Potassium intercalation of carbon onions ‘opened’ by carbon dioxide treatment

Abstract

The potassium intercalation of onion-like carbon (OLC) samples consisting of aggregates of carbon onions is studied with photoemission spectroscopy. OLC samples were initially prepared by annealing nanodiamonds (3–20 nm in diameter) at 1800 K in vacuum. The resulting OLC consists of closed fullerene-like shells. The ‘closed’ OLC was subsequently treated with carbon dioxide at 1020 K in order to open the carbon shells by partial oxidation to create ‘opened’ OLC. Core level and valence band photoelectron spectroscopy have been employed in characterizing the changes in electronic structure of the samples. Upon intercalation of the closed OLC with K the C1s core level and valence band features shift to higher binding energies and the density of states at the Fermi level increases, while this effect is significantly smaller for intercalated opened OLC. These results indicate that opening the shells of carbon onions allows potassium to penetrate inside the particles and thus opens up a possible route to fill carbon onions with desired substances and their application as nanocapsules.

Full text

Page 1

Potassium intercalation of carbon onions ‘opened’ by
carbon dioxide treatment
Yu.V. Butenkoa,*, Amit K. Chakrabortyb,c,1, N. Peltekisa, S. Krishnamurthya,
V.R. Dhanakd,e, M.R.C. Huntb, L. Sˇillera,*
aSchool of Chemical Engineering and Advanced Materials, The University of Newcastle upon Tyne, Newcastle upon Tyne, NE1 7RU, UK
bDepartment of Physics, Durham University, Durham, DH1 3LE, UK
cSchool of Physics and Astronomy, The University of Nottingham, Nottingham, NG7 2RD, UK
dCCLRC, Daresbury Laboratory, Warrington, Cheshire, WA4 4AD, UK
ePhysics Department, University of Liverpool, Liverpool, L69 3BX, UK
A R T I C L E I N F O
Article history:
Received 13 November 2007
Accepted 7 April 2008
Available online 12 April 2008
A B S T R A C T
The potassium intercalation of onion-like carbon (OLC) samples consisting of aggregates of
carbon onions is studied with photoemission spectroscopy. OLC samples were initially pre-
pared by annealing nanodiamonds (3–20 nm in diameter) at 1800 K in vacuum. The result-
ing OLC consists of closed fullerene-like shells. The ‘closed’ OLC was subsequently treated
with carbon dioxide at 1020 K in order to open the carbon shells by partial oxidation to cre-
ate ‘opened’ OLC. Core level and valence band photoelectron spectroscopy have been
employed in characterizing the changes in electronic structure of the samples. Upon inter-
calation of the closed OLC with K the C1s core level and valence band features shift to
higher binding energies and the density of states at the Fermi level increases, while this
effect is significantly smaller for intercalated opened OLC. These results indicate that open-
ing the shells of carbon onions allows potassium to penetrate inside the particles and thus
opens up a possible route to fill carbon onions with desired substances and their applica-
tion as nanocapsules.
? 2008 Elsevier Ltd. All rights reserved.
1.Introduction
Synthesis of carbon onions filled with different substances is
of interest for several applications. In recent years there have
been a number of publications demonstrating that magnetic
nanoparticles encapsulated into graphite-like layers may find
applications as components of magnetic recording systems
[1,2], magnetic fluids [1,2], electromagnetic shielding materi-
als [3] or magnetic resonance imaging agents [4]. Carbon
onions can even serve as nanocapsules for high pressure
experiments [5–7]. External graphite layers in such materials
provide protection to the inner substances [8] and can serve
for attachment of desirable functional groups [9] for further
applications, for example, as drug delivery agents as in case
of carbon nanotubes [10]. Existing methods of synthesis of
filled carbon onions such as arc-discharge [5,11–15], anneal-
ing [2,16,17], electron bombardment [11,18] of carbonaceous
material in the presence of catalytic metal particles, and
chemical vapour deposition techniques [19] require the pres-
ence of all substances in the reaction media at high tempera-
ture. Therefore the substances which can be encapsulated
into carbon onions are limited to metals, metal oxides and
0008-6223/$ - see front matter ? 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2008.04.012
* Corresponding authors:
E-mail addresses: yuriy.butenko@ncl.ac.uk (Yu.V. Butenko), lidija.siller@ncl.ac.uk (L. Sˇiller).
1Present address: Department of Chemistry, Durham University, Durham, DH1 3LE, UK.
C A R B O N 4 6 ( 20 0 8 ) 1 1 3 3–11 4 0
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/carbon

