Formation of spherical voids in the oxidation of Co nanoparticles

Article abstract of DOI:10.1134/S0036024408040341

A high-resolution transmission microscopy study showed that the oxidation of Co nanoparticles deposited on montmorillonite resulted in the formation of nanoshells consisting of cobalt oxides and containing spherical voids. A linear dependence was found between the initial diameter of Co particles and the oxide shell thickness.

Full text of DOI:10.1134/S0036024408040341

ISSN 0036-0244, Russian Journal of Physical Chemistry A, 2008, Vol. 82, No. 4, pp. 690–693. © Pleiades Publishing, Ltd., 2008.
Original Russian Text © P.A. Chernavskii, G.V. Pankina, V.I. Zaikovskii, N.V. Peskov, V.V. Lunin, 2008, published in Zhurnal Fizicheskoi Khimii, 2008, Vol. 82, No. 4, pp. 796–800.
SHORT
COMMUNICATIONS
Formation of Spherical Voids in the Oxidation of Co Nanoparticles
P. A. Chernavskiia, G. V. Pankinaa, V. I. Zaikovskiib, N. V. Peskovc, and V. V. Lunina
a Faculty of Chemistry, Moscow State University, Leninskie gory, Moscow, 119992 Russia
b Boreskov Institute of Catalysis, Siberian Branch, Russian Academy of Sciences,
pr. Akademika Lavrent’eva 5, Novosibirsk, 630090 Russia
c Faculty of Computational Mathematics and Cybernetics, Moscow State University, Leninskie gory, Moscow, 119992 Russia
e-mail: chern@kge.chem.msu.ru
Received April 23, 2007
Abstract—A high-resolution transmission microscopy study showed that the oxidation of Co nanoparticles
deposited on montmorillonite resulted in the formation of nanoshells consisting of cobalt oxides and containing
spherical voids. A linear dependence was found between the initial diameter of Co particles and the oxide shell
thickness.
DOI: 10.1134/S0036024408040341
INTRODUCTION
EXPERIMENTAL
Co nanoparticles on montmorillonite were prepared
by impregnating the support with a solution of cobalt
nitrate of kh. ch. (chemically pure) grade. Grade F-160
montmorillonite with a specific surface area of 300 m2/g
and the mean pore size 6.6 nm was used as a support.
After impregnation, the sample was dried in a flow of
air at 100°ë for 3 h. As a result, a 20% Co/montmoril-
lonite sample (denoted below as Co/M) was obtained.
The dry sample was reduced under hydrogen in a flow
microreactor, which was simultaneously the measuring
cell of a vibration magnetometer [7]. Reduction was
performed in a temperature-programmed mode (TPR)
at temperatures of up to 700°ë at a heating rate of
0.47 K/s until the degree of reduction ceased to change.
The degree of reduction was monitored by measuring
changes in magnetization. Magnetization constant in
time indicated that the limiting degree of reduction was
achieved at the given temperature. The sample was
cooled to 200°ë in a flow of ç2; ç2 was then replaced
with Ar to remove adsorbed hydrogen, and the sample
was cooled to room temperature.
As is known, high-temperature oxidation of metals
during the formation of thick oxide layers at the metal–
oxide interface causes the formation of voids as a result
of the association of vacancies, which appear because
of the difference between the diffusion coefficients of
metal ions and oxygen [1–3].
Recently, a similar phenomenon was observed in the
oxidation of Co [4] and Fe [5] nanoparticles at rela-
tively low temperatures. High-resolution electron
microscopy showed that the low-temperature oxidation
of Co and Fe nanoparticles led to the accumulation of
vacancies at the metal–oxide interface and formation of
spherical voids within metal oxide shells.
This phenomenon has the following explanation.
Oxide films grow either at the metal–oxide interface
(oxygen ion transport inside particles) or at the oxide–
gas boundary (metal ion transport to the surface), which
depends on the type of migrating lattice defects. When
the difference in the diffusion coefficients is large
(Kirkendall effect), regions with vacancies are formed
at the metal–oxide interface, and the accumulation of
vacancies leads to the formation of voids.
Cobalt nanoparticles were subjected to temperature-
programmed oxidation (TPO) in a 1% é2/ç flow
(heating rate 0.47 K/s). After magnetization decreased
to zero, the sample was cooled in an Ar flow to ~20°C
and removed from the reactor. It was then examined by
high-resolution transmission electron microscopy on a
JEM-2010 electron microscope (JEOL, Japan) with a
grid resolution of 0.14 nm at an accelerating voltage of
200 kV. The sample was fixed on a standard copper
grid.
It was found [5] that, for Fe nanoparticles, there was
a critical size (8 nm) at which they were completely
oxidized at 20°C, and spherical voids formed inside
oxide shells; for large-sized particles, only separate
voids, however, formed at the metal–oxide interface.
The critical sizes of metal particles at which particle
oxidation resulted in the formation of spherical voids
was studied theoretically [6]. The thermal instability of
spherical nanoshells was considered, and the results of
In addition, the Co/M sample was studied by the
TPR method, and changes in magnetization and hydro-
Monte Carlo simulation of the kinetics of nanoshell for- gen absorption rate were recorded concurrently. This
mation were reported.
combination of methods allowed us to obtain additional
690

