ArticlePDF Available

Abstract and Figures

We investigated the local order in individual iron nanoparticles (NPs) embedded in carbon nanotubes (CNTs). The NPs directly come from the CNT growth on stainless steel without addition of external metal catalyst. The structural analysis has been obtained through nanoscale transmission extended electron energy loss fine structure (EXELFS) measurements above the iron L2,3 edge. A theoretical simulation of the EXELFS features has been performed within the multiple scattering theory. By comparing the experimental data with the simulations, we found that pure γ-Fe and Fe3C nanoparticles are the catalysts of the CNT synthesis on the stainless steel. Moreover, from the analysis of the fine details of the EXELFS oscillations, we also estimated the value of the fcc Fe NP lattice parameter to be a = 3.61 ± 0.03 Å. This last finding suggests a high magnetic moment of the fcc Fe NPs.
Content may be subject to copyright.
RESEARCH PAPER
Probing the structure of Fe nanoparticles in multiwall
carbon nanotubes grown on a stainless steel substrate
L. Camilli P. Castrucci M. Scarselli
E. Gautron S. Lefrant M. De Crescenzi
Received: 17 May 2013 / Accepted: 8 July 2013 / Published online: 24 July 2013
ÓSpringer Science+Business Media Dordrecht 2013
Abstract We investigated the local order in indi-
vidual iron nanoparticles (NPs) embedded in carbon
nanotubes (CNTs). The NPs directly come from the
CNT growth on stainless steel without addition of
external metal catalyst. The structural analysis has
been obtained through nanoscale transmission
extended electron energy loss fine structure (EXELFS)
measurements above the iron L
2,3
edge. A theoretical
simulation of the EXELFS features has been per-
formed within the multiple scattering theory. By
comparing the experimental data with the simulations,
we found that pure c-Fe and Fe
3
C nanoparticles are the
catalysts of the CNT synthesis on the stainless steel.
Moreover, from the analysis of the fine details of the
EXELFS oscillations, we also estimated the value of
the fcc Fe NP lattice parameter to be a=
3.61 ±0.03 A
˚. This last finding suggests a high
magnetic moment of the fcc Fe NPs.
Keywords Fcc iron Extended electron energy
loss fine structure (EXELFS) spectroscopy Iron
nanoparticles Carbon nanotubes
Introduction
Face-centered cubic (fcc) bulk iron (the so called c-Fe)
phase is paramagnetic and stable at temperatures
between 1,183 and 1,667 K (Massalski and Okamoto
1992). Nevertheless, the possibility to form iron fcc
structures at room temperature has attracted much
interest. This was in part stimulated by some calcu-
lations predicting, for bulk fcc iron with a lattice
parameter, a, between 3.4 and 3.7 A
˚, a ferromagnetic
behavior with atomic moments, for aC3.59 A
˚,
significantly enhanced (C2.5 l
B
/atom) if compared
to the value in bulk body-centered cubic (bcc) iron (a-
Fe) (*2.2 l
B
/atom) (Moruzzi et al. 1986; Bagayoko
and Callaway 1983). This theoretical indication made
the iron fcc structure appealing from the point of view
of potential applications. To obtain such a c-Fe phase,
several attempts have been made: (i) alloying Fe with
other elements such as Ni (like in stainless steel); (ii)
epitaxially growing Fe ultra thin films on Cu (Shen
et al. 1998) or Cu/Si (Gubbiotti et al. 1999) substrates;
(iii) forming fcc Fe precipitates in Cu (Hines et al.
2009)orCu
1-x
Au
x
(Klein et al. 1991) matrix through
heat treatments; (iv) synthesizing c-Fe nanoparticles
and nanowires on a carbon film (Ling et al. 2008,
L. Camilli P. Castrucci (&)M. Scarselli
M. De Crescenzi
Dipartimento di Fisica and Unita
`CNISM, Universita
`
Roma Tor Vergata, Via della Ricerca Scientifica 1,
00133 Rome, Italy
e-mail: paola.castrucci@roma2.infn.it
E. Gautron S. Lefrant
Institut des Mate
´riaux Jean Rouxel (IMN), Universite
´
de Nantes, CNRS, 2 rue de la Houssinie
`re, BP 32229,
44322 Nantes Cedex 3, France
123
J Nanopart Res (2013) 15:1846
DOI 10.1007/s11051-013-1846-4
2009); and (v) very recently embedding iron fcc
nanoparticles in carbon nanotubes (CNT) (Lyubutin
et al. 2012). These last studies are particularly
appealing because (i) it is possible to grow highly
oriented CNTs, leading to anisotropic magnetic
response of the device (Lyubutin et al. 2012), and
(ii) the surrounding CNT walls protect the metallic
nanostructures from the action of ambient oxygen
(Camilli et al. 2011). For these reasons, extensive
investigations have been devoted to CNTs filled with
ferromagnetic metals to be used as superdense mag-
netic recording (Li et al. 2000), as magnetic force
microscope tips (Winkler et al. 2006), magnetic field
sensors or nanocontainers for targeted drug delivery
(Mo
¨nch et al. 2007). Different experimental
approaches have been employed to fabricate metal-
filled or partially filled CNTs: e.g., by means of
capillarity (Borowiak-Palen et al. 2006), catalytic
chemical vapor deposition (CVD) (Lyubutin et al.
2012), and chemical vapor deposition (He et al. 2011).
