Advances in Chemical Engineering and Science, 2012, 2, 423-434
http://dx.doi.org/10.4236/aces.2012.24052 Published Online October 2012 (http://www.SciRP.org/journal/aces)
Two-Dimensional Lithium-Ion Battery Modeling with
Electrolyte and Cathode Extensions
Glyn F. Kennell, Richard W. Evitts
Department of Chemical and Biological Engineering, University of Saskatchewan, Saskatoon, Canada
Received July 3, 2012; revised August 6, 2012; accepted August 18, 2012
ABSTRACT
A two-dimensional model for transport and the coupled electric field is applied to simulate a charging lithium-ion cell
and investigate the effects of lithium concentration gradients within electrodes on cell performance. The lithium con-
centration gradients within electrodes are affected by the cell geometry. Two different geometries are investigated: ex-
tending the length of the electrolyte past the edges of the electrodes and extending the length of the cathode past the
edge of the anode. It is found that the electrolyte extension has little impact on the behavior of the electrodes, although
it does increase the effective conductivity of the electrolyte in the edge region. However, the extension of the cathode
past the edge of the anode, and the possibility for electrochemical reactions on the flooded electrode edges, are both
found to impact the concentration gradients of lithium in electrodes and the current distribution within the electrolyte
during charging. It is found that concentration gradients of lithium within electrodes may have stronger impacts on
electrolytic current distributions, depending on the level of completeness of cell charge. This is because very different
gradients of electric potential are expected from similar electrode gradients of lithium concentrations at different levels
of cell charge, especially for the LixC6 cathode investigated in this study. This leads to the prediction of significant elec-
tric potential gradients along the electrolyte length during early cell charging, and a reduced risk of lithium deposition
on the cathode edge during later cell charging, as seen experimentally by others.
Keywords: Lithium-Ion Cell; Mathematical Modeling; Cathode Extension; Electrolyte Extension; Current Distributions;
Electric and Concentration Fields
1. Introduction
Lithium-ion cells store and release energy via the emis-
sion, transport and insertion of lithium-ions from/into
electrode materials at different electrochemical potentials.
This difference in potential may be because the elec-
trodes are comprised of different materials, because of an
externally applied electric potential, and/or may also be
because of the stoichiometric coefficient of lithium al-
ready present in the electrodes. At the ends of electrodes
are edges. Lithium-ions may be produced and consumed
at these edge regions if they are in contact with the elec-
trolyte, such as when the electrolyte is flooded. At
flooded electrode edges the geometry of the edges may
cause multi-dimensional effects, such as concentration
gradients in the electrolyte and the electrodes, and also
electric potential gradients in the electrolyte. These ef-
fects are the focus of this paper.
It has previously been found that negative conse-
quences to cell performance may arise due to the con-
centration gradients associated with the flooded electrode
edges. These consequences include the increased risk of
lithium deposition at the cathode edge region. Therefore,
a cathode edge may be extended past the anode edge to
reduce the likelihood of lithium deposition at the cathode
edge region; however, this may result in other problems.
Some of these were experimentally observed by Scott et
al. [1,2], and include a relatively large electric potential
drop along the length of the electrolyte, parallel to the
electrodes, and associated with the extended cathode
edge.
West et al. [3] developed a one-dimensional model
accounting for the transport in the electrolyte and elec-
trode phases of a cell with porous insertion electrodes
and a liquid electrolyte. It was demonstrated how elec-
trolyte depletion was the principal factor limiting the
discharge capacity of the system. Doyle et al. [4] pre-
sented a model for a lithium-ion cell that was imple-
mented by considering one-dimensional transport for a
galvanostatic current. It was found that the decreased
lithium concentration in the composite cathode illustrated
the need for higher lithium concentrations. This model
was expanded by Fuller et al. [5] who considered a po-
rous insertion anode instead of a lithium foil anode.
C
opyright © 2012 SciRes. ACES

G. F. KENNELL, R. W. EVITTS
424
Transport was considered one-dimensionally. Arora et al.
[6] used the model of Fuller et al. [5] for one-dimen-
sional lithium-ion battery predictions. They concluded
that lithium deposition may occur in cells with lower
excess negative electrode capacity.
Tang et al. [7] presented a two-dimensional model for
the investigation of lithium deposition. They utilized a
COMSOL Multiphysics™ model (based on dilute solu-
tion theory) to explain why extending the cathode edge
may decrease the tendency for lithium deposition during
cell charging. Some of the assumptions made by Tang et
al. were: constant and uniform electrolyte concentrations
and conductivity, uniform anode concentration with re-
spect to position, linearized Tafel kinetics, solid film
electrodes, and electrolyte electroneutrality. Tang et al.
showed that for their model a cathodic extension of 0.4
mm is sufficient to prevent the onset of lithium deposi-
tion. Eberman et al. [8] used a two-dimensional model
based on concentrated solution theory to model the ef-
fects of a cathode under-lap (the opposite to a cathode
extension). Eberman et al. used this model to conduct a
sensitivity analysis of various factors on the risk for lith-
ium deposition. They found the three most significant
factors affecting the risk of deposition to be: the open-
circuit potential, the size of the underlap, and the charge
rate. Kennell and Evitts [9] presented a two-dimensional
model for the concentrations, current distributions, and
electric field as a function of time, in a lithium-ion cell.
They demonstrated that it is possible to predict not only
the lithium deposition at the cathode edge at later charg-
ing times, but also the large electric gradients that were
experimentally observed by Scott et al. [1,2] along the
electrolyte during early charging. It can be noted that all
of the models described above incorporate simplified
insertion kinetics when compared with models that focus
on the insertion and transport of lithium inside electrode
particles [10]. The research presented in this current pa-
per uses the model of Kennell and Evitts [9] to continue
the study of a lithium-ion cell and numerically predict the
effects associated with equal and extended electrodes/
electrolyte.
2. Theory and Model Implementation
This paper is focused on results from a numerical model
implemented using C++. Lithium-ion cells were modeled
using two governing equations where fluid bulk velocity
has been neglected [9]:

i
ii i
CzuF C
t


2
ii i
DC S (1)
2
ii iii
ii
FF
zD CzS

  

Equation (1) may be used to describe the transport of
species due to diffusion and electro migration. Equation
(1) also contains a term for the source or sink of species
due to reactions. Because Equation (1) was developed
using the Control Volume technique, for use with an
up-winding scheme, it omits one term that would be pre-
sent in an equation developed for use with alternative
methods:


2
zuF C
ii i
. This application of Equation
(1), using the Control Volume technique, also ensures the
conservation of charge and mass due to transport. Equa-
tion (2) describes the Laplacian of potential due to diffu-
sion potential, spatially separated anodic and cathodic
reactions, and charge density. When this equation is ap-
plied over a time interval an assumption that the electric
field will promote electroneutrality is incorporated that
ensures a system of i + 1 equations are available for solv-
ing for i species concentrations and the electric field.
This system of equations is advantageous when com-
pared to equation sets containing Poisson’s equation,


ii
i
ii
i
F
zC
t





(2)
F
zC   

, because Equation (2) is not nu-
merically stiff. The validity and development of these
equations were demonstrated by Kennell who applied
these equations to several different electrochemical sys-
tems [11]. These equations were solved using C++ and
numerical techniques described elsewhere [11]. The ef-
fective diffusion coefficient was calculated [9]:
1.5
0
DD


22
iii
i
(3)
Values for diffusion coefficients and other model pa-
rameters are presented in Table 1.
Conductivity was assumed non-uniform with time and
position inside the electrolyte:
F
zuC  

(4)
For the calculation of electro mobilities the Nernst-
Einstein equation was used:
i
DRTu
i

(5)
Predictions were conducted for a lithium-ion cell that
consisted of an electrolyte sandwiched between two solid
electrodes. Figure 1 shows the cell geometry and the
aspects that were modified in the simulations; the exten-
sion of the cathode edge beyond the anode edge, and the
electrolyte edge beyond the cathode edge, were varied.
These edge extensions were conducted in the x-dimen-
sion, or along the length of the cell. The numerical do-
main was split into three parts: the cathode, the anode,
and the electrolyte. These three domains were solved
concurrently, where charged species were assumed to
exist in the electrolyte, but not in the solid electrodes.
Thus Equation (1) is written for the interior of an elec-
trode as:

G. F. KENNELL, R. W. EVITTS
ACES
425
Table 1. Cell parameters.
Electrode parameters LixC6 LixCoO2
Lithium insertion rate constant, k, m2.5·mol–0.5·s–1 4.9 × 10–11 [6] 2.8 × 10–10 [12]
Initial stoichiometric coefficient 0.01 [7] 1 [7]
Maximum concentration, Ct, mol·m–3 30,540 [7] 56,250 [7]
Diffusion coefficients, D, m2·s–1 5.5 × 10–14 [12] 1.0 × 10–11 [12]
Transfer coefficients, αa, αc 0.5 [7] 0.5 [7]
Electrolyte parameters
Volume fraction, ε 0.55 [13]
LiPF6 initial concentration, mol·m–3 1200 [7]
Li+ diffusion coefficient in liquid phase, D0, m2·s–1 8.39 × 10–11 [14]
Figure 1. Cell configuration (not to scale). The x-dimension
corresponds to the length of the cell and the y-dimension
corresponds to the cell height.
2
ii
CDCS
t
 

 
i
i
(6)
The electric potential field inside each electrode was
assumed to be uniform and equal to an applied value,
a or c. Conductivity was also assumed constant
inside each electrode. The boundary condition for all
boundaries of each of the three numerical domains for
Equation (1) or (6) was:
0
i
C
x

 (7)
and the boundary condition used for Equation (2) at all
electrolyte boundaries was:
0
x
 
 (8)
During charge and discharge, charge and mass were
calculated to move between the three numerical domains
via the source terms in Equations (1), (2) and (6). The
following development will describe how the source
terms were calculated as being dependent on electro-
chemical reactions, how charge and mass were conserved,
and how the overall cell potential was determined.
The rates of electrochemical reactions were assumed
to follow Tafel kinetics for the anode and cathode:

ae a
U

 


,
,exp ak aoa
F
ii RT

 (9)