Page 2

carbides. Another disadvantage of these techniques is that
the size and number of shells of the filled carbon onions de-
pend on the properties of the encapsulated substances. This
can restrict the synthesis of filled carbon onions with struc-
tures needed for a particular application.
Another approach to the synthesis of filled carbon onions
is to use a two stage process: initial opening of the carbon
onions and subsequent filling of the opened onions with de-
sired substances. This type of approach has been successfully
used for carbon nanotubes. It has been shown that opening of
initially closed carbon nanotubes can be achieved by treat-
ment with carbon dioxide [20], nitric acid [21] and oxygen
[22]. Opened carbon nanotubes have been successfully filled
with different substances [22–26]. Such a two stage process
would also allow one to have a larger choice of carbon onions,
enabling selection of material with particular structures (par-
ticle size, number of shells, size of inner space), which better
satisfy final applications.
Tsang et al. used high resolution transmission electron
microscopy (HRTEM) to study changes in multi-walled carbon
nanotube structures brought about by their partial oxidation
with carbon dioxide [20]. It was demonstrated that this treat-
ment results in the partial or complete destruction of the tube
caps, i.e. the formation of holes at the nanotube ends. How-
ever, it is rather more difficult to observe the formation of
such holes in the shells of the carbon onions by HRTEM, since
these holes have lower contrast on HRTEM images. In this pa-
per we employ carbon dioxide treatment to open up the car-
bon shells of closed carbon onion structures, such as those
present in onion-like carbon (OLC), and use potassium inter-
calation as a means to probe this opening. Suzuki et al.
showed that in case of closed single-walled carbon nanotubes
potassium intercalates between the individual carbon nano-
tubes within their bundles [27]; whereas potassium pene-
trates inside opened carbon nanotubes [25]. In our previous
studies we demonstrated, by means of photoelectron spec-
troscopy, that potassium intercalation of closed OLC results
in core level and valence band states shifting to higher bind-
ing energies, and an increase of the density of states at the
Fermi level, both of which are associated with charge transfer
from potassium to the OLC [28]. Since the electronic proper-
ties of potassium intercalated graphitic structures strongly
depend on the location of the potassium atoms (whether on
external surfaces and/or inside the graphitic structures)
[25,27,29–31], the electronic states of intercalated ‘closed’
and ‘opened’ OLC have been investigated using photoelectron
spectroscopy to test the efficacy of our opening and filling
procedure.
2.Experimental
Samples of OLC were prepared by heat treatment of nanodia-
monds (diamond crystallites with diameters of 3–20 nm) at
1800 K under a vacuum of 10?5mbar for 1 h: details of the
preparation methods are described elsewhere [32]. Studies
using HRTEM [32,33], density measurements [32], and X-ray
photoelectron spectroscopy [34] have shown that annealing
nanodiamond in this way results in complete transformation
of diamond nanocrystallites into graphitic OLC containing
carbon onions with closed outer fullerene-like shells. HRTEM
images showed that the sample consisted of OLC carbon par-
ticles mainly composed of 3–10 fullerene-like spherical shells
and extended curved graphitic layers between them, binding
the carbon onions together. The OLC sample containing
closed carbon onions is referred to throughout this paper as
‘closed OLC’.
Opening the carbon onions in the closed OLC sample was
achieved by treatment in a flow of carbon dioxide as follows:
(1) the closed OLC sample was placed in a ceramic boat and
inserted into a ceramic tube; (2) the sample was heated to
1020 K under nitrogen flow (40 cm3/min); (3) after the sample
temperature reached 1020 K the nitrogen was replaced with
carbon dioxide; (4) the sample was maintained at 1020 K un-
der carbon dioxide flow (40 cm3/min) for 1 h at atmospheric
pressure; (5) the sample was cooled down to room tempera-
ture under nitrogen flow. This sample is referred to through-
out the paper as ‘opened OLC’.
Potassium intercalation of the two types of samples and
characterization by X-ray and ultraviolet photoelectron spec-
troscopies were carried out at Beamline 4.1 of the Synchro-
tron Radiation Source (SRS), Daresbury Laboratory, UK. Both
samples (closed and opened OLC) were ultrasonicated in iso-
propanol and drops of the resulting suspension were depos-
ited onto silicon substrates with a native oxide layer until a
macroscopically thick film was produced. Once dried, the
samples were mounted in a sample holder using tantalum
retaining clips in good electrical contact with the OLC films
and introduced into an ultra-high vacuum (UHV) chamber
with a base pressure below 2 · 10?10mbar. Before potassium
intercalation the samples were annealed at 1270 K for
10 min to remove adsorbates such as condensed water, traces
of isopropanol and any oxygen-containing groups bound to
the OLC. Potassium intercalation was performed by evaporat-
ing potassium onto the OLC films from commercial (SAES)
‘getter’ sources while keeping the samples at room tempera-
ture. For each sample of closed and opened OLC a new getter
source was used.
Valence band photoelectron spectra were measured using
a SCIENTA SES-200 analyzer at a fixed pass energy of 40 eV
and a photon energy of 40 eV. The binding energy scales of
the valence-band spectra were calibrated by measuring the
position of the Fermi edge obtained from a platinum foil in
good electrical contact with the sample. For the valence band
data an overall energy resolution of 0.15 eV was determined
from the width of the Fermi cutoff. The C1s and K2p core-le-
vel spectra were acquired with a VG Scientific CLAM2 ana-
lyzer and Mg Ka radiation (1253.6 eV) from a conventional
X-ray source. The C1s binding energy of a highly oriented
pyrolytic graphite reference sample (284.4 eV) was utilized
for calibration of the core-level photoelectron spectra. An
overall energy resolution of 0.75 eV was determined from
the Gaussian width of a Au 4f line from a gold foil in good
electrical contact with the samples. The C1s photoemission
spectra of sp2bound carbon were fitted using a Doniac–Sˇunjic ´
lineshape [35] convoluted with a Gaussian broadening. The p
plasmon peak of graphitic carbon was fitted with a Lorentzian
peak convoluted with the same Gaussian employed for broad-
ening the sp2-related peak. The background photoelectron
intensity was subtracted by the Shirley method [36].
1134
C A R B O N 4 6 (2 0 0 8) 1 1 33 –1 1 40