As a matter of facts, iron nanoparticles filling the
CNTs have been found to possess different nature,
namely a-Fe, Fe
3
C and/or c-Fe, depending on the
synthesis conditions. X-ray diffraction as well as
electron diffraction performed in a transmission
electron microscopy (TEM) apparatus are the most
used techniques to study the structural properties of
these nanoparticles. While XRD investigation pro-
vides information averaged on huge number of
nanoparticles, electron diffraction has the advantage
to give the structure of the individual nanostructures
imaged by TEM. In this framework, a useful comple-
mentary experimental technique is the electron energy
loss spectroscopy (EELS) and its associated extended
electron energy loss fine structure (EXELFS) spec-
troscopy. This technique, like electron diffraction, has
nanoscale spatial resolution, and also allows to
perform a study of the chemical nature and local
atomic ordering in the selected area exposed to the
electron beam spot. In fact, within the limit of the
dipole approximation, the EXELFS oscillations can be
analyzed as the extended X-ray absorption fine
structure (EXAFS) data (Sto
¨hr 1988), thus allowing
to determine the local atomic structure around the
ionized atom, such as bond lengths and coordination
number, and the dynamic disorder (i.e., the atomic
thermal vibration around the equilibrium position). On
the other hand, since the amount of inelastically
scattered electrons drastically decreases with the
increasing energy loss, giving rise to small peaks
superimposed on a strongly decreasing background,
only edges with energies lower than 1,000 eV can
be easily recorded. This means that, in the case of
iron, only the EXELFS oscillations around the L
2,3
edges (above 708 eV) can be investigated. The K
edge (around 7,100 eV) is indeed too far in energy
loss from the elastic peak thus exhibiting a very
low inelastic cross section (*DE
-4
, where
DEbeing the energy loss) (Auerhammer and Rez
1989). However, analyzing the EXELFS (as well as
EXAFS) oscillations over the iron L edges is quite
challenging because of substantial difficulties due
to the spin–orbit splitting of the 2pstate. Studies of
EXAFS spectra performed on thin film of bcc iron
around the Fe L edges have been already published
(Kurde et al. 2007). In the case of nanoparticles,
the rather low number of atoms makes this
experiment a true challenge in terms of signal-to-
noise ratio of the EXELFS spectra. In addition, the
encapsulation of the nanoparticle inside the CNT
walls further reduces the signal to be collected.
Here, we report on an EXELFS study with the
aim at investigating the structure of the metal
nanoparticles embedded in the CNTs that have
catalyzed the CNT growth process. Our analysis of
the experimental EXELFS oscillations ensured us on
the c-Fe phase nature of several nanoparticles
imaged by TEM and allowed us to estimate the
lattice parameter of the fcc iron structure to be
around 3.61 A
˚. Noticeably, such a value of the c-Fe
phase lattice parameter makes them possible candi-
dates to be ferromagnetic at room temperature and
have atomic moment higher than that of bulk a-Fe.
In addition, we were able to detect the presence of
iron carbide nanoparticles, too.
Experimental details
The CNTs are synthesized by chemical vapor depo-
sition from acetylene directly on AISI-316 stainless
steel (Fe 70 %, Cr 18 %, and Ni 10 %). Once the
stainless steel is heated up to 1,003 K, acetylene
(200 sccm) is introduced in the chamber for 10 min in
dynamic conditions, thus allowing the CNT growth.
More details on the synthesis of iron-filled CNTs are
described by Camilli et al. (2011). a-Fe microparti-
cles, used as reference for the EXELFS measurements,
Page 2 of 9 J Nanopart Res (2013) 15:1846
123
have been obtained from an iron powder (from R.
P. Normapur) after reduction in H
2
at 973 K for 10 h
followed by a subsequent rapid introduction in the
TEM apparatus.
Iron-filled CNT samples for TEM experiments
were prepared by scratching the synthesis product
from the stainless steel surface directly on a holey
carbon-coated TEM grid. Imaging and EXELFS data
acquisition were performed with a cold FEG Hitachi
HF2000 transmission electron microscope (TEM)
equipped with a modified Gatan PEELS 666 spec-
trometer. All the EXELFS measurements were
recorded in the same experimental conditions: accel-
erating voltage 100 kV, diffraction coupling mode,
energy dispersion of 0.3 eV/channel, convergence
angle 2.8 mrad, and collection angle 9.1 mrad. This
collection angle was chosen very close to the so-called
‘magic angle’’ (7.8 mrad at 710 eV), so that spectra
do not vary with sample orientation (Jouffrey et al.
2004).
1
Deconvolution was performed with PEELS
Cambridge software (Fallon and Walsh 1996)to
remove the instrument contribution and the plural
scattering effects. TEM images were acquired before
and after each electron energy loss spectroscopy
(EELS) measurement to ensure that the area under
investigation had not suffered any morphological
change and/or damage that may arise from its long
exposure to the high-energy electron beam. Various
sizes of nanoparticles were successfully analyzed
following this procedure. However, EXELFS spectra
with satisfactory signal-to-noise ratio were quite
challenging to obtain on individual small nanoparti-
cles. This difficulty is mainly due to the low amount of
atoms to detect. Moreover, near-edge EELS spectra
were systematically recorded, with suitable energy
range, to detect the presence of any feature of the Cr
and Ni L
2,3
or O K edges. As far as Ni content regards,
very few of the observed iron nanoparticles presented
small traces of Ni. The present EXELFS analysis has
been limited to those nanostructures without Ni
detected in EELS spectra. On the other hand, the
majority of the EELS measurements showed no traces,
within the detection limit, of any oxygen impurities
that might have oxidized the iron nanoparticles.
