Reducing reaction of Fe3O4 in nanoscopic reactors of a-CNTs

Article abstract of DOI:10.1021/jp0661037

A reduction of Fe₃O₄ nanowires in nanoscopic reactors of amorphous C:H nanotubes (a-CNTs) was taken to understand features of the chemical reaction mechanism in nanoscale reactors. Fe₃O₄ nanowires encapsulated in a-CNTs were reduced into iron at a relatively low temperature of 570°C, producing iron nanoparticles encapsulated in CNTs accompanied by the crystallization of the a-CNT shell. It was found that carbon in the a-CNT shell rather than hydrogen (5.5 wt % in it) reduced Fe ₃O₄, showing features different from those in a macroscopic system. The possible mechanisms behind this phenomenon are discussed.

Full text of DOI:10.1021/jp0661037

1724
J. Phys. Chem. B 2007, 111, 1724-1728
Reducing Reaction of Fe3O4 in Nanoscopic Reactors of a-CNTs
†
†
‡
§
‡
,†
Fangyu Cao, Kaifu Zhong, Aimei Gao, Changle Chen, Quanxin Li, and Qianwang Chen*
Hefei National Laboratory for Physical Sciences at Microscale and Department of Materials Science &
Engineering, UniVersity of Science & Technology of China, Hefei 230026, P. R. China
ReceiVed: September 18, 2006; In Final Form: December 14, 2006
A reduction of Fe
3
O
4
nanowires in nanoscopic reactors of amorphous C:H nanotubes (a-CNTs) was taken to
nanowires encapsulated
understand features of the chemical reaction mechanism in nanoscale reactors. Fe
3
O
4
in a-CNTs were reduced into iron at a relatively low temperature of 570 °C, producing iron nanoparticles
encapsulated in CNTs accompanied by the crystallization of the a-CNT shell. It was found that carbon in the
3 4
a-CNT shell rather than hydrogen (5.5 wt % in it) reduced Fe O , showing features different from those in
a macroscopic system. The possible mechanisms behind this phenomenon are discussed.
Introduction
nanocables by acid disposal. In this work, we present experi-
mental evidence of facile reduction of Fe3O4 nanowires located
in the cavities of a-CNTs to Fe nanoparticles at around 570 °C,
vs reduction of powdered Fe3O4 by a-CNTs mechanically mixed
together at higher temperature (>700 °C), as an example of
chemical reactions of obtaining CNT-confined metallic nano-
particles through direct reduction of incorporated oxide nanow-
ires in nanometer-sized reactors.
Nanoscopic reactions have gained wide interest because of
their mechanism and high activity different from macroscopic
reactions. A carbon nanotube (CNT), like a nanometer-sized
test tube, could be used as a nanoscopic reactor with motion
1
restricted to a quasi-one-dimensional environment. Several
successful applications have been reported, including synthesis
of nanostructured materials in the cavity of CNTs under mild
2
-4
5
5
6
conditions (e.g., SiC, â-zeolite, CoFe2O4, GaN, SiC-SiOx
Experimental Section
7
8
biaxial nanowires, Si-B-C-N nanocables, and heterostruc-
tures of CNTs and carbide nanorods ), focusing on making
9
3 4
Heat treatment of the Fe O /a-CNT nanocables was taken in
quartz boat heating by a tube furnace in N2 stream, with a flow
rate of 100 mL/min. The rate of heat increase of the system
was 10 °C/min, and the temperature was held for 10 min when
it came to certain temperature points. Gas effluents were
collected and injected into a Finnigan Trace 2000-MS Gas
Chromatography/Mass Spectrometer for analysis, and the solid
products were saved for further characterizations. In control
experiments, a-CNTs and a mixture of 0.06 g of a-CNTs and
different kinds of nanometer-sized heterostructures by filling
the cavity of CNTs with heterogenities or by decorating the
outside surfaces of them. Some chemical reactions confined
within these nanometer-sized channels exhibited higher activi-
ties.4,5,10
However, chemical reactions that take place in the filled
CNTs or heterostructures of them with other materials have
rarely been reported. As an example, Bao et.al.11 found that
R-Fe2O3 nanoparticles in the cavity of CNTs were reduced by
them at 600 °C, and those adhered at the outside wall were
reduced at 800 °C, while R-Fe2O3 powder was reduced by
graphite when the temperature was higher than 1000 °C,
implying that the reactions exhibited higher activities.