Page 3

3. Results and discussion
Fig. 1A shows core level photoelectron spectra of the closed
OLC sample after successive treatments. The spectra were fit-
ted with a component for sp2carbon and a component asso-
ciated with p–p*transitions in graphitic materials (p plasmon)
[37]. The fitting curves are presented only for the as-intro-
duced sample, as shown in Fig. 1A (a). The fitting parameters
of the sp2component for all spectra are presented in Table 1.
Fitting the C1s spectrum of the as-introduced sample shows
(Fig. 1A (a)) the absence, within an experimental error of
?0.5%, of components at higher binding energies related to
sp3carbon and carbon atoms bound to oxygen-containing
groups. Typical binding energies of C1s electrons in sp3
hybridized carbon lie in the range 285.7–286.1 eV and for oxy-
gen-containing groups binding energies lie between 286.5 eV
and 287.7 eV [34]. Analysis of the O1s photoelectron spectrum
(not shown) of the as-introduced closed OLC also indicated
negligible oxygen concentration (less than 0.5%).
Fig. 2A presents valence band photoelectron spectra of the
closed OLC after different successive treatments of annealing
and potassium intercalation. The key features of the valence
band spectrum of the as-introduced sample (Fig. 2A (a)) are
two prominent peaks at 2.9 ± 0.1 eV and 7.9 ± 0.1 eV, and a
shoulder at 4.2 ± 0.2 eV. The peaks at 2.9 ± 0.1 eV, and
7.9 ± 0.1 eV are related to p bonding and r bonding states in
graphite,respectively[38,39]
4.2 ± 0.2 eV is associated with mixed r–p states [39]. We be-
lieve that the broad peaks at approximately 14.1 and 16.6 eV
are associated with r bonding states of carbon atoms in the
sample. The slight increase in intensity of the peak at
14.1 eV after annealing at 1270 K for 10 min is related to the
desorption of water and residual solvent from sample sur-
faces. Also of interest is the region in the vicinity of the Fermi
level where an increase in the density of states (DOS) as a
function of potassium dose is clearly visible. For clarity this
region is plotted in a separate graph in Fig. 3A which shows
the near Fermi level valence band spectra of the closed OLC
after different successive treatments of annealing and potas-
sium intercalation.
These results also show that annealing the as-introduced
closed OLC at 1270 K for 10 min in UHV does not significantly
affect the C1s and valance band spectra (see Table 1 and com-
pare Fig. 1A (a) and (b), and Fig. 2A (a) with (b)). Intercalation
of the sample by potassium results in the appearance of new
peaks at binding energies of 294.6 ± 0.1 eV and 297.4 ± 0.1 eV,
which become most prominent after the highest potassium
dose (see inset in Fig. 1A). These peaks are assigned to the
K2p doublet [40]. The atomic percentages of potassium in
the sample after the different potassium doses were deter-
mined from the C1s and K2p peak areas taking into account
the corresponding sensitivity factors, and are presented in Ta-
ble 1. For adsorption of potassium on HOPG the K2p3/2peak
was previously observed at the same binding energy of
294.6 ± 0.1 eVand was attributed to ‘‘elemental’’, not oxidized,
potassium [40]. Therefore, we can conclude that the potas-
sium deposited on the OLC sample is not oxidized. In the va-
lence band spectra the presence of potassium in the OLC
sample as a result of potassium dose becomes evident by
the emergence of a rather broad feature at 20.4 ± 0.2 eV
whiletheshoulderat
Fig. 1 – (A) Evolution of C1s core-level spectra of closed OLC
as a function of successive treatments: (a) as-introduced; the
points are experimental data and solid lines are the fit
components into which the spectra were decomposed; the
resulting fit is superimposed on the data as a solid line, (b)
after annealing at 1270 K for 10 min. After potassium
deposition for: (c) 90 min, (d) 115 min, (e) 140 min, (f)
190 min; (g) after final annealing at 1270 K for 10 min. The
inset shows the K2p doublet (2p3/2= 294.6 eV; D = 2.8 eV) of
the sample after 190 min of potassium deposition and the p
plasmon peak (shake-up satellite) at a binding energy of
291.0 ± 0.1 eV expanded from spectrum (f)). (B) Evolution of
C1s core-level spectra of opened OLC as a function of
various successive steps of annealing and potassium doses:
(a) as-introduced; the points are experimental data and the
solid lines underneath are the fit components into which
the spectra were decomposed; the resulting fit is
superimposed on the data as a solid line; (b) after annealing
at 1270 K for 10 min; after potassium deposition for: (c)
90 min, (d) 115 min, (e) 140 min, (f) 190 min, (g) 250 min; and
(h) after final annealing at 1270 K for 10 min. The inset
shows the K2p doublet (2p3/2= 294.6 eV; D = 2.8 eV) of the
sample after 250 min of potassium deposition and the p
plasmon peak (shake-up satellite) at a binding energy of
291.0 ± 0.1 eVexpanded from spectrum (g). The spectra were
recorded using a photon energy of 1253.6 eV in normal
emission geometry.
C A R B O N 4 6 ( 20 0 8 ) 1 1 3 3–11 4 0
1135

