Electrochemical behavior of carbon alloy CxN prepared by CVD using a nickel catalyst

Article abstract of DOI:10.1016/S0013-4686(97)00263-6

Carbon-nitrogen layered compounds, C₁₄N to C₆₂N have been synthesized by thermal decomposition of acetonitrile at 800-1100°C in the presence of nickel catalyst. The C_xN samples prepared with nickel had higher crystallinity and less pyridine-type nitrogens existing at the edge of graphene layers than C_xN prepared in the absence of nickel. With an increase in the deposition temperature of C_xN, the cycleability for electrochemical intercalation-deintercalation of lithium ions was improved and the profile of the charge-discharge curve approached that of natural graphite powder due to an increase in the crystallinity and a decrease in the incorporated nitrogens. Surface oxidation of the C_xN by nitric acid solutions gave rise to an increase in the surface oxygens and a slight decrease in nitrogens without any change of crystallinity. The surface modification improved the cycleability for lithium ion intercalation-deintercalation reaction.

Full text of DOI:10.1016/S0013-4686(97)00263-6

Electrochimica Acta, Vol. 43, No. 8, pp. 883±891, 1998
# 1997 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: S0013-4686(97)00263-6
0013±4686/98 $19.00 + 0.00
Electrochemical behavior of carbon alloy CxN
prepared by CVD using a nickel catalyst
Tsuyoshi Nakajima,a* Meiten Koha and Masayuki Takashimab
aDivision of Polymer Chemistry, Graduate School of Engineering, Kyoto University, Sakyo-ku,
Kyoto 606-01, Japan
bDepartment of Material Science and Engineering, Faculty of Engineering, Fukui University, Bunkyo,
Fukui 910, Japan
(Received 17 January 1997; in revised form 7 May 1997)
AbstractÐCarbon±nitrogen layered compounds, C14N to C62N have been synthesized by thermal de-
composition of acetonitrile at 800±11008C in the presence of nickel catalyst. The CxN samples prepared
with nickel had higher crystallinity and less pyridine-type nitrogens existing at the edge of graphene layers
than CxN prepared in the absence of nickel. With an increase in the deposition temperature of CxN , the
cycleability for electrochemical intercalation±deintercalation of lithium ions was improved and the pro®le
of the charge±discharge curve approached that of natural graphite powder due to an increase in the crystal-
linity and a decrease in the incorporated nitrogens. Surface oxidation of the CxN by nitric acid solutions
gave rise to an increase in the surface oxygens and a slight decrease in nitrogens without any change of
crystallinity. The surface modi®cation improved the cycleability for lithium ion intercalation±deintercala-
tion reaction. # 1997 Elsevier Science Ltd. All rights reserved.
Key words: CxN, CVD, lithium ion battery, surface modi®cation.
INTRODUCTION
potentials in the lithium ion deintercalation process,
however, the potential sharply increases at the end
of the process. On the other hand, such carbon ma-
terials with lower crystallinity than natural graphite
as petroleum coke or boron- or nitrogen-substituted
carbon exhibit gradual increase in potential in
lithium ion deintercalation process. It was reported
that carbon±nitrogen compound CxN was not suit-
able as a negative electrode because a signi®cant
amount of nitrogen existing at the edge of graphene
layers was responsible for the irreversible capacity
of CxN, whereas the boron±carbon compound BCx
showed the high performance without such an irre-
versible capacity [6±8]. Substution of boron for car-
bon is more easily carried out than nitrogen
substitution because of reduced structural strain in
BCx. This may be the main reason why nitrogen
atoms are often replaced at the edge of graphene
layers. It was also found that air oxidation of car-
bon materials at high temperatures e€ectively
improved their electrochemical properties as nega-
tive electrodes in several cases [11±14]. Detailed stu-
dies were performed on the catalytic growth of
A graphite intercalation compound of lithium is
employed as a negative electrode for the lithium ion
secondary battery to avoid the dendrite formation
at the negative electrode. Carbon material with var-
ious morphologies and crystallinity have been
examined from the electrochemical and structural
viewpoints [1, 2]. It was discovered that highly crys-
tallized graphites and low crystallinity carbons have
high discharge capacities, whereas the capacities are
less in carbons with intermediate crystallinity [1, 2].