Data analysis
EELS probes unoccupied electron states above the
Fermi level in the energy range of core edge levels.
This allows measuring the number of scattered
electrons of the primary beam which have excited
core electrons of the medium to the same interference
final state of the X-ray absorption process, assuming
that the dipole transition selection rule dominates the
matrix element of the scattering process (Kincaid et al.
1978). Within this approximation, EXELFS spectra
can be analyzed with the same data procedure as for
the EXAFS ones. Therefore, the experimental EX-
ELFS features, i.e., the fine structure which extends
for several hundreds of electron volts beyond an inner-
shell ionization edge after the excitation of an s (K or
L
1
) core electron or of 2pand 3p(L
2,3
and M
2,3
,
respectively) core electrons, are described in terms of
av(k) function, also called ionization coefficient
(Stern and Rehr 1983), using the EXAFS formula
vkðÞ¼XNjAje2k2r2
je2Rj=ksin 2kRjþUjðkÞ

kR2
j
with jbeing the coordination shell of atoms surround-
ing the excited atom, and N
j
the coordination number
of atoms located at an average distance R
j
from the
central atom. The first exponential term accounts for
the motion of atoms due to thermal vibrations and to
static disorder. The parameter r
j
is the mean-square
relative displacement for the jth shell and the electron
wave vector kis related to the emitted electron energy
by k=H2m
e
(E-E
0
)/h
2
, where E
0
is the edge onset
energy, m
e
is the electron mass, and hthe Planck
constant divided by 2p.kis a phenomenological mean
free path that corresponds to a finite lifetime of the
excited electron state, A
j
(k,p) is the backscattering
amplitude of the atoms in the jth shell, /
j
is the phase
shift experienced by the electron on going through the
potential of the central and backscattering atoms. In
the present paper, we simulated the EXELFS data by
using the FEFF8.2 software that calculates the
extended X-ray absorption fine structure taking into
account the multiple scattering processes (Ankudinov
et al. 1998; Ankudinov et al. 2002).
1
The magic angle was calculated from the Fig. 2 of the article
by Jouffrey et al. (2004): for an incident energy of 100 keV, the
magic angle value is 2.25 times that of the characteristic
scattering angle.
J Nanopart Res (2013) 15:1846 Page 3 of 9
123
For the EXELFS simulations, we used the param-
eters that provided the best fit of the EXAFS spectrum
of the Fe bcc film reported in Kurde et al. (2007).
Figure 1a shows the experimental EXAFS Fe L
2,3
signal taken from Kurde et al. (2007) and the one we
simulated. The good accordance between the two
spectra is clearly visible. This is also highlighted in
their corresponding Fourier transform (FT) reported in
Fig. 1b. Actually, the Fourier transform of the
kv(k) oscillations into the real space allows to visually
obtain information about the average distance trav-
elled by the electron between the absorbing (ionized,
in the EXELFS case) and the various backscattering
atoms (shifted by the scattering phase shift). When
accounting for only single scattering paths, i.e., paths
involving only two scattering atoms, the FT of the
EXAFS (EXELFS) oscillations is strictly related to the
radial pair distribution functions of each scattering
shell. However, even when multiple scattering events
are not negligible, at least the first peak of the FT can
be associated to the first neighbor atom distance
corrected by the phase shift. In particular, starting
from a Fe bcc lattice with a=2.87 A
˚, we found a
good reproduction of the experimental data consider-
ing a cluster of radius R=6A
˚, all paths with at least a
mean amplitude 3.5 % of largest path (namely, the one
relative to the nearest neighbor atom) and a maximum
of four scattering atoms.
Results and discussion
Figure 2reports the TEMimages of three nanoparticles.
The lowest panel (Fig. 2a) shows a a-Fe microparticle,
while the images in Fig. 2b, c regard two nanoparticles
encased in carbon nanotubes. Figure 3shows the
corresponding collected EELS spectral features. The
spectra are dominated by two intense and few eV wide
peaks due to the transition from the Fe 2p
3/2,1/2
core
levels to s and d final unoccupied states. These structures
are the so-called white lines corresponding to theL
3
and
L
2
edges, respectively. The other features (i.e., oscilla-
tions) above 740 eV correspond to the scattering
experienced by the 2p
3/2,1/2
electrons in their final
unoccupied states. They extend for several hundreds of
eV above the edges. As a consequence, the experimental
signal is a superposition of the L
3
and L
2
spectra. The L
2
component is shifted in energy with respect to the L
3
one, with an energy separation DE
SO
=-13.1 eV,
arising from spin–orbit splitting. Figure 4shows the
corresponding EXELFS signals, kv(k), for the three
nanoparticles after background subtraction and edge
Fig. 1 a bcc Fe film kv(k)FeL
2,3
signals measured through
EXAFS (lower, red line) and that calculated (upper, black line)
for Fe bcc lattice with a=2.87 A
˚considering a cluster of radius
R=6A
˚, all paths with at least mean amplitude 3.5 % of largest
path and a maximum of four scattering atoms. bCorresponding
Fourier transform obtained for a k-window ranging between 3
and 11.5 A
˚
-1
. EXAFS oscillations are adapted with permission
from Fig. 2of Kurde et al. (2007), copyright (2007) by the
American Physical Society. (Color figure online)
Page 4 of 9 J Nanopart Res (2013) 15:1846
123
step normalization of the experimental spectra reported
in Fig. 3,overakrange extending up to about 9.0 A
˚
-1
.
