Journal of Power Sources 288 (2015) 359e367

Contents lists available at ScienceDirect

Journal of Power Sources

journal homepage: www.elsevier.com/locate/jpowsour

Influence of dry mixing and distribution of conductive additives

in cathodes for lithium ion batteries
Werner Bauer a, *, Dorit No
€ tzel a, Valentin Wenzel b, Hermann Nirschl b
a
Karlsruhe Institute of Technology, Institute for Applied Materials (IAM-KWT), Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen,
Germany
b
Karlsruhe Institute of Technology, Institute for Mechanical Process Engineering and Mechanics, Strasse am Forum 8, 76131 Karlsruhe, Germany

h i g h l i g h t s

A model is presented that explains the conductivity drop after dry mixing of active material and carbon black.

It is shown how the conductivity drop can be avoided by an admixture of carbon materials.

Recommendations are given in order to optimize the carbon distribution.

a r t i c l e i n f o a b s t r a c t

Article history: Conductive additives, like carbon black or graphite, are essential components of lithium ion batteries due
Received 24 December 2014 to the limited electrical conductivity of most electrode materials. However, there is still a lack of
Received in revised form knowledge about the optimized distribution of these materials within the electrode. A dry mixing
13 March 2015
process is used in order to prepare a conductive coating by depositing carbon black on the surface of
Accepted 14 April 2015
Available online
Li(Ni1/3Mn1/3Co1/3)O2 (NMC) cathode particles. It is demonstrated that this e from a theoretically point of
view e favorable distribution does not allow the preparation of working electrodes without taking into
account the role of the binder. After adding an organic binder to the slurry, the polymer deposits on top
Keywords:
Electrode manufacturing
of the carbon shell during drying and inhibits the conductive contact between the particles. This can be
Dry mixing avoided by a fraction of distributed carbon particles which are associated with the binder phase
Conductivity providing conductive paths through the isolating organic material. It is shown that carbon black and
Binder graphite are principally fulfilling this task, but both materials are leading to varying processing behavior
Carbon black and electrode properties.
Graphite © 2015 Elsevier B.V. All rights reserved.

1. Introduction electronic transport through the composite which is essential to

Electrodes of lithium ion batteries (LIB) primarily consist of a
particulate electrochemical active material (AM) that is mixed with
electro-conductive additives and a polymer binder. The polymeric
component is added to bind the particles together and to fix these
on a metallic current collector foil. The conductive additives are
necessary in order to compensate the limited conductivity of most
electrode materials. Especially the poor electrical conductivity of
common cathode materials (ranging for commercial materials from
109 S/cm for LiFePO4 to 103 S/cm for LiCoO2 [1]) hampers the
* Corresponding author.
E-mail address: werner.bauer@kit.edu (W. Bauer).
provide the overall charge neutrality during the charge transfer.
Carbon black (CB) or graphite powders are usually chosen as
conductive additives at the cathode side [2]. Although other types
of conductive fillers with a higher aspect ratio, like carbon fibers or
carbon nano tubes, have a lower percolation threshold and are
more effective to form conductive pathways between multiple
particles [3,4], carbon black is still preferred due to its slurry sta-
bilizing capability [5], high chemical inertness and low costs [6].
The role of graphite is slightly different as the conductivity
enhancing potential is less marked due to the larger particle size
and the plate-like morphology. However, graphite is also an effi-
cient compaction aid that allows improving the direct particle
contacts throughout the electrode, providing also a lower electrical
resistance and a higher energy density [6e8].