Page 4

(Fig. 2A (c)–(f)) associated with K3p electrons [28]. The position
of this peak is higher than the reported value of 19.3 eV for
adsorption of potassium on graphite [30] and is closer to the
value of 19.8 eV observed for this peak in potassium interca-
lated OLC produced by annealing nanodiamonds at a differ-
ent temperature (2140 K) to the present sample [28].
A closer inspection of Fig. 1A reveals that upon potassium
intercalation the C1s peak gradually shifts towards higher
binding energies and becomes broader (see Table 1). The max-
imum shift of 0.15 ± 0.03 eV is registered after 190 min of
potassium deposition (see Table 1 and Fig. 1A (f)). Similar
shifts of main spectral features towards higher binding ener-
gies upon potassium intercalation are also observed in the va-
lence band spectra (Fig. 2A), but are of larger magnitude. At
the maximum potassium dose of 190 min the valence band
shift reaches 0.30 ± 0.05 eV (compare spectra (b) and (f) in
Fig. 2A). The upward shift of the spectral features is also
accompanied by a significant increase in the density of states
(DOS) at the Fermi level (see Figs. 2A and 3A) upon potassium
deposition.
The increase of DOS at the Fermi level and upward shift of
features in photoelectron spectra upon potassium intercala-
tion has been previously reported for graphite [29,31,40,41],
carbon nanotubes [25,42] and OLC produced by annealing
nanodiamonds at 2140 K [28]. This behaviour is explained by
the charge transfer from the donor potassium atoms into
unoccupied states of graphitic samples. Filling of the unoccu-
pied states near Fermi level results in a continuous move-
ment of the Fermi level ‘‘up’’ the DOS [28]. Since the binding
energies of the photoelectrons are presented with reference
to the Fermi level, the upward movement of the Fermi level
causes shifts in the position of photoemission peaks towards
higher binding energies. In Fig. 3A this is manifested by an
apparent change in curvature of the rising intensity on the
high binding energy side of the Fermi level. The upward shift
of the Fermi level is not visible because the binding energy
scale is referenced to this level (Fig. 3A).
Desorption of potassium upon final annealing of the
closed OLC at 1270 K for 10 min results in the vanishing of
charge transfer and in the recovery of core level and valence
band spectra, as indicated by the reduction of DOS at the Fer-
mi level to its original intensity, a loss of the K3p peak in the
valence band spectra (compare Fig. 2A (g) with (b) and Fig. 3A
(g) with (b)), and a loss of the K2p peak and associated shifts
in the C1s core level spectra (compare Fig. 1A (g) with (b)).
Fig. 1B presents the C1s photoelectron spectra of opened
OLC sample after different successive treatments of anneal-
ing and potassium doses. The spectra were again fitted with
a component for sp2carbon and a component accounting
for the p plasmon peak. The fitting parameters for the sp2
component are presented in Table 2. Treatment of the closed
OLC sample with carbon dioxide resulted in a shift of the sp2
peak by 0.05 ± 0.03 eV to higher binding energies (compare
positions for the non-intercalated samples in Tables 1 and
2). Taking into account that the experimental error of
0.03 eV is very close to the observed shift, we repeated this
experiment with three different closed OLC samples treated
with CO2and found that this shift is reproducible. Analysis
of C1s (Fig. 1B) and O1s (not shown) spectra of the sample re-
vealed no significant presence of oxygen-containing groups
on its surfaces. To explain this we note that the reaction of
carbon with carbon dioxide at 1070 K at atmospheric pressure
results in oxidation of carbon according to the following reac-
tion [43]:
Csolidþ CO2;gas( ) 2COgas:
As a result of the above reaction some carbon atoms are
removed from the graphite layers. The observed shift towards
higher binding energy of the sp2component can be explained
by the appearance of carbon atoms which have lost their
neighbours. These carbon atoms can be present in defects
such as the edges of voids produced in graphite-like layers
as a result of the CO2treatment. It is notable that for a sin-
gle-wall carbon nanotube sample subjected to Ar ion bom-
bardment the sp2-related component of the C1s peak
shifted to lower binding energy and there were strong
changes in the valence band shape near the Fermi level [44].
Unlike the case of high fluence Ar ion bombardment, no evi-
dence of any feature associated with sp3hybridized carbon
was found for the opened OLC. The difference between the re-
sults of Ar ion bombardment of graphitic material and oxida-
tion with carbon dioxide suggest a qualitative difference
Table 1 – Fitting parameters for the sp2-related C1s photoemission line (Doniac–Sˇunjic ´ lineshape) of closed OLC (Fig. 1A)
and apparent percentage of potassium in the sample after different potassium doses deduced from the photoemission
spectra
Different successive stages
of the treatment of closed OLC
Fitting parameters for sp2components of
C 1 s spectra presented in Fig. 1A
Binding energy, eVLorentzian width, eVFWHMa, eVPercentage of potassium,%
(a) Initial (as-introduced) sample
(b) Annealing at 1270 K for 10 min
(c) 90 min of K deposition
(d) 115 min of K deposition
(e) 140 min of K deposition
(f) 190 min of K deposition
(g) Final annealing at 1270 K for 10 min
284.40 ± 0.03
284.40 ± 0.03
284.49 ± 0.03
284.51 ± 0.03
284.52 ± 0.03
284.55 ± 0.03
284.40 ± 0.03
0.26 ± 0.01
0.26 ± 0.01
0.29 ± 0.01
0.29 ± 0.01
0.30 ± 0.01
0.32 ± 0.01
0.27 ± 0.01
1.23 ± 0.01
1.22 ± 0.01
1.28 ± 0.01
1.32 ± 0.01
1.34 ± 0.01
1.35 ± 0.01
1.25 ± 0.01
0
0
0.3 ± 0.2
0.4 ± 0.2
0.6 ± 0.2
0.7 ± 0.2
0
Gaussian broadening, representing instrumental broadening and sample inhomogeneity, is fixed for all samples at 0.75 eV. The singularity
index is fixed at 0.15 [34]. The position of the shakeup peak for all spectra is 291.0 ± 0.1 eV.
a Full width at half maximum.
1136
C A R B O N 4 6 (2 0 0 8) 1 1 33 –1 1 40