The three oscillatory features present many differences,
suggesting that the two iron nanoparticles embedded in
the carbon nanotube could not have a bcc structure.
Moreover, some differences can be also observed
especially at the high oscillation frequency (k[
6A
˚
-1
) between the two EXELFS signals associated
to the iron NPs inside the nanotubes. This should mean
that the major differences between the two structures
involve atoms close to the ionized one. In order to givea
closer look at the structure of the nanoparticles, we
performed simulations of the EXELFS oscillations
starting fromknown crystallographiclattices. In order to
take into account for the presence of the signals due to
both the L
3
and L
2
edges, we calculated the total
ionization coefficient v(E)asfollowing:
vEðÞ¼2=3vL3EðÞþavL3EþDESO
ðÞ

where vL3(E) is the L
3
ionization coefficient and the
right part of the equation represents the L
2
one. The
intensity ratio, a, between the two signals has been
assumed equal to , on the basis of 2j ?1 degeneracy
in a one-electron model.
2
Figure 5, lower and upper
panels, shows the comparison between the experi-
mental and the best simulated v(k) oscillations of a-
iron microparticle and the NPs encased in the CNT.
Their corresponding FTs (i.e., F(R)) are also dis-
played. The krange window used to obtain the Fourier
transforms of Fig. 3b extends between 2.5 and 9 A
˚
-1
and phase shift corrections have been taken into
account. Calculations have been obtained by consid-
ering in the former case (lower panel) an iron bcc
cluster with lattice parameter a=2.87 A
˚and in the
latter one (upper panel) a Fe fcc structure with a lattice
parameter a=3.61 A
˚. The accordance between the
experimental and simulated EXELFS signals is quite
good, thus indicating the cphase of this iron
nanoparticle embedded in the carbon nanotube. It is
worth noting that the lattice parameter in a fcc Fe
lattice has been reported to assume values between 3.4
and 3.7 A
˚. Therefore, in order to establish the avalue
that best reproduces the experimental data, we calcu-
lated EXELFS spectra considering c-Fe cluster of
atoms with different lattice parameters. The
accordance between theory and experiment has been
quantified by the value of the Rfactor, defined as the
sum of the absolute values of the differences between
theory and experiment normalized by the number of
experimental points.
3
Figure 6displays a comparison
among the kv(k) oscillations calculated for the avalues
of 3.51 A
˚(black, left curve), 3.61 A
˚(red, central
curve), and 3.67 A
˚(green, right curve). The corre-
sponding Rvalues are 0.055, 0.043, and 0.093,
Fig. 2 (Top to bottom) TEM images of the two nanoparticles
embedded in carbon nanotubes and of a commercial a-Fe
microparticle used as standard
2
The same approach has been considered for the simulation of
the Fe bcc film reported in the ‘Data analysis’’ section.
3
The intensities of the simulated curves have been multiplied
by a factor that has been varied between 0.1 and 1 to account a
posteriori for an amplitude reduction factor S2
0, present in the
EXELFS formula used by FEFF8.2, not equal to 1. The lowest R
factor values were found to occur for S2
0equal to 0.25. The R
factor values reported in the paper have been obtained for the
best S2
0value.
J Nanopart Res (2013) 15:1846 Page 5 of 9
123
respectively. From this result, the obtained lattice
parameter of the Fe fcc nanoparticle embedded in the
carbon nanotubes is 3.61 ±0.03 A
˚. Very interest-
ingly, such a value of the c-Fe lattice parameter is
compatible with a ferromagnetic behavior for the NPs
at room temperature with atomic moment higher than
that of bulk a-Fe (Zhou et al. 1995). This result
confirms the X-ray diffraction analysis performed on a
large amount of carbon nanotubes removed from the
stainless steel surface reported in Camilli et al. (2012).
In addition, our study gives another important piece of
information about the structural and dynamical disor-
der of the nanoparticle, i.e., we can evaluate the
Debye–Waller factor, r
2
. In Fig. 7, we display a
comparison among the EXELFS oscillations calcu-
lated for r
2
=0 (black, dot-dashed curve), 0.004 (red,
dashed curve), and 0.008 A
˚
2
(blue, solid curve),
showing how much this parameter influences the
kv(k) fine structure, above all the amplitude of high
frequency oscillation. The corresponding Rfactors are
0.069, 0.052, and 0.044, respectively. This result is a
little bit higher than the Debye–Waller factor obtained
for c-Fe thin films deposited on copper substrate and
analyzed at room temperature (RT) (Butterfield and
Crapper 2000). Nonetheless, such finding is absolutely
reasonable. In fact, if on one hand the atoms in the NP
are compressed by the multiwall CNT, which should
lead to a low value of Debye–Waller factor, on the
other hand during EXELFS measurements, the system
could be at a temperature little bit higher than RT due
to the exposure to the high-energetic electron beam,
which in turn increases the atomic motion in the NP
and thus the Debye–Waller factor’s value. As far as
the EXELFS features reported in Fig. 3c, we know
from X-ray diffraction data reported in Camilli et al.