0.04 g of commercial Fe3O4 (CP) powder were also treated with
high temperature by a tube furnace in N2 stream, respectively.
Heat treatment of a-CNTs was taken in temperature-programmed
desorption (TPD) with heating rate of 10 °C/min. In the heating
process, gas effluents such as H2 (m/e ) 2), H2O (m/e ) 18),
and CO2 (m/e ) 44) were continuously monitored by a Blazers
GSD300 Omnistar quadrupole mass spectrometer. The powder
X-ray diffraction (XRD) analyses were performed on a Rigaku
D/MAX-γA X-ray diffractometer equipped with Cu KR radia-
tion (λ ) 1.542 Å) over the 2θ range of 10-70×bc. Transmis-
sion electron microscopy (TEM) analyses were performed on a
Hitachi H-800 transmission electron microscope with electron
diffraction, and the accelerating potential is 200 kV. High-
resolution TEM (HRTEM) images were taken on a JEOL-2010
with an accelerating voltage of 200 kV. Magnetic measurements
were carried out on a Riken BHV-55 vibrating sample magne-
trometer (VSM).
Recently, we developed a one-pot method for the synthesis
of Fe3O4/amorphous C:H nanotube (a-CNT) coaxial nanocables
12
by pyrolysis of ferrocene in Sc-CO2 at 400 °C. In the as-
prepared products, single-crystal Fe3O4 nanowires with a
diameter of around 40 nm are continuously covered by a-CNT
shell. The a-CNT shell is rich in hydrogen (elemental analysis
shows that H% in the a-CNTs’ shell is as high as 5.5 wt %.),
which releases in the form of molecular H2 from 400 °C. This
cable-like nanostructure provides a natural nanometer-sized
a-CNT test tube filled with Fe3O4 nanowires. a-CNTs with a
hollow cavity can also be obtained from the Fe3O4/a-CNT
*
Towhomcorrespondenceshouldbeaddressed.E-mail: cqw@ustc.edu.cn.
Fax: (+86)551-3631760.
†
Hefei National Laboratory for Physical Sciences at Microscale and
Results and Discussion
Department of Materials Science & Engineering, University of Science &
Technology of China.
Heat treatment of the Fe3O4/a-CNT nanocables at various
temperatures was performed under high-purity N2 stream. From
XRD patterns of the solid products (Figure 1), it can be found
that the main solid components changed from Fe3O4 and a-CNTs
‡
Department of Chemical Physics, Lab of Biomass Clean Energy,
University of Science & Technology of China.
§
Department of Chemistry, The University of Chicago, 5735 South Ellis
Avenue, Chicago, Illinois.
1
0.1021/jp0661037 CCC: $37.00 © 2007 American Chemical Society
Published on Web 02/01/2007

Fe3O4 in Nanoscopic Reactors of a-CNTs
J. Phys. Chem. B, Vol. 111, No. 7, 2007 1725
Figure 1. XRD patterns of the Fe
3
O
4
/a-CNT nanocables heated to (a)
4
09 °C, (b) 468 °C, (c) 533 °C, (d) 570 °C, and (e) 595 °C and then
2
cooled down in an N stream. Peaks of the compounds were labeled
on the patterns.
Figure 3. XRD patterns of (a) the a-CNTs heated to (A) 400 °C, (B)
6
00 °C, (C) 700 °C, and (D) 800 °C and (b) the mixture of commercial
Fe powders and a-CNTs heated to (E) 700 °C, (F) 800 °C, and (G)
00 °C and then cooled down in N stream. Peaks of the compounds
were labeled on the patterns.