Page 5

between the nature of the defects formed in the two pro-
cesses, and that a low level of overall damage results from
OLC opening.
To explain the negligible presence of oxygen-containing
groups in the opened OLC sample we note that after CO2
treatment the sample was cooled under nitrogen flow from
1020 K to room temperature. Therefore, if during CO2oxida-
tion any oxygen-containing groups were formed on sample
surfaces they would have been removed by the subsequent
heating in inert nitrogen atmosphere. We also note that be-
fore insertion into the UHV chamber the opened OLC sample
was exposed for approximately 1 h to atmosphere at room
temperature. However, the apparent absence of oxygen-con-
taining groups (as revealed by photoemission data) in the
opened OLC sample indicates that carbon atoms occurring
in different defect structures in graphite-like layers are rela-
tively inert and do not react with atmospheric oxygen and
water at room temperature. Further annealing of the opened
OLC at 1270 K for 10 min in UHV resulted in only a slight nar-
rowing of the main sp2component of the C1s spectrum, while
its position remained the same, within experimental error,
compared with the sample before annealing (see Table 2).
Valence band spectra of the opened OLC are presented in
Fig. 2B. The spectrum of the opened OLC sample after anneal-
ing in vacuum at 1270 K (Fig. 2B (a)) contains the same fea-
tures as the spectrum of the closed OLC after annealing
under the same conditions (Fig. 2B (b)). The similarity bet-
ween the spectra from opened and closed OLC (compare
Fig. 2 – (A) Evolution of valence band spectra of closed OLC
as a function of successive steps of annealing and
potassium dose: (a) as-introduced; (b) after annealing at
1270 K for 10 min; after potassium deposition for: (c) 90 min,
(d) 115 min, (e) 140 min, (f) 190 min; and (g) after final
annealing at 1270 K for 10 min. (B) Evolution of valence band
spectra of opened OLC as a function of successive steps of
annealing and potassium dose: (a) after annealing at 1270 K
for 10 min; after potassium deposition for: (b) 90 min, (c)
115 min, (d) 140 min, (e) 190 mins, (f) 250 min; and (g) after
final annealing at 1270 K for 30 min. Spectrum (a) is
replotted as (h) for direct comparison with (g). Spectra were
recorded using a photon energy of 40 eV in normal emission
geometry.
Fig. 3 – (A) Evolution of valence band spectra of closed OLC in
the near Fermi level region as a function of successive steps
of annealing and potassium dose: (a) as-introduced; (b) after
annealing at 1270 K for 10 min; after potassium deposition
for: (c) 90 min, (d) 115 min, (e) 140 min, (f) 190 min; and (g)
after final annealing at 1270 K for 10 min. (B) Evolution of
valence band spectra of opened OLC in the near Fermi level
region as a function of successive steps of annealing and
potassium dose: (a) after annealing at 1270 K for 10 min;
after potassium deposition for: (b) 190 min, (c) 250 min; and
(d) after final annealing at 1270 K for 10 min. The dashed
line marks the Fermi level position. Spectra were recorded
using a photon energy of 40 eV in normal emission
geometry.
C A R B O N 4 6 ( 20 0 8 ) 1 1 3 3–11 4 0
1137