Recently boron- and/or nitrogen-substituted car-
bons BCxN, BCx and CxN were synthesized by
CVD and their electrochemical behavior as negative
electrodes were investigated [3±10]. These com-
pounds constitute a new class of candidates for
negative electrodes with modi®ed electronic struc-
tures, having di€erent chemical interactions with
lithium ions. On one hand, graphite with high crys-
tallinity such as natural graphite shows ¯at and low
*Corresponding author.
883

884
T. Nakajima et al.
Table 1.
Compositions, morphologies and XRD data of CxN and carbon samples
Reaction
temperature (8C) Composition
Morphology
Diameter (mm)
d(002) (nm)
FWHM (8)
Lc(002) (nm)
1100
1050
1000
900
C62Na
C33Na
C20Na
C21Na
Particle
Particle
Particle
Particle
Fiber
±
±
0.335
0.335
0.336
0.335
0.575
0.663
0.775
0.900
17
15
13
11
2±3
2±3
10.5
13, 0.3
800
C40Na
C14NHb0.6
Cc
Fiber
0.337
0.340
0.336
1.250
2.850
0.938
8
3
1000
900
±
±
±
±
10
aPrepared from acetonitrile (partial pressure: 9.2Â 103 Pa) with Ni catalyst.
bPrepared from acetonitrile (partial pressure: 9.2Â 103 Pa) without catalyst.
cPrepared from benzene (partial pressure: 9.2Â 103 Pa) with Ni catalyst.
carbon ®laments on iron±nickel alloy [15±18]. with ultrasonic agitation for 1 h. The oxidized
Growth of carbon ®laments are in¯uenced by the samples were dried at 1208C under a vacuum for
composition of iron±nickel alloy and additives. The half a day prior to use.
CxN ®laments and particles with high crystallinity
have been synthesized by thermal decomposition of
acetonitrile or pyridine using nickel or cobalt cata-
lyst at 800±11008C [19]. The electrochemical beha-
Measurement of speci®c surface areas of CxN and
carbon samples
Speci®c surface areas of as-prepared and surface-
vior of CxN samples thus prepared has been also oxidized CxN and carbon samples were measured
reported in a short communication [20]. After that, by nitrogen gas adsorption (BET method) using
we have obtained more better electrochemical data Micromeritics ASAP 2000. Samples were pretreated
by using new CxN electrodes instead of pellet elec- at 1208C under 0.26 Pa for 3 h prior to nitrogen gas
trodes. In this paper, we report the synthesis of a adsorption.
carbon±nitrogen compound CxN with high crystal-
linity and the electrochemical behavior of various
CxN samples.
Electrochemical measurement
The working electrode was prepared as follows.
CxN powder (ca. 30 mg) was dispersed into poly
(vinylidene ¯uoride) (PVDF)-dissolved 1-methyl-2-
pyrrolidone (CH3NC4H6O ca. 86 mg) so that the
concentration of PVDF might be 8 wt% in CxN
and PVDF mixture without solvent. The CxN-sus-
pended solution was then painted on a nickel grid,
which was dried at 1008C for 2 h and further at
1808C overnight to remove the solvent. The amount
of CxN stuck on a nickel grid was calculated from
the weight increase. The electrochemical measure-
ment was conducted using a beaker-type three-elec-
trodes cell with a cap in a dry box ®lled by argon.
The electrolyte solution was 1 M LiClO4±EC/DEC
(1:1) supplied by Mitsubishi Chemical Corporation.
The electrochemical behavior was evaluated by
charge±discharge cycling at a current density of
EXPERIMENTAL
Synthesis and surface oxidation of CxN samples
The CxN samples were synthesized by thermal
decomposition of acetonitrile (purity: 99%) at 800±
11008C. A mixture of acetonitrile and nitrogen gas
(partial pressure of acetonitrile: 9.2Â 103 Pa) was
allowed to ¯ow over metallic nickel powder in an
alumina boat, which was placed in an alumina reac-
tor, with total ¯ow rate 55 mL min 1. The 400±
500 mg of the CxN sample was obtained by a half
day reaction. The CxN samples were analyzed by el-
emental analysis, X-ray di€ractometry, SEM, TEM
and XPS. Binding energies for XPS peaks were
determined relative to that of C1 s electron of
graphite, 284.3 eV without charging correction. The
spectrometer used was Ulvac Phi Model 5500. The
following carbon materials were used for compari-
son: natural graphite powder (average diameter: ca.