http://dx.doi.org/10.1016/j.jpowsour.2015.04.081
0378-7753/© 2015 Elsevier B.V. All rights reserved.
360 W. Bauer et al. / Journal of Power Sources 288 (2015) 359e367
Carbon black is delivered as a powder with low bulk density. It
consists of colloidal particles with a primary particle size below
100 nm, which are chemically bound to form aggregates with a
more or less complex structure [9]. The aggregates itself form ag-
glomerates which are held together by weak physical forces, e.g.
van-der-Waals forces, forming branched clusters with a size up to
the micron range. Within the electrode, the carbon black must be
distributed in order to produce continuous conduction pathways
from the current collector to each particle of the active component
with a minimum of added amount. As the particle size of many
active materials is in the range of 10 mm, it is also beneficial to
provide supplementary contact points and cover the particle sur-
face with additional conduction paths. It is a common agreement in
literature and in practice that the carbon black clusters must
therefore be broken into much smaller fragments which have to be
homogeneously distributed within the electrode. But, although
theoretical and experimental studies already have been carried out
to investigate the influence of the spatial distribution of the
conductive aids [1,10e24], there is still a lack of knowledge about
the optimized distribution and the therefore required processing
routes of the carbon black which also takes into account the other
components within the electrode.
From a theoretical point it should be beneficial to deposit the
carbon black particles as a thin coating on the surface of the active
material [25]. In this configuration only a small amount of
conductive additive is required to provide an electrical connection
between all particles. Dominko, Drofenik and coworkers e.g.
propose a homogenous coating of the carbon black particles by
fixing them on the surface of the active material with the help of
gelatin [26,27]. With such uniformly distributed carbon black it
was demonstrated for LiCoO2 cathodes to yield better perfor-
mance for merely 2 wt.% of carbon black as with 10 wt.% of non-
uniformly distributed conductive aids [28]. However, in their in-
vestigations the electrodes were only pressed and did not contain
an additional binder phase. The binder is necessary to provide
good mechanical strength and sufficient adherence to the current
collector in order to prevent disintegration of the particles during
cycling [29].
An alternative to produce a carbon coating on the active mate-
rial can be realized by high-shear dry mixing. In this process a blend
of active material and carbon black is agitated by peripheral pad-
dles with high speeds whereby the blend is centrifuged to the outer
shell, getting stressed in a small gap between the shell and the
paddles [30,31]. Compared to a commonly used planetary mixer, a
dry mixer reveals a number of process-related differences. Mainly,
two effects influence carbon black crushing and distribution during
the process. First is the excessive specific energy input of the dry
mixer and the resulting high mechanical stresses in the mode of
compaction, shear, and impact. Such conditions induce tens of
thousands of collisions per second for a particle with comparable
impact strength in both the normal and tangential direction [32].
Second is a crushing effect of the carbon black caused by the active
material (analogous to ball mills). Due to this, comprehensive
deagglomeration of the carbon black and the achievement of a
homogeneous mixing state is possible. With increasing load and
stress duration, more and more fragments are adhering to the
surface of the active material, forming a layer around the particles.
After appropriate energy input even mechanofusion occurs, i.e.
penetration of the fines into the open pores of the larger particles
and thermo-mechanical induced fusion of particles, forming a
dense shell of carbon around the active material [33]. In this paper
the consequences of this process for the manufacturing of LIB
electrodes are investigated, i.e. properties and processing of the
slurries and the electrical and electrochemical behavior of the
electrodes are presented.
2. Experimental
2.1. Electrode and cell preparation
Li(Ni1/3Mn1/3Co1/3)O2 (NM-3100, Toda America, USA) with a
mean particle size of 8.9 mm, measured by electroacoustic spec-
trometry (DT1200, Dispersion Technology, USA), and a BET surface
of 0.4 m2/g, (measured by Flowsorb II 2300, Micromeritics, USA),
was used as the active cathode material (AM). Carbon black (Super
C65) with a BET surface of 62 m2/g and a mean TEM primary par-
ticle size of 35 nm [17] was acquired from Imerys Graphite & Car-
bon, Switzerland.
Dry mixing of the active material and the carbon black was
performed in a Nobilta NOB-130 (Hosokawa Micron Corporation,
Japan). Maximum loading with this machine is 250 g of powder. In
the current investigation fixed rotation speeds at 3600 min1 or
4000 min1 and mixing intervals up to 70 min were chosen.
Slurries, electrodes and cells were prepared in order to inves-
tigate the influence of the dry mixed powders on the electrode
manufacturing process and on the properties of the final product.
First of all, a polyvinylidene difluoride (PVDF) binder (Solef 5130,
Solvay Solexis S.p.A., Italy) was dissolved in N-methyl-2-pyrroli-
done (Sigma Aldrich, Germany) at ambient temperature. The dry
mixed particle blend was dispersed in the PVDF solution in a vac-
uum equipped dissolver (VMA Getzmann, Germany) for 30 min at a
maximum speed of 2000 min1. The resulting base slurry contains
around 92 wt.% of Li(Ni1/3Mn1/3Co1/3)O2 (NMC), 4 wt.% of carbon
black and 4 wt.% of PVDF binder with a solid content (NMC þ CB) of
22.5 vol.%.
Furthermore the influence and necessity of additional conduc-
tive material was investigated. Therefore, an extra of up to 5 wt.%
untreated carbon black or up to 10 wt.% of graphite powder (Timrex
KS6L, Imerys Graphite & Carbon, Switzerland, mean particle size of