,
,exp ck coc cec
F
ii U
RT


  


(10)
where e

represents the electric potential of the elec-
trolyte adjacent to the electrode. e was calculated
from Equation (2), and was not assumed uniform with
time and position. a
U and c U represent the equilib-
rium potential of the anode and cathode respectively, as a
function of lithium stoichiometric coefficient. The equi-
librium potential in the anode was calculated by [12]:

2
,,
34
,
511
4
,,
3.85521.2473 111.1521
42.8185167.711 1
42.508 16.13210exp7.657
ss
a
ta ta
ss
ta t
ss
ta ta
CC
UCC
CC
CC
CC
CC


 
 
 

 
 
 

 
 

 

 


  

 
 
 


 


 
 

 

5




(11)
The equilibrium potential in the cathode was calcu-
lated by [6]:
0.5
11.5
0.7222 0.138680.028952
0.0171890.0019144
0.28082exp150.06
0.79844exp0.446490.92
ss
c
tt
ss
tt
s
t
s
t
CC
UCC
CC
CC
C
C
C
C

 
 
 
 
 

 
 





  

 






 









 
ac
a
otss
iFkCCC C

(12)
The exchange current density from Equations (9) and
(10) was calculated by [7]:

(13) 
Figure 2 shows the equilibrium potentials of the elec-

G. F. KENNELL, R. W. EVITTS
426
trodes described by Equations (11) and (12) for an anode
fabricated from LiyCoO2 and for a cathode from LixC6.
The solid lines in Figure 2 represent the portion of the
equilibrium potentials that correspond to the stoichio-
metric coefficient that is likely to exist in a lithium-ion
cell. In other words, the solid part of the line for LiyCoO2
corresponds to the stoichiometric coefficient of between
0.99 and 0.58 (in the anode) and the solid part of the line
for LixC6 corresponds to the stoichiometric coefficient of
between 0.01 and 1 (in the cathode).
The electrochemical reactions were treated as source
terms. If it is assumed that at the surface of the electrode
the currents described by Equations (9) and (10) are per-
pendicular to the electrode surface, the current vector
describing the electrochemical reaction rate, i, is pro-
duced. This current vector may be converted into a
source term for use with Equations (1), (2), and (6):
i
i
i
SzF


set electrode
isIl
c 
cella c
 
(14)
Therefore, the species produced by electrochemical
reactions were introduced into the numerical procedure
via the source terms in Equations (1), (2) and (6). The
conservation of charge and mass across numerical boun-
daries was guaranteed by ensuring the sum of each elec-
trochemical reaction along the length of the electroactive
surface was equal to a prescribed current:
electrode
d
Li
 (15)
Equation (15) was satisfied by modifying the rates of
electrochemical reactions by varying the applied electric
potentials, a and . Then, the overall cell potential
was calculated by:
V (16)
3. Results and Discussion
Lithium-ion cells depend upon the transport of Li+. It
may be transported through an electrolyte because of
different potentials of electrodes. Different potentials
may be caused by different equilibrium potentials and
different potentials applied to each electrode. Equilib-
rium potentials depend on electrode materials and on the
stoichiometric coefficient of inserted lithium. The two
electrode materials investigated in this paper are LiyCoO2
and LixC6. The equilibrium potentials corresponding to
different stoichiometric coefficients in each of these ma-
terials are presented in Figure 2. Simulations presented
in this paper investigate both the electric potentials estab-
lished between the anode and cathode of a lithium-ion
cell (along the y-dimension), and also the electric poten-
tials that may exist parallel to the electrodes along the
cell length (x-dimension). The cell geometries investi-
gated in the simulations presented in this paper are
shown in Figure 1. The electrodes are considered to be
solid film electrodes and the separator consists of an
electrolyte with the following composition: 1.2 M LiPF6
in a 1:2 v/v mixture of ethylene carbonate and dimethyl
carbonate (EC:2DMC). Further details are given in Fig-
ure 1, including the locations of electrode edges and in-
terior surfaces.
3.1. Electrolyte Extension
In this section predictions for lithium-ion cells with a
flooded electrolyte extended past the edges of the elec-
trodes are presented. In this section, the anode and cath-
ode edges are located at the same cell length (x-dimen-
sion). This geometry exposes the edges of both the anode
and cathode to the electrolyte. Section 3.1.1 explores
simulations where no electrochemical reactions occur on
these electrode edges and Section 3.1.2 explores the case
where electrochemical reactions occur on the electrode
edges in contact with the flooded electrolyte.
3.1.1. Equal Length Electrodes without Edge Reactions
Figure 3 shows the predicted electric field for the case
where the anode and cathode were of equal length (x-
dimension) and the electrolyte length was extended past
the edges of the electrodes by 25 µm and the cell under-
went 4.37 Am−2 charging for 60 seconds. The length of
the electrodes incorporated into the simulation was 70
µm. It was found that this length encompassed com-
pletely the multi-dimensional edge effects, for this case.
It was assumed that only the interior surfaces of the elec-
trodes were electro active. In other words, the edges of
the electrodes did not emit or insert any lithium-ions.
This means that the increased surface area due to elec-
trode edges did not have an impact on the overall rate of
electrochemical reactions around the edge. Therefore, in
Figure 2. Equilibrium potential of electrodes as a function
of stoichiometric coefficient, x or y. Solid lines indicate
stoichiometric coefficient ranges assumed in this paper.

