Journal of Power Sources 283 (2015) 132e150

Contents lists available at ScienceDirect

Journal of Power Sources

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

Non-isothermal electrochemical model for lithium-ion cells with

composite cathodes
Suman Basu a, Rajkumar S. Patil a, Sanoop Ramachandran a, Krishnan S. Hariharan a, *,
Subramanya Mayya Kolake a, Taewon Song b, Dukjin Oh b, Taejung Yeo b,
Seokgwang Doo b
a
Computational Simulations Group (SAIT-India), Samsung R&D Institute India-Bangalore, #2870 Phoenix Building, Bagmane Constellation Business Park,
Outer Ring Road, Doddanekundi Circle, Marathahalli Post, Bangalore 560 037, India
b
Energy Material Lab, SAIT, Samsung Electronics, Republic of Korea

h i g h l i g h t s

Non-isothermal electrochemical model developed for lithium composite cathode cell.

Experimentally obtained composite cathode property implemented in the model.

Model validated for a wide range of temperature and discharge rate.

Results analyzed to determine the heat generation pattern in the cell.

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

Article history: Transition metal oxide cathodes for Li-ion batteries offer high energy density and high voltage. Com-
Received 19 December 2014 posites of these materials have shown excellent life expectancy and improved thermal performance. In
Received in revised form the present work, a comprehensive non-isothermal electrochemical model for a Lithium ion cell with a
16 February 2015
composite cathode is developed. The present work builds on lithium concentration-dependent diffu-
Accepted 23 February 2015
sivity and thermal gradient of cathode potential, obtained from experiments. The model validation is
Available online 24 February 2015
performed for a wide range of temperature and discharge rates. Excellent agreement is found for high
and room temperature with moderate success at low temperatures, which can be attributed to the low
Keywords:
Lithium-ion battery
fidelity of material properties at low temperature. Although the cell operation is limited by electronic
Composite cathode conductivity of NCA at room temperature, at low temperatures a shift in controlling process is seen, and
Electrochemical operation is limited by electrolyte transport. At room temperature, the lithium transport in Cathode
Thermal appears to be the main source of heat generation with entropic heat as the primary contributor at low
Modeling discharge rates and ohmic heat at high discharge rates respectively. Improvement in electronic con-
Nickelecobaltealuminum ductivity of the cathode is expected to improve the performance of these composite cathodes and pave
way for its wider commercialization.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction applications, the push for the adaption of lithium-ion batteries as

energy source is gaining momentum. The ultimate aim is the large
Lithium-ion batteries have become ubiquitous in the field of scale electrification of automobiles. Significant challenges still exist
consumer electronics. These high energy density and light-weight to this end and form the focus of intense research. Lithium-ion
batteries have found application in laptops, mobile phones, cam- battery has to exhibit high energy density, high rate capability,
eras and other consumer products [1]. With the growing demand thermal stability and long life to meet the standard expected by the
for the use of greener alternatives in automotive and aerospace automotive industry [2,3].
Oxides of first row transition metals (Mn, Fe, Co, Ni) offer high
energy density due to high lithium intercalation site density. The
* Corresponding author. first commercial lithium-ion cell, introduced by Sony in 1991 was a
E-mail address: [email protected] (K.S. Hariharan).

http://dx.doi.org/10.1016/j.jpowsour.2015.02.127
0378-7753/© 2015 Elsevier B.V. All rights reserved.
 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150 133

Nomenclature ε porosity
k ionic conductivity (S m1)
A fitting coefficients for NCA OCP r density (kg m3)
a volumetric area (m2 m3) s electronic conductivity (S m1)
B fitting coefficients for NCA entropy generation f potential (V)
C specific heat/heat capacity (J kg1 K1)
c concentration of Li (mol m3) Subscripts
D diffusivity (m2 s1) a anode
E potential (V) c cathode
F Faraday constant (96485 Qmol1) e electrolyte phase
f± mean molar salt activity coefficient eq equilibrium
h heat transfer coefficient (W m2 K1) exp experimental
i current density (A m2) i particle interface/surface
k thermal conductivity (W m1 K1) irr irreversible
N number of data points mod model
R radius (m) n intercalation
Rg universal gas constant (8.3145 J mol1 K1) P constant pressure
r radial coordinate (m) p particle
Q heat source (W m3) rev reversible
T temperature (K) s solid (electrode) phase
t time (s) tot total
tþ0 transference number of the electrolyte ohm Ohmic heat generation
V cell voltage (V) 0 initial condition, exchange
x through plane coordinate (m)
y cathode state of charge Superscripts
eff effective
Greek letters max maximum
a transfer coefficient ref reference

LiCoO2/C cell [4]. Due to high cost of cobalt and low cycle-life both from lattice disorder and oxygen loss perspective [6]. High
however, LiCoO2 is not suitable for large scale applications like temperature instability of nickel-based cathodes has been investi-
electric vehicle. Nickel-based cathode has a similar structure with gated by various researchers [7e10]. The reason for thermal
lower cost and higher energy density as compared cobalt-based instability is attributed to oxidation of organic solvent by the oxy-
cathode [5]. On the other hand, LiNiO2 is more prone to lattice gen produced due to degradation of lithium nickelate [10]. Doping
disorder and hence, difficult to adapt for large scale production. The the lithium nickel cobalt oxide with aluminum or magnesium im-
addition of cobalt in the LiNiO2 structure stabilizes the compound, proves the thermal behavior and cyclic capacity [11]. The initial

Fig. 1. Schematic of LiNCA computational model.

134 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150

Fig. 2. a. Cell and electrode potentials with depth of discharge. b. Cathode (NCA) potentials at various temperatures.

capacity decreases for x > 0.05 in LiNi0.8Co0.2xAlx O2 [12] and it acts Lithium Nickel Cobalt Aluminum oxide (Li-NCA or NCA) cathode is
as a limit on the percentage of aluminum. In this manuscript, the studied.
electrochemical and thermal response of a lithium ion cell with

Table 2
Table 1 Coefficients of polynomial for NCA OCP gradient.
Coefficients of polynomial for NCA OCP.
Power Coefficient
Power Coefficient
10 2743.15209404874
9 46362.7343756619 9 16813.5629208811
8 263375.938332227 8 45715.7813601628
7 654258.515520722 7 72552.9322361997
6 932223.373684673 6 74367.8731240061
5 839169.403627969 5 51403.6264287225
4 494680.844740510 4 24247.5438829565
3 190876.920785182 3 7702.84959051092
2 46466.7085535694 2 1576.40557740756
1 6473.78346450481 1 187.617161272597
0 397.475541425268 0 9.85971414355597
 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150 135

Fig. 3. Thermal gradient of NCA OCP as a function of cathode SOC (experimental values and the fitted model are shown).

Fig. 4. Solid phase diffusivity of NCA.

