Journal of Power Sources 97±98 (2001) 78±82

Relation between surface properties, pore structure and ®rst-cycle charge

loss of graphite as negative electrode in lithium-ion batteries
Felix Johoa,*, Beat Rykarta, Andreas Blomea, Petr NovaÂka,
Henri Wilhelmb, Michael E. Spahrb
a
Laboratory for Electrochemistry, Paul Scherrer Institute, CH-5232 Villigen PSI, Switzerland
b
TIMCAL GROUP, CH-6743 Bodio, Switzerland
Received 19 June 2000; accepted 31 December 2000

Abstract

Cycling of graphite in lithium-ion batteries is reversible except for the ®rst-cycle, where some charge is ``lost'' due to irreversible side
reactions. The irreversible charge loss of different TIMREX1 graphites was found to be a linear function, both of their speci®c BET surface
areas and of the double-layer capacitance of electrodes manufactured from these graphites. Nitrogen-adsorption measurements and
differential-porosity calculations indicate that the pore structure of the graphite particles mainly consists of mesopores (2±50 nm). The
surface area of these mesopores is part of the electroactive surface area determined by impedance spectroscopy, and contributes to the
irreversible capacity of the graphite negative electrode in the ®rst-cycle. # 2001 Elsevier Science B.V. All rights reserved.

Keywords: Lithium intercalation; Graphite; Irreversible capacity; BET surface area; Double-layer capacitance; Graphite pore structure; Lithium-ion battery

1. Introduction experimental conditions such as the cell electrolyte and

the design of the electrodes and test cells [4]. In order to
During recent years extensive work has been carried out learn more about the relationship between graphite material
toward increasing the energy density of lithium-ion bat- parameters and the performance of these materials as nega-
teries. Progress has mainly been made by improving the tive electrode materials, we characterised the surface proper-
carbonaceous material of the negative electrode [1] and ties of different graphites by gas adsorption measurements
especially by replacing amorphous coke materials by more and tried to correlate the results with electrochemical
highly crystalline graphitic materials. A major problem of charge/discharge and impedance measurements.
the carbonaceous materials is their ``charge loss'' during the
®rst-cycle, which consumes lithium that must be brought
into the cell in the form of a relatively heavy lithium metal 2. Experimental
oxide such as LiCoO2. It is generally accepted that this
irreversible charge loss or irreversible capacity seen during The electrochemical charge/discharge measurements
the ®rst Li‡ insertion into graphite electrodes is caused by were carried out in the laboratory cells described elsewhere
the formation of a solid electrolyte interphase (SEI) on the [4]. Electrodes were made, either by doctor-blade coating or
graphite particles that prevents further electrolyte decom- by spraying of different TIMREX1 graphite materials
position at the negative electrode [2,3]. Several research (TIMCAL GROUP, Bodio, Switzerland), no differences
groups began to improve the reversible capacity and reduce in the electrochemical behaviour of these two types of
the charge loss of graphitic electrodes. But care must be electrode were seen. PVDF was used as binder material.
taken when comparing their results, because both the irre- Electrochemical charge/discharge measurements were per-
versible capacity and reversible speci®c charge obtained formed galvanostatically. We chose relatively low current
for a graphitic material in a lithium-ion cell depend on densities of 10 mA g 1 to complete the formation of the
solid electrolyte interphase in the ®rst Li‡ insertion cycle.
*
Corresponding author. Tel.: ‡41-56-310-21-61;
When during the ®rst electrochemical Li‡ insertion a poten-
fax: ‡44-56-310-44-15. tial of 5 mV versus Li/Li‡ was reached, the charging was
E-mail address: [Link]@[Link] (F. Joho). continued potentiostatically until the current had dropped

0378-7753/01/$ ± see front matter # 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 7 7 5 3 ( 0 1 ) 0 0 5 9 5 - X
 F. Joho et al. / Journal of Power Sources 97±98 (2001) 78±82 79