G. F. KENNELL, R. W. EVITTS 427
Figure 3. Predicted electric potential field for the case of
equal length flooded electrodes without edge reactions after
60 seconds as (A) a surface plot and (B) a contour plot with
cell geometry overlay. The height of the cathode was 8 µm.
Simulated electrode length to bulk conditions was 70 µm.
this case, the main edge effect was to increase the effec-
tive conductivity of the electrolyte towards the edge,
caused by the electrolyte length (x-dimension) extension.
The effect of this electrolyte extension on the predicted
electric current distribution is shown in Figure 4. This
figure demonstrates how the electric current tended to
move in the y-dimension across the electrolyte directly
from the anode to the cathode in the bulk interior of the
cell (at larger distances from the edge); however, in the
electrolyte nearer the edges of the electrodes it is shown
that the electric current did not take the shortest path
from the anode to cathode. Instead, the current tended to
spread out along the cell length (x-dimension) into the
extended electrolyte region that would otherwise have
contained no current density. In other words, the ex-
tended electrolyte region had the effect of increasing the
effective conductivity of this area. This increase in effec-
tive conductivity and reduced current densities near the
edge region lead to a lower electric potential gradient, as
seen in Figure 3.
The rates of electrochemical reactions (Equations (9)
and (10)) occurring on the electrode interior surfaces
were dependent upon the electric potential adjacent to the
electrode, . As described above, Figure 3 shows how
the electric potential adjacent to the electrodes was re-
duced in the proximity of the edge region, including on
the interior electrode surfaces, due to the increased effec-
tive conductivity of the extended electrolyte region. In
other words, an increased effective conductivity due to
an extended electrolyte may decrease the potential gra-
dient across the electrolyte (in the y-dimension) and this
may cause an increase in electrochemical reaction rates
on the interior surface near the electrode edges. This ef-
fect was seen in the simulations. However, this effect
was extremely small, as can be seen in Figure 5, which
shows the concentration of lithium inserted into the
cathode after a full hour of 1 C charging; the increase in
lithium concentration near the cathode edge/tip was so
small that it is unobservable in this figure. Hence, it can
be concluded that the electrolyte extension does not have
a perceivable effect on electrode concentrations at the
conditions examined, but it does increase the effective
conductivity near the edge region.
Figure 4. Predicted electric current distribution near elec-
trode edges for the case of equal length flooded electrodes
without edge reactions after 60 seconds.
Figure 5. Predicted cathode concentration for the case of
cathode height of 8 µm and equal length flooded electrodes
after 1 hour of charging and no edge reactions.
e


G. F. KENNELL, R. W. EVITTS
428
3.1.2. Equal Length Electrodes with Edge Reactions
Figure 6 shows the predicted electric field for the case of
two equal height (5 µm in y-dimension) and equal length
electrodes with an electrolyte length extension of 25 µm
after 60 seconds of 4.37 Am–2 charging. For this simula-
tion, it was assumed that the electrochemical reactions
occurred along the complete surfaces of the electrodes in
contact with the electrolyte, including the electrode inte-
rior surfaces and electrode edges. It is important to note
that the scale changes along the abscissa in Figure 6.
From this figure it may be seen that the effect of the edge
reactions (when the height of both electrodes was 5 µm)
was to raise the electric potential near the edges above
that in the bulk region. The reason for this increase was
associated with gradients of lithium stoichiometric coef-
ficient in the electrodes towards the edges. This phe-
nomenon is explained in the next paragraph.
The electric potential was elevated along the electro-
lyte length towards the electrode edges because the over-
all rate of anodic half reactions in this region was pre-
dicted to be greater than the overall rate of cathodic reac-
tions. In other words, in this edge region, the anode was
producing more electric current than the cathode was
Figure 6. Predicted electric potential field for the case of
equal length flooded electrodes with edge reactions after 60
seconds of charging as (A) a surface plot and (B) a contour
plot with cell geometry overlay.
consuming. This excess electric current then had to mi-
grate along the length of the electrolyte, parallel to the
electrodes, towards the bulk cell. This predicted electric
current distribution is shown in Figure 7. This figure
shows electric current emanating from the anode interior
surface and the anode edge. A significant portion of this
electric current flowed into the extended electrolyte re-
gion, taking advantage of the increased effective conduc-
tivity in this area. This electric current flowed towards
the cathode interior surface and edge, where lithium was
inserted into the cathode. The relative magnitudes of the
currents associated with the tips of the anode and cathode
edges can also be observed in Figure 7.
In Figure 7, the arrow depicting the current flow-
ing/inserting into the cathode tip is approximately half of
the size of the arrow depicting the current emanating
from the anode. The difference in magnitude of these two
currents is due to the dependence of the rates of electro-
chemical reactions on the equilibrium potentials, a U
and c. When the charging of the cell is started, the
stoichiometric coefficient of lithium in the anode is very
close to unity and is represented by the right-hand por-
tion of the solid line in Figure 2. This represents only a
slight gradient of potential for a gradient in stoichiomet-
ric coefficient. In other words, this equilibrium potential
does not change significantly for a concentration gradient
inside the electrode, and electrochemical reaction rates
will not change drastically for a gradient in potential.
Therefore, for this case a higher current will emanate
from the anode tip that is more dependent on the tip sur-
face area than it is by the concentration gradient within
the electrode. The opposite is true of the cathode tip. At
the start of cell charging, the stoichiometric coefficient
U
Figure 7. Predicted electric current distribution near elec-
trode edges for the case of equal length flooded electrodes
with edge reactions after 60 seconds. The height of both
electrodes is 5 µm.

