Evaluation of different separators in lab-scale lithium-ion batteries

Viet Nguyen
ABSTRACT: The rate performance of the separator FS2190 with different layer thickness and in different cell
designs was investigated and compared to the performance of the separator FS2226. The highest discharge capacities were observed for cells built with 3 layers of FS2190 (thickness around 695 m). Increased layer thickness
led to decreasing performance and lower layer thickness resulted in cell failure. The theoretical discharge capacity
of graphite 372 mAhg1 was nearly reached (350 mAhg1) with FS2190 separators at low current rates. However,
discharging the cell at the rate of 1C led to great discharge capacity loss. A better performance of Swagelok cells
in comparison to coin cells with same cell arrangement could be observed.

1. Introduction
The lithium-ion battery (LIB) is at the focus of recent
research and thus all cell components are still under
permanent development.1 An important component in
battery cells is the separator, although it does not contribute directly to the electrochemical reactions inside
the cell. Separators have two basic functions to fulfill:
on the one hand, separators have to prevent the physical contact between the anode and the cathode, which
would cause a short circuit. On the other hand, they
have to provide fast Li+-ion transport through the
separators pores, which are filled with electrolyte.2
There are also several other requirements to meet if
possible, because of the influence on many properties
in cell. The separator has to provide chemical, mechanical and thermal stability and safety, also under
demanding charging and discharging conditions.3
This includes mechanism like the shut-down effect
(prevention of LIB overheating through separator
layer melting and thus pore sealing and stop of cell
reaction) or zero shrinkage (prevention of shrinkage
of separator at higher temperatures).3 A low separator
layer thickness has a positive effect on the conductivity within the cell and the energy consumption
(overpotentials) could be reduced. A compromise
between separator stability properties and a low layer
thickness has to be found. The porosity, pore size and
the wettability are important for the absorption of the
electrolyte and for appropriate ion conductivity.3
Separators for LIBs could be divided in different classes, whereby nonwoven mat separators, inorganic
composite separators and porous polymeric membrane separators are the currently most common.4
Porous polymer separators are the most common separators4 because of their combination of important
properties like relatively low costs, good Li+-ion
transport properties and chemical stability.5 This kind

of separators most often consist of the low-cost polymers polyethylene, polypropylene or mixture of both.3
The used separator for this work is the FS2190 separator. It is a polypropylene woven separator. In SEM
micrographs, it reveals large fibers of 10 m diameter
and is described to have slow wetting behavior even
with hydrophobic electrolytes.2
In this work, the C-rate test of the separator FS2190
was realized in two different battery designs with
different layer thickness. The results were compared
with results from C-rate test of coin cells built with
the separator FS2226 and Celgard 2500.

2. Experimental
Two different battery designs were tested: coin cell
(CR2032, Hohsen Corp.) and Swagelok cell design.
Materials used for these cells were Li metal (Rockwood Lithium) as counter and reference electrodes
and LP30-electrolyte (1 M lithium hexafluorophosphate in ethylene carbonate : dimethyl
carbonate 1:1, LP30 Selectilyte, BASF). Graphite
electrodes were prepared in-house. They were composed of 90 wt.% SLP30 graphite (IMERYS) as the
active material, 3 wt.% carboxymethyl cellulose
(CMC) binder, 2 wt.% styrene-butadiene rubber
(SBR) and 5 wt.% carbon black Super C65. The separator FS2190 was commercially available (from
Freudenberg and Co. KG. For the use in the cells, the
electrodes and separator were punched; producing
material with circular form and 12 mm diameter. The
punched electrodes were dried under vacuum for two
days at 120 C and the separators at 40 C, respectively. Cells were built using auxiliary materials (spacer:
0.3 and 0.5 mm, spring: 0.7 and 1.4 mm) to reach
uniform cell thickness (3.2 mm for coin cell). 80 l
LP30-electrolyte was added on the separator and dis-

tributed until the whole separator was optically wetted

with electrolyte. For both cell designs the FS2190
separator was tested using different layer thickness.
The investigated separator thicknesses are shown in
Table 1. The C-rate testing procedure was executed
with battery test equipment from Maccor Inc. and is
described in Section 7. Supporting Information. The
discharge capacities were recorded at different C-rates
(0.1, 0.2, 0.5, 1, 2, 5 and 10 C) and every measurement at a certain C-rate was done three times. For
validation of reproducibility, every cell was built at
least three times. The average values of the three
measurements of the cells are plotted. The corresponding standard deviations are plotted as error bars.
Data for the separator FS2226 and Celgard 2500 were
provided by other groups within the master module
using the same procedure.
Table 1: Thickness of used separators
Layer thickness / m
1 layer FS2190

