Vacancy coalescence during oxidation of iron nanoparticles

Article abstract of DOI:10.1021/ja072574a

In the present work, we analyze the geometry and composition of the nanostructures obtained from the oxidation of iron nanoparticles. The initial oxidation of iron takes place by outward diffusion of cations through the growing oxide shell. This net material flow is balanced by an opposite flow of vacancies, which coalesce at the metal/oxide interface. Thus, the partial oxidation of colloidal iron nanoparticles leads to the formation of core-void-shell nanostructures. Furthermore, the complete oxidation of iron nanoparticles in the 3-8 nm size range leads to the formation of hollow iron oxide nanoparticles. We analyze the size and temperature range in which vacancy coalescence during oxidation of amine-stabilized iron nanoparticles takes place. Maghemite is the crystallographic structure obtained from the complete oxidation of iron nanoparticles under our synthetic conditions. Copyright

Full text of DOI:10.1021/ja072574a

Published on Web 08/03/2007
Vacancy Coalescence during Oxidation of Iron Nanoparticles
Andreu Cabot,† Victor F. Puntes,‡ Elena Shevchenko,§ Yadong Yin,§,| Llu´ıs Balcells,
Matthew A. Marcus,¶ Steven M. Hughes,# and A. Paul Alivisatos*,†,§,#
Materials Sciences DiVision, Molecular Foundry, and AdVanced Light Source, Lawrence Berkeley National
Laboratory, Berkeley, California 94720, Institut Catala` d’ Estudis i Recerca AVanc¸at and Institut Catala´ de
Nanotecnologia, Bellaterra 08193, Barcelona, Spain, Institut de Cie`ncia de Materials de Barcelona, (ICMAB-CSIC),
Campus UAB, Bellaterra 08193, Spain, and Department of Chemistry, UniVersity of California at Berkeley,
Berkeley, California 94720
Received April 12, 2007; E-mail: alivis@berkeley.edu
Iron oxide nanoparticles are widely studied because they occur
naturally, they are readily synthesized artificially, and they have
interesting chemical and magnetic properties as well as applications
in in vivo magnetic imaging. One of the most established ap-
proaches to producing such particles is to first prepare elemental
iron nanoparticles and subsequently oxidize them. A wide range
of morphologies and compositions have been observed when iron
nanoparticles are oxidized, ranging from iron/iron oxide core-shell
structures1 to iron/iron oxide core-void-shell structures,2 iron
oxide solid spheres,3,4 or even iron-based hollow structures.2,5 There
is still no clear picture of the diffusion processes that accompany
the chemical transformations from an elemental nanoparticle of iron
to the corresponding oxides. Here we demonstrate that there is a
large intermediate temperature regime (T < 250 °C), in which
hollow iron oxide nanoparticles of 4-11 nm are spontaneously
formed during iron nanoparticle oxidation in solution, due to the
nanoscale Kirkendall effect. These observations can help to
systematize our understanding of the temperature dependence of
the inward and outward diffusion processes in iron nanoparticle
oxidation, enabling improved morphology and composition control.
Iron nanoparticles can be obtained by decomposition of iron
pentacarbonyl in organic solvents containing amines.2 In order to
prepare iron nanoparticles, 0.4 mL of Fe(CO)5 was injected at
200 °C into 10 mL of air-free octadecene (C18H38) containing 0.67
mmol of oleylamine under vigorous stirring. The resulting solution
was reacted for 20 min. By changing the reaction parameters, the
particle size could be effectively tuned (Supporting Information,
SI). As synthesized, iron nanoparticles are quasi-amorphous, as
indicated by X-ray diffraction (XRD) analysis. X-ray absorption
spectroscopy (XAS) studies identify iron as their main component.
A shift of the spectrum to lower energies points toward the presence
of some carbon in the iron particles (SI).6
In order to oxidize iron nanoparticles, we flow a dry 20% oxygen
mixture in argon through the colloidal solution (20 mL/min). Figure
1 shows the conversion of iron into iron oxide nanoparticles under
a range of temperatures and times. The formation of a first thin
oxide layer on the iron surface is very rapid, even at room
Figure 1. TEM micrographs of iron/iron oxide nanoparticles exposed to
temperature. Iron nanoparticles processed in air-free conditions, but
exposed to the atmosphere only during rapid transfer to a grid and
introduction into a transmission electron microscope (TEM), already
show a thin oxide shell (Figure 1A). The subsequent oxide
dry 20% oxygen: (A) <1 min at room temperature; (B) 1 h at 80 °C; (C)
12 h at 80 °C; (D) 5 min at 150 °C; (E) 1 h at 150 °C; (F) 1 h at
350 °C on a substrate; (G,H) high-resolution of partial and full oxidized
iron nanoparticles. Low- and high-resolution scale bars correspond to 100
and 6 nm, respectively. Two-dimensional projections of the cross sections
for electron scattering in iron/iron(III) oxide core-void-shell nanospheres
are shown as insets. The simulated evolution of the particle size, core
diameter, and shell thickness corresponds to an oxide growth at the oxide/
solution interface.