3.4 mm and BET surface of 21 m2/g), was added to the base slurry in
the dissolver mixer.
Reference slurries were prepared by mixing untreated AM and
CB powders directly in the dissolver mixer. Without a preceding dry
mixing, the slurries were found to yield a much higher viscosity.
Hence, the solid content had to be reduced by additional solvent
down to 18 vol.% in order to get a slurry that can be operated for
electrode preparation.
Electrodes were cast on a 20 mm thick aluminum foil using a
continuous coater (KTF-S, Mathis AG, Switzerland) with knife
coating device at a gap of 200 mm and a speed of 0.2 m/min. The
electrodes were dried in a first stage at 80  C air temperature, fol-
lowed by a second stage at 120  C, in each case for 5 min.
Compaction of the electrodes was done in a heated calendar at
60  C with adjusted clamping force so that an electrode porosity
around 30 vol.% was achieved.
Pouch cells with ceramic coated separator foil (Separion®,
Evonik, Germany) and graphite anode were prepared for electro-
chemical tests. Electrode size was 50  50 mm2 for the cathode and
54  54 mm2 for the anode. Before assembling all electrodes and
separators were dried overnight in a vacuum furnace at 130  C. All
pouch cells were assembled in a dry room at a dew point of 50  C
to 60  C. The electrolyte was 1 M LiPF6/ethylene carbonate
(EC) þ diethyl carbonate (DEC) (1:1 by weight ratio) obtained from
BASF, Germany. After assembling, all cells were stored around
20 h at 40  C to facilitate a homogeneous distribution of the elec-
trolyte within the electrodes.
2.2. Characterization
Electrical resistivity of the AM/CB blends was measured in a
compaction tool with metallic punches and an isolating die. By
 W. Bauer et al. / Journal of Power Sources 288 (2015) 359e367
using the punches as electrodes, the resistance can be measured as
a function of compaction. For resistance measurements, 1 g of
powder was filled into the die. A constant voltage of 1 V was pre-
defined and the current through the sample measured by an RLC
meter, (RLC 200, Grundig GmbH, Germany). Simultaneously to the
blend resistance, the compaction was determined from the gap
between the punches [34].
Rheological tests were carried out using steady state flow and
oscillation measurements (amplitude sweeps). A rheometer (MCR
300, Paar Physica, Austria) with plateeplate geometry (PP50) at a
temperature of 25  C was used for this. Measurements were per-
formed immediately after finishing the slurry mixing process,
starting with steady state flow and followed by the amplitude
sweep. Prior to all measurement runs, the inserted samples were
exposed to a two-minute equilibrium period for the steady state
flow and a five-minute period for the oscillation measurement
which ensures the correct sample temperature and an acceptable
level of the residual normal force. A reference shear rate of 50 s1
was chosen for comparing the slurry viscosities as this value ap-
proximates typical conditions in the coating device.
The electrode resistivity was determined after calendaring.
Discs with a diameter of 16 mm were punched out of the electrodes
and placed between two polished copper cylinders with a diameter
of 16 mm. A pressure of 6.5 kPa was applied on the cylinders while
the across-the-electrode resistance was measured by a milli-
ohmmeter (MO 199, Conrad Elektronik).
Samples for SEM investigations were either prepared by
rupturing the electrodes or by ion milling using a triple ion beam
cutter system (EM TIC-3X, Leica Microsystems, Austria) with argon
as working gas. Typically ion energies of 7 keV were generated,
resulting in preparation times of 5e8 h, depending on the sample
thickness.
The electrochemical performance was measured with a self-
made battery cycler using galvanostatic mode in a temperature
controlled room at 20  C. At first, all cells were formatted by going
through two cycles at 0.05C charging and 0.5C discharging rates.
Following cycles were performed symmetrically at 0.5C, 1C, 2C and
3C, each step repeated 10 times, and finished with 50 cycles at 0.5C.
Cycling voltage limits were set at 3.0 V and 4.2 V. The C-rate was
calculated based on the theoretical capacity of the NMC cathode
material (155 mAh/g).
3. Results
The subsequent results can be divided in two parts. First the
results of the dry mixing step itself and the impact of the dry mixed
blend on conductivity (powder and coating), rheological behavior
and binder distribution are investigated. The second aspect com-
prises the addition of further conductive additives in order to
Fig. 1. SEM micrographs of NMC particles, after 7 min at 4000 min1 (left) and 70 min at 4000 min1 (middle) dry mixing with carbon black, pristine material (right).
361
comprehend the previous results and to prepare a functioning cell.
3.1. Dry mixing
Even after a moderate processing time (e.g. 7 min at a rotation
speed of 4000 min1) nearly all of the carbon black is adhering at
the surface of the NMC particles (Fig. 1, left). Primary particles and
agglomerates of carbon black are still recognizable and the coating
also does not completely cover the NMC surface so that also the
subjacent surface is detectable. After an extended processing time
of 70 min at 4000 min1 the carbon layer is dense and smooth
(Fig. 1, middle). Some remaining individual carbon black particles
still adhere on the surface, but the coating is so thick and close that
the morphology of the underlying NMC structure is hardly
detectable. For better comparability a non-treated active material is
depicted in Fig. 1, right side.
3.1.1. Rheology
Dry mixing and consequential deposition of the carbon black on
the active material significantly influences the handling of the
powder blends and the slurry preparation. An extended powder
preprocessing allows the enhancement of the solid concentration
up to 22.5 vol.% at the preferred viscosity (3 Pa$s at shear rate
50 s1), while for the untreated powder blend this viscosity level is
already achieved at a solid concentration of 18 vol.% (Fig. 2 left).
The absence of free carbon black particles also influences the