277

2 layers FS2190

470

3 layers FS2190

695

6 layers FS2190

1400

1 layers FS2226

177

2 layers FS2226

347

3 layers FS2226

520

6 layers FS2226

1048

Figure 1: Comparison of wettability of the separator

FS2190, FS2226 and Celgard 2500 with 2 layers
It could be seen, that FS2226 was completely wetted
after addition of the electrolyte, whereas FS2190 and
Celgard 2500 are not wetted. This is the reason, why
attention was paid to wet the whole separator surface
during cell assembly. Cells build with the Celgard
separator get an additional rest step of 24 h to provide
a better wettability of the separator.
3.2. Comparison of different layer thicknesses

3. Results
3.1. Cells with 1 layer FS2190, FS2226 and
Celgard 2500 separator
Swagelok cells and Coin cells, which were built with
1 layer FS2190 and FS2226, did not show evaluable
result in the C-rate test. Also all cells, which were
built with Celgard 2500, did not reversible cycling.
To evaluate the wettability of the different separators,
a picture was taken immediately after adding some
drops of LP30-electrolyte on the separators as shown
in Figure 1.

3.3. Discharge capacity loss and coulombic efficiency

Figure 2: Comparison of discharge capacities of the

cells with different separator layer thicknesses and at
different C-rates: a) Coin cells with FS2190, b)
Swagelok cells with FS2190 and c) Coin cells with
FS2226
The discharge capacity as function of the cycle number for different layer thicknesses is shown in Figure
2. In general, a decrease of the discharge capacity
with increasing Charge/discharge C-rate is observed.
By means of Figure 2a), coin cells with 3 layers
FS2190 separator show the best discharging behavior
over all C-rates. This observations match partially
with the results taken from Swagelok cells (Figure
2b). For Swagelok cells, the discharge behavior of
cells with 2 and 3 layers of FS2190 are similar in
nearly the whole range. Only at 2C, the Swagelok
cells with 2 layers show better discharge capacities
than the 3 layered cells, but the standard deviation for
this measurement is high. The Swagelok cells with 6
layers of FS2190 shows apparently lower discharge
capacities for the different C-rates than the cells with
less separator layers. In comparison, the coin cells
with 6 layers have nearly similar performance like the
cells with 2 layers. The coin cells with FS2226 have
discharge capacities for cells with 6 layer separator
similar to the performance of the 2 and 3 layered cells
and even the highest for 0.1 and 0.2 C. FS2226 coin
cells with 3 layers show the worst performance. For
some cells, especially for the 2 layered FS2190 coin
cell, an increase of discharge capacity during the first
cycles is observed. This is probable caused by an
inappropriate wetting of the separator, which is completed after a certain time (after first or second cycle).
Also within the measurement of one cell, it could
often be seen a slight decrease in discharge capacity,
although it is measured at the same C-rate.

Figure 3: discharge capacity loss under different Crates

In Figure 3, the discharge capacity loss (DCL) under
different C-rates for all investigated cells is shown.
The second data point for every cell at each C-rate is
plotted. The discharge capacity for a certain C-rate
(DCC) is set in relation to the discharge capacity at
C=0.1 (DCC=0.1).

It is assumed that for the low C-rate of 0.1 C, the system has enough time to discharge completely. As seen
in figure 3, the coin cells (of both FS2190 as well as
FS2226) have a similar DCL behavior. At the rate of
1C, they show already a DCL of around 80%. All
cells show at 10C almost no discharge capacity anymore. Swagelok cells with 2 and 3 layers FS2190
show significantly less DCL over all C-rates with a
DCL of around 40% at 1C. At first sight over all layer
thicknesses, the FS2190 coin cells show slightly lower DCL than the coin cells with FS2226. For 0.5 C it
could be shown that the best FS2190 coin cell (3 layers) have a DCL of about 45%, whereas the best
FS2226 coin cell (2 layers) have 2-3 % higher DCL.
The coulombic efficiency is the discharge capacity
DCC for a certain C-rate in relation to the charge capacity CCC:

Table 2: First cycle coulombic efficiency of the investigated cells

2 layers
3 layers
6 layers
2 layers
3 layers
6 layers
2 layers
3 layers
6 layers

FS2190 coin cell

Coulombic
efficiency / %
77,6

FS2190 coin cell

84,6

FS2190 coin cell

80,3

75,3

79,3

FS2190 Swagelok

81,7

FS2226 coin cell

84,7

FS2226 coin cell

86,3

FS2226 coin cell

87,1

FS2190 Swagelok
FS2190 Swagelok

Table 2 list the first cycle for all investigated cells.