In general, research indicates that cathodes developed from Electrochemical models for lithium ion cells with blended
single materials do not exhibit all of the desirable characteristics. cathodes are reported [22,23] in literature. It is shown that pulse
Therefore, the recent trend is to make use of multiple active ma- response of NCA/C cells can be accurately represented [24] by the
terials to form a composite electrode to complement the individual electrochemical model and the solid state diffusion limitations
shortcomings of the various components [13e21]. Many other studied in detail [25]. The solid phase diffusivity for a composite
blended or composite cathodes e.g. LiNi1/3Mn1/3Co1/3O2 or NMC, cathode is expected to vary with lithium concentration. Lithium
LiMn1.5Ti0.5O4 are studied for various favorable properties [13e16]. concentration in an electrode, represented as a fraction of
A detailed review of composite and blended cathodes is reported maximum lithium concentration, is defined as electrode state of
[13]. Blend of spinel type cathode (LiMn2O4) with layered cathode charge (SOC). Empirically determined functions of cathode SOC has
(LiNi0.8Co0.15Al0.05O2) has been studied [17] recently to optimize the been used for LiNCA diffusivity [25]. Models to analyze stress pro-
cost and life-expectancy. The capacity fade of batteries with com- files in NCA electrodes [26], impedance response of NCA particle
posite cathodes at different conditions are also widely investigated [27] and 3D electrochemical model [28] are also developed.
[18e21] to meet the stringent requirements of automotive industry. Industrial and research interest notwithstanding, a non-
136 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150

Fig. 5. Electrolyte phase ionic conductivity.

isothermal electrochemistry based model is not available for NCA/C Macro-dimension has been discussed earlier and it consists of the
cells. Low thermal stability of NCA, leading to thermal runaway, 1D through-plane domain with anode, separator and cathode
poses further safety hurdles in industrial application of these types subdomains. The micro-dimension consists of one representative
of cells. Therefore, understanding and control of thermal perfor- particle within which the solid phase diffusion of lithium in active
mance of Li-NCA battery is the way forward for large scale material is computed. There are two separate micro-dimensions
commercialization of this battery. A detailed electrochemical- representing the active material particles in anode and cathode
thermal model for Li-NCA battery can offer insight about the re- respectively. The current generation reaction follows Butler-
action, heat generation and species inside the cell resulting in a eVolmer equation and is computed in the macro dimension. The
comprehensive understanding about the cell operation and opti- lithium concentration in the solid phase comes as an input from the
mum control strategy can be devised out of it. The aim of this work micro-dimension. On the other hand, current generation in the
is to develop one such electrochemical-thermal model for Li-NCA macro-dimension is used as the lithium flux boundary condition in
battery which can predict the cell performance at various oper- the micro-dimension. As a result, this model is known as pseudo-
ating conditions with high accuracy and suggest insights be used to 2D (P2D) model.
reach the optimal operating condition. In the next section, the It is known that many of the critical electrochemical properties
mathematical model is described. The experimental details and the like open circuit potential, diffusivity in both solid and electrolyte
measured electrode properties are discussed subsequently. In the phase are functions of temperature. Heat is generated during the
results and discussion section, after validation of the model, results electrochemical reaction and transport of ion and electron. Thus,
are analyzed and extended for drawing further conclusions. generation of heat during the cell operation affects the electro-
chemical processes. Coupled electrochemical thermal model do
2. Mathematical model exist on the P2D framework [31,32]. Energy equation is solved in
the macro-dimension for all subdomains taking into account the
The well-established pseudo-2 dimensional (P2D) model reaction heat and transport heat generation. The P2D model and
framework [29,30] forms the basis of the electrochemical model thermal integration has been described in detail in the open liter-
used in this work. The geometry of the cell, which is divided into ature [29e34] and so, is mentioned briefly here. The following
separator and two electrode regions, is described in Fig. 1. The coupled partial differential equations are solved, unless specified
assumption of uniformity of properties within the plane of the otherwise.
electrode enables the governing equations to be written in 1D. The The current conservation and modified ohms law for the solid
electrode regions are assumed to be composed of active materials, phase in the electrode region (macro-dimension) [29e31]:
binder and electrolyte. Intercalation/de-intercalation reaction takes
place on the surface of active material particles. The lithium ion is
transported to/from the surface of the active material particle via
the electrolyte pathway, present in the porous electrode domain.

The electron passes through the active material particle to/from the V$ seff
s Vfs  as in ¼ 0 (1)
current collector, which in turn connects to the outer circuit. The
lithium molecule is stored in the interstitial space in the active As the boundary conditions, the potential of the solid phase in
material lattice structure. The lithium storage takes place at the the electrode region is set to zero at the current collector end of the
active material particle length scale (~mm), whereas the ion and anode. Similarly, the discharge current is specified at the current
electron transport length scales within the cell are of the order of collector end of the cathode.
100 mm. Therefore, the total operation is divided into two inter- Current conservation and modified ohms law for the electrolyte
connected domains e macro-dimension and micro-dimension. (macro-dimension) [29e31]:
 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150 137

a b

c d

e f

Fig. 6. a. Experimental validation of discharge at 298 K. b. Experimental validation of discharge at 318 K. c. Experimental validation of discharge at 333 K. d. Experimental validation
of discharge at 273 K. e. Experimental validation of discharge at 263 K. f. Experimental validation of discharge at 253 K.

! The reaction current density is defined through the Butler-

2Rg Tkeff
 d lnf±

eVolmer kinetics [31]:
e
V$ keff
e Vfe þ 0
tþ 1 1þ Vln ce þ as in ¼ 0
F d ln ce     
aa F   ac F  
(2) in ¼ i0 exp fs  fe  Eeq  exp  fs  fe  Eeq
Rg T Rg T
Mass conservation of Lithium salt in the electrolyte (macro- (4)
dimension) [29e31]:
The quantity ðfs  fe  Eeq Þ is termed as over potential. It de-
vðεe ce Þ
 1  t0 notes the voltage lost to discharge the cell at a particular rate. It is
þ
 V$ Deff
e Vce  as in ¼ 0 (3) clear from Eq. (4) that for greater discharge rate the over-potential
vt F
will be greater. The exchange current density is defined [29e31] as
138 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150

Fig. 7. Discharge curves at different discharge rates at 298 K, isothermal condition.