(2012) that iron carbide nanoparticles can be also
formed during the carbon nanotube synthesis. There-
fore, the cementite theoretical v(k) signal was
obtained by making a weighted average
4
of the two
EXELFS spectra calculated by considering as central
Fig. 3 Near and far L
2,3
edge energy loss spectra of the three
nanoparticles reported in Fig. 2.(a), (b) and (c) curves
correspond to the same labelled particles shown in Fig. 1
Fig. 4 EXELFS kv(k) signals extracted from the experimental
spectra reported in Fig. 3, after background subtraction and
edge step normalization. (a), (b) and (c) curves correspond to the
same labelled particles shown in Fig. 1
4
This means that the EXELFS signal calculated for each
ionizing Fe atom was weighted for the number of such atoms
present in the unit cell.
Page 6 of 9 J Nanopart Res (2013) 15:1846
123
Fig. 5 Lower panel left, comparison between the experimental
EXELFS oscillations (black,dotdot dashed) of the a-Fe
microparticle shown in Fig. 2a, and those calculated (red,solid
line) for a bcc lattice with a=2.87 A
˚. The corresponding
Fourier transforms obtained in a k-window of 2.2–9.0 A
˚
-1
are
reported on the right.Upper panel left, comparison between the
experimental EXELFS oscillations (black,dotdot dashed)of
the nanoparticle shown in Fig. 2b and those calculated (red,
solid line) for a fcc lattice with a=3.60 A
˚. The corresponding
Fourier transforms obtained in a k-window of 2.2–9.0 A
˚
-1
are
reported on the right. (Color figure online)
Fig. 6 Comparison among the experimental EXELFS features
(blue,dotdot dashed) of the nanoparticle shown in Fig. 2b and
those calculated for a fcc lattice with a=3.51 A
˚(black, left
line), 3.60 A
˚(red, central line), and 3.67 A
˚(green, right line).
(Color figure online)
Fig. 7 Comparison among the calculated EXELFS signal for a
fcc lattice with a=3.60 A
˚and r
2
=0A
˚
2
(black,dot dashed
line), 0.004 A
˚
2
(red,dashed line), and 0.008 A
˚
2
(blue,solid
line). (Color figure online)
J Nanopart Res (2013) 15:1846 Page 7 of 9
123
atoms the two Fe basis atoms of the Fe
3
C unit cell
according to the Pnma space group symmetry. The
lattice parameters used are a=5.048 A
˚,b=
6.731 A
˚, and c=4.513 A
˚. In Fig. 8, we report the
experimental and simulated kv(k) oscillations (left
panel) and the corresponding Fourier transforms (right
panel). The accordance between theory and experi-
ment is quite good for both the curves, thus evidencing
the Fe
3
C nature of the nanoparticle.
Conclusions
We performed a detailed structural analysis of the Fe
catalytic nanoparticles encapsulated into CNTs
directly grown on stainless steel. We demonstrated
that EXELFS can be successfully used to obtain
information on the local structure around the ionizing
atom. In fact, we were able (a) to distinguish between
a-Fe, c-Fe, and iron carbide nanoparticles, (b) to give
an estimation of the fcc Fe nanoparticle lattice
parameter, which results in being 3.61 ±0.03 A
˚.
Such a value of the fcc iron lattice parameter suggests
that these NPs can be ferromagnetic at room temper-
ature with an atomic moment higher than that of bulk
a-Fe.
Acknowledgments This work was supported by the Air Force
Office of Scientific Research Material Command, USAF, under
grant no. FA8655-11-1-3036.
References
Ankudinov AL, Ravel B, Rehr JJ, Conradson SD (1998) Real-
space multiple-scattering calculation and interpretation of
X-ray-absorption near-edge structure. Phys Rev B 58:7565
Ankudinov AL, Bouldin CE, Rehr JJ, Sims J, Hung H (2002)
Parallel calculation of electron multiple scattering using
Lanczos algorithms. Phys Rev B 65:104107
Auerhammer JM, Rez P (1989) Dipole-forbidden excitations in
electron-energy-loss spectroscopy. Phys Rev B 40:2024
Bagayoko D, Callaway J (1983) Lattice-parameter dependence
of ferromagnetism in bcc and fcc iron. Phys Rev B 28:5419
Borowiak-Palen E, Mendoza E, Bachmatiuk A, Rummeli MH,
Gemming T, Nogues J, Skumryev V, Kalenczuk RJ,
Pichler T, Silva SRP (2006) Iron filled single-wall carbon
nanotubes: a novel ferromagnetic medium. Chem Phys Lett
421:129
Butterfield MT, Crapper MD (2000) Extended X-ray absorption
fine structure investigation of the structure of iron over-
layers on Cu(111). Surf Sci 454(456):719
Camilli L, Scarselli M, Del Gobbo S, Castrucci P, Nanni F,
Gautron E, Lefrant S, De Crescenzi M (2011) The syn-
thesis and characterization of carbon nanotubes grown by
chemical vapor deposition using a stainless steel catalyst.