G. F. KENNELL, R. W. EVITTS 429
inside the cathode is small and is shown by the left side
of the LixC6 solid line in Figure 2. As is evident, there is
a very large gradient of potential associated with a small
change in stoichiometric coefficient in this region.
Therefore, the rates of electrochemical reactions would
be greatly affected by a gradient of lithium concentration
in the electrode. Although the larger surface area at the
cathode tip may promote an increased rate of lithium
insertion into the tip area, the large change in equilibrium
potential that this would cause prevents this increase.
Instead, the extra current produced at the anode tip and
edge must flow along a different path, along the electro-
lyte length (in the x-dimension) adjacent to the cathode,
in a manner that balances transport and concentration
gradients in the electrolyte with concentration and equi-
librium potentials in the electrodes. The resulting con-
centration gradients within the electrodes are discussed
below.
Figure 8 shows the concentration gradients in the re-
gion of the cathode edge after 60 seconds of cell charg-
ing. The lengthwise gradients were restricted to within 5
µm of the edge. This is a smaller region than for the gra-
dients in the anode, shown in Figure 9, where the
lengthwise gradients extended 25 µm from the anode
edge. This reinforces the concept that stoichiometric gra-
dients occur in the anode during early cell charging, but
not in the cathode because of the large equilibrium po-
tential gradients that this would cause. In other words,
the electrochemical reaction rate of lithium production at
the anode is not greatly affected by the anodic lithium
stoichiometric coefficient; the opposite is true of the
cathode (during early charging). Because the numerical
model used a non-uniform mesh, the length of the bulk
cell that was modeled was long enough so that the mi-
croscopic phenomena occurring at the electrode edges
would not have a significant effect on the macroscopic
bulk cell and overall cell potential. Because the model
balances the two-dimensional potential field at the edges
with the bulk cell potential, the model predicts where the
excess current from the anode edge is inserted into the
cathode. Figure 10 shows the current density that oc-
curred on the surfaces of both the anode and cathode as a
Figure 8. Predicted cathode concentration for the case of
equal length flooded electrodes with edge reactions after 60
seconds of 4.37 Am−2 charging.
Figure 9. Predicted anode concentration for the case of
equal length flooded electrodes with edge reactions after 60
seconds of 4.37 Am−2 charging.
Figure 10. Predicted electrochemical reaction rate of lith-
ium dissolution or insertion along surface for the case of
equal length flooded electrodes with edge reactions after 60
seconds of 4.37 Am−2 charging.
function of distance from the electrode edge (positive
distances correspond to locations on the electrode interior
surface, away from the edge, and negative distances cor-
respond to distances away from the electrode tip, along
the electrode edge itself). Figure 10 shows a reduction in
the cathodic current density at the electrode tip, of ap-
proximately 65% of the bulk value, and this is necessary
in order to avoid large stoichiometric lithium gradients in
the cathode (as described above). The anodic reaction
rate suffers only a small reduction at the tip, as shown in
Figure 10. However, since the overall production and
consumption of Li+ must be equal for the entire cell, this
excess Li+ produced at the electrode edges must be con-
sumed elsewhere. Figure 10 shows at distances of ap-
proximately 0.01 mm to 0.02 mm from the edge the an-
odic reaction rate was predicted to be less than the ca-
thodic reaction rate. Therefore, it is in this region that the
excess current from the edge region was inserted into the
cathode. This avoided steep lithium concentration gradi-
ents towards the cathode edge, and instead balanced
these gradients with iR drops and concentration effects in
the electrolyte.

