Synthesis and structural characterization of carbon powder by electrolytic reduction of molten Li2Co3-Na2Co3-K2CO3

Article abstract of DOI:10.1149/1.1464884

A new method was proposed for the preparation of carbon powders based on the electrochemical reduction of a fused eutectic mixture of lithium-sodium-potassium carbonates at 450°C. Transmission electron microscopy observations have revealed the presence of three forms of carbon in the powder samples depending on the deposition potential values: amorphous carbon, graphite, and fibers. The presence of graphite and amorphous carbon was confirmed by X-ray diffraction measurements. The d₀₀₂ values were rather constant whatever the sample and close to 0.34 nm. Moreover, surface analyses by X-ray photoelectron spectroscopy revealed the presence of nearly metallic and ionic lithium belonging to lithium oxides. The specific areas were measured by the Brunauer-Emmett-Teller (BET) method. Whatever the deposit potential values, the total BET specific surface area is much higher than that corresponding to external one because of the pore walls contribution. It has been shown that, for carbon powder dried at 400°C under vacuum, the specific surface area decreased as the deposition potential became more cathodic: from about 850 m² g^-1 at -2.4 V vs. CO₂-O₂ to about 500 m² g^-1 at -6.0 V vs. CO₂-O₂. With increasing negative values of the potential, the nucleation and the growth of carbon competes with the formation of nearly metallic lithium and/or lithium oxides. The presence of the latter induces the closure of some nanopores.

Full text of DOI:10.1149/1.1464884

D72
Journal of The Electrochemical Society, 149 ͑5͒ D72-D78 ͑2002͒
0013-4651/2002/149͑5͒/D72/7/$7.00 © The Electrochemical Society, Inc.
Synthesis and Structural Characterization of Carbon Powder
by Electrolytic Reduction of Molten Li2CO3-Na2CO3-K2CO3
,z
B. Kaplan,a H. Groult,a, A. Barhoun,a F. Lantelme,a T. Nakajima,b,
*
*
c,
c,
b,**
*
*
V. Gupta,
S. Komaba, and N. Kumagai
a
´
Universite Pierre and Marie Curie, Centre National de la Recherche Scientifique, UMR 7612,
75252 Paris Cedex 05, France
bDepartment of Applied Chemistry, Aichi Institute of Technology, Yakusa, Toyota 470-0392, Japan
cFaculty of Engineering, Iwate University, Morioka 020-8551, Japan
A new method was proposed for the preparation of carbon powders based on the electrochemical reduction of a fused eutectic
mixture of lithium-sodium-potassium carbonates at 450°C. Transmission electron microscopy observations have revealed the
presence of three forms of carbon in the powder samples depending on the deposition potential values: amorphous carbon,
graphite, and fibers. The presence of graphite and amorphous carbon was confirmed by X-ray diffraction measurements. The d002
values were rather constant whatever the sample and close to 0.34 nm. Moreover, surface analyses by X-ray photoelectron
spectroscopy revealed the presence of nearly metallic and ionic lithium belonging to lithium oxides. The specific areas were
measured by the Brunauer-Emmett-Teller ͑BET͒ method. Whatever the deposit potential values, the total BET specific surface area
is much higher than that corresponding to external one because of the pore walls contribution. It has been shown that, for carbon
powder dried at 400°C under vacuum, the specific surface area decreased as the deposition potential became more cathodic: from
about 850 m2 gϪ1 at Ϫ2.4 V vs. CO2-O2 to about 500 m2 gϪ1 at Ϫ6.0 V vs. CO2-O2 . With increasing negative values of the
potential, the nucleation and the growth of carbon competes with the formation of nearly metallic lithium and/or lithium oxides.
The presence of the latter induces the closure of some nanopores.
© 2002 The Electrochemical Society. ͓DOI: 10.1149/1.1464884͔ All rights reserved.
Manuscript received April 18, 2001. Available electronically March 29, 2002.
Carbonaceous materials are extensively used in electrochemical
ing the deposition of carbon is the reduction of the carbonate ions,
and therefore, the amount of carbon obtained is strictly dependent
on the carbonate ion content.
processes.1 For example, carbon is used as anodes during the elec-
trolysis of molten KF-2HF for the preparation of fluorine,2 in re-
chargeable lithium-ion batteries,3-6 and in supercapacitors.7
In the present work, a new approach is proposed based on the
deposition of carbon powder by electrolytic reduction of the liquid
ternary system Li2CO3-Na2CO3-K2CO3 at 450°C. This reduction
process could give rise to substantial deposits of carbon powder.17
The reaction mechanism is briefly discussed based on the analysis of
cyclic voltammograms ͑CVs͒ depending on the electrode substrate
͑carbon and nickel͒. Structural characteristics of the carbon pow-
ders, performed by X-ray photoelectron spectroscopy ͑XPS͒, X-ray
diffraction ͑XRD͒, transmission electron microscopy ͑TEM͒, and
scanning electron microscopy ͑SEM͒ are presented. Finally, the spe-
cific surface area and the pore texture of the carbon powder samples
are determined by nitrogen gas adsorption using the Brunauer-
Emmett-Teller ͑BET͒ method.