Page 6

Fig. 2A (b) and Fig. 2 B (a)) supports the assertion that the
treatment with CO2did not significantly damage the sample.
To compare the influence of potassium intercalation on
the electronic structure of closed and opened OLC we depos-
ited potassium on the sample of opened OLC to the same
doses as the sample of closed OLC. As with the closed OLC
sample, potassium intercalation of the opened OLC results
in the appearance of a K2p doublet, with components at
294.7 ± 0.1 eVand 297.5 ± 0.1 eV (see inset in Fig. 1B), the posi-
tion of which, within experimental error, is the same as that
for the potassium intercalated closed OLC (see inset in
Fig. 1A). However, the atomic percentages of potassium ac-
quired from the relative K2p and C1s peak areas in the sample
after different potassium doses is lower than that after the
approximately same potassium doses on closed OLC (see Ta-
ble 2). For example, the atomic percentage of potassium in the
sample of opened OLC after the dosing for 190 min is
0.4 ± 0.2% which is lower than that after the approximately
same potassium dose in closed OLC, which is 0.7 ± 0.2% (see
Table 1). We believe that this can be explained by penetration
of potassium deep inside the carbon onions in the opened
OLC sample and therefore a reduction in sensitivity to potas-
sium atoms situated deeper in the carbon onions than the
photoelectrons escape depth (?15 A˚[45]). Further potassium
dosing for a total of 250 min resulted in an increase of the
potassium concentration to 1.4 ± 0.2%.
Comparison of the positions of K3p peaks in the valence
band spectra of the opened and closed OLC after potassium
deposition revealed a rather surprising result: the binding en-
ergy of the K3p peak of the opened OLC was found to be
approximately 19.4 eV (see Fig. 2B (b)–(f)). This value of the
binding energy is lower than that for the K3p peak observed
in closed OLC (see Fig. 2A (c)–(f)) and is almost the same as
was reported for potassium intercalated graphite [30].
Further differences between closed and opened OLC can
be found when one compares the spectral features in the pho-
toemission spectra of closed and opened OLC upon potassium
intercalation of an approximately equal dose of 190 min. A
smaller shift of the C 1s level in the photoemission spectrum
of the opened OLC (Fig. 1B (f)) is observed compared with that
for the closed OLC sample (Fig. 1A (f)): for closed OLC this shift
is 0.15 ± 0.03 eV, while for opened OLC this shift is only
0.10 ± 0.03 eV (see Tables 1 and 2). Even an additional potas-
sium dose for 60 min on the opened OLC did not produce sig-
nificant changes in the C1s spectrum (Fig. 1B (g), Table 2). As
can be seen from Fig. 2A and B, the valence band spectral fea-
tures of the opened OLC are also less affected by potassium
intercalation compared with those for the closed OLC. While
190 min potassium dose results in a shift of 0.30 ± 0.05 eV to-
wards higher binding energies for the valence band features
(Fig. 2A (f)) of closed OLC, the corresponding shift at an even
larger potassium dose (250 min) for the opened OLC is only
0.11 ± 0.05 eV (compare Fig. 2B (a) and (f)). Inspection of the
valence band spectra near Fermi level of the potassium inter-
calated opened OLC showed a very minor rise of the DOS
when compared with that for the potassium intercalated
closed OLC (see Fig. 3A and B).
To explain the observed differences in the response of the
valence band and core level features of opened and closed
OLC we have to take into account that at room temperature
potassium atoms are quite mobile [30] and penetrate not only
deep into the film of OLC (between OLC particles) but also, in
the case of the opened OLC, inside the individual carbon
onions to occupy their interior space and intercalate between
graphite-like layers [31,41]. In the latter case, at the same
potassium doses potassium coverage of the external graph-
ite-like layer will be less for opened OLC compared with that
for closed OLC. The difference in potassium concentration be-
tween closed and opened OLC was found in our X-ray photo-
electron measurements, see Tables 1 and 2. As a result of this
difference of potassium concentration, the amount of trans-
ferred charge per carbon atom in the graphite layers from
the intercalated potassium is different in the two samples.
Bennich et al. [30] showed that for 0.1 monolayer potassium
coverage on a graphite surface the amount of transferred
charge from potassium atoms to carbon atoms in different
graphite layers decreases from the first layer down to the bulk
layers. As amount of the transferred charge decreases, the
shift of photoemission spectral features for carbon atoms in
these graphite layers also decreases [30].
If we assume that in our experiments the amount of trans-
ferred charge per carbon atom in a graphite-like layer is pro-
portional to the amount of the potassium present on that
particular graphite-like layer, we can expect that the amount
of charge transfer, and thus the shift of core level and valence
band features, will be less for carbon atoms in the external
layers of opened OLC compared with those of closed OLC,
due to the smaller density of potassium present on the exter-
Table 2 – Fitting parameters for the sp2-related C1s photoemission line (Doniac–Sˇunjic ´ lineshape) of opened OLC (Fig. 1B)
and apparent percentages of potassium in the sample after different potassium doses
Different successive stages
of the treatment of opened OLC
Fitting parameters for sp2components of C 1s spectra presented in Fig. 1B
Binding energy, eV Lorentzian width, eVFWHM, eV Percentage of potassium,%
(a) Initial (as-introduced) sample
(b) After annealing at 1270 K for 10 min
(c) 90 min of K deposition
(d) 115 min of K deposition
(e) 140 min of K deposition
(f) 190 min of K deposition
(g) 250 min of K deposition
(h) Final annealing at 1270 K for 10 min
284.45 ± 0.03
284.46 ± 0.03
284.55 ± 0.03
284.54 ± 0.03
284.55 ± 0.03
284.56 ± 0.03
284.57 ± 0.03
284.44 ± 0.03
0.27 ± 0.01
0.25 ± 0.01
0.28 ± 0.01
0.29 ± 0.01
0.29 ± 0.01
0.29 ± 0.01
0.30 ± 0.01
0.27 ± 0.01
1.26 ± 0.01
1.22 ± 0.01
1.28 ± 0.01
1.31 ± 0.01
1.31 ± 0.01
1.31 ± 0.01
1.34 ± 0.01
1.25 ± 0.01
0
0
0.3 ± 0.2
0.3 ± 0.2
0.4 ± 0.2
0.4 ± 0.2
1.4 ± 0.2
0
Gaussian broadening, representing instrumental broadening and sample inhomogeneity, is fixed for all samples at 0.75 eV. The singularity
index is fixed at 0.15 [34]. The position of the shakeup peak for all spectra is 291.0 ± 0.1 e V.
1138
C A R B O N 4 6 (2 0 0 8) 1 1 33 –1 1 40