the following:  
vðrCP TÞ v vT
!aa ¼ k þ Qtot (8)
ce  aa vt vx vx
i0 ¼ Fðkc Þac ðka Þaa cmax
s  cs cas c (5)
cref
e Convection boundary condition is used at both ends of the
electrode. Total heat generated (Qtot) during the battery operation is
Lithium species conservation in the electrode phase (micro-
due to three main reasons e (a) entropic or reversible heat gener-
dimension) is shown below.
ation, (b) finite rate of reaction or irreversible heat generation and
  (c) transport resistance to electron transport in solid phase and ion
vðcs Þ 1 v vcs
¼ 2 Ds r 2 (6) transport in electrolyte phase or ohmic heat generation.
vt r vr vr Irreversible heat (Eq. (9)) is generated due to the over-potential
The boundary condition for Eq. (6), at particle surface is given by required to drive the reaction at a finite rate [32].
the following:  
Qirr ¼ as in fs  fe  Eeq (9)

vcs  in
Ds  ¼ (7) If the current demand is high, the over-potential is also high.
vr r¼Rp F This results in a non-linear increase in irreversible heat generation
in with discharge rate.
Zero flux condition is applied at the center of the particle while
Reversible heat generation (Eq. (10)) is due to the change in
the initial concentration is calculated from the electrode open cir-
open circuit potential of the electrode with temperature [32], as
cuit potential (OCP) curves.
shown below:
The diffusivity of lithium-ion in the NCA particles (Ds ) is not
constant. A lower diffusivity value has been attributed to a thin
vEeq
resistive layer, a few nm thick, at the surface of the NCA particle Qrev ¼ as in T (10)
[24]. Although, a thin NiO like cubic layer forms on the surface of vT
NCA particle, it has been reported to be porous [35]. The pores The change in open-circuit potential with temperature can also
would allow electrolyte to come in contact with NCA phase in the be linked with entropy rise and thus, this heat generation
presence of NiO-like phase. Thus, the assumption of anisotropic component is also called entropic heat generation. This quantity
lithium diffusion is not adopted. A thermodynamically consistent needs to be determined experimentally and it is different for almost
approach reported in literature suggests lithium concentration every electrode. Only a few references are available in open liter-
dependent solid phase diffusivity for lithium inside NCA particle ature about the entropic heat generation in NCA. Experimental
[25]. result from two references [36] and [37] report values which differ
As discussed before, the boundary condition used for lithium by an order of magnitude. For the present study, experimentally
species conservation (Eq. (7)) in micro-dimension, is obtained from obtained entropic heat generation term is used, whose details are
the Eq. (4), which is calculated in the electrode phase of the macro- discussed in the next section.
dimension. The surface concentration of lithium used in Eq. (5), on The ohmic heat generation [32] occurs in the electrolyte and
the other hand is obtained from the solution of Eq. (6) at the par- electrode phases. This contribution is quantified by the product of
ticle surface. Thus, the micro and macro dimensions are connected the current and the corresponding potential gradient in both the
to form the coupled PDE system of the P2D model. phases, as shown in Eq. (11) below. In the electrolyte, ionic current
The energy equation is solved in the macro dimension with heat constitutes of conduction and migration components, given by the
generation terms calculated from the battery operation [31e33]. first and second terms of Eq. (12). In the electrode phase, the
 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150 139

Fig. 8. a. Cathode SOC evolution at 298 K isothermal condition. b. Evolution of currents in solid (S) and electrolyte (EL) phase in cathode (6C, 298 K).

electronic current is due to electronic conductivity resulting in a In the present work ‘Battery and Fuel Cell Module’ of commer-
potential gradient (Eq. (13)). cially available finite element package COMSOL Multiphysics® 4.4
    has been used [34]. The electrode models have been modified to
vfel vfs include the multi-phase coexistence [40]. Modification of the
Qohm ¼ ie  is (11)
vx vx COMSOL model is achieved by coupling the 1-D lithium ion cell
model with the 2-D solid phase diffusion model for the electrodes
!
2Rg Tkeff
 d lnf±
 (as shown in Fig. 1).
e
ie ¼  keff
e Vfe þ 0
tþ 1 1þ Vln ce (12)
F d ln ce 3. Experiments

3.1. Experimental setup

is ¼ seff
s Vfs (13)
Usually, these components are small until and unless operated Cycling experiments are done in Solartron 1470E cell test sys-
at a high current density. Values reported for the heat capacity, tem. It is a multichannel (8 channels), 10 V (0.1% range), 4A (0.1%
density and thermal conductivity of NCA in the literature are not range) cycler (SH 642). It has 1470E (multi-channel potentiostats),
unique [36e39]. Therefore, numbers within the reported range 1455 (FRA), 14703A (temperature measurement option) and
have been used. 14702A (auxiliary voltage channels). OCP of electrodes are obtained
140 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150

Table 3
Material properties and cell parameters.

Anode Separator Cathode

Length (mm) 93 69 17
εl 0.3825 e 0.3382
εs 0.5802 e 0.4483
aa 0.5 e 0.5
ac 0.5 e 0.5
ka ðms1 Þ 7.77e-9 e 5.7e-8
kc ðms1 Þ 7.77e-9 e 5.7e-8
rp ðmÞ 12.5e-6 e 4.0e-6
cmax
s ðmol m3 Þ 28200 e 49195
c0s ðmol m3 Þ 0:647cmax
s e 0:30cmax
s
ss ðSm1 Þ 16700 e 0.04
Ds ðm2 s1 Þ 1e-13 Eq. 16
rðkg m3 Þ 2500 1200 3000
Cp ðJ kg1 K1 Þ 800 800 1000
kðWm1 K1 Þ 1.04 1 2
c0e ðmol m3 Þ 1150
ref
ce ðmol m3 Þ 1
0 0 1 1
De ðm2 s1 Þ

B B C C
1e@  4:43  @

54

A  0:22 ce
1000 A$1e  4
ce
T 229þ5 1000



sl ðSm1 Þ  8:2488 þ 0:053248T  2:9871e  5T 2 þ 0:26235 ce
 9:3063e
1000

c 
c 
c 2
c 2 2
c 
e e e e e
3 T þ 8:069e  6 T 2 þ 0:22002  1:765e  4 T $
1000 1000 1000 1000 10000

0
tþ 0.38
0:5
1:5
  ce ce
0  1Þ 1 þ d lnf±
n ¼ ðtþ 0:601  0:24 1000 þ 0:982ð1  0:0052ðT  294ÞÞ 1000
d ln ce

through three electrode experiments performed on commercially polynomial function is fitted to the experimental data at 298 K for
available LiNCA using Li reference electrode. The solid phase the NCA (cathode) OCP and is given in Eq. (14).
diffusivity of Lithium in NCA particle is estimated using electro-
chemical impedance spectroscopy (EIS) and galvanostatic inter- X
9
mittent titration technique (GITT) experiments. For this Eeq;NCA ¼ An yn (14)
experiment, the frequency range used is 1 MHze10 mHz while the n¼0