Several varieties of carbon, from highly crystalline graphite to
strongly disordered carbons, are commercially available. The use of
carbon materials depends on their structural characteristics; for ex-
ample, graphite cannot be used as anode material for the electrolytic
preparation of fluorine since exfoliation occurs. However, many
types of carbon have been tested as alternative anodes for recharge-
able lithium batteries: graphite,5,8-10 carbon nanotubes,11 disordered
soft ͑graphitizable͒,12 hard carbons ͑ungraphitizable͒,13 etc. The use
of such carbon anodes allows prevention of the well-known problem
of dendritic shape of metallic lithium deposited during charge steps.
In that specific example, graphite is still one of the most desirable
candidates in lithium-ion batteries due to its excellent intercalation
properties: good electric conductivity, low irreversible capacity, and
low electrochemical voltage vs. Li/Liϩ. Its structure is formed
by planar carbon layers stacked with a sequence of the type
ABABAB... This particular lamellar structure is particularly well
adapted to the intercalation of lithium occurs between the carbon
sheets.
Experimental
The electrochemical cell is described in Fig. 1. The electrolyte
was composed of the eutectic mixture of 43.5 mol % of Li2CO3 ,
31.5 mol % of Na2CO3 , and 25 mol % of K2CO3 ͑Prolabo R. P.
Normapur͒. The constituent salts were previously dehydrated under
Except natural graphite, carbon can be obtained by pyrolysis of
organic compounds. This reaction which takes place at temperatures
up to 1000°C can be followed by a calcination at 1600°C for remov-
ing gases included into the crystalline structure. Carbon can also be
obtained by chemical vapor deposition from the thermal decompo-
sition of hydrocarbons. Carbon coatings or carbon with special elec-
trochemical properties can also be obtained by an electrochemical
route.14-18 Recently, Kawamura and Ito18 studied the electrodeposi-
tion of carbon films on aluminum in fused LiCl-KCl-K2CO3 at
500°C. These authors have shown that, depending on the electrolytic
conditions, the morphology and the cohesiveness of the deposited
carbon film are changed. The electrochemical reaction involved dur-
* Electrochemical Society Active Member.
** Electrochemical Society Student Member.
z E-mail: groult@ccr.jussieu.fr
Figure 1. Electrochemical cell used for the deposition of carbon.
Downloaded on 2014-10-16 to IP 134.71.135.126 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 149 ͑5͒ D72-D78 ͑2002͒
D73
Figure 3. SEM micrograph of sample 3.
mixture of dry CO2 :O2 ͑2/3:1/3͒ gases was maintained above the
solution to have an atmosphere in equilibrium with the melt. A small
hole in the bottom of the tube allows the ionic contact. This refer-
ence electrode is denoted CO2-O2 /CO23Ϫ ; all potentials are referred
to that reference, henceforth. The counter electrode was a sheet of
gold.
The XRD patterns of carbon powders were obtained using a
Shimadzu XD-610 X-ray diffractometer with Cu K␣ radiation
(␭ ϭ 1.54178 Å͒ and a nickel filter. XPS analysis using a Ulvac-Phi
model 5500 spectrometer with a Mg K␣ excitation source was per-
formed in order to determine the chemical interaction between car-
bon, lithium, and oxygen. The spectrometer was calibrated with cop-
per at binding energies 75.1 and 932.6 eV for Cu 3p3/2 and Cu 2p3/2
respectively.
,
TEMs were obtained with a JEOL JEM 100 CX II transmission
electron microscope equipped with a JEOL high resolution scanning
attachment ͑STEM-SEM ASID 4D͒ and an energy dispersive X-ray
detector ͑Li-doped Si with an ultrathin window͒.
SEM images were obtained with a Leica S 440 camera.
The specific area of carbon samples were deduced from nitrogen
adsorption experiments ͑BET method͒ at boiling temperature
͑77 K͒.
Figure 2. ͑a͒ Voltammograms at
a GC electrode at 450°C in
Li2CO3-Na2CO3-K2CO3 . Sweep rate: 0.1 V sϪ1. S Ϸ 5 cm2. Reference
electrode: CO2-O2 /CO23Ϫ . Marks on the voltammograms during the reduc-
tion step correspond to the limit of the potential window. ͑b͒ Voltammograms
at a nickel electrode at 450°C in Li2CO3-Na2CO3-K2CO3 . Sweep rate: 1 V
s
Ϫ1. S Ϸ 0.4 cm2. Reference electrode: CO2-O2 /CO23Ϫ
.
vacuum at 150°C and slowly heated to 450°C under a carbon diox-
ide atmosphere to prevent the rapid decomposition of the melt.19
The cell is made of an outer Pyrex envelope at the base of which
was an alumina ͑Degussa͒ crucible as described in Ref. 20. During
each experimental, run carbon dioxide was kept flowing through the
cell. The mechanism of the carbon deposit on a rod of glassy carbon
´
͑GC, V25 Le Carbone Lorraine, France͒ and nickel ͑Weber Metaux,
France͒ electrodes was studied by cyclic voltammetry ( ϭ 0.1
v
V sϪ1) using a Tacussel PRT 20-10X potentiostat.