The DC and CC values of the first cycle were used to
calculate . In general, is around 75-88% for the
investigated cells. It could be observed that coulombic
efficiency increase with increasing layer thickness for
the FS2226 coin cell and the FS2190 Swagelok cell.
Comparing the different layer thickness, the FS2226
coin cells have the highest of the investigated cells;
with the 6 layered FS2226 having the highest . The
lowest efficiency is observed for Swagelok cells.
3.4. Comparison of different cell designs and separators

Figure 4: Comparison of discharge capacities of the

cells with different cell designs and separators: a)
Cells with 2 layers separator b) Cells with 3 layers
separator c) Cells with 6 layers separator

Figure 4 plots the discharge capacity as function

of the cycle number for different cell designs.
Figure 4a) and 4b) shows that the performance
regarding discharge capacities of the Swagelok
cells is the best. The Swagelok cells with 2 and
3 layers reach a discharge capacity of around 350
mAhg1 at 0.1 C. This is little less than the theoretical capacity of 372 mAhg1. Like already recognizable in figure 3, the Swagelok cells show a
less dramatic drop of discharge capacity than the
coin cells over increasing C-rates. The coin cells
with FS2190 have compared to the coin cells
with FS2226 a better discharge capacity performance. The FS2226 coin cells with 2 and 3 layers
do not even reach 300 mAhg1 during discharge
4

with 0.1 C. The performances of the 6 layered

cells are more similar, but the Swagelok cell is
also slightly better than the both coin cells.
4. Discussion
4.1. Cells with 1 layer FS2190, FS2226 and
Celgard 2500 separator
As shown in Figure 1, Celgard 2500 separators exhibited poor wettability and without an appropriate wetting of the separator no reversible cycling is possible.
Because many Celgard cells did not even show a performance for the first cycle, a wetting problem is
probable. Although, cells with FS2190 also show
poor wettability, a wetting problem is excluded, since
the cells with greater layer thickness show measurable
discharge capacities. The fact that all cells, built with
1 layer FS2190, did not show evaluable results is an
indication that the thickness of 1 layer FS2190 is not
sufficient to ensure a reversible cell operation. It is in
contrast to 2 layers of FS2190, which provides good
results in the C-rate test. Possibly, short circuits could
occur, because of the inadequate stability (mechanical
or electrochemical) of too thin FS2190 separator.
Another evidence is that also cells with FS2226,
which show good wettability, did not show a performance. Therefore it is reasonable to assume that a
minimum layer thickness for a certain separator type
exists. Below this minimum thickness, cells do not
work properly.
4.2. Comparison of different layer thicknesses
The comparison of different layer thickness shows
best discharge capacity performance for coin cells
with FS2190 with 3 layers of FS2190, for Swagelok
cells with FS2190 the performance of 2 layers and 3
layers are almost similar. For FS2226 coin cells, best
results were measured for the cells with 6 layered
separators. This is surprising; cells with greater layer
thickness should show less discharge capacity, because of decreasing ion conductivity and increasing
internal resistance respectively. It could be assumed a
trade-off between increasing stability and decreasing
ion conductivity. This optimum is reached for FS2190
separators with 3 layers; these cells could show the
best stability at relatively low resistance. FS2226
separators are non-woven and thus could have greater
porosity or greater pore size. An indication for this is
that the wettability with the electrolyte for FS2226 is
much better than for FS2190. Additionally, it should
be kept in mind that the FS2226 separators are thinner
than the FS2190 separators (comparing the same layer

number). It is possible that increasing the layer thickness of this separator do not increase the resistance in
a relevant way, because the porosity is still sufficient
for appropriate ion conductivity. On the other hand,
the stability of the separator increases. This could lead
in sum to a better discharge capacity of 6 layered
FS2226.

4.3. Discharge capacity loss and Coulombic efficiency

The capacity loss could be seen as an indicator for the
resistance in the cell, especially at high C-rate. In an
ideal cell, doubling the C-rate of the discharge process should halve the time needed to discharge completely. In sum the discharge capacity should stays the
same different C-rates. In real cells, the internal resistance turns some of the energy into Joule heat,
which decreases the available discharge capacity. The
3 layered FS2190 coin cell with the highest discharge
capacity also shows the lowest DCL compared with
the other FS2190 coin cells (for 0.2 and 0.5 C). The
same observation could be done for the Swagelok
cells. For the coin cells with FS2226 separator, the
lowest DCL is obtained with the 2 layered FS2226.
This could mean that the cell has a lower internal
resistance than the 3 or 6 layered cells, which is plausible. The 3 and 6 layered FS2226 coin cells show
similar DCL behavior, which means a similar resistance for different C-rates. This insight strengthens
the argumentation, why 6 layered FS2226 coin cells
have this good discharge capacity performance. It
provides a similar resistance like the 3 layered cells,
but has twice the separator layer thickness, which
provides higher stability.
An important reaction at the first cycle in LIB cells is
the SEI formation.6 SEI formation accompanies with
redox reactions and thus current flow, which built the
SEI and is detected as additional charge capacity. So a
lower coulombic efficiency could mean a better SEI
formation. Thus the coulombic efficiency could be
seen as a measure of the formation of the SEI. The
cells, which have the lowest coulombic efficiency
compared by layer thickness, are the Swagelok cells
with FS2190. It could be assumed a certain effect of
the separator on the SEI formation and stability. Regarding the difference of the coulombic efficiencies,
the worse performance of FS2226 compared with the
cells of FS2190 could be explained. The coulombic
efficiencies of the coin cells with FS2226 are 2-7 %
higher than the efficiencies for same layer numbers of
FS2190 coin cells. So the SEI in FS2226 cells is only