7 mm), CxN prepared at 10008C in the absence of
nickel, and carbon, prepared by pyrolysis of ben-
zene at 9008C in the presence of nickel.
1
30 mA g at 258C.
RESULTS AND DISCUSSION
Compositions, structure and crystallinity of as-
prepared and surface-modi®ed CxN samples
Table 1 shows the typical examples of compo-
The surface oxidation of CxN samples was con- sitions, morphologies and X-ray di€raction data of
ducted in 3, 6 or 10 M (M: mol dm 3) HNO3 sol- CxN samples prepared at 800±11008C with nickel
ution under stirring for half a day and afterward catalyst in comparison with those of CxN prepared

CxN prepared by CVD
885
incorporation in graphene layers. This e€ect would
lead to the decrease in nitrogen content in CxN.
The CxN samples prepared were apparently black
powder. However, SEM and TEM observations
revealed that the products obtained at 8008C were
®brous CxN with diameters of ca. 3 and 0.3 mm,
those at 9008C were mixtures of ®brous CxN with
diameters of ca. 0.5 mm and ®ne CxN particles with
diameters of 2±3 mm, and those at 10008C were
CxN particles with diameters of 2±3 mm. Typical
SEM images of CxN samples prepared at 900 and
10008C are shown in Fig. 1.
It was also found by X-ray di€raction and el-
emental analysis that ca. 7 wt% nickel was mixed
with CxN. The CxN samples obtained using nickel
catalyst exhibited the same (002), (004), (100), (101)
and (110) di€raction lines as usual carbon ma-
terials, which indicates that CxN has the same
layered sheet structure as carbon or graphite. In
this paper, we use the same indexes for CxN
samples as used for carbon materials. The d(002)
values and half widths of (002) di€raction lines
decreased from 0.337 to 0.335 nm and from 1.250
to 0.5758, respectively, with increasing thermal de-
composition temperature of acetonitrile from 800 to
11008C as given in Table 1. Both d(002) values and
half widths were smaller than those for C14NH0.6
prepared without nickel, indicating that the CxN
samples synthesized using nickel have much higher
crystallinity than that prepared without nickel.
Therefore, the crystallite sizes along the c-axis, that
is Lc(002) values calculated from the half widths of
(002) di€raction lines were also much larger than
that of C14NH0.6, increasing with an increase in
temperature. On the other hand, the carbon pre-
pared by pyrolysis of benzene at 9008C with nickel
catalyst shows similar values in d(002), half width
and crystallite size to those of C21N prepared at
9008C with nickel. The obtained results indicate
that nickel e€ectively facilitates the growth and
crystallization of CxN ®laments and particles as
reported in the literatures [15±19].
Fig. 1. SEM images of C20N (a) and C21N (b) prepared
with nickel catalyst at 10008C and 9008C, respectively.
without the catalyst and carbon prepared from ben-
zene with nickel. The catalytic e€ect of iron±nickel
alloy on carbon ®lament growth was already inves-
tigated [15±18]. It was also found that metallic
nickel or cobalt powder well catalyzes the growth
of high crystalline CxN ®laments and particles [19].
It is known that iron is a good catalyst for the
growth of well crystallized carbon ®bers, however,
the use of metallic iron powder was unsuccessful
for the preparation of CxN. Iron was deactivated
during CVD probably due to the reaction with
nitrogen [21].