FORMATION OF SPHERICAL VOIDS IN THE OXIDATION OF Co NANOPARTICLES
691
information about the sequence of stages during reduc-
tion [7]. We used a 5% ç2 + Ar mixture instead of ç2.
RESULTS AND DISCUSSION
Figures 1 and 2 show the photomicrographs of the
20%Co/M sample after reduction in the TPR mode to
700°ë followed by oxidation in the TPO mode to
200°ë. The photomicrographs clearly show ring struc-
tures, which are nanospherical shells consisting of Co
oxide.
The reactions
2ëÓ + é2 = 2ëÓé,
(I)
100 nm
6ëÓé + é2 = 2ëÓ3é4
(II)
Fig. 1. Electron photomicrograph of the 20% Co/M sample
after oxidation in the TPO mode to 200°ë.
occur sequentially during the oxidation of Co nanopar-
ticles; reactions
ëÓ3é4 + ç2 = 3ëÓé + ç2é,
ëÓé + ç2 = ëÓ + ç2é
(III)
(IV)
occur during reduction.
To determine exactly what oxide was formed in our
experiments, the starting sample obtained after calcin-
ing and the sample obtained after the sequence of oxi-
dation and reduction processes were examined by the
TPR method using a 5% ç2 + Ar mixture as a reducing
agent. Figure 3 shows the temperature dependences of
the hydrogen absorption rate and changes in the magne-
tization of the starting sample, preliminarily calcined
under Ar at 400°ë for 1 h.
The peak on the hydrogen absorption rate curve in
the temperature range 200–300°ë is not accompanied
by a change in magnetization and corresponds to reac-
tion (III). At 350°ë, magnetization increases, which
corresponds to the beginning of reaction (IV). The non-
monotonic character of hydrogen absorption in this
temperature range is possibly caused by mass transfer
in support micropores [8].After the completion of TPR,
the sample was oxidized in a 1% é2 + ç flow at a tem-
perature increased linearly from 20 to 250°ë and then
again subjected to reduction in the TPR mode in
5% ç2 + Ar. The results of repeated reduction are given
in Fig. 4.
20 nm
Fig. 2. Photomicrograph of a Co particle on montmoril-
lonite after complete oxidation in the TPO mode in an air
flow.
Figure 5 shows the transmission electron micros-
copy data, namely, the dependence of the thickness δ of
nanospherical shells on the inside diameter of voids d.
It follows from the data shown in Fig. 4 that, over
the range 200–300°ë, we again observe the hydrogen
absorption peak corresponding to reaction (III). This
shows that the oxidation under the conditions described
above produces ëÓ3é4. It, however, follows from the
ratio of the peak areas in the TPR spectrum that the sys-
tem contains not only ëÓ3é4, but also CoO in both
cases. The CoO fraction noticeably increased from 26
to 33% after repeated TPR. We can therefore assume
that spherical oxide shells observed in photomicro-
graphs consist mainly of ëÓ3é4, but CoO is also
present, and the ratio between CoO and ëÓ3é4 is
approximately 1/3.
It follows from the stoichiometry of reactions (I)
and (II) that
nCo = nCoO or nCo = (1/3)nCo O
,
4
(1)
3
where ni is the number of moles of Co, CoO, and ëÓ3é4
in the starting nanoparticle and spherical oxide shell. The
spherical Co particle of diameter d0 contains nëÓ moles,
πρCo
6MCo
3
d0,
------------
nCo
=
(2)
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY A Vol. 82 No. 4 2008

692
CHERNAVSKII et al.
12
10
8
18
16
14
12
10
8
6
4
6
4
2
2
0
0
100
200
300
400
500
600
700
t, °C
Fig. 3. TPR spectrum and temperature dependence of magnetization (κ) for the starting Co/M sample.
12
10
8
35
30
25
20
15
10
5
6
4
2
0
0
100
200
300
400
500
600
700
t, °C
Fig. 4. TPR spectrum and temperature dependence of magnetization for the Co/M sample after the redox process.
where ρ is the density and M is the atomic weight. In a where D is the outside diameter of the particle after
similar way, for CÓ3é4, we can write
complete oxidation and d is the diameter of the
πρ
6M
Co3O4
3
3
spherical void formed in place of the Co particle of
diameter d0.
------------------
nCo O
=
4
(D – d ),
(3)
3
Co3O4
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY A Vol. 82 No. 4 2008