stability of the prepared slurry which can be deduced from oscil-
lation measurements, e.g. amplitude sweeps (Fig. 2 right). With the
preprocessed powder the loss modulus G00 is superior to the storage
modulus G0 , i.e. a fluid-type system with a negligible yield strength
is formed, which is prone to sedimentation. On the other hand, a gel
structure can be obtained with free carbon black in combination
with the binder, as it can be derived from a dominating storage
modulus G0 at low deformation numbers. In this case, the resulting
yield strength effectively prevents sedimentation even for the large
NMC particles.
3.1.2. Electrical resistance
A low electrical resistance should be expected in electrodes
which were manufactured from dry mixed powder blends due to
the homogeneous carbon black distribution. However, these elec-
trodes reveal an electrical resistance that is approximately three to
four magnitudes higher compared to the initial powder blends and
compared to reference slurries from dissolver mixed blends (Fig. 3).
A subsequent calendaring step usually reduces porosity and re-
sistivity of the electrode. A slight resistance drop is actually
observable with the moderate dry mixed blend (at 7 min processing
time), but the resistivity still remains at an inappropriate high level.
After extended dry mixing for 70 min the compression behavior of
362 W. Bauer et al. / Journal of Power Sources 288 (2015) 359e367
Fig. 2. Flow curves (left) and amplitude sweeps (right) of slurries with untreated or dry mixed (70 min at 4000 min1) powder blends. Solid concentrations were adjusted in order
to set up a viscosity level suited for the electrode deposition process.
the electrode becomes even nearly independent on compaction.
3.1.3. Binder distribution
SEM micrographs of the electrodes reveal substantial differ-
ences for the binder distribution. In a typical dissolver based elec-
trode the binder forms a compound with associated and enclosed
carbon black particles. Binder and carbon black are not distributed
homogeneously on the NMC surface but locally concentrated on
spots, while at some regions no trace of an additive can be observed
on the NMC particles (Fig. 4 left column). Often the binder/carbon
black clusters are located in narrow gaps between the NMC parti-
cles. Cross sections of the electrode reveal the porous structure of
the clusters that allows a very space-consuming distribution of the
carbon black. However, after using a dry mixed powder blend, the
binder forms a polymer shell (Fig. 4 right column) which encloses
the NMC particles after drying. Although in the cross sections the
binder shell can be hardly distinguished from the subjacent carbon
layer, it seems likely that the binder completely enwraps the NMC
particles, preventing a direct contact of the carbon coatings.
3.2. Admixture of conductive additives
An extra of conductive additives was admixed to the base slurry
composition in order to clarify and to refine the former results. The
further material, carbon black or graphite, was added to the dry
mixed blends of NMC and carbon black (3600 min1/10 min)
Fig. 3. Resistivity of powder blends and electrodes after dry preprocessing. Reference
electrode was made with untreated powder blend. Porosity was manipulated by
compaction or calendaring.
during the homogenization step in the dissolver.
3.2.1. Rheology
Fig. 5 depicts the impact of the additional particles on the slurry
rheology. With an addition of 1, 3 or 5 wt.% carbon black the vis-
cosity at a shear rate of 50 s1 increases from 1.6 Pa$s to 2.7, 4.5 or
7.1 Pa$s, respectively. While the lower amounts can still be handled
by knife coating, the addition of 5 wt.% carbon black hits the edge of
processing. The larger graphite particles, on the other hand, do not
significantly change the rheology of the slurry even after an addi-
tion up to 10 wt.%.
With 1 wt.% of extra carbon black the slurry still keeps its fluidic
behavior, while a yield point and gel like behavior is detectable in
amplitude sweeps for the larger carbon black additions of 3 or
5 wt.%. The addition of graphite up to 10 wt.% does not change the
viscoelastic behavior, i.e. all slurries conserve their fluidic character.
3.2.2. Electrical resistance
The electrode resistance is significantly reduced by the addi-
tional carbon additive (Fig. 6). With an extra amount of 3 wt.%
carbon black, added to the dry mixed blends (already containing
4 wt.% carbon as the particle coating), the specific resistance is still
slightly higher than the resistance of the dissolver processed
reference electrode containing only 4 wt.% carbon black, while with
5 wt.% it is lower. The slope of the compaction related curves is very
similar to the reference electrode ones. However, adding an extra of
graphite to the dry mixed blends provides an altered result as in
this case the resistivity becomes independent on the porosity.
Reducing the resistivity can only be achieved by additional graphite
but not by electrode compaction. For example, an amount of
10 wt.% graphite is necessary to obtain the same resistivity as a
highly compacted electrode with additional 3 wt.% carbon black,
but without calendaring the resistivity of carbon black containing
electrode is one decade higher.
3.2.3. Electrode morphology
Thick binder coatings which completely enwrap the NMC par-
ticles (see Fig. 4 right column) are no longer observable in the SEM
micrographs after the addition of the extra carbon black. Only a
small amount of the binder is fixed directly on the surface of the
NMC particles, while a large fraction seems to be embedded within
carbon black agglomerates (Fig. 7 left). Cross sections of the sample
reveal a microstructure which is comparable to samples without
dry preprocessing (Fig. 7 right). The localization of the binder
within the carbon black agglomerates is hard to recognize. Never-
theless, the compact domains which cross the carbon black ag-
glomerates are an indication of a binder rich phase. In contrast,
 W. Bauer et al. / Journal of Power Sources 288 (2015) 359e367 363