The reference electrode was constituted of a gold wire placed in
an alumina tube containing molten Li2CO3-Na2CO3-K2CO3 .21
A
Figure 4. XRD patterns of samples 2, 3, and 4.
Downloaded on 2014-10-16 to IP 134.71.135.126 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D74
Journal of The Electrochemical Society, 149 ͑5͒ D72-D78 ͑2002͒
Table II. Results of XPS investigations on deposited carbon pow-
Table I. Values of the interlayer distance, d002 , between adjacent
graphene sheets in graphite deduced from XRD patterns.
der samples. The carbon, oxygen, and lithium contents were cal-
culated from the intensity of the photoelectron peaks of Li 1s, O
1s, and C 1s peaks „surface area under each peak…, respectively,
displayed from the survey spectra.
d002
Substrate
Nickel
Samples
͑nm͒
2
3
4
5
0.337
0.339
0.335
0.339
C 1s
O 1s
Li 1s
Samples
͑%͒
͑%͒
͑%͒
Carbon
1
2
3
4
5
85.3
83.1
73.0
88.5
76.2
7.5
6.2
15.8
7.7
7.2
10.7
11.2
3.8
Results and Discussion
11.0
12.8
Mechanism of carbon deposit in fused alkali carbonates.—The
mechanism of carbon deposition, partially reported elsewhere,17 is
studied briefly hereafter from the analysis of CVs. A typical CV
obtained with a GC electrode in carbonate melts is presented in
Fig. 2a. The electroactivity range is about 2.7 V wide. The cathodic
limit17,22 corresponds to the reduction of carbonate ions into carbon
and oxide ions
2͒ were observed, as in the case of a deposit on a GC electrode. The
anodic limit corresponds to the oxidation of nickel. Indeed, in oxi-
dizing atmosphere, the oxidation of carbonate ions interferes with
the nickel oxidation. As proposed by several authors,24-28 one can
assume that NiO appears at the surface of the nickel electrode ac-
cording to the reactions
CO23Ϫ ϩ 4eϪ → C ϩ 3 O2Ϫ
͓1͔
Ni → Ni2ϩ ϩ 2eϪ
͓6a͔
͓6b͔
͓6c͔
The formation of lithium carbide compounds is also highly prob-
ably and is associated to that of lithium oxide anions according to
the following reaction
and
Ni2ϩ ϩ CO23Ϫ → NiO ϩ CO2
Li CO ϩ 2x Ϫ 1͒C ϩ 6eϪ → 3 O2Ϫ ϩ 2LiCx
͓2͔
͑
2
3
The overall reaction of nickel oxidation can be written
with 1 р x р 60.
In contrast with the CVs obtained from other metallic
substrates,17 no reduction peaks were observed before the cathodic
limit was reached. It means that, in the accessible potential range,
GC remains an inert material which is not involved in the electro-
chemical reaction.
Ni ϩ CO23Ϫ → NiO ϩ CO2 ϩ 2eϪ
It explains shoulders on the previous oxidation peaks ͑Reactions
2, 4, and 5͒ observed in Fig. 2b.
Pure stoichiometric NiO is an insulator. However, lithium ions
present in the electrolyte can be incorporated into the NiO structure
to form an electronic conductor Li-Ni-O.27,28 Therefore, during the
reduction step, the peaks ascribed to the reduction of nickel oxide
and/or Li-Ni-O compounds, and to the intercalation of lithium inside
NiO were also observed.