3 4
O
9
2
3 4
Figure 2. HRTEM images of the Fe O
/a-CNT nanocables (a)12 before
and (d) after annealing at 570 °C. Parts (b) and (c) are partial enlarged
images of (a), and (e) is a TEM panorama of (d).
sharper and shaper, and the center of them moves toward 26.6°,
which is the position of the graphite (002) crystal plane, showing
better crystallization of the a-CNTs by heating under higher
temperature. The crystallization temperature of the hollow
a-CNTs is much higher than that of a-CNTs in Fe3O4/a-CNT
nanocables, implying that encapsulated Fe3O4 nanowires may
contribute to the crystallization, producing Fe/Fe3C catalysts.
Figure 3b shows that the Fe3O4 powders in the mixture were
not reduced until heated to 800 °C, indicating that reduction of
the unencapsulated Fe3O4 is harder than encapsulated ones. The
σ-electron density in CNTs shifting from the concave inner to
the convex outer surface caused by the deviation from planarity
to iron and graphitic CNTs, with some Fe3C as byproducts.
Reduction of Fe3O4 to Fe happened at a temperature (called
Tc) between 532 and 570 °C, much lower than the temperature
needed to reduce pulverous Fe3O4 by graphite (>1100 °C);
carbonization of the a-CNT shells happened at the same
temperature, also much lower than the reported temperature
1
3
(
>1700 °C), indicating a simultaneous quick reduction of
Fe3O4 nanowires and crystallization of a-CNTs. The HRTEM
image (Figure 2d) shows that the whole a-CNT shell has
changed from amorphous carbon to graphitic CNTs at 570 °C,
with a d value of 0.33 nm, very close to that of graphite (002).
Heat treatment of a-CNTs and a mixture of commercial
Fe3O4/a-CNTs were both taken as control experiments. In Figure
14,15
may contribute to the temperature difference.
The interaction
between the electron-deficient concave surface of the nanotube
and the anionic oxygen in encapsulated Fe3O4 nanowires could
lead to weakened bonding strength of Fe3O4 and hence lower
the activation energy for the reduction in the cavity.11 Thus, Fe
could form at a very low temperature of Tc. In contrast, for the
3
a, each of the XRD patterns exhibit a broad peak, typical of
a-CNTs. With the temperature elevated, the peaks become

1726 J. Phys. Chem. B, Vol. 111, No. 7, 2007
Cao et al.
Figure 4. Ratios of (a) O
2
, (b) H
2
O, and (c) CO
2
fractions in gas
effluents of heat treatment of the Fe
temperatures.
3 4
O /a-CNT nanocables at different
2 2 2
Figure 5. (a) H , (b) H O, and (c) CO TPD of a-CNTs.
outer particles, the interaction of the electron-density-enriched
outer nanotube surface with the oxygen in Fe3O4 should be
11
weakened, and a higher temperature to reduce Fe3O4 is needed,
as observed. Defects in a-CNTs may also contribute to the
facile reduction of Fe3O4 nanowires, lowering the reaction
temperature.
Both the reduction of unencapsulated Fe3O4 powder and the
crystallization of a-CNTs are imperfect at 800 °C, which is
higher than Tc. It is inferred that the activation energy of
reduction of Fe3O4 nanowires and crystallization of a-CNTs
in the Fe3O4/a-CNT nanocables are lowered in the Fe3O4/a-C
cable-like system, so that the two happen together at a low
temperature. There are various reports regarding Fe catalyzing
the crystallization of graphite and/or CNTs in the processes of
synthesizing or annealing, first forming Fe3C as an intermediate
16-19
product and then obtaining CNTs from decomposing them.
In our experiment, the final solid products of annealing the
nanocables are graphitic CNTs, Fe, and a little Fe3C.