below 5 mA g 1. Metallic lithium was used as the reference

and counterelectrode, 1 M LiPF6 in DMC/EC (LP 30,
Merck) was used as the electrolyte solution.
The double-layer capacitance was calculated from the
imaginary part of impedance measured at 0.1 Hz. Impe-
dance spectra were recorded at 2.5 V versus Li/Li‡ and a
temperature of 298 K with a Zahner IM6 impedance spectro-
meter at an ac signal amplitude of 10 mV. The cell compo-
nents were arranged as in button cells, and ®tted with a
lithium reference electrode as published elsewhere [5]. The
compression spring was isolated from the electronic circuit
with PTFE caps, and instead of the spring, a copper wire was
used to make the electronic contact.
Nitrogen adsorption measurements were performed,
either with a StroÈhlein Areameter 2 apparatus for the usual
BET surface area determination or a Micromeritics ASAP Fig. 1. Irreversible capacities of the first electrochemical Li‡ intercalation
into different synthetic and natural TIMREX1 graphites as a function of
2010 apparatus for determination of the pore size distribu-
the specific BET surface areas measured at 10 mA g 1 in a 1 M LiPF6,
tion. Owing to differences in the experimental conditions, EC:DMC (1:1) electrolyte.
the surface areas measured with the StroÈhlein apparatus are
systematically higher than those measured with the Micro-
meritics ASAP 2010 apparatus. Samples were evacuated chose a small speci®c current (about C/36) and room
until a static vacuum of less than 7 mm Hg was reached. The temperature for galvanostatic charging of the cell in the
Brunauer±Emmett±Teller (BET) transform was calculated ®rst-cycle in order to complete formation of the SEI layer
with ASAP software. We also used DFT‡ software, an and maximise the irreversible capacity in the ®rst-cycle of
enhanced data reduction application provided by Micro- the electrode.
meritics. In the isotherm deconvolution according to density The ®rst-cycle of an electrode containing TIMREX1
functional theory (DFT), either the porosity or the surface E-SLX 50 as the active material is presented as an example
energy of the solid is taken into account. These calculations in Fig. 2. It shows a galvanostatic charge/discharge curve
have been successfully applied to isotherms obtained on typical for highly crystalline graphite electrode materials,
graphitised materials [6,7]. with reversible Li‡ intercalation/deintercalation occurring
at potentials below 0.2 V versus Li/Li‡. The SEI formation
takes place at potentials between 0.8 and 0.2 V versus Li/Li‡
3. Results and discussion in the Li‡ insertion (charging) half cycle. The speci®c BET
surface area of this sample was 3.9 m2 g 1. The ®rst char-
The irreversible capacity seen during the ®rst reduction ging of the electrode consumed 385 Ah kg 1, while the
cycle affects the available capacity of a lithium-ion cell [8]. reversible capacity was 360 Ah kg 1. The charge loss or
On the other hand, an ideal solid electrolyte interphase that is
formed as a surface ®lm is essential for the functioning of the
graphite negative electrode. This means that SEI formation
must be as effective as possible so as to keep the charge
losses as small as possible. It is almost universally accepted
in the literature [9,10] that the irreversible charge loss
occurring during the ®rst reduction cycle depends on the
electrode surface area and on the speci®c BET surface area
of the graphite material used in the negative electrode,
because the SEI covers all the surface area exposed to the
electrolyte solution.
Fig. 1 shows the irreversible capacities of the ®rst charge/
discharge cycle plotted as a function of the BET surface
areas determined for synthetic TIMREX1 graphites of the
KS, SFG, T, E-SLX and SLM type as well as for the natural
graphite TIMREX1 E-NP 15. It can be seen that this
function is linear. A linear regression ®t of the experimental
data resulted in a correlation factor, R2 of 95%. It must be Fig. 2. First galvanostatic charge/discharge cycle of a TIMREX1 E-SLX
pointed out, however, that the experimental conditions are 50 negative electrode at 10 mA g 1 graphite in a 1 M LiPF6, EC:DMC
very important for consistent and reproducible results. We (1:1) electrolyte.
80 F. Joho et al. / Journal of Power Sources 97±98 (2001) 78±82