During the potentiostatic deposition of carbon on GC and nickel
electrodes, an increase of the current during the electrolysis was
generally observed. This phenomenon was attributed to the increase
in the electroactive surface of the electrode due to the formation
The anodic limit corresponds to the oxidation of carbonate ions
into carbon dioxide and oxygen according to
2CO23Ϫ → 2CO2 ϩ O2 ϩ 4eϪ
͓3͔
The oxidation peak observed at Ϫ0.4 and ϩ0.2 V were attributed
to the oxidation of the deposited carbon17,23
2 O2Ϫ ϩ C → CO2 ϩ 4eϪ
C ϩ 2CO23Ϫ → 3CO2 ϩ 4eϪ
͓4͔
͓5͔
of
a carbon deposit. After electrolysis of molten Li2CO3-
Na2CO3-K2CO3 , the deposited carbon powders were immersed into
hot water. After immersion, the hard gray carbon deposit is disinte-
grated. In fact, solidified salt constitutes about 90% of the total
deposit, the remainder being the carbon. As described in Reaction 2,
the formation of lithium carbide and lithium oxide competes with
that of carbon. It is well known that lithium carbide strongly reacts
with water to give lithium hydroxide and acetylene. Therefore, the
gas evolution observed during the washing procedure of carbon
powder deposited at potential values lower than Ϫ1.8 V is probably
due to the decomposition of lithium carbide. The titration of the
aqueous washing solution by hydrochloric acid coupled with atomic
and to the oxidation of the LiCx compounds formed during the re-
duction step ͑Reaction 2͒. When the cathodic limit of the potential
sweep became more and more negative ͑Fig. 2͒, the intensities of
the two oxidation peaks increased in the same manner. It confirms
that these peaks are due to the oxidation of carbon and of LiCx
compounds.
In the case of nickel electrodes, the electroactive range is around
1.5 V wide ͑Fig. 2b͒. The two peaks related to the oxidation of the
carbon deposit according to Reactions 4 and 5, and to the oxidation
of the LiCx compounds formed during the reduction step ͑Reaction
Table III. Results of XPS investigations on deposited carbon powder samples. Fitting parameters of C 1s region: BE, FWHM peak height, and
relative contributions of each group in the sample „Ii….
Sample 2
Sample 3
Sample 4
Sample 5
BE
͑eV͒
FWHM
͑eV͒
BE
͑eV͒
FWHM
͑eV͒
BE
͑eV͒
FWHM
͑eV͒
BE
͑eV͒
FWHM
͑eV͒
Ii
͑%͒
Ii
͑%͒
Ii
͑%͒
Ii
͑%͒
Chemical States
C—C
284.3
285.5
-
288.8
291.7
1.4
3.0
-
3.1
2.2
49.8
33.9
-
13.3
3.1
284.5
285.6
-
288.5
291.7
1.4
3.0
-
3.1
2.2
47.6
30.7
-
16.6
5.1
-
-
-
93.7
3.6
2.7
-
285.5
287.1
288.9
-
1.8
1.4
1.5
-
93.3
3.9
2.8
-
285.6
287.3
288.9
-
1.8
1.4
1.5
-
CvO
-COOH
-CO23Ϫ
Satellite
Downloaded on 2014-10-16 to IP 134.71.135.126 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 149 ͑5͒ D72-D78 ͑2002͒
absorption analysis of the solution have also shown that the solidi-
D75
fied salt is not only composed of alkali carbonates but also of
lithium oxide.17
Analysis of the carbon deposit.—The carbon deposits were ana-
lyzed using several techniques such as XRD, XPS, TEM, SEM, and
BET. No significant difference was pointed out between the powder
deposited on nickel and GC electrodes. Therefore, the results ob-
tained for carbon deposited on nickel or GC will be presented alter-
natively henceforth.
The influence of the potential applied to the working electrode on
the nature of the deposited carbon powders was studied. Hereafter,
the different samples were numbered as follows
1: Carbon deposited on nickel at E ϭ Ϫ1.6 V
2: Carbon deposited on nickel at E ϭ Ϫ2.4 V
3: Carbon deposited on nickel at E ϭ Ϫ6.0 V
4: Carbon deposited on GC at E ϭ Ϫ2.4 V
5: Carbon deposited on GC at E ϭ Ϫ6.0 V.
After deposition, the carbon powders were washed in hydrochlo-
ric acid in order to remove traces of electrolyte, and dried under
vacuum at 400°C.
Structural characterization by SEM.—The morphology of the
carbon deposits, studied by SEM, does not seem to depend on the
values of the applied potential, because the size of the particles is
roughly the same whatever the sample. As an example, Fig. 3 rep-
resents a typical SEM image of carbon deposit ͑sample 3͒. It is
comprised of very fine and ‘‘quasi-spherical’’ particles which are
aggregated on the image. The size of one particle is less than
100 nm.
Structural characterization by XRD.—The X-ray diffraction pat-
terns of samples 2-4 are presented in Fig. 4. For samples 2 and 3, the
bump observed at 2␪ Ϸ 20° is characteristic of amorphous carbon.
For sample 3, the XRD pattern was recorded between 15 and 100°
to identify all the peaks. In spite of the presence of amorphous
carbon, the well-defined peaks related to all the diffraction lines of
graphitized samples were observed. For sample 4, the sharp shape of
the ͑002͒ peak and the lower intensity of the bump at 2␪ Ϸ 20°
related to the presence of amorphous peak indicate that the carbon
powder is much more crystallized on a carbon electrode ͑sample 4͒
than on a nickel electrode ͑samples 2 and 3͒.
The crystallinity of all these samples is also attested by the value
of the interlayer distance, d002 , deduced from the Bragg relation;
whatever the sample, these values ͑Table I͒ are close to the theoret-
ical value obtained in the case of pure graphite, i.e., 0.335 nm.