Carbon 49:3307
Camilli L, Scarselli M, Del Gobbo S, Castrucci P, Lamastra FR,
Nanni F, Gautron E, Lefrant S, D’Orazio F, Lucari F, De
Crescenzi M (2012) High coercivity of iron-filled carbon
nanotubes synthesized on austenitic stainless steel. Carbon
50:718
Fallon P, Walsh CA (1996) Computer code PEELS. University
of Cambridge, England
Gubbiotti G, Albini L, Tacchi S, Carlotti G, Gunnella R, De
Crescenzi M (1999) Structural and magnetic properties of
epitaxial Cu/Fe/Cu/Si(111) ultrathin films. Phys Rev B
60:17150
Fig. 8 Left panel, comparison between the experimental
EXELFS oscillations (black,dotdot dashed) of the nanopar-
ticle shown in Fig. 2c and those calculated (red,solid line) for a
Fe
3
C lattice. The corresponding Fourier transforms obtained in a
k-window of 2.2–9.0 A
˚
-1
are reported on the right. (Color figure
online)
Page 8 of 9 J Nanopart Res (2013) 15:1846
123
He Z, Maurice JL, Gohier A, Lee CS, Pribat D, Cojocaru CS
(2011) Iron catalysts for the growth of carbon nanofibers:
Fe, Fe
3
C or both? Chem Mater 23:5379
Hines WA, Shanthakumar P, Huang T, Budnick JI, Miller RL,
Pease DM, Perry DM (2009) Magnetic and structural study
of fcc c-Fe precipitates in Cu. Phys Status Solidi B
246:2154
Jouffrey B, Schattschneider P, He
´bert C (2004) The magic
angle: a solved mystery. Ultramicroscopy 102:61
Kincaid BM, Meixner AE, Platzman PM (1978) Carbon K edge
in graphite measured using electron-energy-loss spectros-
copy. Phys Rev Lett 40:1296
Klein J, Campbell SJ, Aubertin F, Gonser U, Schneeweiss O
(1991) c-Fe precipitation in Cu
97
Fe
3
and Cu
75
Au
24
Fe
1
.
Phys Status Solidi B 166:87
Kurde J, Ponpandian N, Luo J, Weis C, Baberschke K, Sri-
vastava P, Wende H (2007) Scattering-path analysis and
magnetic scattering properties of Fe/Ag(100) films: a
temperature-dependent magnetic EXAFS study. Phys Rev
B 76:224418
Li DC, Dai L, Huang S, Mau AWH, Wang ZL (2000) Structure
and growth of aligned carbon nanotube films by pyrolysis.
Chem Phys Lett 316:349
Ling T, Yu H, Liu X, Shen Z, Zhu J (2008) Five-fold twinned
nanorods of fcc Fe: synthesis and characterization. Cryst
Growth Des 8:4340
Ling T, Zhu J, Yu H, Xie L (2009) Size effect on crystal mor-
phology of faceted face-centered cubic Fe nanoparticles.
J Phys Chem C 113:9450
Lyubutin IS, Anosova OA, Frolov KV, Sulyanov SN, Okotrub
AV, Kudashov AG,Bulusheva LG (2012) Iron nanoparticles
in aligned arrays of pure and nitrogen-doped carbon nano-
tubes. Carbon 50:2628
Massalski TB, Okamoto H (1992) Binary alloys phase diagrams.
ASM International, Materials Park, OH
Mo
¨nch I, Leonhardt A, Meye A, Hampel S, Kozhuharova-
Koseva R, Elefant D, Wirth MP, Bu
¨chner B (2007) Syn-
thesis and characteristics of Fe-filled multi-walled carbon
nanotubes for biomedical application. J Phys Conf Series
61:820
Moruzzi VL, Marcus PM, Schwarz K, Mohn P (1986) Ferro-
magnetic phases of bcc and fcc Fe, Co, and Ni. Phys Rev B
34:1784
Shen J, Ohresser PCh, Klaua M, Barthel J, Kirschner J (1998)
Magnetic moment of fcc Fe(111) ultrathin films by ultra-
fast deposition on Cu(111). Phys Rev Lett 80:1980
Stern EA, Rehr JJ (1983) Many-body aspects of the near-edge
structure in X-ray absorption. Phys Rev B 27:3351
Sto
¨hr J (1988) X-ray absorption: principles, applications, tech-
niques of EXAFS, SEXAFS and XANES. In: Konings-
berger DC, Prins R (eds) Chemical analysis: a series of
monographs on analytical chemistry, vol 92. Wiley, New
York, p 443
Winkler A, Mu
¨hl T, Menzel S, Koseva RK, Hampel S, Leon-
hardt A, Bu
¨chner B (2006) Magnetic force microscopy
sensors using iron-filled carbon nanotubes. J Appl Phys
99:104905
Zhou YM, Zhang WQ, Zhong LP, Wang DS (1995) Theoretical
prediction of ferrimagnetism in face-centered cubic iron.
J Magn Magn Mater 145:L273
J Nanopart Res (2013) 15:1846 Page 9 of 9
123
... Recently, we have shown [43][44][45][46] that the direct growth of high quality MWCNTs on stainless steel in the absence of any external catalysts is possible. Moreover, acid treatments and oxidation-reduction stages on this type of surface are not necessary because of the native nano-scale roughness of the substrate and the iron-rich substrate surface both act as an efficient catalyst or template in the synthesis of MWCNTs. ...