Fig. 4. SEM micrographs of ruptured cathodes (upper line) and ion etched cross sections (lower line) after conventional dissolver mixing (left column) and dry preprocessing (right
column) of NMC and carbon black particles.

Fig. 5. Flow curves (left) and amplitude sweeps (right) of slurries from dry preprocessed (3600 min1, 10 min) NMC/CB blends after addition of various amounts of conductive
additives.

regions of solely carbon black which is deposit on the surface or in
the pits of the NMC particles still retain a particulate structure with
visible porosity (see magnified section in Fig. 7 right).
Adding an extra of graphite particles to the dry preprocessed
blends does not prevent binder deposition on the NMC particles.
But in this case also the flakes of the delaminated graphite particles
are embedded in the polymer shell (Fig. 8). The flakes are able to
penetrate the shell due to their extension and facilitate electrical
contact between the NMC particles.
3.2.4. Electrochemical properties
With additional conductive material the resistivity of all sam-
ples is sufficiently reduced to enable the implementation of cycling
tests. For that, pouch cells were prepared using cathodes with and
without calendaring. Before calendaring the electrode sheets
exhibit porosities in the range of 42e47 vol.%, after calendaring the
porosity was reduced to a target value of approximately 30 vol.%
(see Table 1).
Fig. 9 compares the discharge voltage profiles in the 12th and
the 42nd cycle, corresponding to a rate of C/2 and 3C, before and
after calendaring. Similar voltage profiles at C/2 rate and a negli-
gible capacity improvement by the calendaring process are an
indication for a conductive additive distribution that enables high
utilization of the active material. However, at 3C rate the dis-
charging capacity is reduced by incomplete charging due to high
cathode polarization induced by the higher current and the
different internal resistance of the cells.
All cells show very similar rate capabilities for C-rates below 1
(Fig. 10). At higher C-rates the discharge capacities deviate distin-
guishable. The highest discharge capacities of all samples at a C-rate
of 3 are measured for the addition of 10 wt.% graphite, while with
1 wt.% carbon black and 5 wt.% graphite the lowest values are
observed. Calendaring does not influence the discharge capacity up
to a C-rate of 1. However, in most cases the discharging capacity is
improved by the calendaring step at higher C-rates. Only the
calendared 10 wt.% graphite sample somewhat deteriorates as it
364 W. Bauer et al. / Journal of Power Sources 288 (2015) 359e367
Fig. 6. Resistivity of electrodes manufactured from untreated or dry preprocessed
(3600 min1, 10 min) powder blends with extra conductive material. Porosity was
manipulated by calendaring.
reveals slightly lower discharging capacities above 1C after
calendaring.
4. Discussion
Blending of active material and carbon black in a dry mixer is a
simple and industrially compatible method for applying a carbon
coating on active electrode particles. Depending on the intensity of
the mixing process a porous layer of weakly adhering carbon black
aggregates up to a dense and homogeneous carbon coating can be
formed. Although the massive energy input of the dry mixer in-
duces some structural damage, the powder conductivity of the
blends still remains in the same order of magnitude and is com-
parable to mixtures which are prepared in a conventional dissolver
mixer. However, after preparing electrodes with the blends, the
layer conductivity is drastically reduced. Therefore, it was not
feasible to manufacture operative cells with the dry preprocessed
powders.
For understanding this effect the role of the binder has to be
taken into account. In Ref. [35] Zheng et al. describe the interaction
between active material, conductive carbon additive and binder for
LiNi0.8Co0.15Al0.05O2 cathodes. In a composite electrode, there is
always a competition between active material and conductive ad-
ditive particles for the available polymer binder. And although the
mass fraction of the conductive aid is much smaller, carbon black or
acetylene black have a significant higher specific surface which
often allows them to dominate the absolute surface area. According
Fig. 7. SEM micrographs of an ion milled NMC cathode from a moderate dry mixed blend (3600 min1, 10 min) with additional 3 wt.% carbon black. The arrows indicate spots where
carbon was deposited on the central particle during the dry mixing process.
Fig. 8. SEM micrograph of a ruptured NMC cathode from a moderate dry mixed blend
(3600 min1, 10 min)) with additional 5 wt.% graphite.
to the concept of fixed binder layers, proposed by Liu et al. [36], a
significant proportion of the available binder is therefore bound
and immobilized by the conductive aid particles. In a typical elec-
trode, this leads to the formation of clusters consisting of the binder
phase with enclosed or attached conductive fillers [37]. The struc-
ture of these clusters plays a vital role for the performance of the
electrode. Binder clusters with a robust percolation structure of
carbon black particles provide both binding and conducting func-
tionality [38]. By that, the binder not only fixes the active material
in the electrode, it also fastens the conductive additives and pre-
vents washing out of the fine carbon black particles (Fig. 11 left).
From this model, further recommendations can be derived. First,
all of the binder should be incorporated in the cluster structure as
surplus binder is deposited on the surface of the active material. A
layer of pure binder will not only reduce the electronic conduc-
tivity, but also block the lithium ion transfer. Second, the embedded
carbon black particles must be able to establish a conductive matrix
that spans the whole cluster volume. From this view it should even
be beneficial to distribute the carbon black particles in less
dispersed form, i.e. with a fractal morphology [39] that needs a
Table 1
Porosity of electrodes with extra conductive material before and after calendaring.
Extra conductive material Porosity before calendaring After calendaring
1 wt.% carbon black 42.0% 32.1%
5 wt.% carbon black 44.9% 29.8%
5 wt.% carbon black 46.5% 31.1%
5 wt.% graphite 41.4% 30.4%
10 wt.% graphite 44.5% 28.9%
 W. Bauer et al. / Journal of Power Sources 288 (2015) 359e367 365