irreversible capacity Cirrev of this sample as calculated by the

formula
Cint Crev
Cirrev ˆ
Cint
was 6.5%; here Cirrev is the irreversible capacity in %, Cint
the specific charge recorded in the first Li‡ intercalation,
and Crev the reversible specific charge recorded in the
corresponding Li‡ deintecalation.
Electrodes containing the different TIMREX1 graphites
all exhibited reversible capacities of more than 360 Ah kg 1
at a speci®c current of 10 mA g 1 (both ®gures being
referred to the mass of graphite in the negative electrode).
The data obtained for the irreversible capacity in the galva-
nostatic charge/discharge experiments as well as for the
double-layer capacitance obtained from impedance mea- Fig. 3. Irreversible capacities of the first electrochemical reduction of
surements are summarised in Table 1. different synthetic and natural TIMREX1 graphite electrodes as a function
of their double-layer capacitance calculated from impedance spectroscopic
A linear function is also obtained when the charge losses
measurements at 2.5 V vs. Li/Li‡.
or irreversible capacities of graphite electrodes containing
the same TIMREX1 graphites are plotted against their
double-layer capacitance as determined by impedance spec- with results obtained in our laboratory some years ago [11].
troscopy (Fig. 3). Here the correlation factor was 96%. For a Differences in electrode manufacturing and in the experi-
reliability test of our results, the double-layer capacitance mental set-up may be responsible for these differences. For
found in all 38 measurements were divided by the corre- the present studies we had optimised the structure of the
sponding speci®c BET surface areas. From this calculation negative electrode so as to obtain the maximum reversible
we obtained an average double-layer capacitance of ca. speci®c capacity. By improved electrode manufacturing we
4 mF cm 2. could ensure that all particles of the electroactive material in
There are some differences between the two plots: irre- the electrodes fully participated in the electrochemical
versible charge loss versus speci®c BET surface areas reaction. Also, the drying conditions for the electrodes
(Fig. 1) and irreversible charge loss versus double-layer before cell assembly were improved, since water contam-
capacitance (Cdl) (Fig. 3), but the linear regression ®t is ination of the electrolyte has a signi®cant in¯uence on the
equivalent. The instances of departure of the data from the irreversible charge loss seen during the ®rst-cycle [4].
straight-line ®t might be attributed to a material parameter In this context, it is remarkable that TIMREX1 graphites
other than the BET surface areas. However, it is not certain of the SFG type have relatively high irreversible capacities,
whether these deviations are signi®cant. Moreover, in some while graphites of the E-SLX type have relatively low
cases the results reported in the present paper are inconsistent irreversible capacities.

Table 1
Surface properties and results of electrochemical charge/discharge measurements and impedance spectroscopic studies of different synthetic and natural
TIMREX1 graphites

Product Specific BET surface area (m2 g 1) Charge loss (%) Specific capacitance, Cdl (F g 1) d50 (mm)

KS 6 19.4 24.8 0.769 3.6

KS 10 13.8 20.0 5.6
KS 15 12.3 18.0 0.587 7.5
KS 25 12.0 14.6 8.2
KS 44 8.3 12.5 0.315 16.2
SFG 6 15.2 21.0 0.688 3.4
SFG 10 10.0 16.0 5.8
SFG 15 8.5 14.0 0.337 7.5
SFG 44 4.3 9.6 0.186 23.6
T 15 11.2 15.2 0.397 8.6
T 44 8.1 10.7 0.273 20.2
SLM 44 4.5 8.0 24.0
E-SLX 20 7.7 10.7 0.331 10.2
E-SLX 30 4.8 7.5 0.157 18.5
E-SLX 50 3.9 6.5 0.138 28.0
E-NP 15 8.4 11.8 0.347 9.1
 F. Joho et al. / Journal of Power Sources 97±98 (2001) 78±82 81

Fig. 4. Nitrogen gas adsorption isotherm of TIMREX1 SFG 15. The

shape of this isotherm is similar to that found for the other TIMREX1 Fig. 5. Pore structures of different synthetic and natural TIMREX1
samples, and typical for materials with a pore structure containing mainly graphites: incremental surface areas as a function of the pore width
mesopores. calculated from DFT porosity calculations based on data taken from
nitrogen gas adsorption measurements.

For a better understanding of the mechanism of SEI surface energy theory. These results indicate that the
formation at the graphite particle surface, differential- mesopores of TIMREX1 E-NP 15 make a smaller con-
porosity measurements were performed using different tribution to the BET surface area and, thus, to the irrever-
TIMREX1 graphite types having approximately the same sible capacity of the ®rst reduction cycle than those of the
particle-size distributions. Fig. 4 shows a typical isotherm synthetic graphites examined. However, the total BET
obtained in the gas absorption measurements of these gra- surface area measured for the natural graphite TIMREX1
phites. The shape of this isotherm is typical for materials E-NP 15 is in the same range as the BET surface areas
with a high content of mesopores (typical pore widths of 2± obtained for the synthetic graphites TIMREX1 E-SLX 20
50 nm). Two DFT models (porosity and surface energy) and SFG 15.
have been applied to the isotherms. The pore-size distribu- The irreversible capacity found for natural graphite nicely
tions obtained with the porosity model indicate that the adheres to the linear relations of Figs. 1 and 3 linking the
graphite materials mainly consist of mesopores. The pore charge loss to BET surface areas and double-layer capaci-
structures of the synthetic and natural graphite samples tance. This indicates that the geometrical surface area of the
examined are similar. Interestingly, however, the DFT natural graphite particle excluding the mesopores contri-
porosity calculations indicate that mesoporosity is more butes more to the BET surface area and to the irreversible
pronounced in the synthetic TIMREX1 graphite samples capacity of the ®rst electrochemical reduction than that of
(E-SLX, SFG, KS and T) than in natural graphite TIM- the synthetic graphites. Obviously, natural graphite differs
REX1 E-NP 15 (Fig. 5). This result was con®rmed, when from the synthetic graphites examined, not only in its
comparing surface area values calculated with DFT por- mesoporosity, but also in its surface morphology, i.e. rough-
osity model and with BET transforms (Table 2). The largest ness of the particle surface and defects on the particle
deviation is seen for the natural graphite sample, indicating surface. The synthetic graphite samples exhibit more highly
that here the porosity model is de®nitively inappropriate. developed mesoporosity, while the natural graphite sample
For the natural graphite TIMREX1 E-NP 15, the surface has a rougher surface, so that similar values are recorded for
energy model appears to be more appropriate, since it leads their BET surface areas and for the irreversible capacities in
to a satisfactory agreement between the measured BET the ®rst reduction of the corresponding graphite negative
value and the surface area calculated by means of DFT electrodes.