To understand the mechanism of the reaction, gas effluents
released at the selected temperature points in the heating process
of the Fe3O4/a-CNT nanocables were analyzed with mass
spectrometry. By setting the intensity of the N2 fractions peak
Figure 6. M-H hysteresis loops for the samples after annealing,
measured at room temperature. The insert shows detail of the remnant
magnetization and coercive field.
heating Fe3O4/a-CNT nanocables, indicating that the released
CO2 in the latter experiment should be the product of the
reaction of encapsulated Fe3O4 nanowires with an a-CNT shell,
different from that released from heating pure a-CNTs. H2
became the obvious gas product after 400 °C, accompanied by
a little water vapor possibly produced by the reaction between
H2 and released CO2 or air leaked into the instrument.
(
m/e ∼ 28) in the gas mixture as 100%, the intensities of the
peaks of different fractions are shown in Figure 4. CO2 fractions
(
m/e ∼ 44) started to be released before 400 °C, reached their
maximum at around 594 °C, close to Tc, and then decreased
soon. H2 fractions (m/ e ∼ 2) cannot be detected because of
their lower molecular weight than He carrier used in the
instrument. O2 fractions (m/e ∼ 32) that were detected came
from air leaked into the collection, whose amount was almost
invariable. The other composites, such as H2O fractions (m/e
The M-H hysteresis loop (Figure 6) measured at room
temperature shows that the powder sample after annealing
exhibits ferromagnetism, with a saturation magnetization, Ms
) 93 emu/g; remnant magnetization, Mr ) 3 emu/g; and
coercive field, Hc ) 85 Oe. The Ms value is much lower than
∼
18), fluctuated inconspicuously by temperature, implying that
they were not main products of the reaction but came from other
routes such as desorption of planar water or decomposition of
a-CNT.
2
1
220 emu/g of the corresponding bulk iron, which could be
due to the considerable mass of carbon and the nanoscopic size
of the iron particles. Correspondingly, the data of the sample
before annealing is Ms ) 27.5 emu/g, Mr ) 12.7 emu/g, and
For comparison, the TPD of the a-CNTs was carried out in
vacuum and the slowly released gas effluents were monitored
by an online quadrupole mass spectrometer (TPD-MS, Figure
1
2
Hc ) 324.5 Oe. In addition, iron nanoparticles with oxide
impurity are known to exhibit Hc values larger than 1000 Oe,22
while the Hc values of our products are relatively low. This
implies that a thorough transformation from iron oxide to iron
was happening during the annealing process.
5
). CO2, formerly presented as adsorbed CO2 or carbonyl, ether,
2
0
and/or other surface oxygen groups in the carbon shell, was
released between 300 and 600 °C. The peaking temperature of
the releasing rate of CO2 is about 460 °C, far from that of

Fe3O4 in Nanoscopic Reactors of a-CNTs
J. Phys. Chem. B, Vol. 111, No. 7, 2007 1727
and prevents H2 from diffusing through it. The released H2 was
soon expelled from the experimental system by an N2 stream.
Experimentally, H2 was detected to be released from a-CNTs
between 300 and 600 °C, as shown in Figure 5. As a result, the
concentration of H2 in the system remained at a low level out
of the a-CNT shell, and there was almost no H2 in the cavity of
it. Therefore, it was hardly possible for H2 to meet Fe3O4
nanowires by diffusing through the a-CNT shell. In contrast,
there is no barrier between a-CNTs and Fe3O4 nanowires in
the nanometer-sized reactor, which makes the reaction more
kinetically favored. Thus, the Fe3O4 nanowires were reduced
by carbon rather than hydrogen in the reaction. Further research
is needed to better understand the reaction.
Figure 7. Mechanism of the nanoscopic reaction: a f b shows the
process of releasing gas effluents form a-CNTs at low temperature; b
Conclusions
3 4
f c shows the process of reducing Fe O by carbon accompanied with
the graphitization of a-CNTs; 532 °C < Tc < 570 °C.