CxN samples with compositions C14N to C62N
were obtained in the temperature range 800±
10008C. No hydrogen was detected by elemental
analysis for all the samples prepared using nickel
catalyst, whereas less crystallized C14NH0.6 having
hydrogen was obtained at 10008C in the absence of
the catalyst. The products prepared at 9008C and
10008C had the highest nitrogen contents, as C14N
to C21N. Nitrogen concentrations were much less in
the products deposited at the lower and higher tem-
peratures, 800, 1050 and 11008C (C40N, C33N and
The results of surface analysis by XPS are given
in Figs 2 and 3 and Table 2. Figure 2 shows Nls
spectra for CxN samples prepared at 10008C with
and without nickel catalyst. CxN samples prepared
at 9008C have the same pro®les as those shown in
Fig. 2. C20N prepared with nickel exhibits three
peaks at 398.8, 400.9 and 402.3 eV, which are
C62N, respectively). Two di€erent factors should be ascribed to pyridine-type nitrogen existing at the
considered for the temperature dependency of nitro- edge of graphene layer, quarternary nitrogen incor-
porated in graphene layer and pyridine-N-oxide,
gen concentration. Nitrogen solubility in nickel is
estimated to increase with increase in the tempera-
ture from the positive values of enthalpy and
entropy changes for nitrogen dissolution in nickel
[22], which would enhance the incorporation of
nitrogen in CxN. On the other hand, the tempera-
ture raise facilitates the crystallization of CxN with
+
–
N
O
at the edge of the graphene layer, respectively [23±
26]. A small amount of oxygen detected may have
reducing the structural strain induced by nitrogen been introduced into CxN by exposure to the air.

886
T. Nakajima et al.
Fig. 3. O1 s XPS spectra of CxN samples prepared at
10008C. (a) C20N as-prepared with nickel, (b) C20N oxi-
dized by 6 M HNO3 solution.
between those for quarternary and pyridine-N-
oxide. The spectrum was observed at slightly lower
binding energy probably by the loss of charging due
to the decrease of nitrogen in the sample. The new
peak at 399.2 eV in Fig. 2(b) may be ascribed to
amine (0NH2), which may be formed along with a
carboxyl group by carbon±nitrogen bond rupture
of pyridine-N-oxide. In the Ols spectra shown in
Fig. 3, two peaks indicating carbonyl and hyroxyl
groups were observed at 531.0±531.2 and 532.4±
532.8 eV, respectively, in both as-prepared and sur-
face-modi®ed samples with a small peak attributed
to adsorbed water molecules at ca. 535 eV.
Corresponding to Ols spectra, the shifted peaks
were observed in Cls spectra at 285.6 and 287.9 eV,
indicating carbonyl and hydroxyl groups, and car-
boxyl group, respectively. The contribution due to
carbon±nitrogen bonds is included in the Cls peak
Fig. 2. N1 s XPS spectra of CxN samples prepared at
10008C. (a) C20N as-prepared with nickel, (b) C20N oxi-
dized by 6 M HNO3 solution and (c) C14NH0.6 as-pre-
pared without nickel.
The peak at 400.9 eV indicating nitrogen atoms in-
corporated in the graphene layer has a higher inten-
sity than other two peaks located at 398.8 and
402.3 eV, corresponding to nitrogen atoms existing
at the edge of the graphene layer when CxN is pre-
pared using nickel catalyst [Fig. 2(a)], whereas CxN
prepared without nickel has a large amount of pyri-
dine-type nitrogens [Fig. 2(c)]. The use of nickel cat-
alyst is thus important not only because the
crystallinity of CxN is signi®cantly improved as
mentioned above, but also because the pyridine-
type nitrogens existing at the edge of graphene
layers are highly reduced. On the other hand, C20N
oxidized by 6 M HNO3 solution shows a new peak located at 285.6 eV.
Table 2.
N/C and O/C ratios at the CxN and carbon sample surfaces, obtained from peak areas of XPS spectra
C
20Na
3 M HNO3
6 M HNO3
10 M HNO3
N/C (Â102)
O/C (Â102)
5.95
2.80
5.49
10.6
5.61
10.1
5.18
10.9
C21Nb
3 M HNO3
6 M HNO3
10 M HNO3
N/C (Â102)
O/C (Â102)
6.49
2.60
4.52
10.1
3.70
11.9
5.32
11.0
Cc
±
±
6 M HNO3
4.29
±
O/C (Â102)
4.20
aPrepared from acetonitrile at 10008C with Ni catalyst.
bPrepared from acetonitrile at 9008C with Ni catalyst.
cPrepared from benzene at 9008C with Ni catalyst.

CxN prepared by CVD
887
Table 3.