Fig. 9. Galvanostatic discharging profiles for electrodes with dry preprocessed powder blends and extra conductive material in the 12th and the 42nd cycle.

Fig. 10. Rate capability of dry preprocessed powder blends with extra conductive material without (left) and with (right) calendaring.

Fig. 11. Schematic description of typical particle e binder distribution without (left) and after dry preprocessing (middle). The right scheme depicts the binder distribution after the
addition of extra carbon black to the dry preprocessed blend.

lower amount to form a percolating network within the binder
phase [40], whereas well de-agglomerated or even primary parti-
cles require a higher number of particles as they are packed in a
denser configuration.
During the intensive dry mixing process an immobilization of
the carbon black particles occurs, i.e. they are no longer free
movable but attached to the surface of the active material. This has
consequences for the slurry processing primarily, since the carbon
black becomes rheologically inactive. Without a fraction of inter-
acting small particles, the viscosity of the slurry is significantly
reduced. Another implication for the slurry processing is the
absence of a stabilizing gel structure which is established by the
interaction of carbon black particles and binder. Instead of that, a
fluidic behavior is detectable in the viscoelastic system as it is
known for CB free slurries containing only large particles [5].
The immobilization of carbon black particles also suppresses the
formation of a percolating conductive network within the binder
phase. Instead of interpenetration, during drying a stacked
configuration is formed as the solved binder deposits on the carbon
coating as a wrapping polymer shell. Although the binder still
fulfills its bonding function, it also interrupts the electrical
connection between the particles (Fig. 11 middle). In this configu-
ration the electrical resistance of the electrode is increased by the
polymer shell to the measured high level, in spite of an adequate
conductivity of the embedded particles. Compressing these elec-
trodes in a calendar reduces the porosity but does not increase the
conductivity because the polymer layer still separates the particles.
A slight reduction of the resistivity can be observed only in case of
366 W. Bauer et al. / Journal of Power Sources 288 (2015) 359e367
moderate dry mixing where a rough carbon coating allows the
formation of some additional connecting paths by puncturing the
polymer layer.
The crucial relevance of the localization of the conductive ad-
ditive can be proven by admixing an extra portion of carbon black
particles to the dry premixed blend. The rheological properties as
well as the stability behavior of these slurries are virtually identical
to conventional preparations using electrode particles without a
dry mix carbon coating, i.e. the viscosity is comparable and the
slurries rebuild a gel structure. This also becomes evident for the
resistivity of the electrodes since the metered values for reference
electrodes (with 4 wt.% of carbon black) is located between the
resistivity of admixtures with 3 wt.% and 5 wt.% extra carbon black
respectively. Even the slope of the appropriated compaction curves
is comparable to the behavior of the reference electrodes. From this
point of view the original carbon coating is obviously transparent
for the properties of the electrode.
Graphite is less common as conductive additive due to its larger
particle size. However, it delaminates under mechanical stress,
creating platelet shaped fragments which also have the potential to
form a connective framework. Although the number and the sur-
face area of the platelets is much lower than for carbon black, they
also interact with the binder and generate electrical pathways
through the binder phase after becoming deposited on the surface
of the particles. The addition of graphite is therefore an alternative
to carbon black in order to reduce the resistivity after the dry
mixing process.
In contrast to carbon black, the graphite framework does not
reveal a conductivity enhancement by the calendaring process. This
is mainly due to a lower number and due to the incompressible
nature of the graphite particles, while during the compaction of the
carbon black electrode agglomerates a large quantity of agglom-
erates are compressed, so that a lot of new particle contacts are
formed. Compressing graphite agglomerates also does not generate
the same improvement due to the lower conductivity of the basal
plane surfaces. As consequence of the pressure independence of the
graphite containing electrode, a much larger amount of graphite is
required to achieve the internal conductivity of a compressed car-
bon black containing electrode.
After adding extra carbon black or graphite it is feasible to
prepare stable functioning cells even with dry mixed powder
blends. This can be seen as evidence that a conductive network is
formed within the electrode. The best performance at higher C-
rates is yielded with an uncompressed graphite containing elec-
trode, but a relatively high graphite amount is required for that.