A facile reducing reaction of Fe3O4 nanowires in the Fe3O4/
a-CNT nanocables to metallic Fe nanoparticles by the a-CNTs
at Tc (532 °C < Tc < 570 °C) was observed. Though H2 started
to release at a low temperature of 400 °C, it did not take the
role of reductant to reduce Fe3O4 nanowires because it tends to
diffuse to the exterior of the nanotubes and then escape form
the reaction system rather than to diffuse to the interior part of
them. In contrast, carbon in a-CNTs functioned as a reductant,
reducing Fe3O4 to metallic Fe, and carbon itself was oxidized
to CO2. Graphitization of a-CNTs occurred simultaneously with
the reducing reaction at relatively low temperature of Tc. The
studies reveal that a gas such as hydrogen cannot easily enter
into a nanotube reactor to take part in reactions, which could
provide information for reactions occurring in a confined system
and help to develop nanoscopic reactors and synthesize new
nanostructured materials.
According to the experimental results, one clearly sees that
it is carbon rather than H2 released from a-CNTs reduced Fe3O4
nanowires, with itself oxidized to CO2
Tc
Fe O + a-CNTs f Fe + C(graphitic) + H + CO
2
3
4
2
A model (as shown in Figure 7) of the reduction of Fe3O4
coupled with the crystallization of a-CNT is proposed. As shown
in the HRTEM images (Figure 2), the a-CNTs consist of small
graphene sheets. First, the graphene sheets start to decompose
at a temperature lower than Tc, releasing gas products such as
CO2, H2, water vapor, and so forth, and leaving a-CNTs full of
defects and unsaturated carbon atoms. When the temperature
rises to Tc, the a-CNTs reduce Fe3O4 to Fe metal, releasing
CO2 as a byproduct. CO is also a possible product of the
reaction.11 However, because the molecular weight (Mw ∼ 28)
of CO is the same as N2, the carrier gas used in the system, the
existence of CO cannot be verified by the mass spectrometric
method. Catalyzed by the as-produced Fe metal nanoparticles,
the graphene sheets change to small graphitic crystallites, with
Fe3C as an intermediate product. Due to the catalyzation of the
metallic Fe, the crystallization of a-CNT proceeds faster. As a
result, once the reaction starts, it goes along fast and comes to
an end soon, within several minutes. To reduce the surface
energy, the iron atoms start bonding together. At last, metallic
Fe nanoparticles grow and small graphitic crystallites reorient
and merge to form a three-dimensional graphitic nanotube order,
as a process expected to bring about by itself a negative free
energy change.1 These Fe particles are separated instead of
continuous in CNTs, different from the encapsulated Fe3O4
nanowires, because the volume of iron is not enough to fill the
cavity of a-CNTs. If only a few iron nucleating points form at
the beginning of the growth process, the Fe atoms may
accumulate on them and form particles of large sizes, while
many nucleating points dispersing in one nanocable may lead
to small dispersed particles.
Acknowledgment. This work was supported by the Natural
Science Foundation of China (20125103 and 90206034).
Supporting Information Available: Synthesis, magnetic
properties, XRD, TEM, FESEM, FTIR, and elemental analysis
of the sample before annealing and details of GC-MS data are
available. This material is available free of charge via the
Internet at http://pubs.acs.org.
References and Notes
(
1) Dujardin, E.; Ebbesen, T. W.; Hiura, H.; Tanigaki, K. Science 1994,
65, 1850.
2) Dai, H. J.; Wong, E. W.; Lu, Y. Z.; Fan, S. S.; Lieber, C. M. Nature
1995, 375, 769.
(3) Pan, Z.; Lai, H. L.; Au, F. C. K.; Duan, X. F.; Zhou, W. Y.; Shi,
W. S.; Wong, N.; Lee, C. S.; Wong, N. B.; Lee, S. T. AdV. Mater. 2000,
2, 1186.
4) Sun, X. H.; Li, C. P.; Wong, W. K.; Wong, N. B.; Lee, C. S.; Lee,
S. T.; Teo, B. K. J. Am. Chem. Soc. 2002, 124, 14464.
5) Nhut, J. M.; Pesant, L.; Tessonnier, J. P.; Win e´ , G.; Guille, J.; Pham-
Huu, C.; Ledoux, M. J. Appl. Catal. A. 2003, 254, 345.