frequency resolution is 1 in 65,000,000. The ionic conductivity of In Eq. (14), ‘y’ represents the cathode state-of-charge (SOC). The
the electrolyte is measured using a Metrohm conductivity meter coefficients are listed in Table 1.
(856 conductivity module) at various temperatures (ranging from The cell OCP is measured at various temperatures (0.01C) and
60  C to 20  C) and concentrations. At each temperature, the NCA OCP at those temperatures are calculated. The NCA OCPs at
limiting molar conductivity is estimated. Diffusivity is estimated different temperatures are plotted in Fig. 2b against discharge ca-
using the NernsteEinstein equation. pacity. The NCA OCPs at different temperatures are standardized
against cathode SOC. These values are then used to calculate the
3.2. Electrode properties temperature dependence of NCA OCP. It is found that this quantity
can be represented with good accuracy by a polynomial of cathode
Open-circuit potential (OCP) and the gradient of open-circuit SOC as shown in Eq. (15). The fitted equation gives less than 7%
potential with temperature for NCA electrode are required as a normalized root mean square error.
function of lithium concentration for the numerical model. The
ratio of lithium concentration to the maximum lithium concen- vEeq;NCA X 10
¼ Bn yn (15)
tration, in an electrode, is defined as the electrode state of charge vT n¼0
(SOC). On the other hand, the cell SOC is considered 1.0 at fully
charged condition and 0.0 at fully discharged condition, and is The coefficients of Eq. (15) are listed in Table 2. The fitted curve
related to electrode SOC. The depth of discharge (DOD) of a cell is along with the experimental data is shown in Fig. 3. Widely used
defined as the compliment of cell SOC i.e. sum of cell SOC and cell entropic heat generation equations for graphite electrode is adop-
DOD is always one. It is convenient to express the electrode OCP as ted in this study from literature [42].
a function of the electrode SOC. Commercially available 18650 cells The solid phase diffusion of NCA particles are plotted in Fig. 4. It
are discharged at 0.01C (33 mA) to obtain the cell and electrode is found that the diffusivity of lithium decreases as its concentration
potentials. At this low rate of discharge, the experiment lasted for increased in NCA. For the operating range of lithium concentration
99.85 h and data is procured at an interval of 1 s. Discharge at such a however, the variation is within about one order of magnitude. It is
rate ensures that all the voltage drop due to transport and reaction seen that the concentration dependence of diffusion coefficient is
are negligible, and the resultant voltage corresponds to the ther- represented by a rectangular hyperbolic equation (Eq. (16)):
modynamic OCV of NCA material. The cell OCP and electrode OCPs
at 298 K are plotted in Fig. 2a against the cell DOD (¼1-SOC). The 3:3e  13
Ds;NCA ¼ (16)
negative electrode potential agrees well with the equation available ðy þ 0:02Þ2
in literature [41] and the same equation is adopted here. A
 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150 141

Fig. 9. a. Anode SOC evolution at 298 K isothermal condition. b. Evolution of currents in solid (S) and electrolyte (EL) phase in anode (6C, 298 K).

In Eq. (16), ‘y’ represents the cathode state-of-charge (SOC). The LiNCA cell at different temperatures, mentioned earlier at varying
ionic conductivity of electrolyte are reported in Fig. 5, and are in discharge rates, from very low (0.01C) to moderate (1C). The tem-
agreement with literature [43]. perature from the experiment is used as the surrounding temper-
ature in the model along with a very high heat transfer coefficient
4. Results and discussion (h ¼ 10,000 W m2 K1). Comparison with experiments at 298 K is
shown in Fig. 6a. A contact resistance of 0.00756 U m2 is estimated
Charge-discharge experiments are performed on the commer- for the cycler and has been implemented in the model for discharge
cially available NCA cells at room temperature (25  C), high tem- voltage validation. Plots show excellent agreement between
perature (45  C, 60  C) and low temperature (0  C, 10  C, 20  C). experiment and simulation. In Fig. 6b, comparison of discharge
Constant current chargeedischarge cycling, within the voltage curves at 318 K is shown. Agreement is excellent in the lower rates
limits of 4.2 V and 2.5 V, are performed at 0.01C, 0.1C, 0.5C and 1C (0.01C and 0.1C). At 1C discharge rate, the model seems to over-
for all the temperatures mentioned above. predict the experiments by about 0.1 V. Comparison of discharge
curves for 333 K is shown in Fig. 6c. An over prediction of about
4.1. Validation 0.1 V is can be observed at 1C discharge for this case also. The ac-
curacy of the model is defined with respect to normalized root
The proposed model is validated against constant surface tem- mean square error (NRMSE) and is shown in Eq. (17). It is found to
perature experiments performed on a commercially available be >93% at these temperatures at all the rates analyzed.
142 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150

Fig. 10. a. Evolution of heat sources at 1C discharge at 298 K isothermal condition. b. Evolution of heat sources at 2C discharge at 298 K isothermal condition. c. Evolution of heat
sources at 4C discharge at 298 K isothermal condition. d. Evolution of heat sources at 6C discharge at 298 K isothermal condition.

Also, the diffusivity of the electrolyte is increased to stabilize the

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ffi
1
PN
exp mod
numerical solution of the model. The minimum average accuracy,
N i¼1 V i
 V i as per Eq. (17) is 85%, for low temperatures. The rapid loss of voltage
NRMSE ¼ exp exp 0Accuracy ¼ 1  NRMSE and performance recovery however, could not be predicted effec-
Vmax  Vmin
tively using this model. A more detailed investigation focusing on
(17) low temperature prediction will be required for this. From the
All the cells are charged from the fully discharged state at the comparisons as discussed, it can be concluded that the present
controlled temperature for experiment. The initial conditions for model predicts voltage with very high accuracy at all temperatures,
the low temperature conditions are suitably changed based on the except at high discharge rates under low temperatures.
charging time. Model prediction at 273 K (Fig. 6d), is very good at
low discharge rates. At higher discharge rates (1C @5 Ah m2) 4.2. Analysis of reaction rates
however, there is a rapid drop in the voltage, which could not be
predicted with the same accuracy. Similarly, the model prediction is The discharge rate of a battery depends on the type of appli-
good at low rates and not so good at high rates when discharged at cation. For example, in mobile phones discharge rate usually do not
263 K (Fig. 6e). At very low temperature (253 K), as shown in Fig. 6f, cross 1C. The electric vehicle battery, on the other hand, will be
there is a recovery in cell voltage during high rate discharge (0.5C discharged at 6C rate for peak power demand. Therefore, four
and 1C). This result is, most probably due to self-heating as the cell discharge conditions (1C, 2C, 4C and 6C) at room temperature
temperature increases by about 8 K during this experiment. The (298 K, isothermal) are chosen to investigate the thermal perfor-
issue of self-heating for very low temperature operation for Li-NMC mance of the cell. The cell voltages for the discharge rates
(Nickel Manganese Cobalt) cell has been discussed in detail in mentioned above are plotted in Fig. 7. Although, the starting voltage
literature [44]. Experiments reported in Ref. [44] showed similar is the same for all the conditions, the discharge capacity increases
recovery in voltage during discharge, which could be replicated by about 20% as the discharge rate is decreased from 6C to 1C.
using self-heating condition. On the other hand, in-house experi- During discharge, the cathode (NCA) fills up with lithium. The
ments showed that the electrolyte tends to freeze at 253 K and electrochemical model enables understanding of the intercalation
therefore, the validity of extrapolated transport properties can be process in the electrode domains and identifies the utilization. The
questioned. Considering the experimentally observed increase in cathode SOC, at the separator end and the current collector end for
cell temperature, adiabatic boundary condition is used for this case. 1C and 6C discharge rates are plotted in Fig. 8a. As mentioned
 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150 143