After calendaring the advance of graphite drops away, although the
conductivity is not influenced by the compaction. This is presum-
ably due to a steric effect, which was already identified with planar
graphene sheets [41]. At high content the compressed graphite
forms dense stacks with an aligned structure, which retard a fast
Liþ transport.
5. Summary
From a theoretical point of view, a thin layer of carbon which is
deposited on the surface of the active material should be an opti-
mum configuration as it provides sufficient electrical connection
between all particles with a minimum of conductive material.
However, this approach ignores the existence and function of other
components in the electrode. Especially the interaction of
conductive additives with the binder plays a dominant role for the
performance of electrochemical cells.
The existence of a mobile fraction of carbon particles is a vital
prerequisite for the preparation of well performing electrodes
containing a polymer binder. Carbon black particles or aggregates
not only stabilize the slurry against segregation effects, e.g. sedi-
mentation or agglomeration, and therefore support the formation
of a homogeneous electrode. They are also required to form con-
ducting paths through the binder phase and to prevent the creation
of isolating polymer layers between the active particles. Graphite as
conductive aid has a less expressed potential for slurry stabiliza-
tion, but can be used for the creation of a conductive network as
well. However, due to the larger particle size of graphite compared
to carbon black, more material is required for sufficient electrical
connection.
Although it is not feasible to prepare functional cells without
further carbon black or graphite addition, the existence of a thin
carbon coating has some advantages. It enables a homogeneous
current distribution on the surface of large sized active materials
and it can reduce reactions between the active material and the
electrolyte. Therefore, investigations are recommended in order to
fully integrate the dry mixing process into the process chain and to
evaluate the potential of a reduction of inactive components in this
configuration.
From these results it can be deduced that the optimized spatial
distribution of the conductive additives in the electrode varies with
the regarded scale length. On a short scale a homogeneous and thin
coating of the active material might be beneficial as it provides
extensive electrical contacting of the particle surface without
affecting lithium diffusion. A homogeneous distribution of carbon
black or graphite is essential when looking at the complete elec-
trode in order to realize an equivalent contribution of the entire
active material. However, on a medium scale a spotted distribution
is preferable, concentrating the conductive additive with the binder
in clusters localized next to the contact area of the active material
particles, which simultaneously provide binding and conducting
functionality.
References
[1] G. Liu, H. Zheng, A.S. Simens, A.M. Minor, X. Song, V.S. Battaglia, J. Electrochem.
Soc. 154 (12) (2007) A1129.
[2] M.E. Spahr, Carbon-conductive additives for lithium-ion batteries, in:
M. Yoshio, R.J. Brodd, A. Kozawa (Eds.), Lithium-ion BatteriesdScience and
Technology, Springer, New York, 2009, pp. 117e154.
[3] X.M. Liu, Z.D. Huang, S.W. Oh, B. Zhang, P.C. Ma, M.M.F. Yuen, J.K. Kim,
Compos. Sci. Technol. 72 (2012) 121.
[4] K. Wang, Y. Wu, S. Luo, X. He, J. Wang, K. Jiang, S. Fan, J. Power Sources 233
(2013) 209.
[5] W. Bauer, D. No €tzel, Ceram. Int. 40 (2014) 4591.
[6] M.E. Spahr, D. Goers, A. Leone, S. Stallone, E. Grivei, J. Power Sources 196
(2011) 3404.
[7] J.K. Hong, J.H. Lee, S.M. Oh, J. Power Sources 111 (2002) 90.
[8] H.Y. Tran, G. Grecoa, C. T€ aubert, M. Wohlfahrt-Mehrens, W. Haselrieder,
A. Kwade, J. Power Sources 210 (2012) 276.
[9] X. Qi, B. Blizanac, A. DuPasquier, M. Oljaca, J. Li, M. Winter, Carbon 64 (2013)
334.
[10] S. Ahn, Y. Kim, K.J. Kim, T.H. Kim, H. Lee, M.H. Kim, J. Power Sources 81e82
(1999) 896.
[11] S. Mandal, J.M. Amarilla, J. Ib ~ ez, J.M. Rojo, J. Electrochem. Soc. 148 (1) (2001)
an
A24.
[12] S. Kuroda, N. Tobori, M. Sakuraba, Y. Sato, J. Power Sources 119e121 (2003)
924.
[13] R. Dominko, M. Gaberscek, J. Drofenik, M. Bele, S. Pejovnik, J. Jamnik, J. Power
Sources 119e121 (2003) 770.
[14] D. Guy, B. Lestriez, D. Guyomard, Adv. Mater. 16 (2004) 553.
[15] A. Sever Skapin, M. Gaberscek, R. Dominko, M. Bele, J. Drofenik, J. Jamnik, Solid
State Ionics 167 (2004) 229.
[16] D. Guy, B. Lestriez, R. Bouchet, V. Gaudefroy, D. Guyomard, J. Power Sources
157 (2006) 438.
[17] E. Ligneel, B. Lestriez, O. Richard, D. Guyomard, J. Phys. Chem. Solids 67 (2006)
1275.
[18] Y.H. Chen, C.W. Wang, G. Liu, X.Y. Song, V.S. Battaglia, A.M. Sastry,
J. Electrochem. Soc. 154 (10) (2007) A978.
[19] G. Liu, H. Zheng, S. Kim, Y. Deng, A.M. Minor, X. Song, V.S. Battaglia,
J. Electrochem. Soc. 155 (12) (2008) A887.
[20] T.J. Patey, A. Hintennach, F. La Mantia, P. Nova k, J. Power Sources 189 (2009)
590.
[21] C.C. Chang, L.J. Her, H.K. Su, S.H. Hsu, Y.T. Yen, J. Electrochem. Soc. 158 (5)
 W. Bauer et al. / Journal of Power Sources 288 (2015) 359e367 367