6) Han, W. Q.; Fan, S. S.; Li, Q. Q.; Hu, Y. D. Science 1997, 277,
287.
(7) Wang, Z. L.; Dai, Z. R.; Gao, R. P.; Bai, Z. G.; Gole, J. L. Appl.
Phys. Lett. 2000, 77, 3349.
2
(
3
1
(
(
(
1
Though it was possible that released H2 might reduce Fe3O4
at even lower temperature (e.g., 260 °C) based on thermody-
(
73.
8) Zhang, Y.; Suenaga, K.; Colliex, C.; Iijima, S. Science 1998, 281,
9
2
3
namic data, no obvious quantitative change of H2O in gas
effluents was found, showing that H2 was not the reductant in
our experiment. A possible reason is that hydrogen cannot easily
enter into a nanotube reactor to take part in reactions. It is
reported that a-C:H coatings may decrease the permeation
properties of H2, N2, O2, and CO2.24 As mentioned above, Fe3O4
nanowires were encapsulated in a-CNTs. At high temperature,
the released H2 tend to diffuse to the exterior rather than the
cavity of the a-CNT shell, which lies over the Fe3O4 nanowires
(9) Zhang, Y.; Ichihashi, T.; Landree, E.; Nihey, F.; Iijima, S. Science
1999, 285, 1719.
10) Zhang, Y.; Zhang, H. B.; Lin, G. D.; Chen, P.; Yuan, Y. Z.; Tsai,
K. R. Appl. Catal., A 1999, 187, 213.
11) Chen, W.; Pan, X. L.; Willinger, M.-G.; Su, D. S.; Bao, X. H. J.
(
(
Am. Chem. Soc. 2006, 128, 3136.
(12) Cao, F. Y.; Chen, C. L.; Wang, Q.; Chen, Q. W. Carbon 2007,
doi: 10.1016/j.carbon.2006.11.030.
(13) Ci, L. J.; Wei, B. Q.; Xu, C. L.; Liang, J.; Wu, D. H.; Xie, S. S.;
Zhou, W. Y.; Li, Y. B.; Liu, Z. Q.; Tang, D. S. J. Cryst. Growth 2001,
233, 823.

1728 J. Phys. Chem. B, Vol. 111, No. 7, 2007
Cao et al.
(
(
(
(
14) Haddon, R. C. Science 1993, 261, 1545.
15) Ugarte, D.; Chatelain, A.; de Heer, W. A. Science 1996, 274, 1897.
16) Oya, A.; Otani, S. Carbon 1981, 19, 391.
(20) Boehm, H. P. Carbon 2002, 40, 145.
(21) Klabunde, K. J.; Zhang, D.; Glavee, G. N.; Sorensen, C. M.;
Hadjipanayis, G. C. Chem. Mater. 1994, 6, 784.
(22) Gangopadhyay, S.; Hadjipanayis, G. C.; Dalc, B.; Sorensen, C. M.;
Klaubunde, K. J.; Papaefyhymjou, V.; Kostikas, A. Phys. ReV. B 1992, 45,
9778.
(23) Tiernan, M. J.; Barnes, P. A.; Parkes, G. M. B. J. Phys. Chem. B
2001, 105, 220.
17) Hernadi, K.; Fonseca, A.; Nagy, J. B.; Bernaerts, D.; Lucas, A. A.
Carbon 1996, 34, 1249.
18) M u¨ ller, T. E.; Reid, D. G.; Hsu, W. K.; Hare, J. P.; Kroto, H. W.;
Walton, D. R. M. Carbon 1997, 35, 951.
19) Cantoro, M.; Hofmann, S.; Pisana, S.; Scardaci, V.; Parvez, A.;
(
(
Ducati, C.; Ferrari, A. C.; Blackburn, A. M.; Wang, K. Y.; Robertson, J.
Nano Lett. 2006, 6, 1107.
(24) Vasquez-Boruchi, S.; Jacob, W.; Achete, C. A. Diamond Relat.
Mater. 2000, 9, 1971.