Fig. 11. a. Comparison of Ohmic Source in Anode at 298 K isothermal condition. b. Comparison of Ohmic Source in Cathode at 298 K isothermal condition. c. Components of Ohmic
source in cathode for 1C at 298 K isothermal condition. d. Components of Ohmic source in cathode for 6C at 298 K isothermal condition. e. Comparison of reversible heat source at
anode at 298 K isothermal condition. f. Comparison of reversible heat source at cathode at 298 K isothermal condition.

earlier, the cathode SOC is defined as the ratio of lithium concen- implies, that during cell discharge, the lithium ion reaches the
tration at the surface of the particles, to the maximum lithium separator end of the cathode first but it does not intercalate there. It
concentration, is defined as the electrode state of charge should be noted that for intercalation to occur, availability of the
ðSOC ¼ cs; r¼RP =cmax
s Þ. It can be seen from Fig. 8a, that at low rate (i.e. lithium ion and electrons are necessary. For the operating range,
1C) the whole cathode lithiates uniformly. In other words, cathode the ionic conductivity of the electrolyte is of the order of 1 S m1
SOC (or lithiation) has small spatial gradient. As the discharge rates [43], but the electronic conductivity of the NCA cathode is only
increase, the cathode SOC within the cathode varies. At 6C, it can be 0.04 S m1 [24]. Thus, the resistance to electron transport is much
seen from Fig. 8a that the cathode SOC is found to be significantly greater than resistance to ion transport, in the NCA cathode. Total
greater at current collector end than the separator end. This transport resistance in cathode is the sum of ion transport
144 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150

Fig. 12. Comparison of total heat generation at different cooling conditions and discharge rates with room temperature initial condition.

resistance in electrolyte phase and electron transport resistance in the sum of electronic (or solid phase) and ionic (or electrolyte
the solid/electrode phase. Since the electron transport resistance is phase) current is equal to the external current demand. Spatial
much greater than the ion transport resistance in cathode, elec- gradient of electrolyte phase current is proportional to the reaction
trochemical reaction is more where the electron transport resis- rate (Eqs. (1) and (4)).
tance is least. The electron transport path is shortest at the current The solid and electrolyte phase currents along the cathode are
collector end and so the resistance to electron transport is least at plotted at different expended capacities (i.e. time instants) for 6C
that location. Therefore, the rate of lithium intercalation is higher at discharge rate in Fig. 8b. At the beginning of the discharge, the
the current collector end than the separator end in the cathode. As gradient of electrolyte current is steeper at the current collector end
lithium intercalates in NCA however, the diffusivity of the NCA and gradual at the separator end due to the low electronic con-
phase decreases (Eq. (16)). Finally the rate of lithium intercalation ductivity of NCA. As the discharge progresses, cathode SOC near the
at separator end and current collector end becomes equal to each current collector becomes greater than 0.5. From Eq. (5), it is clear
other, while the difference in cathode SOC between these points that the exchange current density is highest at cathode SOC equal to
reaches a constant as the discharge proceeds. This differential 0.5 (as the exponents aa and ac are equal to each other, Table 3).
intercalation results in the cathode SOC gradient within the NCA Therefore, the exchange current density starts to reduce, resulting
electrode and increases with discharge rate. In the case of 6C in slower reaction. Also, the solid phase diffusivity continues to
discharge rate, the difference in cathode SOC of cathode current reduce with increasing cathode SOC (Eq. (16)). This results in rate of
collector and the cathode-separator interface is much greater. reaction decreasing at the current collector end of the cathode
Although the difference stabilizes, it never decreases. Therefore, for increasing at the separatorecathode interface. Finally the electro-
6C discharge condition, when the cathode SOC corresponding to lyte current reduces linearly from separator to current collector,
cutoff voltage is reached at the current collector end, there remains indicating constant gradient or in other words, uniform reaction
significant amount of unused capacity at separator end of the rate. As the discharge continues however, again the gradients near
cathode. As the operating temperature is decreased, the Li salt the separator end become more gradual with a steeper section
diffusivity in the electrolyte reduces by order of magnitude. Thus, inside the cathode. This is due to the reduction in exchange current
the transport resistance in electrolyte phase and solid phase can density as cathode SOC crosses 0.5 at the separator end. As the
become comparable in the cathode at low temperature. At very low exchange current density reduces in the separator side, the total
temperature, the controlling transport mechanism may reverse. resistance to reaction at the separator end again becomes greater.
Low temperature focused investigation is required for better un- Considering the non-uniformity of the reaction rate in the cathode,
derstanding of this phenomenon, and it is done in section 4.5. it might be a good idea to design the cathode in such a way that
Inside a lithium cell, both electronic and ionic currents exist. more active material is stacked close to the cathode current col-
Both electrons and lithium ions are produced at the anode during lector. This would help to extract more capacity and effective uti-
discharge. The electrons are transported to the outer circuit lization from the battery at high rates.
through the active material particles and the lithium ions are The electronic conductivity of graphite anode is much greater
transported to the cathode through the electrolyte network present than ionic conductivity of the electrolyte. In the graphite anode of
in the porous electrodes and separator. Similarly, in the cathode the cell the electrode SOC gradient shows a trend opposite to NCA
during discharge, the electrons reach the NCA particle surface from cathode (Fig. 9a). At the electrode e separator interface, due to the
outer circuit through the electrode phase while the lithium ions high solid (electrode) phase current conductivity, electron can
reach through the electrolyte phase. The electrochemical reaction easily travel to the anode current collector. The lithium ion at this
takes place at the NCA particle surface and the product lithium interface has minimum electrolytic resistance. At the current col-
atom intercalates into the NCA particle. At any point inside the cell, lector end, however, the ionic resistance in the electrolyte phase is
 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150 145

Fig. 13. a. Discharge curves at 6C discharge rate for different heat transfer coefficients and 298 K initial temperature. b. Cell temperature at 6C discharge rate for different heat
transfer coefficients.