† Materials Science Division, Lawrence Berkeley National Laboratory.
‡ Institut Catala` d’ Estudis i Recerca Avanc¸at & Institut Catala´ de Nanotecno-
logia.
§ Molecular Foundry, Lawrence Berkeley National Laboratory.
Institut de Cie`ncia de Materials de Barcelona.
¶ Advanced Light Source, Lawrence Berkeley National Laboratory.
# University of California at Berkeley.
formation is substantially slower; estimations based on the Cabrera-
Mott and Fromhold-Cook models7 have shown that the growth of
4 nm thick oxide layers in iron films can take ∼600 years at room
| Current address: Department of Chemistry, University of California at
Riverside, Riverside, CA 92521. E-mail: yadong.yin@ucr.edu.
9
10358
J. AM. CHEM. SOC. 2007, 129, 10358-10360
10.1021/ja072574a CCC: $37.00 © 2007 American Chemical Society

C O M M U N I C A T I O N S
Figure 2. TEM micrographs of iron nanoparticles with different sizes oxidized at 250 °C in solution (A-E), and at 350 °C while supported on a substrate
(F). Scale bar corresponds to 100 nm.
temperature.8 This process can be accelerated by taking advantage
of the exponential temperature dependence of the iron diffusivities.
As a result, the reaction temperature and oxidation time allow to
precisely tune the thickness of an initial oxide shell. The second
stage of the oxidation process can be seen clearly in Figure 1B-F.
Of particular interest here is the clear observation, even at very
early stages, of a very thin low-density region between the initial
iron particle and the outer oxide layer. This is due to the coalescence
of voids at the interface and is a clear signature of iron diffusing
outward through the initial oxide shell, leaving vacancies behind,
in the nanoscale Kirkendall effect.9,10 TEM micrographs of the
partially oxidized particles show three differentiated contrast regions
(Figure 1). The darker inner region corresponds to the iron core.
The outermost shell, corresponding to a lower-density material, is
the iron oxide. As the chemical transformation proceeds, the oxide
shell grows thicker due to the continual appearance and subsequent
oxidation of iron atoms on the outermost surface of the oxide. The
disappearance of the spherical iron core in the center of the hollow
particle can be clearly observed. When iron atoms diffuse outward,
the vacancies left behind ultimately coalesce into a single central
void (Figure 1F). These results are consistent with a previous study
on the room-temperature oxidation of iron clusters supported on
carbon grids.8 The nanoscale Kirkendall effect was initially
described for reactions of Co nanoparticles with oxygen, sulfur,
and selenium.10 Careful TEM investigation of cobalt nanoparticles
oxidized with selenium revealed thin filamentous connections
between the shell and the central metal particle. These metal bridges
provide a fast transport path for diffusion of metal atoms from the
core to the oxide shell.
contact potential of the chemisorbed oxygen results in a strong
electric field able to drive the ion diffusion.7,11,12 At temperatures
below 150 °C, the dominant electron transport mechanism is by
tunneling through the thin oxide layer. Electron tunneling is
operative only up to a thickness of ∼1-3 nm.11,12 Above this
temperature, an additional electron supply is provided by thermoe-
mission of electrons from the iron into the oxide conduction band.12
This additional source of electrons allows the oxide growth to
proceed further, up to ∼10 nm thick oxide layers.
The location of the iron-oxygen reaction front is determined
by the relative diffusion rates of iron and oxygen ions, as well as
the shell thickness, microstructure, and temperature. Our experi-
mental results show the coalescence of vacancies at the metal/oxide
interface, which points toward an initial net outward material flow.
This observation is consistent with faster outward diffusion of iron
cations than inward flow of oxygen anions.8 The oxide growth
strongly slows down for shells thicker than ∼2.0 ( 0.4 nm at
temperatures lower or equal to 250 °C. In this temperature range,
only iron particles smaller than ∼8 nm could be completely
converted into hollow oxide particles. The oxidation rate of larger
particles becomes imperceptible for ∼3 ( 0.4 nm thick shells,
trapping a core inside the oxide shell (Figure 2B-E).
When heating the particles in solution at a higher temperature
(T ∼ 250-300 °C), cracks are formed on the oxide shells (SI).