G. F. KENNELL, R. W. EVITTS
430
Figure 11 shows the predictions for a cell with similar
geometry to the one used in the previous predictions, but
with a cathode height of 8 µm. This increase in cathode
height decreases the concentration gradients within the
cathode caused by lithium diffusion from the surfaces.
The prediction showed that an increase in cathode height
decreases the values of electric potential towards the
electrode edges. Figure 12 shows the electric current
distribution for this case. Figure 12 displays the current
that was drawn parallel to the electrodes (along its length)
into the edge area to compensate for the additional lith-
ium inserted into the cathode in this area. Hence, during
early cell charging, the rate of lithium insertion into the
cathode is primarily determined by the concentration
gradients within the cathode, and the rate of lithium
emission from the anode is determined by anode surface
area and electrolyte potential gradients.
3.2. Extended Cathodes
Large potential drops due to concentration gradients
within the cathode have been seen experimentally when
Figure 11. Predicted electric potential field for the case of
equal length flooded electrodes with edge reactions after 60
seconds of charging as (A) a surface plot and (B) a contour
plot with cell geometry overlay. The height of the cathode
was 8 µm. The electrolyte length was extended past the edge
of the electrodes by 25 µm.
Figure 12. Predicted electric current distribution near elec-
trode edges for the case of equal length flooded electrodes
and 8 µm height cathode after 60 seconds of charging.
the cathode edge is extended (in the x-dimension) sig-
nificantly past the anode edge (Scott et al. [1,2]). The
cathode edge may be extended in order to prevent higher
levels of lithium concentration at the tip/edge that may be
detrimental to the cell. The model presented in this paper
does predict these damaging levels of lithium concentra-
tion, and the resulting lithium deposition, at the cathode
tip/edge; however, these high concentrations of lithium
are more likely to occur in the cathode towards the end of
cell charging when the stoichiometric coefficient of lith-
ium in LixC6 is almost unity, rather than at the beginning
of cell charging. The relationship between the equilib-
rium potential and stoichiometric coefficient of lithium in
LixC6 of an almost completely charged cell is shown in
Figure 2 (for stoichiometric coefficients approaching
unity). The potential gradient caused by the LixC6
stoichiometric coefficient gradient is much less for coef-
ficients approaching unity (a fully charged cell) than for
coefficients approaching zero (an uncharged cell). These
different equilibrium potential gradients for an uncharged
and charged cathode result in the possibility for larger
lithium concentration gradients in a cathode approaching
a full charge. In other words, if the difference in
stoichiometric coefficient of lithium in an electrode
causes a large equilibrium potential gradient, a large
electric potential gradient may be apparent in the elec-
trolyte, as seen by Scott et al. [1,2]. If the cell is in a state
of charge whereby a large difference in stoichiometric
coefficient (with length) does not cause a large equilib-
rium potential gradient, then large electric potential gra-
dients may not be seen in the electrolyte; however, large
concentration gradients in the electrode may then be pos-
sible, along with electrode over-saturation and lithium
deposition at regions of high surface area, as seen in the
numerical simulations of Tang et al. [7]. The model pre-
sented in this paper predicts both such phenomena. For

G. F. KENNELL, R. W. EVITTS 431
example, Figure 13 shows the concentration profile for
the cathode of 6 µm height and equal electrode length
cell having undergone 4.37 Am−2 charging for one hour.
Figure 13 shows that although much of the cathode con-
centration was significantly below the maximum concen-
tration of 30.5 M, the tip region was above this concen-
tration and lithium deposition here is likely. Figure 14
shows the predicted electric field for the case where the
length of the cathode was extended past the edge of the
anode by 1.75 cm, after 100 seconds of charging at 2
Am-2, corresponding to the current density utilized by
Scott et al. [1]. The height of the cathode was 8 µm and
the initial stoichiometric coefficient in the cathode was
0.0025. Figure 14 shows a predicted electric field that
has a minimum of approximately –0.4 V with respect to
the bulk values. This is about half of the maximum ex-
perimental value seen by Scott et al. [1] after 90 seconds
for a lithium ion cell with a cathode extension undergo-
ing charging of a similar current density. However, many
cell parameters were unpublished by Scott et al. and the
predictions given here are for solid electrodes, not porous
ones. Of interest in this case is not a direct comparison
with experimental data, but instead, the trends caused by
lithium gradients in the electrodes and resulting effects
are examined below. For the simulation presented in Fig-
ure 14 the concentration of lithium in both electrodes
changed with position and time. Because the concentra-
tion of lithium is non-uniform with position, potential
gradients were evident in the electrolyte. The effect of
the lithium electrode gradients on the cell are investi-
gated through an examination of the rates of electro-
chemical reactions predicted to occur on electrode sur-
faces. Figure 15 shows the rate of anodic currents that
emanated from the anode from the simulation presented
in Figure 14 as a function of position at different times.
Positive distance values represent the interior surface of
the anode and negative values the small edge region.
Because at larger distances along the cell length, away
from the anode edge, the current was predicted to remain
constant, this data was not presented as part of this figure.
Figure 15 shows that at early charging times (100 s and
500 s) the highest rates of current were drawn from re-
gions close to the anode edge. The anode emanated cur-
Figure 13. Predicted cathode concentration for the case of
equal length flooded electrodes with edge reactions after 1
hour of 4.37 Am−2 charging. The cathode height was 6 µm.
Figure 14. Predicted electric potential field (V) in the elec-
trolyte for the case of a 1.75 cm cathode extension after 100
seconds of charging as (A) a surface plot and (B) a contour
plot with cell geometry overlay. The cathode height was 8
µm and the cathode length extension was 1.75 cm.
Figure 15. Predicted electric current emanating from anode
surface at different times for the case of a 1.75 cm cathode
extension. Positive distance values represent the interior
anode surface and negative distance values represent the
distance along the anode edge itself, away from the anode
tip.
rent from this region for two reasons: this region was the
closest to the extended cathode and this region had a
greater surface area due to the anode edge. Also, from
Figure 2 it is evident that significant concentration gra-
dients were possible in the initially charging anode that