... Moreover, MWCNTs come with a wide distribution of tube diameters, with average value 58.19 ± 18.35 nm. Also, we have shown in our past works [43][44][45][46] transmission electron microscopy (TEM) images confirming the multi-walled nature of the as-grown carbon nanotubes. Furthermore, in Figure 1b,d it may be observed that MWCNTs are mostly capped and often they present carbonaceous nanostructures (amorphous and/or graphitic carbon) around the tips (Figure 1d) and close to the stainless steel surface (Figure 1c), with a characteristic dimension of hundred nanometers, as also reported by other authors [42]. ...
Article
Full-text available
We have taken advantage of the native surface roughness and the iron content of AISI 316 stainless steel to directly grow multi-walled carbon nanotube (MWCNT) random networks by chemical vapor deposition (CVD) at low-temperature ([Formula: see text]) without the addition of any external catalysts or time-consuming pre-treatments. In this way, super-hydrophobic MWCNT films on stainless steel sheets were obtained, exhibiting high contact angle values ([Formula: see text]) and high adhesion force (high contact angle hysteresis). Furthermore, the investigation of MWCNT films with scanning electron microscopy (SEM) reveals a two-fold hierarchical morphology of the MWCNT random networks made of hydrophilic carbonaceous nanostructures on the tip of hydrophobic MWCNTs. Owing to the Salvinia effect, the hydrophobic and hydrophilic composite surface of the MWCNT films supplies a stationary super-hydrophobic coating for conductive stainless steel. This biomimetical inspired surface not only may prevent corrosion and fouling, but also could provide low friction and drag reduction.
... Carbon nanotubes (CNT) [1,2], carbon nanofibers (or nanowire) [3], and graphene nanoribbons [4,5] are typical one-dimensional (1-D) carbon nanomaterials. Such nanomaterials, which possess high lengthto-width ratios, tend to attract intense interest due to their large theoretical specific surface area, excellent thermal properties, and controllable electrical properties [6,7]. ...
Article
A chemical vapor deposition (CVD) method for the preparation of various graphene-based nanostructures is described. The morphology, microstructure, elemental distribution, and bonding states of the different products obtained during reaction optimization were analyzed by various characterization methods. With a Ni sputtering time of 50 s, growth temperature of 950 °C, pretreatment time of 30 min, and growth time of 10 min, a one-dimensional (1-D) graphene-based nanostructure was successfully grown on a Si substrate. The growth mechanism of the 1-D graphene-based nanostructures was then analyzed experimentally. The obtained 1-D carbon nanostructure was composed of curled or wound graphene sheets, the majority of which exhibit carbon nanotube-like hollow structures and characteristics similar to graphene nanoribbons. This 1-D carbon nanostructure can find potential applications in the development of sensors, capacitors, and catalysts, mainly due to its high aspect ratio, large surface area, and favorable characteristics.
... From this standpoint, stainless steel, as an economic material with a high content of iron as the catalyst for CNT synthesis, is an interesting candidate. In fact, several groups have conducted research on direct growth of CNT on stainless steel [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23]. Generally, the interest behind this area of research is multiple, including electrodes for supercapacitors followed by nitrogen drying, was applied to all samples to remove organic and inorganic contaminants. ...
Article
Full-text available
In this work, the corrosion behavior of three differently treated AISI 316 stainless steel plates is investigated, viz.(1) pristine, (2) coated with a carbon nanotube (CNT) layer and (3) coated with a carbon nanofiber (CNF) layer. CNTs and CNFs were directly grown on stainless steel via a CVD method, using ethylene as the carbon source and without the addition of an external catalyst. The corrosion behavior of these materials was investigated by a combination of microstructural and electrochemical methods. Electrochemical tests included potentiodynamic and potentiostatic techniques in hot sulfuric acid solutions. A strong deterioration in corrosion resistance was revealed by the electrochemical tests and confirmed by microstructural examination of the samples. It was found that carbon diffusion into the substrate material during the CNT/CNF growth process results in chromium depletion of the near-surface region of 316 SS and chromium carbide precipitation at grain boundaries causing accelerated intergranular corrosion. Accordingly, notwithstanding the obvious inability of CNT/CNF layers to provide protection to the substrate due to their porous nature, a real corrosion damage arises from the high temperature exposure of stainless steel to the CVD atmosphere, suggesting that a similar risk may be present even for compact carbon coating deposited by CVD process, in the event of local damage of the coating.
Chapter
An overview of recent progress in high loss electron energy loss spectroscopy is presented. This covers the instrumental aspects of how to best set up a scanning transmission electron microscope and post-column spectrometer combination to provide best performance, a survey of a range of results that are possible in EELS with a good setup, and a brief investigation of the current limitations imposed by the current mainstream detector technology. In the first section, detailed consideration is given to the coupling optics between microscope and spectrometer, the effects on the EELS spectrum, the need for spectrometer refocusing, and the problems with gun anodes in some field emission gun designs. In the second section, a survey of the absolute cross sections for the L edges of 4d transition elements is set out, these are used for absolute quantification in two cases, and the extended energy loss fine structure beyond the Si–K edge is used for atomic structure investigation. The final section comprises an investigation of noise levels on the Rh L3 edge with a standard CCD detector in the spectrometer with a comparison to simulated shot noise, and shows that in this case, RMS noise was 7 times worse than the fundamental limit, illustrating clearly how much improvement would be possible with electron counting detectors.