The growth of the crystal domains creates large stresses that lead
to the shell fragmentation. The temperature range at which cracks
are formed depends on the shell thickness and thus on the particle
size. In large iron nanoparticles, the shell brakes before the oxidation
of the whole particle is complete. The shell fragmentation allows
the solvent and oxygen to reach and oxidize the core. No such
cracks are observed upon heating of the same particles on a substrate
in the absence of a solvent. In this way, the oxidation of iron
nanoparticles at 350 °C leads to the formation of hollow iron oxide
nanoparticles with diameters as large as 20 nm, and ∼4-5 nm thick
shells (Figure 2F).
It is interesting to consider the range of behaviors which can be
expected from this system as a function of the temperature and the
nanocrystal size. Dissociative adsorption of oxygen, followed by
place-exchange with the surface iron is considered the first step in
the oxidation of iron surfaces.11 Once an initial oxide layer is
formed, electron transport from the metal core to the adsorbed
oxygen through the oxide layer allows oxidation to carry on. The
9
J. AM. CHEM. SOC. VOL. 129, NO. 34, 2007 10359

C O M M U N I C A T I O N S
The formation of hollow iron oxide nanoparticles under the TEM
electron beam when analyzing Fe nanoparticles stabilized by
trioctylphosphine oxide has been reported.4 To exclude such an
effect on our particles during TEM characterization, the particles
at different oxidation stages were kept under the electron beam
during periods of more than 30 min. No electron-beam-induced
changes were observed in any of the amine-stabilized particles we
analyzed (SI).
Acknowledgment. This work was supported by the Director,
Office of Science, Office of Basic Energy Sciences, Materials
Sciences and Engineering Division, of the U.S. Department of
Energy under Contract No. DE-AC02-05CH11231 and by the
DARPA/AFOSR DURINT Program Grant No. F49620-01-1-0474.
A.C. thanks the Generalitat de Catalunya, Departament d’Universitats,
Recerca i Societat de l’Informacio´ for financial support. V.F.P.
thanks financial support from MAT2006-13572-C02-02. We thank
Prof. J. Long and his group for the assistance and use of their
SQUID.
As indicated by high-resolution TEM (HRTEM), the oxide shells
around the colloidal Fe particles are polycrystalline (Figure 1F).
Crystal domains extend across the entire shell. However, no
preferential orientation of the lattices across the shell could be
deduced from our data. Regarding the crystal structure, while XRD
data allows us to safely exclude the presence of the hematite (R-
Fe2O3) phase, the line broadening associated with the small crystal
size domains makes it impossible to distinguish between magnetite
(Fe3O4) and maghemite (γ-Fe2O3) phases. Nonetheless, the X-ray
absorption near edge structure (XANES) spectra of the particles
oxidized in the temperature range between 200 and 250 °C resemble
that of a maghemite reference sample (SI). Oxidized nanoparticles
show no band at the pre-edge corresponding to Fe2+ and a shift of
the absorption spectrum to higher energies than that obtained for
magnetite.13 Both features allow us to identify the iron oxide phase
of the completely oxidized particles as maghemite. Further,
maghemite is a semiconductor with a 2.03 eV band gap, while
magnetite is a semimetal with a 0.14 eV band gap. Optical spectra
of hollow nanoparticles show the absorption band at wavelengths
lower than ∼600 nm, consistent with the maghemite phase of the
fully oxidized iron nanoparticles (SI).
Supporting Information Available: Detailed synthesis and ex-
perimental procedures, UV-vis, XRD, and XAS of the iron/iron oxide
nanoparticles, SQUID of the final iron oxide shells, histograms of the
particle sizes, and analysis of the electron beam influence during TEM
imaging and of the high-temperature oxidation of the particles in
solution. This material is available free of charge via the Internet at
http://pubs.acs.org.
References
(1) Carpenter, E. E.; Calvin, S.; Stroud, R. M.; Harris, V. G. Chem. Mater.
2003, 15, 3245-3246.
(2) (a) Peng, S.; Wang, C.; Xie, J.; Sun, S. J. Am. Chem. Soc. 2006, 128,
10676-10677. (b) Peng, S.; Sun, S. Angew. Chem., Int. Ed. 2007, 46,
1-5.
(3) Hyeon, T.; Lee, S. S.; Park, J.; Chung, Y.; Na, H. B. J. Am. Chem. Soc.
2001, 123, 12798-12801.
(4) Latham, A. H.; Wilson, M. J.; Schiffer, P.; Williams, M. E. J. Am. Chem.
Soc. 2006, 128, 12632-12633.