However, the d002 values slightly increase with increasing negative
potential values in fused alkali carbonates. That means that the crys-
tallinity of the carbon particles formed at low negative potential
values ͑samples 2 and 4͒ are slightly higher than that obtained at
high negative potential values ͑samples 3 and 5͒.
Surface characterization by XPS.—XPS is a suitable technique
for studying the elemental composition of surface layers through the
analysis of electron core level peaks.
Analysis of the C, Li, and O contents.—The content of lithium,
oxygen, and carbon in a sample was calculated from the intensity of
the photoelectron peaks of Li 1s, O 1s, and C 1s peaks ͑surface area
under each peak͒, respectively, displayed from the survey spectra.
The values of the peak areas were calculated by including the sen-
sitivity factor corresponding to each element: 0.25 ͑lithium͒, 0.711
͑oxygen͒, and 0.296 ͑carbon͒. The results reported in Table II were
obtained with samples 1-5. As shown above, the lithium content in
the carbon deposit is dependent on the potential values applied
to the electrode for the electrolysis of molten Li2CO3-
Na2CO3-K2CO3 : more negative deposition potentials resulted in an
Figure 5. XPS analysis. High resolution spectra of C 1s region of ͑a͒ sample
2 and ͑b͒ sample 3. High resolution spectra of ͑c͒ O 1s and ͑d͒ Li 1s regions
of sample 2.
increasing lithium and oxygen content ͑formation of lithium oxide͒,
and a decrease of carbon content of the deposit. Moreover, for a
same deposit potential value, the nature of the substrate has a small
influence on the carbon, the oxygen, and the lithium contents.
Downloaded on 2014-10-16 to IP 134.71.135.126 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D76
Journal of The Electrochemical Society, 149 ͑5͒ D72-D78 ͑2002͒
Table IV. Results of XPS investigations on deposited carbon powder samples, and fitting parameters of the Li 1s region. BE, FWHM peak
height, and relative contributions of each group in the sample „Ii….
Sample 2
Sample 3
Sample 4
Sample 5
BE
͑eV͒
FWHM
͑eV͒
BE
͑eV͒
FWHM
͑eV͒
BE
͑eV͒
FWHM
͑eV͒
BE
͑eV͒
FWHM
͑eV͒
Ii
͑%͒
Ii
͑%͒
Ii
͑%͒
Ii
͑%͒
Chemical States
Nearly metallic
Li2O
Li2CO3 and LiOH
51.5
53.6
-
1.5
2.3
-
68.8
31.2
-
-
-
2.0
2.0
-
51.7
53.4
-
1.5
2.3
-
66.9
33.1
-
-
-
1.8
1.9
-
53.0
55.6
83.3
16.7
52.9
55.3
83.7
16.3
Analysis of high resolution spectra of the C 1s region.—XPS indi-
vidual curve fitting parameters of the C 1s region ͓binding energy
͑BE͒ full width at half maximum ͑FWHM͒ peak height, and relative
contribution of each group in the same sample, Ii͔ for samples 2-5
are given in Table III. The detailed analysis of the high resolution
spectra of C 1s regions for samples 2 and 3 are presented in Fig. 5a
and b, respectively. The main peak observed in the C 1s region for
sample 2 ͑Fig. 5a͒ at 284.3 eV (FWHM ϭ 1.4 eV) is attributed to
C—C bonds.29 For samples 3 and 5, the C 1s peaks shift toward
higher BE ͑Fig. 5b͒; in that case, the main peak located at 285.5
Ϯ 0.1 eV (FWHM ϭ 1.8 eV͒ is resolved into one peak ascribed to
carbon-oxygen double bonds.29 The peak situated around 288.7
Ϯ 0.2 eV is related to -COOH surface groups. An additional peak is
also observed at 288 Ϯ 0.2 eV due to the presence of C—O bonds
into lithium carbonate.30 In addition, a small plasmon satellite is also
visible at 291.0 eV.
internal available space in the ‘‘holes’’ previously observed in Fig.
6b is now occupied by nickel particles. Such a phenomenon was
already observed, for example, during the preparation of carbon
from a gaseous phase ͑CO, CH4 ...͒ in presence of a nickel
catalyst;32 in that example, depending on the temperature and the
concentration of the gases, the metallic catalysts were surroundered
by carbon shells instead of giving rise to carbon filaments. Another
example is the crystallization, at moderate temperature, of carbon
atoms during the thermal decomposition of metal carbide such as
silicon or vanadium carbides.33-35 In our case, the chemical compo-
sition of these particles was determined from the X-ray spectrum
presented in Fig. 6d. These particles were composed of nickel which
could act as catalyst for the formation of graphite. However, one
must notice that only few particles were detected by this technique.
It explains why they were not detected by XPS or by XRD ͑Ͻ1 %
wt͒.