much greater. So at the beginning of discharge, the anode SOC reduces sharply, while the electrolyte phase current increases
drops at a greater rate at the anode-separator interface. This effect sharply, there. As discharge progresses, the gradient of electrolyte
is much greater at 6C rate than in 1C rate. As the discharge pro- current reduces near the separator end while increases near the
gresses, the surface concentration of lithium at the graphite particle current collector. At about 12.75 Ah m2 capacity, the reaction rate
surface near the anode separator interface reduces, resulting in at the separator becomes greater than that at the current collector
lower exchange current density and thus low reaction rate. As a (Fig. 9b). In Fig. 9a, it can be observed that SOC at current collector
result, the reaction rate at the current collector end increases and starts to decreases faster than anode SOC at separator from
steeper reduction is anode SOC is observed. Finally, at the end of 12.75 Ah m2 discharged capacity for 6C discharge rate. High ex-
discharge, the whole of anode is almost uniformly emptied. The change current density at separator (anode SOC close to 0.5)
solid phase and electrolyte phase current for 6C discharge rate compared to current collector end (anode SOC close to 0.1) is
along the anode is plotted at different capacity points in Fig. 9b. As responsible for this. At the end of discharge the anode is found to be
per the boundary condition, the solid phase current is highest at the almost uniformly discharged i.e. empty of lithium, even at high
current collector end while the electrolyte phase current is highest discharge rate of 6C.
at the separator-anode end. At the beginning of discharge, the re-
action is mostly at the separator end and so the solid phase current
146 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150

Fig. 14. a. Ohmic heat source in anode for different heat transfer coefficients e 6C, 298 K. b. Ohmic heat source in separator for different heat transfer coefficients e 6C, 298 K. c.
Ohmic heat source in cathode for different heat transfer coefficients e 6C, 298 K.

4.3. Analysis of heat generation rates during discharge might be helpful, it can be a hindrance for cold
start conditions (10  C or 20  C). The heat source components at
Heat is generated in the Li-NCA cell due to various reasons, 2C discharge rate are plotted in Fig. 10b. The ohmic heat generation
which have been discussed in the previous section. Average heat is proportional to the square of the current, whereas the reversible
source in any section is defined as the following (Eq. (18)). heat generation is proportional to the current and irreversible heat
generation is proportional to the current multiplied by the over-
0 1
lZ
section potential. As the current is doubled, from 1C to 2C, the cathode
1 B C ohmic heat generation increases four fold. As a result, ohmic heat
Q section ¼ @ Qsection dxA (18)
lsection generation in cathode becomes greater than the sum of reversible
0
heat generation at anode and cathode at the first 2/3rd part of the
The average heat sources due to different mechanisms for discharge. Thus, the total heat generation is always positive. In
anode, separator and cathode are plotted in Fig. 10. For 1C discharge other words, the discharge operation is always exothermic for 2C
rate (Fig. 10a), the irreversible heat generation in both anode and discharge rate. At 4C discharge rate (Fig. 10c), the ohmic heat
cathode is found to be orders of magnitude smaller than other generation at the cathode is many times greater than reversible
components. The ohmic heat generation in the anode and separator heat generation of both anode and cathode put together. At 6C
are also small. Due to the low electronic conductivity of NCA, the discharge rate (Fig. 10d) the total heat generation is completely
ohmic heat generation in the cathode is found to the dominant heat controlled by the ohmic heat generation at the cathode. The total
generation component. The reversible heat generation is negative heat generation follows the cathode ohmic heat generation pattern
in both cathode and anode for first 2/3rd part of the discharge, at high discharge rates. Near the end of the discharge, a sudden
which helps in keeping the discharge endothermic in period. In the increase in the heat generation is observed due to the contribution
last 1/3rd part of the discharge the total heat generation becomes from anode and cathode reversible heat generation, both of which
positive as the reversible heat generations also become positive. At increases sharply near the end of discharge.
1C discharge rate, the discharge operation is initially endothermic The dominant heat sources are compared as the discharge rate is
while becoming exothermic near the end. As a result, the total heat varied in Fig. 11. The ohmic heat generation in anode is predomi-
generation over complete discharge at 1C rate is small. Although, at nantly due to resistance to ionic current. As the discharge pro-
high temperature operations, this trait of negative heat generation gresses, more reaction starts taking place towards the current
 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150 147

a b

Fig. 15. Cathode SOC evolution at 273 K isothermal condition. b. Evolution of currents in solid (S) and electrolyte (EL) phase in cathode for 6C at 273 K isothermal condition. c.
Evolution of Lithium salt concentration in the cell at 273 K for 6C discharge rate.

collector end, increasing the average transport length of ions is not attained. Therefore, the reversible heat generation curves are
through the electrolyte. Hence transport resistance increase, not scaled replica of each other. The reversible heat generation in
increasing the ohmic heat generation. It is clear from Fig. 9b that the anode and cathode were found to be of the same order and
the area under the il curve over the anode length increases with generally following the same trend. The heat generation in sepa-
time. Therefore, ohmic heat generation in anode also increase with rator is only due to resistance to ionic transport and the current
time. The same trend is observed in Fig. 11a. In the cathode, the does not change with time. Therefore, the heat generation is largely
dominant component of ohmic heat generation is the resistance to constant in the separator domain for discharge rates (Fig. 10).
electronic current. It is clear from Fig. 8b, that the area under is
curve initially increases and then becomes more or less constant. 4.4. Effect of heat transfer coefficient
Therefore, the ohmic heat generation in cathode also increases
initially then reaches a constant stage (Fig. 11b). As the ohmic heat Thermal management of NCA based Li-ion battery is of utmost
generation is proportional to the square of the discharge current, importance for high power applications. In all the previous cases a
the ohmic heat generation at 6C nearly two orders of magnitude high heat transfer coefficient is assumed to ensure isothermal
greater than that in 1C rate. The ohmic heat generation has two condition. In reality, such high heat transfer coefficients do not
components e 1) due to the resistance to the electronic transport exist. For example, the thermal management cycle of an electric
and 2) due to the resistance to ionic transport. In the cathode side, vehicle battery draws power from the battery output. Therefore, it
the electronic component of ohmic heat generation is by far is obvious that the heat transfer coefficient will have a moderate
dominant for the whole range of discharge rate (Fig. 11c and d). In value and the battery may get heated within the allowable limit.
both the cases the, the electronic heat generation is almost one Thus, effect of heat transfer coefficient on NCA cell operation and
order of magnitude greater than the ionic (electrolyte or liquid heat generation is studied.
phase) heat generation. Therefore, it can be concluded that the Three conditions are considered e adiabatic (h ¼ 0 W m2 K1),
single most important component of heat generation at high air cooling (h ¼ 100 W m2 K1) and istothermal
discharge rate in LiNCA battery is the electronic heat generation in (h ¼ 10000 W m2 K1), whereas the external temperature fixed at
the cathode. 273.15 K. Total heat generations for these cases are compared for 1C
The reversible heat generation rates in anode and cathode are and 6C discharge rates in Fig. 12. It is clear that no appreciable
plotted in Fig. 11e and f respectively. These are proportional to the difference in heat generation exists for the case of 1C. The reason
discharge current. At higher discharge rates, the complete capacity for that is that the total heat generation over the discharge for 1C
148 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150

Fig. 16. a. Evolution of heat sources at 1C discharge at 318 K isothermal condition. b. Evolution of heat sources at 1C discharge at 273 K isothermal condition.