Page 7

nal shells of opened OLC. Photoelectron spectroscopy is only
sensitive to the outermost surface layers of a sample, with
most sensitivity for the external surface layer. For example,
at an excitation energy of 1253.6 eV, the kinetic energy of a
C1s photoelectron is around 970 eV – for this kinetic energy
the inelastic mean free path (IMFP) is 15 A˚ [45], which corre-
sponds to 4–5 graphite-like layers. At the excitation energy
of 40 eV the kinetic energy of photoelectrons originating from
valence band is around 12–37 eV. The IMFP for these electrons
is even lower than that for photoelectrons with kinetic energy
of 970 eV. Therefore, the observed lower shifts of the photo-
emission spectral features and almost unaltered DOS near
Fermi level for opened OLC upon potassium intercalation
can be simply explained by a smaller amount of potassium
present in the external surface layers of the opened OLC com-
pared with that for closed OLC, causing a reduction of charge
transfer to surface carbon atoms in the former sample. We
believe that this potassium deficiency on the external layers
arises from the penetration of potassium atoms deep inside
the carbon onions through holes made by oxidation of the
OLC by carbon dioxide.
Finally, it is interesting to note that the desorption of
potassium from opened OLC at 1270 K does not lead to a com-
plete recovery of the valence band spectrum: the DOS near
Fermi level for the opened OLC after potassium desorption
(Fig. 3B (d)) is higher than that for the initial sample (Fig. 3B
(a)). This behavior is different to that observed for closed
OLC (Fig. 3A) where we observed a complete recovery of all
spectral features after the final annealing step at 1270 K (com-
pare Fig. 3A (b) and (g)). A similar increase of DOS near Fermi
level was observed by photoelectron spectroscopy for single-
wall carbon nanotubes irradiated by argon ions [44]. This indi-
cates a possible irreversible structural disorder, which may
have been introduced into the opened carbon onions upon
potassium intercalation inside them.
4. Conclusions
Core level and valence band photoelectron spectroscopy have
been employed to study the effect of potassium intercalation
on the electronic structure of closed OLC, prepared by anneal-
ing of nanodiamonds at 1800 K, and opened OLC, prepared by
carbon dioxide treatment of closed OLC. The results show
that, for a given potassium dose, core level and valence band
features of opened OLC are less affected by potassium inter-
calation than those of closed OLC. This difference in behavior
of OLC samples upon potassium intercalation has been ex-
plained by penetration of potassium inside carbon onions in
the opened OLC. The diffusion of potassium inside the carbon
onions results in a reduction of the amount of transferred
charge to carbon atoms in external graphite-like layers, and
therefore in smaller shifts of core level and valence band fea-
tures in photoemission spectra of the opened OLC. Diffusion
of potassium inside OLC is indicative of their successful open-
ing by carbon dioxide oxidation. Thus, we envisage that the
opening of carbon onions can be used as a method to prepare
nanocapsules for different substances. Since the process of
the opening of carbon onions is performed separately from
their filling, the range of these substances can be increased
almost indefinitely.
Acknowledgments
We thank George Miller for his valuable technical support
throughout the experiment. We also wish to thank Ray Jones
and Chris Corrigan at the Materials Science Laboratory (SRS,
Daresbury Laboratory, UK) for their help in preparation of
the samples. YVB is grateful to the European Community’s
Sixth Framework Programme for award of a Marie Curie
Incoming International Fellowship (MIF1-CT-2005-021528).
AKC thanks the University of Nottingham, and EPSRC for
the financial support in the form of a Ph.D. studentship. This
work has also received support from the Royal Society and the
Council for the Science and Technology Facilities Council
(STFC), UK.
R E F E R E N C E S
[1] Subramoney S. Novel nanocarbons – structure, properties,
and potential applications. Adv Mater 1998;10(15):1157–71.
[2] Caiulo N, Yu CH, Yu KMK, Lo CCH, Oduro W, Thiebaut B, et al.
Carbon-decorated FePt nanoparticles. Adv Funct Mater
2007;17(8):1392–6.
[3] Che RC, Peng LM, Duan XF, Chen Q, Liang XL. Microwave
absorption enhancement and complex permittivity and
permeability of Fe encapsulated within carbon nanotubes.
Adv Mater 2004;16(5):401–5.
[4] McHenry ME, Majetich SA, Artman JO, Degraef M, Staley SW.
Superparamagnetism in carbon-coated co particles produced
by the Kratschmer carbon-arc process. Phys Rev B
1994;49(16):11358–63.
[5] Banhart F, Ajayan PM. Carbon onions as nanoscopic pressure
cells for diamond formation. Nature (London)
1996;382(6590):433–5.
[6] Banhart F, Redlich P, Ajayan PM. The migration of metal
atoms through carbon onions. Chem Phys Lett 1998;292(4–
6):554–60.
[7] Li JX, Banhart F. The deformation of single, nanometer-sized
metal crystals in graphitic shells. Adv Mater
2005;17(12):1539–42.
[8] Dravid VP, Host JJ, Teng MH, Elliot B, Hwang JH, Johnson DL,
et al. Controlled-size nanocapsules. Nature (London)
1995;374(6523):602–3.
[9] Ma YW, Hu Z, Yu LS, Hu YM, Yue B, Wang XZ, et al. Chemical
functionalization of magnetic carbon-encapsulated
nanoparticles based on acid oxidation. J Phys Chem B
2006;110(41):20118–22.
[10] Pantarotto D, Briand JP, Prato M, Bianco A. Translocation of
bioactive peptides across cell membranes by carbon
nanotubes. Chem Commun 2004;1:16–7.
[11] Ugarte D. How to fill or empty a graphitic onion. Chem Phys
Lett 1993;209(1–2):99–103.
[12] Saito Y. Nanoparticles and filled nanocapsules. Carbon
1995;33(7):979–88.
[13] Huang HJ, Yang SH, Gu G. Preparation of carbon-coated cobalt
nanocrystals in a new gas blow arc reactor and their
characterization. J Phys Chem B 1998;102(18):3420–4.
[14] Seraphin S, Zhou D, Jiao J. Filling the carbon nanocages. J
Appl Phys 1996;80(4):2097–104.
C A R B O N 4 6 ( 20 0 8 ) 1 1 3 3–11 4 0
1139