Table 2
Comparison of the specific BET surface areas with the surface areas obtained from the DFT porosity and surface energy model calculationsa

BET surface area Surface area (porosity model) Surface area (surface energy model)
(m2 g 1) (m2 g 1) (m2 g 1)

E-SLX 20 6.8 7.8 14.1

SFG 15 8.1 8.7 16.0
T 15 10.7 10.7 18.8
KS 15 11.5 10.4 18.0
E-NP 15 7.7 4.0 7.7
a
The listed BET surface areas were obtained from nitrogen adsorption measurements, and differ slightly from the BET surface areas in Fig. 1 measured
by single point nitrogen adsorption.
82 F. Joho et al. / Journal of Power Sources 97±98 (2001) 78±82

4. Conclusion graphites. The surface of the natural graphite particles is

rougher, thus, makes a larger contribution to the BET surface
Results of nitrogen gas absorption measurements per- area and to the irreversible capacity of the electrochemical
formed with different synthetic and natural TIMREX1 reduction cycle than that of the synthetic graphite materials.
graphites indicated that the porosity of highly crystalline
graphites mainly consists of mesopores. The irreversible
capacities of the ®rst Li‡ intercalation in graphite negative References
electrodes linearly depend on both the speci®c BET surface
areas of the graphite materials and the double-layer capa- [1] A. Yoshino, Abstract ISIC 10 (1999) 31A5, Okazaki, Japan.
citance of the corresponding graphite electrodes. Since the [2] E. Peled, in: J.-P. Gabano (Ed.), Lithium Batteries, Academic Press,
London, 1983, p. 43.
speci®c BET surface areas and double-layer capacitance
[3] J.R. Dahn, A.K. Sleigh, H. Shi, B.M. Way, W.J. Weydanz, J.N.
also exhibit a linear relationship, we conclude that the Reimers, Q. Zhong, U. von Sacken, in: G. Pistoia (Ed.), Lithium
surface area of the graphite mesopores contributes to the Batteries, New Materials, Developments, and Perspectives, Elsevier,
electroactive surface, i.e. almost all graphite mesopores Amsterdam, 1994, p. 1.
determined by gas adsorption are wetted by the liquid [4] F. Joho, B. Rykart, R. Imhof, P. NovaÂk, M.E. Spahr, A. Monnier, J.
Power Sources 81/82 (1999) 243.
electrolyte solution used in lithium-ion batteries.
[5] P. NovaÂk, W. Scheifele, F. Joho, O. Haas, J. Electrochem. Soc. 142
The results of DFT calculations based on nitrogen gas (1995) 2544.
adsorption measurements indicate that the pore structures [6] J.P. Olivier, J. Porous Mater. 2 (1995) 9.
and surface morphologies of the synthetic graphites inves- [7] J.P. Olivier, Carbon 36 (1998) 1469.
tigated are signi®cantly different from those of the natural [8] M. Broussely, P. Biensan, B. Simon, Electrochim. Acta 45 (1999) 3.
[9] R. Fong, U. von Sacken, J.R. Dahn, J. Electrochem. Soc. 137 (1990)
graphite sample. The contribution of mesoporosity to the
2009.
irreversible capacity of a graphite electrode is larger in the [10] B. Simon, S. Flandrois, A. FeÂvrier-Bouvier, P. Biensan, Mol. Cryst.
case of synthetic graphites. However, the surface morphol- Liq. Cryst. 310 (1998) 333.
ogy of the natural graphite differs from that of the synthetic [11] M. Winter, P. NovaÂk, A. Monnier, J. Electrochem. Soc. 145 (1998) 428.