Speci®c surface areas of as-prepared and surface-oxidized CxN and carbon samples
Reaction
temperature (8C)
Speci®c surface areas
1
1
Composition
As-prepared sample (m2g
)
Oxidized samplea (m2g
)
1100
1000
900
C62Nb
C20Nb
C21Nb
C40Nb
1.55
1.50
2.30
10.5
±
3.20
4.40
±
800
1000
C14NHc0.6
Cd
16.0
6.90
±
900
5.40
aOxidized by 6 M HNO3 solution.
bPrepared from acetonitrile (partial pressure: 9.2Â 103 Pa) with Ni catalyst.
cPrepared from benzene (partial pressure: 9.2Â 103 Pa) without Ni catalyst.
dPrepared from benzene (partial pressure: 9.2Â 103 Pa) with Ni catalyst.
Table 2 summarizes the surface concentrations of large dependency was found on the concentrations
nitrogen and oxygen of as-prepared and surface- of nitric acid solutions used for the sample oxi-
dation.
Table 3 shows the speci®c surface areas of as-
modi®ed samples and carbon derived from benzene.
The amounts of nitrogens were somewhat decreased
and at the same time those of oxygen were largely
increased in both C20N and C21N, whereas the
amounts of oxygen in benzene-derived carbon were
nearly the same before and after the oxidation. No
prepared and surface-oxidized CxN and carbon
samples. The speci®c surface areas of CxN samples
prepared with nickel catalyst were signi®cantly
smaller than that of C14NH0.6 prepared without cat-
alyst, having a trend to decrease with increasing de-
position temperature. The carbon prepared at
9008C from benzene had a larger surface area than
C21N prepared at the same temperature. These
trends well coincide with the di€erence in crystalli-
nity of the CxN and carbon samples, shown by the
X-ray di€raction data in Table 1, that is, the sur-
face area is reduced with increasing crystallinity of
the sample. The surface areas of C20N and C21N
were enlarged approximately twice by the oxidation.
The increased surface oxygens and areas given in
Tables 2 and 3 suggest that CxN samples were
highly oxidized in addition to the formation of
amine (0NH2) and carboxyl groups by carbon±
nitrogen bond rupture of pyridine-N-oxide. On the
other hand, the speci®c surface area of benzene-
derived carbon slightly decreased by oxidation. In
addition, the amount of surface oxygen was nearly
constant before and after the oxidation as shown in
Table 2. These results would imply that the disor-
dered parts of benzene-derived carbon were lost as
gaseous carbon oxides.
Charge±discharge behavior of CxN samples
Figures 4, 5, 6 and 7 show typical charge±dis-
charge curves for as-prepared and surface-modi®ed
CxN samples. For comparison, those of natural
graphite powder and benzene-derived carbon are
also shown in Figs 8 and 9. In the electrochemical
cell used in this study, the electrodes of carbon ma-
terials are positive electrodes since the counter elec-
trode is lithium metal, therefore, the deintercalation
of lithium ions is a charge process. Flat potential
curves and nearly constant charge capacities were
observed for natural graphite electrode as seen in
Fig. 4. Charge±discharge curves for C21N prepared from
acetonitrile at 9008C with nickel catalyst. (a) As-prepared
C21N, (b) C21N oxidized by 6 M HNO3 solution. lst, 5th
1
and 10th cycles at 30 mAg
.

888
T. Nakajima et al.
Fig. 6. Charge±discharge curves for C33N prepared from
acetonitrile at 10508C with nickel catalyst. lst, 5th and
1
10th cycles at 30 mAg
.
deposition temperature. The charge capacities some-
what decreased with increasing cycle number in
both C20N and C21N as-prepared at 1000 and
9008C, respectively. The cycleability was, however,
improved for C33N and C62N deposited at 1050
and 11008C probably due to the increase in the
crystallinity. Comparing the charge±discharge
curves of C20N prepared at 10008C with those of
C21N prepared at 9008C, both charge and discharge
capacities were larger in C20N than in C21N. This
may be because C20N has slightly higher crystalli-
nity than C21N as given in Table 1.
Fig. 5. Charge±discharge curves for C20N prepared from
acetonitrile at 10008C with nickel catalyst. (a) As-prepared
C20N and (b) C20N oxidized by 3 M HNO3 solution. lst,
1
Surface oxidation of CxN samples by nitric acid
solutions signi®cantly improve the cycleability and
changed the charge-discharge pro®les as shown in
Figs 4(b) and 5(b). The surface-modi®ed C21N and
C20N gave the similar charge±discharge curves to
each other in spite of the di€erent concentrations of
nitric acid solutions employed for the oxidation. In
Figs 4(b) and 5(b), the smaller polarization is
observed for the ®rst discharge curves and the lar-
ger one for the charge curves. The smaller polariz-
ation for the ®rst discharge curves would be
5th and 10th cycles at 30 mAg
.