(5) Farrell, D.; Majetich, S. A.; Wilcoxon, J. P. J. Phys. Chem. B 2007, 107,
11022-11030.
(6) van Wonterghem, J.; Mørup, S.; Charles, S. W.; Wells, S.; Villadsen, J.
While our experimental results point toward the formation of
maghemite by oxidation of the iron nanoparticles at temperatures
in the range of 200-250 °C, other authors report the total or partial
formation of magnetite by oxidation of iron at lower temperatures
or using other oxygen precursors.2,13,14 Maghemite and magnetite
are both thermodynamically metastable phases. However, the pre-
eminence of the maghemite phase over magnetite is consistent with
kinetic considerations: while the small kinetic barrier for the
transformation of magnetite to maghemite allows this phase
transition to take place at low temperatures, the maghemite to
hematite transformation is inhibited at temperatures lower than
300-400 °C because of the very different crystal structures.13,15
The slight difference between magnetite and maghemite crystalline
lattices might explain their strong dependence on the oxidation
conditions. The presence of water has also an important influence
on the oxidation procedure and the final nanostructures obtained.16
In summary, we detailed a synthetic route, based on the nanoscale
Kirkendall effect, to produce colloidal solutions of monodisperse
iron/iron oxide core-void-shell nanostructures and hollow
maghemite nanoparticles with controllable particle size and shell
thickness. A limitation on the size of the hollow nanostructures
that can be synthesized in solution exists. This depends on the iron
diffusivities and the growth of the crystal size domains in the
polycrystalline oxide shells.
Phys. ReV. Lett. 1985, 55, 410-412.
(7) (a) Fromhold, A. T., Jr.; Cook, E. L. Phys. ReV. 1967, 158, 600-612. (b)
Fromhold, A. T., Jr.; Cook, E. L. Phys. ReV. 1967, 163, 650-664. (c)
Cabrera, N.; Mott, N. F. Rep. Prog. Phys. 1949, 12, 163-184.
(8) Wang, C. M.; Baer, D. R.; Thomas, L. E.; Amonette, J. E.; Antony, J.;
Qiang, Y.; Duscher, G. J. Appl. Phys. 2005, 98, 094308.
(9) Smigelskas, A. D.; Kirkendall, E. O. Trans. AIME 1947, 171, 130-142.
(10) (a) Yin, Y.; Rioux, R. M.; Erdonmez, C. K.; Hughes, S.; Somorjai, G.
A.; Alivisatos, A. P. Science 2004, 304, 711-714. (b) Yin, Y.; Erdonmez,
C. K.; Cabot, A.; Hughes, S.; Alivisatos, A. P. AdV. Funct. Mater. 2006,
16, 1389-1399.
(11) (a) Grosvenor, A. P.; Kobe, B. A.; McIntyre, N. S. Surf. Sci. 2004, 565,
151-162. (b) Grosvenor, A. P.; Kobe, B. A.; McIntyre, N. S. Surf. Sci.
2005, 574, 317-321.
(12) (a) Leibbrandt, G. W. R.; Hoogers, G.; Habraken, F. H. P. M. Phys. ReV.
Lett. 1992, 68, 1947-1950. (b) Graat, P. C. J.; Somers, M. A. J.;
Vredenberg, A. M.; Mittemeijer, E. J. J. Appl. Phys. 1997, 82, 1416-
1422. (c) Roosendaal, S. J.; Vredenberg, A. M.; Habraken, F. H. P. M.
Phys. ReV. Lett. 2000, 84, 3366-3369.
(13) Signorini, L.; Pasquini, L.; Savini, L.; Carboni, R.; Boscherini, F.; Bonetti,
E.; Giglia, A.; Pedio, M.; Mahne, N.; Nannarone, S. Phys. ReV. B 2003,
68, 195423.
(14) Leibbrandt, G. W. R.; Spiekman, L. H.; Habraken, F. H. P. M. Surf. Sci.
1993, 287/288, 250-254.
(15) (a) Schimanke, G.; Martin, M. Solid State Ionics 2000, 136-137, 1235-
1240. (b) Ayyub, P.; Multani, M.; Barma, M.; Palkar, V. R.; Vijayaragha-
van, R. J. Phys. C: Solid State Phys. 1988, 21, 2229-2245.
(16) (a) Grosvenor, A. P.; Kobe, B. A.; McIntyre, N. S. Surf. Sci. 2004, 572,
217-227. (b) Roosendaal, S. J.; Bakker, J. P. R.; Vredenberg, A. M.;
Habraken, F. H. P. M. Surf. Sci. 2001, 494, 197-205.
JA072574A
9
10360 J. AM. CHEM. SOC. VOL. 129, NO. 34, 2007