(2011) A481. [32] W. Chen, R.N. Dave, R. Pfeffer, O. Walton, Powder Technol. 146 (2004) 121.
[22] M. Zhu, J. Park, A.M. Sastry, J. Electrochem. Soc. 158 (10) (2011) A1155. [33] K. Tanno, Kona 8 (1990) 7.
[23] K.C. Kil, G.Y. Kim, C.W. Cho, M.D. Lim, K. Kim, K.M. Jeong, J. Lee, U. Paik, [34] V. Wenzel, R.S. Moeller, H. Nirschl, Energy Technol. 2 (2014) 176.
Electrochim. Acta 111 (2013) 946. [35] H. Zheng, R. Yang, G. Liu, X. Song, V.S. Battaglia, J. Phys. Chem. C 116 (2012)
[24] W. Haselrieder, S. Ivanov, H.Y. Tran, S. Theil, L. Frobo €se, B. Westphal, 4875.
M. Wohlfahrt-Mehrens, A. Kwade, Prog. Solid State Chem. 42 (2014) 157. [36] G. Liu, H. Zheng, X. Song, V.S. Battaglia, J. Electrochem. Soc. 159 (3) (2012)
[25] C.M. Julien, A. Mauger, H. Groult, K. Zaghib, Thin Solid Films 572 (2014) 200. A214.
[26] R. Dominko, M. Gaberscek, J. Drofenik, M. Bele, Pejovnik, Electrochem. Solid- [37] S. Babinec, H. Tang, A. Talik, S. Hughes, G. Meyers, J. Power Sources 174 (2007)
State Lett. 4 (11) (2001) A187. 508.
[27] J. Drofenik, M. Gaberscek, R. Dominko, M. Belea, S. Pejovnik, J. Power Sources [38] H. Wang, T. Umeno, K. Mizuma, M. Yoshio, J. Power Sources 175 (2008) 886.
94 (2001) 97. [39] C.L. Barrie, P.C. Griffiths, R.J. Abbott, I. Grillo, E. Kudryashov, C. Smyth, J. Colloid
[28] R. Dominko, M. Gaberscek, J. Drofenik, M. Bele, J. Jamnik, Electrochim. Acta 48 Interface Sci. 272 (2004) 210.
(2003) 3709. [40] D. Guy, B. Lestriez, R. Bouchet, D. Guyomard, J. Electrochem. Soc. 153 (4)
[29] B. Lestriez, C. R. Chim. 13 (2010) 1341. (2006) A679.
[30] A. Du Pasquier, C.C. Huang, T. Spitler, J. Power Sources 186 (2009) 508. [41] F.Y. Su, Y.B. He, B. Li, X.C. Chen, C.H. You, W. Wei, W. Lv, Q.H. Yang, F. Kang,
[31] H. Bockholt, W. Haselrieder, A. Kwade, ECS Trans. 50 (26) (2013) 25. Nano Energy 1 (2012) 429.