Fig. 8. The charge capacities of natural graphite
powder up to 3 V vs Li were as high as 360
mAh g 1. The benzene-derived carbon, however,
exhibited a gradual change in the potential with less
1
charge capacity of 310 mAh g than natural graph-
ite powder because of its lower crystallinity, as
shown in Fig. 9. The cycleability of benzene-derived
carbon was comparable to that of natural graphite
powder, although the ®rst coulombic eciency was
lower. The potentials of CxN electrodes gradually induced by the interaction of lithium ions with sur-
decreased and increased with lithium ion intercala- face oxygen groups. In the charging process, the po-
tion and deintercalation. Not only the variation of
the potentials, but also the capacity and cycleability
were dependent on the nitrogen content and crystal-
linity of CxN samples. Nitrogen incorporation in
carbon induced the large polarization for lithium
ion deintercalation process and a small one for the
intercalation process. The polarization observed
suggests that the lithium ion interacts with a lone
pair of the nitrogen atom in CxN. Table 1 indicates
that nitrogen content decreased and crystallinity
increased with increase in the deposition tempera-
ture from 900 to 11008C. According to this trend in
the nitrogen content and crystallinity of CxN, the
pro®le of the charge±discharge curves of CxN
samples approached those of benzene-derived car-
Fig. 7. Charge±discharge curves for C62N prepared from
bon and natural graphite powder as shown in Figs 6
acetonitrile at 11008C with nickel catalyst. lst, 5th and
1
and 7, while the capacity decreased with increasing 10th cycles at 30 mAg
.

CxN prepared by CVD
889
Fig. 8. Charge±discharge curves for natural graphite
(average diameter: 7 mm). lst, 5th and 10th cycles at
1
30 mA g
.
Fig. 10. Change in the charge capacities of as-prepared
and surface-modi®ed C20Ns, as-prepared C14NH0.6 and
1
tential varied almost linearly after ca. 170 mAh g
,
natural graphite as a function of cycle number.
as-prepared at 10008C with nickel catalyst, * C20N oxi-
dized by 3 M HNO3 solution, C20N oxidized by 6 M
C20N
which may be due to the interaction of lithium ions
with amine (0NH2) formed by nitric acid treat-
ment. The charge and discharge capacities were
slightly reduced at the 2nd cycle, but almost no
change in the charge±discharge curves was observed
after 2nd cycle in both C21N and C20N. One reason
for the improvement in cycleability may be the
enlargement of surface areas by oxidation.
HNO3 solution,
out nickel and
C14NH0.6 as-prepared at 10008C with-
natural graphite (average diameter:
7 mm). Current: 30 mA g 1, potential: 0.0±3.0 V.
benzene-derived carbons are lower than that of
natural graphite powder probably because of its
lower crystallinity.
Surface oxidation using nitric acid solution did
not improve the charge±discharge behavior of ben-
zene-derived carbon. The charge±discharge curves
of benzene-derived carbon in Fig. 9 are rather simi-
lar to those for C21N and C20N in the potential
range between 0.0 and 1.0 V, whereas they were
close to those for natural graphite in the range
between 1.0 V and 3.0 V (Fig. 8). Considering that
the crystallinity of benzene-derived carbon is similar
to that of C21N prepared at 9008C, the main di€er-
ences of C21N from benzene-derived carbon would
be the nitrogen incorporation into carbon and
higher surface oxygen concentration. These factors
may give rise to an increase in the chemical inter-
action with lithium ions. The charge capacities of
Figures 10 and 11 show the change in the charge
capacities of CxN samples in comparison with other
materials as a function of cycle number. As shown
in Fig. 10, C20N prepared at 10008C with nickel
exhibited lower charge capacities than C14NH0.6
Fig. 11. Change in the charge capacities of as-prepared
and surface-modi®ed C21Ns and benzene-derived carbon
as a function of cycle number.