TEM analyses have also revealed the presence of carbon fibers as
presented in Fig. 6e for sample 1. The fibers have a diam of ϳ10 nm
and a nearly uniform size, and were self-organized into long crys-
talline ‘‘ropes.’’ The diameter of a rope is between 50 and 100 nm,
which correspond to 5-10 tubes per rope.
Analysis of high resolution spectra of the O 1s region.—The detailed
analysis of the high resolution spectra of O 1s regions of sample 2 is
presented in Fig. 5c. The O 1s region consists of three contributions.
The peaks indicating at 531.4, 533.0, and 535.3 eV are attributed to
carbonyl groups29 ͓denoted (C—O)a in Fig. 5c͔ and/or Li-O bonds
in LiOH,30 to hydroxyl groups29 ͓denoted (C—O)b in Fig. 5c͔, and
to H2O adsorbed,31 respectively.
Surface and pore texture analysis.—The method developed by
Brunauer, Emmett, and Teller in the 1940s36,37 is a powerful and
widely used tool to access to the surface characteristics and the
porous texture of a powder material.38 Thus, the total BET specific
area and the external surface area, as well as the micropore area and
volume, the average pore diameter of the deposited carbon powder
were measured by nitrogen gas adsorption at boiling temperature
͑77 K͒. It is very important to understand the morphological param-
eters of carbon powders and their electrochemical behavior when
they are used, for example, as anode material in lithium-ion batteries
or in supercapacitors.
All the powder samples were degassed prior to the BET mea-
surements to remove traces of moisture. Figure 7 gives the influence
of the applied potential on the above parameters. Whatever the po-
tential values, the total BET specific surface area is higher than that
corresponding to the external one because of the pore walls contri-
bution. Pores can have a regular or, more commonly, an irregular
shape. Usually, three geometric forms are used to represent pore
shape: cylinders ͑in some oxides like alumina and magnesia͒, slits
͑in active carbon and clays͒, and voids between connected solid
spheres ͑in silica and many solids obtained from gels͒.38 According
to the results obtained from TEM and XRD analyses showing the
presence of graphite, the pore shape of the carbon powder can be
represented by slits, i.e., lamellar structure. This representation ex-
plains the high values of the micropore area compared to the exter-
nal one. For example, in the case of sample 4, the micropore and the
external areas are about 649 and 169 m2 gϪ1, respectively.
Analysis of high resolution spectra of the Li 1s region.—XPS indi-
vidual curve fitting parameters of the Li 1s regions ͑BE and FWHM
peak height, relative contribution of each group in the same sample,
Ii͒ for samples 2-5 are given in Table IV. For example, Fig. 5d
gives the high resolution spectrum of the Li 1s region for sample 2.
The Li 1s region is composed of two contributions at BE
ϭ 51.5 and 53.6 eV. The first one is ascribed to nearly metallic
lithium,30 the second one to ionic lithium due to the presence of
Li2O.30 From the analysis of the Li 1s region, the relative contribu-
tion of ionic lithium (Li2O) and nearly metallic lithium in each
sample were also compared. It was found that the first one drasti-
cally changes from 31.2 to 83.3% with an increase of the negative
potential values from Ϫ2.4 to Ϫ6.0 V. Indeed, the formation of
lithium oxides ͑Reaction 1͒ competes with the formation of nearly
metallic lithium and/or lithium carbide ͑Reaction 2͒. In the case of
samples 3 and 5, the peak located at 55.3 eV is related to the pres-
ence of Li into LiOH and Li2CO3 ;30 the latter is also detected in the
C 1s region.
Structural characterization by TEM.—The TEM micrographs of
sample 5 are presented in Fig. 6a-b. As observed by SEM ͑Fig. 3͒,
the electrolysis of molten Li2CO3-Na2CO3-K2CO3 provides carbon
bundles with particle size between 20 and 50 nm. Some of them
were comprised of amorphous carbon ͑Fig. 6a͒. Others were sur-
roundered by graphitized layers with a random orientation ͑Fig. 6b͒,
as in the case of carbon black.32 The fringe spacing deduced from
Fig. 6b is about 0.34 nm and is very similar to the theoretical dis-
tance between adjacent graphene sheets, d002 , in graphite and to that
deduced from XRD analyses ͑Table I͒. In addition, in the case of
carbon deposited on nickel electrodes, very well crystallized par-
ticles embedded in the carbon and surroundered by graphite, were
also observed as illustrated in Fig. 6c for sample 3. The whole
The surface area decreases dramatically with decreasing poten-
tial values; for sample 4, the BET specific surface area is about 850
m2 gϪ1 and reaches a value of around 500 m2 gϪ1 when the depo-
sition potentials are made more negative ͑sample 5͒. As discussed
above, for very negative values of the deposit potential, the forma-
tion of lithium and/or lithium oxides competes with the formation of
graphite. Therefore, the nucleation and the growth of graphite
should be affected greatly by very negative potential values, as
Downloaded on 2014-10-16 to IP 134.71.135.126 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 149 ͑5͒ D72-D78 ͑2002͒
D77
Figure 7. Results of the nitrogen gas adsorption at boiling temperature ͑77
K͒ using the BET method for samples 4 and 5.