G. F. KENNELL, R. W. EVITTS
432
did not cause significant gradients of equilibrium poten-
tial. Figure 16 shows a figure similar to Figure 15, but
describing the current inserted into the cathode. At the
early charging times of 100 s and 500 s, Figure 16 shows
that current densities of approximately 0.4 and 0.2 A/m2
were inserted along the extended region of the cathode,
respectively. At early charging times the gradient of lith-
ium in the cathode caused significant gradients of equi-
librium potential which provided the driving force behind
the large potential drop in the electrolyte (shown in Fig-
ure 14), and caused these significant cathodic reaction
rates towards the edge (shown in Figure 16). Over an
initial period of time, small amounts of lithium were in-
serted into the extended cathode region and the concen-
tration of lithium in the extended cathode region in-
creased such that a large difference in equilibrium poten-
tial no longer existed between the edge region and the
bulk cell region (see Figure 2). This decrease in the dif-
ference in equilibrium potential decreased the available
driving force for the migration and insertion of lithium
along the length of the extended cathode and is visible in
Figure 16 that shows how the current drawn by the ex-
tended cathode decreased for times of 1000 s and later.
This decrease in current drawn by the extended cathode
also affected the current emanating from the anode. Fig-
ure 15 shows that after 1000 s the excess current drawn
from the region towards the edge of the anode had de-
creased and some other effects were visible as the “hook”
shape in the reaction rates towards the edge region. This
hook shape was likely because of the decreased current
being drawn towards the extended cathode and the de-
creased Ohmic drop in the electrolyte making it more
possible for current to be drawn from the anode surface
further into the cell. This current drawn from the anode
further into the cell took advantage of the fact that the
anode edge became more depleted of lithium during the
early charging when the extended cathode was drawing
significant quantities of current. Figure 15 shows that as
time progressed further, less and less excess current was
produced towards the anode edge region, and instead,
because the extended cathode was no longer drawing
significant current, the concentration gradients previ-
ously established in the anode became the dominant
phenomenon impacting the rates of anodic reactions.
This was because, as the cell became more charged and
the stoichiometric coefficient in the anode decreased, a
slight electrochemical potential gradient is evident to-
wards the left of the corresponding solid line in Figure 2.
This significant lithium concentration gradient in the
anode after 1 hour is shown in Figure 17. Figure 18
shows the lithium concentration gradient in the cathode
after 1 hour. It can be seen that the concentration of lith-
ium in the extended region was approximately one tenth
of the maximum value seen in the bulk cell.
Figure 16. Predicted electric current inserted into cathode
surface at different times for the case of a 1.75 cm cathode
extension. Positive distance values represent the interior
cathode surface and negative distance values represent the
distance along the cathode edge itself, away from the cath-
ode tip.
Figure 17. Predicted anode concentration for the case of a
1.75 cm cathode extension after 1 hour of charging 2 A/m2.
Figure 18. Predicted cathode concentration for the case of a
1.75 cm cathode extension after 1 hour of charging at 2
A/m2.
4. Conclusions
This paper explores the edge effects of electrodes in lith-
ium-ion cells undergoing charging, and the effects of
stoichiometric coefficient gradients within electrodes. It
was predicted, for the cases examined, that the increase
in effective conductivity associated with a flooded elec-
trolyte that is extended past the electrode edges does not
have an appreciable effect on the rates of anodic or ca-
thodic reactions near the edge regions. However, it was