C21N as-prepared at
9008C with nickel catalyst, * C21N oxidized by 3 M
HNO3 solution, C21N oxidized by 6 M HNO3 solution,
Fig. 9. Charge±discharge curves for benzene-derived car-
bon prepared at 9008C and oxidised by 10 M HNO3
C21N oxidized by 10 M HNO3 solution and
carbon
prepared from benzene at 9008C with nickel catalyst.
1
solution. lst, 5th and 10th cycles at 30 mA g
.
Current: 30 mA g 1, potential: 0.0±3.0 V.

890
T. Nakajima et al.
Table 4.
Charge and discharge capacities between 0 and 3 V vs Li/Li+ and coulombic eciencies of CxN and carbon samples
1st cycle
Charge
10th cycle
Charge
Composi-
tion
Discharge
1
Q
(%)
Discharge
1
Q
(%)
1
1
(mAh g
)
(mAh g
)
(mAh g
)
(mAh g
)
C62Na
C33Na
C20Na
C20Nb
C21Na
C21Nc
404
487
567
657
560
697
302
332
408
414
390
424
74.7
68.2
72.0
63.0
69.6
60.8
308
325
387
420
367
418
301
320
391
411
362
408
97.7
98.5
98.4
97.8
98.6
97.6
C14NHd0.6
Ce
677
435
430
306
63.5
70.3
393
312
382
306
97.2
98.1
aPrepared from acetonitrile (partial pressure: 9.2Â 103 Pa) with Ni catalyst.
bOxidized by 3 M HNO3 solution.
cOxidized by 6 M HNO3 solution.
dPrepared from acetonitrile (partial pressure: 9.2Â 103 Pa) without Ni catalyst.
ePrepared from benzene (partial pressure: 9.2Â 103 Pa) with Ni catalyst.
prepared without nickel at the same temperature. pared at 9008C. A possible explanation may be that
the stronger oxidation is needed to enlarge the sur-
face area of the ®brous component of C21N for
easier lithium ion intercalation and deintercalation.
Figure 12 shows the coulombic eciencies for
C14NH0.6, C20N, C21N, natural graphite powder
and benzene-derived carbon. The charge and dis-
charge capacities and coulombic eciencies at 1st
and 10th cycles are summarized in Table 4. The
coulombic eciencies of C14NH0.6 were inferior to
those of other samples. A problem in the electro-
chemical characteristics of CxN samples as-prepared
with nickel catalyst is that their coulombic ecien-
cies were 68±75% lower than that of natural graph-
ite powder. Surface oxidation further reduced the
®rst coulombic eciencies by 9% in the case of
C20N and C21N as given in Table 4. The decrease
in the eciencies may be due to an increase in the
surface oxygen groups by oxidation.
However, their capacities approached each other
after the 7th cycle, showing the similar trend with
cycle number. The surface oxidation of C20N by
3 M and 6 M HNO3 solutions improved the cycling
behavior. C20N as-prepared at 10008C with nickel
exhibited higher charge capacities than C21N as pre-
pared at 9008C probably because C20N had slightly
higher crystallinity than C21N. The oxidation of
C21N with 6 M and 10 M nitric acid solutions sig-
ni®cantly improved cycling behavior as shown in
Fig. 11. The charge capacity was increased and
stabilized with increasing concentration of nitric
acid. The optimum concentrations were 3 M and
6 M HNO3 solutions for C20N prepared at 10008C
and 6 M and 10 M HNO3 solutions for C21N pre-
Factors in¯uencing the charge±discharge behavior
The charge±discharge behavior should be dis-
cussed in relation with several structural and com-
positional factors of carbon materials, which would
be the crystallinity, surface area, chemical species
interacting with lithium ions and so on. The in¯u-
ence of the crystallinity on the charge±discharge
behavior was intensively investigated [1, 2]. It is
known that highly crystallized graphites and poorly
crystalline carbons have high discharge capacities
[1, 2]. In the present study, CxN samples with high
crystallinity and less pyridine-type nitrogens were
obtained by the use of nickel powder as a catalyst
for the CVD reaction. With increasing deposition
temperature, the crystallinity of CxN increases,
while its nitrogen concentration and speci®c surface
area decrease. The increase in crystallinity and
decrease in nitrogen concentration improve the
cycleability and induce the approach of the charge±
Fig. 12. Coulombic eciencies for carbon materials as a
function of cycle number. W C14NH0.6 prepared at 10008C
without nickel, * C20N oxidized by 3 M HNO3 solution,
C21N
oxidized by 6 M HNO3 solution,
graphite and benzene-derived carbon.
natural

CxN prepared by CVD
891
8. B. M. Way and J. R. Dahn, J. Electrochem. Soc. 141,
907 (1994).