case is close to zero. Thus, the rise in temperature even for adiabatic increasing temperature. In the cathode, however, the heat gener-
case is small. Hence, for all practical purposes, the three operating ation due to electronic phase transport resistance is dominant in
conditions are same. On the other hand, there is an appreciable ohmic heat generation. So the ohmic heat generation in the cathode
difference in heat generation in 6C discharge rate. The heat gen- (Fig. 14c) does not vary much as the heat transfer coefficient is
eration is lower for adiabatic condition and this operating condition reduced. In effect, the total heat generation reduces as temperature
is further investigated. The cell voltage, for 6C discharge rate con- increases as a result of low heat transfer coefficient.
dition, is greater at the adiabatic condition (Fig. 13a) due to
decreased transport resistance at elevated cell temperature
4.5. Effect of operating temperature
(Fig. 13b). The temperature rise in the case of air cooling condition is
found to be insignificant. The reversible heat generation at anode
Operating temperature is known to have a strong impact on the
and cathode did not vary much as the absolute temperature vari-
operation of the cell. Considering 298 K (room temperature) as the
ation is of the order of 10%. The ohmic heat generation component
ideal condition two other conditions are tested 318 K (high tem-
for 6C discharge rate with previously mentioned heat transfer co-
perature) and 273 K (low temperature). The result of high tem-
efficients are plotted in Fig. 14. The ohmic heat generation in the
perature is close that of room temperature. The reason is that the
anode (Fig. 14a) and separator (Fig. 14b) decreases significantly for
transport resistances reduce as the temperature increases. At low
adiabatic condition as the electrolyte conductivity increases with
temperature (273 K) the transport resistances increase significantly
 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150 149

Fig. 17. a. Comparison of total heat generation at different ambient temperatures and 1C discharge rate. b. Comparison of Ohmic heat generation in cathode at different ambient
temperatures and 1C discharge rate. c. Comparison of reversible heat generation in cathode at different ambient temperatures and 1C discharge rate.

and so the cell capacity reduces. The cathode SOC at separator and collector and the zone with zero lithium salt concentration grows
current collector is plotted for 1C and 6C at 273 K in Fig. 15a. At 1C with discharge. Therefore, it can be concluded that at low tem-
the cathode SOC evolution in the cathode is similar to that at room perature and high rate, lithium from anode cannot reach the
temperature (Fig. 8a). At low temperature and high discharge rate cathode due to low diffusivity and the lithium from the electrolyte
(6C) the cell capacity reduces to almost 30% of the capacity at 1C gets consumed. This way the cutoff voltage is reached even when
discharge rate. Near the current collector end, the cathode SOC the cathode can accommodate more lithium. Thus, the cell capacity
becomes constant at about 0.55 for the 6C case. Constant cathode reduces greatly at high discharge rate and low temperature. It was
SOC indicates that no reaction is taking place at current collector. mentioned in section 4.2 that the cell operations are limited by the
Considering that interstitial space for Li atom is still available in the electronic conductivity of the cathode at 298 K. From the discussion
NCA particle, no reaction at the current collector means Li is not in this section, it can be seen that at 273 K however, the limiting
getting transported there. This observation can be further sub- process is the lithium transport in the electrolyte. Hence it can be
stantiated by analysis of the current densities. The electrolytic and concluded that there is a shift in the controlling process in an NCA/C
solid phase currents in the cathode at different discharged capac- cell at low temperatures.
ities are plotted in Fig. 15b. The solid phase current (also the elec- The components of heat generation can change with ambient
trolyte phase current) is found to be constant near the current temperature. This is because, the electrolyte conductivity is a
collector. As the discharge progresses the zone with constant solid function of temperature and reversible heat generation rate is
phase current increases. From this pattern, it can be inferred that proportional to temperature. The components of heat generation
lithium from the electrolyte is getting consumed and due to low rates for 1C discharge rate at 318 K and 273 K are plotted in Fig. 16a
diffusivity in the electrolyte it cannot reach the cathode from the and b respectively. Heat generation at 318 K (Fig. 16a) do not show
anode. With discharge, more and more part of the cathode elec- much variation compared to the data at 298 K (Fig. 10a). On the
trolyte domain becomes devoid of lithium. This can be attributed to other hand, at 273 K, the total capacity of the cell is much lesser. The
the low electrolytic conductivity at low temperature. To verify, the ohmic heat generation in cathode is even greater at 273 K (Fig. 16b),
salt concentrations are plotted at different discharged capacities due to decreased electrolyte conductivity. On the other hand, total
(Fig. 15c). The lithium salt concentration in the electrolyte phase is heat generation is positive for most part of the discharge. The total
found to increase as the discharge progresses in the anode. In the heat generation rates for 1C discharge rate at the temperatures
cathode, the lithium salt concentration reaches zero at the current 318 K, 298 K and 273 K are compared in Fig. 17a. Total heat
150 S. Basu et al. / Journal of Power Sources 283 (2015) 132e150

