109.
(16) Leon, R.; Margolese, D.; Stucky, G. D.; Petroff, P. M. Phys. ReV.
B 1995, 52, 2285.
(17) Srdanov, V. I.; Alxneit, I.; Stucky, G. D.; Reaves, C. M.; DenBaars,
S. P. J. Phys. Chem. B 1998, 102, 3341.
(18) Agger, J. R.; Anderson, M. W.; Pemble, M. E.; Terasaki, O.; Nozue,
Y. J. Phys. Chem. B 1998, 102, 3345.
(19) (a) Junges, U.; Jacobs, W.; Voiget-Martin, I.; Krutzsch, B.; Schüth,
F. J. Chem. Soc., Chem. Commun. 1995, 2283. (b) Corma, A.; Martı́nez,
A.; Martı́nez-Soria, V. J. Catal. 1997, 169, 480.
(20) Mehnert, C. P.; Ying, J. Y. Chem. Commun. 1997, 2215. (b)
Mehnert, C. P.; Weaver, D. W.; Ying, J. Y. J. Am. Chem. Soc. 1998, 120,
12 289. (c) Koh, C. A.; Nooney, R.; Tahir, S. Catal. Lett. 1996, 163, 148.
(21) Mumukutla, R. S.; Asakura, K.; Namba, S.; Iwasawa, Y. Chem.
Commun. 1998, 1425.
(22) Cullity, B. D. Introduction to Magnetic Materials; Addison-
Wesley: Reading, 1972; Chapters 9 and 11.
(23) Pankhurst, Q. A.; Pollard, R. J. Phys. ReV. Lett. 1991, 67, 248.
(24) Aharoni, A. In Magnetic Properties of Fine Particles; Dormann,
J. L., Fiorani, D., Eds.; Elsevier: North-Holland, 1992; p 3.
(25) Kelly, H. V.; Folen, V. J.; Hass, M.; Schreiner, W. N.; Beard, G.
B. Phys. ReV. 1961, 122, 1447.
(26) Murad, E. In Magnetic Properties of Fine Particles; Dormann, J.
L.; Fiorani, D., Eds.; Elsevier: North-Holland, 1992; p 339.
(27) Mørup, S.; Topsøe, H. J. Appl. Phys. 1976, 11, 63.
(28) de Jongh, L. J.; Miedema, A. R. AdV. Phys. 1974, 23, 1.
(29) Kaufman, K. S.; Papaefthymiou, G. C.; Frankel, R. B.; Rosenthal,
A. Biochim. Biophys. Acta 1980, 522.
Synthesis and Properties of γ-Fe2O3 Nanoclusters
(30) Greenwood, N. N.; Gibb, T. C. Mössbauer Spectroscopy; Chapman
Hall: London, 1971; p 241.
(31) Coey, J. M. D. Phys. ReV. Lett. 1971, 27, 1140.
(32) Teillet, J.; Bouree, F.; Krishnan, R. J. Magn. Magn. Mater. 1993,
123, 93.
(33) Imaoka, Y. J. Electrochem. Jpn. 1968, 36, 15.
(34) Morrish, H. In Crystals: Growth, Properties, and Applications;
Freyhardt, H. C., Mgr Ed.; Springer-Verlag: New York, Vol. 2, 1980; p
171.
J. Phys. Chem. B, Vol. 105, No. 31, 2001 7423
(35) (a) Ziolo, R. F.; Giannelis, E. P.; Weinstein, B. A.; O’Horo, M. P.;
Ganguly, B. N.; Mehrotra, V.; Russell, M. W.; Huffman, D. R. Science
1992, 257, 219. (b) Vassiliou, J. K.; Mehrotra, V.; Russell, M. W.; Giannelis,
E. P.; McMichael, R. D.; Shull, R. D.; Ziolo, R. F. J. Appl. Phys. 1993, 73,
5109.
(36) Sohn, B. H.; Cohen, R. E.; Papaefthymiou, G. C. J. Magn. Magn.
Mater. 1998, 182, 216;
(37) Stoner, E. C.; Wohlfarth, E. P. Philos. Trans. R. Soc. A 1948, 240,
599.