FORMATION OF SPHERICAL VOIDS IN THE OXIDATION OF Co NANOPARTICLES
693
Suppose the diameter of the spherical void and the
diameter of the starting Co particle are related as
δ, nm
7
d = d0 – ε,
6
5
4
3
2
where ε is a parameter. For ε = 0, the oxide film grows
only by metal ion diffusion to the oxide–gas interface;
in other words, the oxide film grows only at the expense
of increasing the oxide film thickness at the gas–oxide
interface. For ε > 0, the oxide film grows not only by
metal ion diffusion, but also by the counterdiffusion of
oxygen ions and the diffusion of vacancies toward the
gas–oxide interface.
Taking into account (2) and (3), we can use (1) to
derive the dependence of the thickness of the spherical
oxide shell on the diameter of a spherical void,
1
4
8
12
16
d, nm
d + ε 3 1
⎞
d
⎝ ⎠
2
d
2
3
1/3
⎛
⎝
⎛ ⎞
--
----------- ---
--
– ,
δ =
+
(4)
⎠
2
C
Fig. 5. Dependence of the oxide shell thickness δ on the
spherical void diameter d for the Co/M sample. The solid
line corresponds to Eq. (4) with the parameters ε = 2.66 and
C = 0.35.
where C = V(Co)/V(CoO) = 0.548 for the CoO
nanoshell and C = V(Co)/3V(Co3O4) = 0.056 for the
ëÓ3é4 shell. Here, V is the molar volume of Co and
oxides. As was shown above, the oxide shell actually
contained ~33% CoO and ~67% ëÓ3é4. In addition, it
should be taken into account that the density of bulk
oxide is obviously different from that of oxide in a
nanoshell, and the exact value of C is therefore
unknown. The unknown ε and C parameters can be
evaluated from the experimental data available using
the method of least squares. We obtained ε = 2.66 and
C = 0.35; the sum of the squares of deviations depended
only slightly on the C parameter.
oxidation is accompanied by the formation of spherical
voids.
ACKNOWLEDGMENTS
This work was financially supported by the Russian
Foundation for Basic Research (project no. 06-03-
32500a) and the Federal Scientific and Technological
Program (grant no. 2005-RI-112.0/001/056).
A comparison of the results of the regression analy-
sis and above calculations suggests that the diameter of
a spherical void is smaller than that of the starting Co
particle. At the first stage of oxidation, several external
Co atomic layers are obviously oxidized in situ by oxy-
gen diffusion inside the particle. The Co ions then dif-
fuse under the action of the Mott potential through the
oxide layer toward the particle surface and are oxidized
there, augmenting the oxide film at the top. Attempts to
take into account the mixed composition of the oxide
film did not significantly change the estimates of the ε
and C parameters in the linear dependence of the oxide
layer thickness on the spherical void diameter.
REFERENCES
1. A. Atkinson, Rev. Mod. Phys. 57, 437 (1985).
2. P. Kofstad, High Temperature Corrosion (Elsevier
Applied Science, London, 1988).
3. T. C. Rojas, J. M. Greneche, A. Conde, and A. Fernan-
dez, J. Mater. Sci. 39, 4877 (2004).
4. Y. Yin, R. M. Riou, C. K. Erdonmez, et al., Science
(Washington, D.C.) 304, 711 (2004).
5. C. M. Wang, D. R. Baer, L. E. Thomas, et al., J. Appl.
Phys. 98, 094308 (2005).
6. A. M. Gusak, T. V. Zaporozhets, K. N. Tu, and U. Gos-
ele, Philos. Mag. 85, 4445 (2005).
Calculations by (4) showed that zero spherical void
diameter, d = 0 (the absence of a spherical void inside a
particle), corresponded to the particle diameter d0 ≈
2 nm. This value was considered an estimate of a cer-
tain critical size of the particle, starting with which the
7. P. A. Chernavskii, A. Y. Khodakov, G. V. Pankina, et al.,
Appl. Catal., A 306, 108 (2006).
8. P. A. Chernavskii, A. S. Lermontov, G. V. Pankina, et al.,
Kinet. Katal. 43 (2), 292 (2002) [Kinet. Catal. 43 (2),
268 (2002)].
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY A Vol. 82 No. 4 2008