generation rates for 298 K and 318 K follow each other closely. In [4] W.V. Schalkwijk, B. Scorsati, in: W.V. Schalkwijk, B. Scorsati (Eds.), Advances in
Lithium Batteries, Kluwer Academic Publishers, New York, 2002, pp. 1e6.
the case of 273 K, the maximum heat generation is about the same
[5] M.G.S.R. Thomas, W.I.F. David, J.B. Goodenough, Mater. Res. Bull. 20 (1985)
as other temperatures and the peak is reached close to the end of 1137e1146.
discharge. The total heat generation rates can almost fall on each [6] T. Ohzuku, A. Ueda, M. Nagayama, Y. Iwakoshi, H. Komori, Electrochim. Acta
other in the last part of discharge. The ohmic heat generation in the 38 (1993) 1159e1167.
[7] J.R. Dahn, E.W. Fuller, M. Obrovac, U. von Sacken, Solid State Ionics 69 (1994)
cathode is significantly greater in the case of 273 K compared to 265e270.
other temperatures (Fig. 17b), whereas the reversible heat gener- [8] H. Arai, S. Okada, Y. Sakurai, J. Yamaki, Solid State Ionics 109 (1998) 295e302.
ation is lesser (Fig. 17c). At low temperature the magnitude of [9] H. Arai, Y. Sakurai, J. Power Sources 81e82 (1999) 401e405.
[10] H. Arai, M. Tsuda, K. Saito, M. Hayashi, Y. Sakurai, J. Electrochem. Soc. 149
reversible heat generation is lesser as it is proportional to the (2002) A401eA406.
temperature. On the other hand, the electrolyte conductivity re- [11] S. Albrecht, J. Kumpers, M. Kruft, S. Malcus, C. Vogler, M. Wahl, M. Wohlfahrt-
duces by one order of magnitude at 273 K. Thus, the component of Mehrens, J. Power Sources 119e121 (2003) 178e183.
[12] C.H. Chen, J. Liu, M.E. Stoll, G. Henriksen, D.R. Vissers, K. Amine, J. Power
ohmic heat generation due to ion transport resistance increases. Sources 128 (2004) 278e285.
Increase in ohmic heat generation is roughly equal to the decrease [13] S.B. Chikkannanavar, D.M. Bernardi, L. Liu, J. Power Sources 248 (2014)
in reversible heat generation and thus the maximum total heat 91e100.
[14] H. Kitao, T. Fujihara, K. Takeda, N. Nakanishi, T. Nohma, Electrochem. Solid-
generations are close. There is noticeable difference in the heat State Lett. 8 (2005) A87eA90.
generation pattern though. [15] S. Wang, Y. Wu, Y. Li, J. Zheng, J. Yang, Y. Yang, Electrochim. Acta 133 (2014)
100e106.
[16] S. Yang, G. Huang, S. Hu, X. Hou, Y. Huang, M. Yue, G. Lei, Mater. Lett. 118
5. Conclusion
(2014) 8e11.
[17] W. Haselrieder, S. Ivanov, H.Y. Tran, S. Theil, L. Frobo € se, B. Westphal,
Electrochemical thermal model for Li Nickel- M. Wohlfahrt-Mehrens, A. Kwade, Prog. Solid State Chem. 42 (2014) 157e174.
eCobalteAluminum battery is developed on the P2D framework. [18] S. Watanabe, M. Kinoshita, T. Hosokawa, K. Morigaki, K. Nakura, J. Power
Sources 258 (2014) 210e217.
The NCA cathode entropy generation and solid phase diffusion are [19] S.H. Lee, C.S. Yoon, K. Amine, Y.K. Sun, J. Power Sources 234 (2013) 201e207.
extracted from experiments and implemented in the model. Model [20] Y. Wang, J. Jiang, J.R. Dahn, Electrochem. Commun. 9 (2007) 2534e2540.
predictions compare very well with the experimental observations [21] Y. Kojima, S. Muto, K. Tatsumi, H. Kondo, H. Oka, K. Horibuchi, Y. Ukyo,
J. Power Sources 196 (2011) 7721e7727.
for a wide range of temperature and discharge rate. Subsequently, [22] P. Albertus, J. Christensen, J. Newman, J. Electrochem. Soc. 156 (2009)
the reaction rate and hence utilization of the electrodes are A606eA618.
analyzed in detail to identify the controlling processes. The results [23] S. Jung, J. Power Sources 264 (2014) 184e194.
[24] D.M. Bernardi, J.-Y. Go, J. Power Sources 196 (2011) 412e427.
indicate that the cell processes are limited by the electronic con- [25] D.M. Bernardi, R. Chandrasekaran, J. Go, J. Electrochem. Soc. 160 (2013)
ductivity of NCA electrode at 298 K and lithium transport in the A1430eA1441.
electrolyte at lower temperatures. Closer look at the heat sources [26] Y. Dai, L. Cai, R.E. White, J. Power Sources 247 (2014) 365e376.
[27] D.W. Dees, K.G. Gallagher, D.P. Abraham, A.N. Jansen, J. Electrochem. Soc. 160
reveal that ohmic heat generation in the cathode is the dominant (2013) A478eA486.
component. The reversible heat generation of anode and cathode [28] K.-J. Lee, K. Smith, A. Pesaran, G.-H. Kim, J. Power Sources 241 (2013) 20e32.
are found to be of the same order of magnitude. At very low [29] M. Doyle, T.F. Fuller, J. Newman, J. Electrochem. Soc. 140 (1993) 1526e1533.
[30] T.F. Fuller, M. Doyle, J. Newman, J. Electrochem. Soc. 141 (1994) 1e10.
discharge rates, these components counterbalance the ohmic heat
[31] J. Newman, Electrochemical Systems, Prentice Hall, Englewood Cliffs, NJ, 1991.
generation, resulting in an almost isothermal operation. At higher [32] W.B. Gu, C.Y. Wang, J. Electrochem. Soc. 147 (2000) 2910e2922.
discharge rates, the cathode ohmic heat generation rate dominates [33] K. Kumaresan, G. Sikha, R.E. White, J. Electrochem. Soc. 155 (2008)
and the total heat generation also largely follows the cathode ohmic A164eA171.
[34] COMSOL 4.4 Documentation, COMSOL Inc., 2014.
heat generation. Low electronic conductivity of the NCA cathode is [35] S. Hwang, W. Chang, S.M. Kim, D. Su, D.H. Kim, J.Y. Lee, K.Y. Chung, E.A. Stach,
the predominant reason of heat generation. Therefore, it can be Chem. Mat. 26 (2014) 1084e1092.
inferred that improving the electronic conductivity of NCA com- [36] M. Shadman Rad, D.L. Danilov, M. Baghalha, M. Kazemeini, P.H.L. Notten,
Electrochim. Acta 102 (2013) 183e195.
posite may hold the key to improved thermal performance of these [37] H. Yang, J. Prakash, J. Electrochem. Soc. 151 (2004) A1222eA1229.
types of cells. Detailed understanding of the cell response under [38] R. Benger, H. Wenzl, H.-P. Beck, M. Jiang, D. Ohms, G. Schaedlich, World Electr.
very low temperature operating conditions and degradation is Veh. J. 3 (2009). ISSN 2032e6653.
[39] H. Maleki, S. Al Hallaj, J.R. Selman, R.B. Dinwiddie, H. Wang, J. Electrochem.
ongoing, and will be part of a future work. Soc. 146 (1999) 947e954.
[40] A. Khandelwal, K.S. Hariharan, V. Senthil Kumar, Gambhire Priya,
References Subramanya Mayya Kolake, Oh Dukjin, Doo Seokgwang, J. Power Sources 248
(2014) 101e104.
[41] M. Safari, C. Delacourt, Modeling of a commercial graphite/LiFePO4 cell,
[1] V. Ramadesigan, P.W.C. Northrop, S. De, S. Santhanagopalan, R.D. Braatz,
J. Electrochem. Soc. 158 (2011) A562eA571.
V.R. Subramanian, J. Electrochem. Soc. 159 (2012) R31eR45.
[42] M. Guo, G. Sikha, R. White, J. Electrochem. Soc. 158 (2011) A122eA132.
[2] Marca M. Doeff, in: R.J. Brodd (Ed.), Batteries for Sustainability: Selected En-
[43] L.O. Valoen, J.N. Reimers, J. Electrochem. Soc. 152 (2005) A882eA891.
tries from the Encyclopedia of Sustainability Science and Technology, Springer
[44] Y. Ji, Y. Zhang, C.Y. Wang, J. Electrochem. Soc. 160 (4) (2013) A636eA649.
Science Business Media, New York, 2013, pp. 1e49.
[3] M. Armand, J. Tarascon, Building better batteries, Nature 451 (2008) 652e657.