9. M. Ishikawa, T. Nakajima, M. Morita, Y. Matsuda,
S. Tsujioka and T. Kawashima, J. Power Sources 55,
127 (1995).
10. K. Nishimura, T. Takahashi, M. Endo, T. Tamaki and
M. S. Dresselhaus, Tanso 172, 89 (1996).
11. M. Hara, A. Satoh, N. Takami and T. Ohsaki, Tanso
165, 261 (1994).
12. T. Nakajima and K. Yanagida, Tanso 174, 195 (1996).
13. E. Peled, C. Menachem, D. Bar-Tow and A. Melman,
J. Electrochem. Soc. 143, L4 (1996).
14. T. Takamura and M. Kikuchi, Denchi Gijutsu 7,
29 (1995).
15. N. M. Rodriguez, M. S. Kim, F. Fortin, I. Mochida
R. T. K. Baker, Appl. Catal. A: General 148,
265 (1997).
discharge curve to that of natural graphite powder,
while the capacity is somewhat reduced. It has been
also shown that surface oxidation is
a useful
method to improve the cycleability of CxN, how-
ever, the increased amount of oxygen reduces the
®rst coulombic eciency.
ACKNOWLEDGEMENTS
The electrolyte solution used in the present study
was supplied by courtesy of Mitsubishi Chemical
Corporation, which the authors gratefully appreci-
ate. This research was partially supported by the
Ministry of Education, Science, Sports and Culture,
Grant-in-Aid for Scienti®c Research on Priority
Areas (Carbon Alloys) No. 288, 1997.
16. R. T. K. Baker and J. J. Chludzinski Jr, J. Catal. 64,
464 (1980).
17. R. T. K. Baker, Carbon 127, 315 (1989).
18. N. M. Rodringuez, J. Mater. Res. 8, 3233 (1993).
19. T. Nakajima and M. Koh, Carbon 35, 203 (1997).
20. T. Nakajima and M. Koh, Denki Kagaku 64,
917 (1996).
21. R. P. Elliott, Constitution of Binary Alloys, First
Supplement, pp. 326, 420, 639. McGraw-Hill, New
York (1965).
22. Phase Diagrams of Binary Nickel Alloys (Edited by
P. Nash), p. 216. ASM International, Ohio (1991).
23. J. R. Pels, F. Kapteijn, J. A. Moulijn, Q. Zhu and
K. M. Thomas, Carbon 33, 1641 (1995).
24. A. N. Buckley, Fuel Process, Technol. 38, 165 (1994).
25. S. R. Keleman, M. L. Corbaty and P. J. Kwiatek,
Energy and Fuels 8, 896 (1994).
REFERENCES
1. A. Mabuchi, Tanso, 165, 298 (194), and references
therein.
2. K. Tatsumi, Denchi Gijutsu, 7, 68 (1995), and refer-
ences therein.
3. M. Kawaguchi and N. Bartlett, Fluorine±carbon and
¯uoride±carbon materials (Edited by T. Nakajima),
Ch. 5. Marcel Dekker, New York (1995).
4. M. Morita, T. Hanada, H. Tsutsumi, Y. Matsuda and
M. Kawaguchi, J. Electrochem. Soc. 139, 1227 (1992).
5. Y. Matsuda, M. Morita, T. Hanada and M.
Kawaguchi, J. Power Sources 43/44, 75 (1993).
6. M. Ishikawa, M. Morita, T. Hanada, Y. Matsuda and
M. Kawaguchi, Denki Kagaku 61, 1395 (1993).
7. W. J. Weydanz, B. M. Way, T. van Buuren and J. R.
Dahn, J. Electrochem. Soc. 141, 900 (1994).
26. K. Stanczyk, R. Dziembaj, X. Piwowarska and S.
Witkowski, Carbon 33, 1383 (1995).