G. F. KENNELL, R. W. EVITTS
433
predicted that lithium concentration gradients inside the
cathode impact the rate of cathodic reactions signifi-
cantly and concentration gradients inside the anode do
not significantly impact the rate of anodic reactions, both
during early cell charging. Instead, the rates of anodic
reactions are significantly affected by the surface area of
the anode contacting the electrolyte, and not the concen-
tration gradient of lithium in the anode. It was also pre-
dicted that during later stages of cell charging, when the
gradient of equilibrium potential due to a gradient in ca-
thodic stoichiometric coefficient is less steep, concentra-
tion gradients within the cathode (for equal electrode
lengths) are more likely and might lead to a possibility
for lithium deposition at the cathode edge region.
Simulations were conducted for the case where the
cathode edge was extended past the anode edge to reduce
the possibility for lithium deposition at the cathode edge
region. The simulations indicate that the stoichiometric
coefficient of lithium in an extended cathode edge would
be reduced in value; however, this extension may cause a
large electric potential drop along the electrolyte length
(during early cell charging) that corresponds to the lith-
ium stoichiometric coefficient gradient in the extended
cathode and also with the Ohmic losses and concentra-
tion gradients within the electrolyte itself. It was ob-
served that this equilibrium potential gradient would de-
crease as charging of the cell proceeded, causing a reduc-
tion in the rate of cathodic reactions occurring along the
extended cathode region. This reduction in the rate of
cathodic reactions along the extended cathode region
reduces the risk for lithium deposition at the cathode
edge region, as desired by many cell manufacturers.
5. Acknowledgements
The authors thank the University of Saskatchewan for
computing facilities and the National Science and Engi-
neering Research Council for a Canada Doctoral Schol-
arship.
REFERENCES
[1] E. Scott, G. Tam, B. Anderson and C. Schmidt, “Anoma-
lous Potentials in Lithium Ion Cells: Making the Case for
3-D Modeling of 3-D Systems,” The Electrochemical So-
ciety Meeting, Orlando, 13 October 2003.
[2] E. Scott, G. Tam, B. Anderson and C. Schmidt, “Obser-
vation and Mechanism of Anomalous Local Potentials
during Charging of Lithium Ion Cells,” The Electro-
chemical Society Meeting, Paris, 29 April 2003.
[3] K. West, T. Jacobsen and S. Atlung, “Modeling of Porous
Insertion Electrodes with Liquid Electrolyte,” Journal of
the Electrochemical Society, Vol. 129, No. 7, 1982, pp.
1480-1485. doi:10.1149/1.2124188
[4] M. Doyle, T. F. Fuller and J. Newman, “Modeling of
Galvanostatic Charge and Discharge of the Lithium/Poly-
mer/Insertion Cell,” Journal of the Electrochemical Soci-
ety, Vol. 140, No. 6, 1993, pp. 1526-1533.
doi:10.1149/1.2221597
[5] T. F. Fuller, M. Doyle and J. Newman, “Simulation and
Optimization of the Dual Lithium Ion Insertion Cell,”
Journal of the Electrochemical Society, Vol. 141, No. 1,
1994, pp. 1-10. doi:10.1149/1.2054684
[6] P. Arora, M. Doyle and R. E. White, “Mathematical Mod-
eling of the Lithium Deposition Overcharge Reaction in
Lithium-Ion Batteries Using Carbon-Based Negative Elec-
trodes,” Journal of the Electrochemical Society, Vol. 146,
No. 10, 1999, pp. 3543-3553. doi:10.1149/1.1392512
[7] M. Tang, P. Albertus and J. Newman, “Two-Dimensional
Modelling of Lithium Deposition during Cell Charging,”
Journal of the Electrochemical Society, Vol. 156, No. 5,
2009, pp. A390-A399. doi:10.1149/1.3095513
[8] K. Eberman, P. M. Gomadam, G. Jain and E. Scott, “Ma-
terial and Design Options for Avoiding Lithium-Plating
during Charging,” ECS Transactions, Vol. 25, No. 35,
2010, pp. 47-58. doi:10.1149/1.3414003
[9] G. F. Kennell and R. W. Evitts, “Charge Density in Non-
Isotropic Electrolytes Conducting Current,” The Canadian
Journal of Chemical Engineering, Vol. 90, No. 2, 2012,
pp. 377-384.
[10] W. Dreyer, M. Gaberscek, C. Guhlke, R. Huth and J. Jam-
nik, “Phase Transition in a Rechargeable Lithium Bat-
tery,” European Journal of Applied Mathematics, Vol. 22,
No. 3, 2011, pp. 267-290.
doi:10.1017/S0956792511000052
[11] G. F. Kennell, “Electrolytic Transport, Electric Fields,
and the Propensity for Charge Density in Electrolytes,”
Ph.D. Dissertation, University of Saskatchewan, Saska-
toon, 2011.
[12] M. Doyle and Y. Fuentes, “Computer Simulations of a
Lithium-Ion Polymer Battery and Implications for Higher
Capacity Next-Generation Battery Designs,” Journal of
the Electrochemical Society, Vol. 150, No. 6, 2003, pp.
A706-A713. doi:10.1149/1.1569478
[13] J. Christensen, V. Srinivasan and J. Newman, “Optimiza-
tion of Lithium-Titanate Electrodes for High-Power Cells,”
Journal of the Electrochemical Society, Vol. 153, No. 3,
2006, pp. A560-A565. doi:10.1149/1.2172535
[14] S. G. Stewart and J. Newman, “The Use of UV/Vis Ab-
sorption to Measure Diffusion Coefficients in LiPF6
Electrolytic Solutions,” Journal of the Electrochemical
Society, Vol. 155, No. 1, 2008, pp. F13-F16.
doi:10.1149/1.2801378

G. F. KENNELL, R. W. EVITTS
434
Nomenclature
C species concentration (mol/m3)
Cs solid phase lithium concentration (mol/m3)
Ct maximum lithium concentration (mol/m3)
D effective diffusion coefficient (m2/s)
D0 diffusion coefficient of liquid phase (m2/s)
F Faraday’s Constant (96,487 C/mol)
i current density (A/m2)
ect
l
current density vector (A/m2)
iLi current density of insertion/dissolution (A/m2)
k lithium insertion rate constant (m2.5mol −0.5s−1)
Iset applied current density (A/m2)
l length (m)
el rode
R gas constant (8.314 JK−1mol−1)
length of electrode (m)
S source term (mol/m3s)
s position (m)
t time (s)
T temperature (K)
u mobility (m2·mol/J·s)
U equilibrium potential (V)
x position along the x-dimension (m)
z charge number
Greek Letters

transfer coefficient

volume fraction
electric potential (V)

 conductivity (C/Vms)
Subscripts/Superscripts
a anode
c cathode
e electrolyte
i species