proved by the increase of the d002 values observed for samples 3 and
5. The evolution of the BET surface area can result from: ͑i͒ a
decrease of the amount of graphite phases; and ͑ii͒ an increase of the
lithium content as revealed by XPS ͑Table II͒ that results in the
closure of some nanopores. Finally, a small decrease of the average
pore diameter is observed when the potential applied to the electrode
for the deposition of carbon is more negative: from 3.25 nm for
sample 4 to 3.06 nm for sample 5.
Conclusion
An original electrochemical route for the preparation of carbon
powders has been proposed based on the deposit of carbon powder
by electrolytic reduction of the liquid ternary system lithium-
sodium-potassium carbonates at 450°C. This reduction process
could give rise to substantial deposits of carbon powder.17 Carbon
powder has been deposited on nickel and GC electrodes from the
electrochemical reduction of carbonate ions in a fused eutectic mix-
ture of lithium, sodium, and potassium carbonates. The analysis of
the reaction mechanism has been obtained from CVs. The reduction
of carbonate ions gives rise not only to a carbon deposit but also to
oxide ions and lithium carbide. Moreover, in the case of carbon
deposited on nickel electrodes, the formation of nickel oxide con-
taining small traces of lithium was also taken into account to get a
better interpretation of the CVs.
After washing in HCl solutions and drying at 400°C to remove
traces of water, the structural characterization of the carbon powders
was performed. TEM analyses have revealed the presence of three
forms of carbon depending on deposit potential values: amorphous,
graphite, and fibers. The fibers have a nearly uniform size with a
diam of ϳ10 nm, and are self-organized into long crystalline ropes.
For increasing negative potential values, graphite was formed pref-
erentially to amorphous carbon. XRD analyses give rise to the dis-
tance between adjacent graphene sheets, d002 , using the Bragg rela-
tion. Whatever the sample, this value is very close to the theoretical
value of about 0.34 nm.
Surface analysis by XPS measurements have also revealed the
presence of nearly metallic and ionic lithium in the carbon lattice.
The latter is due to Li2O, LiOH and/or Li2CO3 .
Finally, the carbon powders prepared in our experimental condi-
tions exhibit large BET specific surface areas due to the presence of
graphite. This surface area decreases with increasing negative de-
Figure 6. TEM images of ͑a,b͒ sample 5, ͑c͒ sample 3, and ͑e͒ sample 1. ͑d͒
X-ray emission spectrum of the particles observed in Fig. 6c.
Downloaded on 2014-10-16 to IP 134.71.135.126 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D78
Journal of The Electrochemical Society, 149 ͑5͒ D72-D78 ͑2002͒
14. M. D. Ingram, B. Baron, and G. J. Janz, Electrochim. Acta, 11, 1629 ͑1966͒.
15. Y. K. Delimarskii, V. I. Shapoval, V. A. Vasilenko, and V. F. Grischenko, Tovarnye
Znaki, 52, 176 ͑1975͒.
16. Y. Ito, T. Shimada, and H. Kawamura, in Molten Salts, R. J. Gale, C. Blomgren,
and H. Kojima, Editors, PV 92-16, p. 574, The Electrochemical Society Proceed-
ings Series, Pennington, NJ ͑1992͒.
17. F. Lantelme, B. Kaplan, H. Groult, and D. Devilliers, J. Mol. Liq., 83, 255 ͑1999͒.
18. H. Kawamura and Y. Ito, J. Appl. Electrochem., 30, 571 ͑2000͒.
19. A. J. Appleby and S. B. Nicholson, J. Electroanal. Chem., 53, 105 ͑1974͒.
20. F. Lantelme, D. Inman, and D. G. Lovering, in Molten Salt Techniques, Vol. 2, p.
109, R. J. Gale and D. G. Lovering, Editors, Plenum Press, New York ͑1984͒.
21. N. Q. Minh and L. Redey, in Molten Salt Techniques: Reference Electrodes for
Molten Electrolytes, Vol. 3, D. G. Lovering and R. J. Gale, Editors, Plenum Press,
New York ͑1987͒.
posit potential values. Finally, the total BET specific surface area
is mainly due to the micropore area because of the pore walls
contribution.
This work is a prelude to the study of the kinetics of the lithium
cations insertion/deinsertion reaction into the carbon powders as
prepared depending on the potential values for the deposit of the
carbon powders into fused carbonates and on the temperature of the
heat-treatment after washing in HCl.