formed in lower extent than for the FS2190 cells. An

insufficient SEI could not prevent the further decomposition of the electrolyte and loss of lithium during
the charging steps. Consequently the discharging capacities will decrease more rapidly than at a cell with
intact SEI.
4.4. Comparison of different cell designs and separators
Comparing the different cell designs, the Swagelok
cells with FS2190 separator show the best results,
whereas the coin cells with FS2226 have the worst
results. The worse performance of the coin cells with
FS2226 could be explained with the SEI formation.
The Swagelok cells with FS2190 have significant
better discharge capacities over almost every layer
thickness and C-rate in comparison to the FS2190
coin cells. Ideally Swagelok and coin cells should
show same performance, because Swagelok cells
should have the same arrangement of cell components
as the coin cells. The difference is that Swagelok
cells have an additional reference electrode, which
should support the research of the cells, but must not
intervene in the cell electrochemistry. It could be possible, that the cell material in coin cells have less
scope to react on mechanical stress than the material
in Swagelok cells. Also plausible, is that the contact
pressure in Swagelok cells are higher, thus the ion
conductivity through the cell is better, leading to
higher values.

Acknowledgement
The author thanks Ralf Wagner, Pia Janssen and Sebastian Brox from MEET- Mnster Electrochemical
Energy Technology for the support.

7. References
1

R. Wagner; N. Preschitschek; S. Passerini; J. Leker;

M. Winter, J Appl Electrochem 2013, 43, 481.
2
Kirchfer, M.; von Zamory, J.; Paillard, E.;
Passerini, S.; Int. J. Mol. Sci. 2014, 15(8), 1486814890
3
Suding, M.T.H.; HHU Dsseldorf, Faculty of Mathematics and Natural Science 2012
4
Huang, X.; J Solid State Electrochem. 2011, 15(4),
649-662
5
Huang, X.; Journal of Power Sources. 2014, 256, 96101
6
M. Winter, Zeitschrift fr Physikalische Chemie
2009, 223, 1395.

5. Outlook
Further investigations could be done to evaluate the
precise layer thickness of FS2190 which provides
optimal discharge capacities of the cell. A wetting
procedure to make use of Celgrad 2500 separators for
cell applications could be developed. It could be seen
that the real discharge capacity for some cells show
nearly the theoretical value, but the strong breakdown
at 1C is a great shortcoming. Further investigations
should be done to evaluate the reasons and to improve
cell performance at even higher C-rates.

6. Sources of error
The measurement error could be estimated with the
standard deviation. An incalculable source of error is
the cell production by different experimenters, which
could lead to less consistent results, even though the
same experimental procedures were used.

8. Supporting Information
Table 3: C-rate test program for measurements of charge and discharge capacities
Step
1

Command
Rest for 2 hours

Remarks

Discharge and charge at 0.1 C

Process was repeated 3 times

Voltage at the end of discharge: 0.01 V
Voltage at the end of charge: 1.5 V

Discharge and charge at 0.2 C

Process was repeated 3 times

Voltage at the end of discharge: 0.01
Voltage at the end of charge: 1.5

Discharge and charge at 0.5 C

Process was repeated 3 times

Voltage at the end of discharge: 0.01
Voltage at the end of charge: 1.5

Discharge and charge at 1 C

Process was repeated 3 times

Voltage at the end of discharge: 0.01
Voltage at the end of charge: 1.5

Discharge and charge at 2 C

Process was repeated 3 times

Voltage at the end of discharge: 0.01
Voltage at the end of charge: 1.5

Discharge and charge at 5 C

Process was repeated 3 times

Voltage at the end of discharge: 0.01
Voltage at the end of charge: 1.5

Discharge and charge at 10 C

Process was repeated 3 times

Voltage at the end of discharge: 0.01
Voltage at the end of charge: 1.5

Discharge and charge at 0,1 C

Process was repeated 3 times

Voltage at the end of discharge: 0.01
Voltage at the end of charge: 1.5