Article
Full-text available
Functional materials have a promising potential for the fabrication of new devices with improved properties to meet many requirements, including environmental issues. Along this idea, a multiphase structure made using a TiO2/MWCNT hybrid nanoscaffold grafted on a metal grating (stainless steel type), acting as a strong, highly durable and heat/thermal inert support, is proposed. The method, consisting firstly of the fabrication of a porous scaffold via catalytic-CVD of a MWCNT forest on stainless steel, followed by the grafting of nanocrystalline TiO2 via the sol–gel method and then calcination, is simple and effective. Morphology, structure and optical properties have been investigated using XRD, SEM, AFM, and HRTEM techniques as well as Raman and UV-visible spectroscopy, and porosity analysis. Interestingly, the TiO2/MWCNT hybrid composite exhibits enhanced photocatalytic activity as compared to nanocrystalline TiO2, obtained by adopting the same preparation. More interestingly, the hybrid system exhibits additional functionalities, such as magnetic, surface and optical properties. The multifunctional approach allows for the combination of enhanced photodegradation with magnetic properties, which can make the recovery of a solution from a photocatalyst easier. Furthermore, it will be shown that, by moving from MWCNT/stainless steel to TiO2/MWCNT/stainless steel composites, the surface character changes from hydrophobic to hydrophilic in nature. Grafting on the stainless steel support allows for the addition of a broad range of features, including combined strength and corrosion resistance in aqueous solutions at ambient temperature, together with enhancement of electrical, optical and photocatalytic properties.
Article
Full-text available
Epitaxial Cu/Fe/Cu heterostructures with Fe film thicknesses between 2.5 and 110 Å were successfully grown on 7×7-Si(111) substrates, using metal-metal epitaxy on silicon. The evolution of their structural and magnetic properties as a function of the Fe film thickness has been studied by a number of complementary techniques. This revealed that fcc pseudomorphic Fe layers, characterized by a low-spin ferromagnetic phase, grow up to about 6 Å. For larger thicknesses one observes the appearance of three-dimensional bcc Fe (110) domains in the Kurdjumov-Sachs orientation characterized by a high magnetic moment. A combined study of the magnetic anisotropy constants has been performed by the simultaneous use of both a static technique (surface magneto-optical Kerr effect) and a dynamic one (Brillouin light scattering). An excellent agreement, in both sign and magnitude, between the results of the two techniques has been found.
Article
Full-text available
Thermally deposited ultrathin Fe films on Cu(111) at room temperature or below have been known to grow in a multilayer mode with a low net magnetic moment  0.5muB~0.5muB. In the present work we have used pulsed laser deposition to produce isotropic fcc Fe/Cu111111 films which grow layer by layer with a magnetic moment of more than 2muB. We attribute the larger magnetic moment of the pulsed laser deposited films to their structural perfection as seen by low energy electron diffraction and scanning tunneling microscopy.
Article
Surface extended X-ray absorption fine structure (SEXAFS) has been used to probe the structure of ultrathin films of iron on Cu(111). With increasing film thickness, the interatomic separation falls from 2.57Å rapidly towards that of bulk iron (2.48Å). The degree of order and coordination in the films also fall over this region. These changes are consistent with a loss of face-centred cubic structure. However, the structure does not exhibit a clear body-centred cubic SEXAFS fingerprint until around eight layers.
Article
Arrays of aligned carbon nanotubes (CNTs) and nitrogen-doped carbon (CNx) nanotubes have been grown on silicon substrates as the result of thermolysis of ferrocene/toluene and ferrocene/acetonitrile mixture. The microstructure of materials was studied by transmission and scanning electron microscopy, and X-ray diffraction was used to control the carbon and iron forms. The composition and properties of iron nanoparticles developed in the CNT and CNx nanotube samples were determined from Mössbauer spectroscopy data. The total iron content in CNx nanotubes was found to be considerably higher than that in CNTs. Three forms of iron nanoparticles α-Fe, γ-Fe, and Fe3C were detected in CNTs and only two last of them in CNx nanotubes. In the interior of CNT channels the α-Fe and Fe3C nanoparticles were observed to be coupled by a strong exchange interaction and to exhibit magnetic behavior at room temperature.
Article
Multiwalled carbon nanotubes synthesized directly on austenitic stainless steel result in being filled with pure iron nanoparticles. X-ray diffraction shows that the nanoparticles are either in the γ- or α-phase, although iron in the original alloy is in γ-phase because of the presence of nickel. This phase transformation is due to the selective extraction of iron, performed by carbon nanotubes during their growth. A high coercivity of iron-filled carbon nanotubes is measured although the starting steel is paramagnetic. The presence of the α-phase, the magnetic anisotropy and the single domain character of the Fe nanoparticles explain their magnetic behavior.
Article
We have measured the carbon K edge in graphite using inelastic-electron-scattering spectroscopy. The extended x-ray absorption fine structure is in good agreement with theory for the first-neighbor atoms at a distance of 1.42 Å. The momentum dependence of the edge structure is in qualitative agreement with a simple band-structure picture. A comparison of the signal counting rates for electron-energy-loss and photoabsorption experiments shows that the energy-loss method is competitive with synchrotron radiation sources up to about 1000 eV.
Article
The magnetic structure of fcc Fe has been investigated in a range of lattice constants from a = 3.50 to 3.75 Å by the first-principles all-electron linearized augmented plane wave method in the local spin-density functional approximation. Besides the previously reported antiferromagnetic phase, one high-spin and two low-spin ferromagnetic phases, a ferrimagnetic phase is predicted to coexist in the range of a = 3.59 to 3.67 Å.