Acknowledgments
The authors are grateful to Dr. J. Blanchard for BET measure-
ments, Dr. P. Beaunier for TEM, Dr. S. Borensztajn for SEM deter-
minations, and Professor D. Devilliers for fruitful discussions. This
work was done within the frame of a France-Japan collaboration
program ͑PICS͒ financially supported by the CNRS in France.
22. Y. K. Delimarskii, O. V. Gorodis’kii, and V. F. Grishchenko, Dokl. Akad. Nauk.
SSSR, 156, 650 ͑1964͒.
23. A. Borucka, J. Electrochem. Soc., 124, 972 ͑1977͒.
24. I. M. Petrushina, L. Qingfeng, F. Borup, and N. J. Bjerrum, J. Appl. Electrochem.,
30, 929 ͑2000͒.
The University of Paris VI assisted in meeting the publication costs of
this article.
25. J. P. T. Vossen, L. Plomp, and J. H. W. de Wit, J. Electrochem. Soc., 141, 3040
͑1994͒.
26. P. L. Cabot, E. Guezalla, J. C. Calpe, M. T. Garcia, and J. Casado, J. Electrochem.
Soc., 147, 43 ͑2000͒.
27. P. Tomczyk, M. Mosialek, J. B. Guillen, and M. Kostolowska, Molten Salt Forum,
5-6, 87 ͑1998͒.
28. M. Cassir, M. Olivry, V. Albin, B. Malinowska, and J. Devynck, J. Electroanal.
Chem., 452, 127 ͑1998͒.
29. H. Groult, D. Devilliers, M. Vogler, C. Hinnen, P. Marcus, and F. Nicolas, Electro-
chim. Acta, 38, 2413 ͑1993͒.
30. K. Kanamura, H. Takezawa, S. Shiraishi, and Z.-I. Takehara, J. Electrochem. Soc.,
144, 1900 ͑1997͒.
31. S. D. Gardner, C. S. K. Singamsetty, G. L. Booth, G.-R. He, C. U. Pittman Jr.,
Carbon, 33, 587 ͑1995͒.
32. A. Oberlin, in Chemistry and Physics of Carbon, Vol. 22, Chap. 1, P. A. Thrower,
Editor, Marcel Dekker, New York ͑1989͒.
33. D. V. Badami, Carbon, 3, 53 ͑1965͒.
References
´
1. S. Flandrois, in Le Carbone dans Tous Ses Etats, P. Bernier and S. Lefrant, Editors,
p. 517, Gordon and Breach Science Publishers, New York ͑1997͒.
2. H. Groult, D. Devilliers, and M. Vogler, in Current Topics in Electrochemistry, Vol.
4, p. 23, Research trends, Poojupura, Trivandrum, India ͑1997͒.
3. J. R. Dahn, Phys. Rev. B, 44, 9170 ͑1991͒.
4. T. Nagaura and K. Tozawa, Prog. Batteries Sol. Cells, 9, 209 ͑1990͒.
5. K. Tokumitsu, A. Mabuchi, H. Fujimoto, and T. Kasuh, J. Electrochem. Soc., 143,
2235 ͑1996͒.
6. D. Aurbach, in Nonaqueous Electrochemistry, D. Aurbach, Editor, Chap. 6, p. 289,
Marcel Dekker, New York ͑1999͒.
7. B. E. Conway, Electrochemical Supercapacitors: Scientific Fundamentals and
Technological Applications, Kluwer Academic/Plenum Publishers, New York
͑1999͒.
8. T. Ohzuku, Y. Iwakoshi, and K. Sawai, J. Electrochem. Soc., 140, 2490 ͑1993͒.
9. S. Flandrois and B. Simon, Carbon, 37, 165 ͑1999͒.
10. A. Funabiki, M. Inaba, T. Abe, and Z. Ogumi, J. Electrochem. Soc., 146, 2443
͑1999͒.
11. A. S. Claye, J. E. Fischer, C. B. Huffman, A. G. Rinzler, and R. E. Smalley, J.
Electrochem. Soc., 147, 2845 ͑2000͒.
34. S. Iijima, J. Solid State Chem., 42, 101 ͑1982͒.
35. A. Oberlin, M. Oberlin, and J. R. Comte-Trotet, J. Microsc. Spectrosc. Electron., 1,
391 ͑1976͒.
36. S. Brunauer and P. H. Emmet, J. Am. Chem. Soc., 60, 309 ͑1938͒.
37. S. Brunauer, The Adsorption of Gases and Vapors, University Press, Oxford
͑1945͒.
38. G. Leofanti, M. Padovan, G. Tozzola, and B. Venturelli, Catal. Today, 41, 207
͑1998͒.
12. J. M. Tarascon and D. Guyomard, J. Electrochem. Soc., 138, 2864 ͑1991͒.
13. K. Sato, M. Noguchi, A. Demachi, N. Oki, and M. Endo, Science, 264, 556 ͑1994͒.
Downloaded on 2014-10-16 to IP 134.71.135.126 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).