Page 8

[15] Jiao J, Seraphin S, Wang XK, Withers JC. Preparation and
properties of ferromagnetic carbon-coated Fe, Co, and Ni
nanoparticles. J Appl Phys 1996;80(1):103–8.
[16] Tsang SC, Qiu JS, Harris PJF, Fu QJ, Zhang N. Synthesis of
fullerenic nanocapsules from bio-molecule carbonisation.
Chem Phys Lett 2000;322(6):553–60.
[17] Harris PJF,Tsang SC.A simple techniquefor the synthesis of
filledcarbonnanoparticles.ChemPhysLett1998;293(1–2):53–8.
[18] Oku T, Schmid G, Suganuma K. Formation of carbon onions
with Pd clusters in a high-resolution electron microscope. J
Mater Chem 1998;8(9):2113–7.
[19] Flahaut E, Agnoli F, Sloan J, O’Connor C, Green MLH. CCVD
synthesis and characterization of cobalt-encapsulated
nanoparticles. Chem Mater 2002;14(6):2553–8.
[20] Tsang SC, Harris PJF, Green MLH. Thinning and opening of
carbon nanotubes by oxidation using carbon-dioxide. Nature
(London) 1993;362(6420):520–2.
[21] Tsang SC, Chen YK, Harris PJF, Green MLH. A simple chemical
method of opening and filling carbon nanotubes. Nature
(London) 1994;372(6502):159–62.
[22] Ajayan PM, Ebbesen TW, Ichihashi T, Iijima S, Tanigaki K,
Hiura H. Opening carbon nanotubes with oxygen and
implications for filling. Nature (London) 1993;362(6420):522–5.
[23] Sloan J, Cook J, Green MLH, Hutchison JL, Tenne R.
Crystallisation inside fullerene related structures. J Mater
Chem 1997;7(7):1089–95.
[24] Hirahara K, Suenaga K, Bandow S, Kato H, Okazaki T,
Shinohara H, et al. One-dimensional metallofullerene crystal
generated inside single-walled carbon nanotubes. Phys Rev
Lett 2000;85(25):5384–7.
[25] Suzuki S, Maeda F, Watanabe Y, Ogino T. Electronic structure
of single-walled carbon nanotubes encapsulating potassium.
Phys Rev B 2003;67(11):115418-1–6.
[26] Sˇvrc ˇek V, Pham-Huu C, Amadou J, Begin D, Ledoux MJ, Le
Normand F, et al. Filling and capping multiwall carbon
nanotubes with silicon nanocrystals dispersed in SiO2-based
spin on glass. J Appl Phys 2006;99(6):064306-1–9.
[27] Suzuki S, Bower C, Zhou O. In-situ TEM and EELS studies of
alkali-metal intercalation with single-walled carbon
nanotubes. Chem Phys Lett 1998;285(3-4):230–4.
[28] Montalti M, Krishnamurthy S, Chao Y, Butenko YuV,
Kuznetsov VL, Dhanak VR, et al. Photoemission spectroscopy
of clean and potassium-intercalated carbon onions. Phys Rev
B 2003;67(11):113401.
[29] Suzuki S, Tomita M. Observation of potassium-intercalated
carbon nanotubes and their valence-band excitation spectra.
J Appl Phys 1996;79(7):3739–43.
[30] Bennich P, Puglia C, Bruhwiler PA, Nilsson A, Maxwell AJ,
Sandell A, et al. Photoemission study of K on graphite. Phys
Rev B 1999;59(12):8292–304.
[31] Caragiu M, Finberg S. Alkali metal adsorption on graphite: a
review. J Phys-Condens Matter 2005;17(35):R995–R1024.
[32] Butenko YuV, Kuznetsov VL, Chuvilin AL, Kolomiichuk VN,
Stankus SV, Khairulin RA, et al. The kinetics of the
graphitization of dispersed diamonds at ‘low’ temperatures. J
Appl Phys 2000;88(7):4380–8.
[33] Kuznetsov VL, Chuvilin AL, Butenko YuV, Malkov IL, Titov
VM. Onion-like carbon from ultra-disperse diamond. Chem
Phys Lett 1994;222(4):343–8.
[34] Butenko YuV, Krishnamurthy S, Chakraborty AK, Kuznetsov
VL, Dhanak VR, Hunt MRC, et al. Photoemission study of
onion-like carbons produced by annealing nanodiamonds.
Phys Rev B 2005;71(7):075420.
[35] Doniach S, Sˇunjic ˇ M. Many-electron singularity in X-ray
photoemission and X-ray line spectra from metals. J Phys C
1970;3(2):285–91.
[36] Shirley DA. High-resolution X-ray photoemission
spectrum of valence bands of gold. Phys Rev B
1972;5(12):4709–14.
[37] Fink J, Mu ¨ller-Heinzerling Th, Pflu ¨ger J, Scheerer B, Dischler B,
Koidl P, et al. Investigation of hydrocarbon-plasma-
generated carbon-films by electron-energy-loss spectroscopy.
Phys Rev B 1984;30(8):4713–8.
[38] Skytt P, Glans P, Mancini DC, Guo JH, Wassdahl N, Nordgren J,
et al. Angle-resolved soft-X-ray fluorescence and absorption
study of graphite. Phys Rev B 1994;50(15):10457–61.
[39] Okotrub AV, Bulusheva LG, Kuznetsov VL, Butenko YuV,
Chuvilin AL, Heggie MI. X-ray emission studies of the valence
band of nanodiamonds annealed at different temperatures. J
Phys Chem A 2001;105(42):9781–7.
[40] Caballero A, Espinos JP, Fernandez A, Soriano L,
GonzalezElipe AR. Adsorption and oxidation of K deposited
on graphite. Surf Sci 1996;364(3):253–65.
[41] Kihlgren T, Balasubramanian T, Wallden L, Yakimova R. K/
graphite: uniform energy shifts of graphite valence states.
Surf Sci 2006;600(5):1160–4.
[42] Rauf H, Pichler T, Knupfer M, Fink J, Kataura H. Transition
from a Tomonaga–Luttinger liquid to a Fermi liquid in
potassium-intercalated bundles of single-wall carbon
nanotubes. Phys Rev Lett 2004;93(9):096805-1–4.
[43] Kuznetsov VL, Butenko YuV, Zaikovskii VI, Chuvilin AL.
Carbon redistribution processes in nanocarbons. Carbon
2004;42(5–6):1057–61.
[44] Chakraborty AK, Woolley RAJ, Butenko YuV, Dhanak VR,
Sˇiller L, Hunt MRC. A photoelectron spectroscopy study of
ion-irradiation induced defects in single-wall carbon
nanotubes. Carbon 2007;45(14):2744–50.
[45] Steinhardt RG, Hudis J, Perlman ML. Attenuation of low-
energy electrons by solids – results from X-ray photoelectron
spectroscopy. Phys Rev B 1972;5(3):1016–20.
1140
C A R B O N 4 6 (2 0 0 8) 1 1 33 –1 1 40