You are on page 1of 12

Heat evolution and temperature increase in lithium

ion cells studied by combined electrochemicalcalorimetric measurements on lithium ion cells


Magnus Rohde, Boxia Lei, Carlos Ziebert, Andreas Melcher, Hans Jrgen Seifert
Karlsruhe Institut of Technology (KIT), Institut of Applied Material Applied Materials Physics (IAM-AWP)

Abstract:
The performance of a lithium ion cell is strongly related to the temperature. Therefore it is important to understand the process of heat generation and dissipation inside a single cell but also in battery packs since this is also
closely coupled to safety issues.
In this study, commercial 18650 lithium ion cells with LiMn2O4 cathodes as well as 20Ah pouch cells with
LiFePO4 cathodes were tested under isoperibolic and adiabatic conditions in an accelerating rate calorimeter
(THT Company) to investigate the heat effects during cycling.
Isoperibolic investigations in the range from 25 to 60C show that the applied environmental temperature does
not largely influence the battery thermal behavior. At 1C rate the maximum temperature increase over three
cycles was 4C almost independent of the environmental temperature. Additionally, the heat capacity and calorimeter constant were measured after calibration using cylindrical dummy cells made of AlMgSi0.5 with the cell
dimensions. By integrating over the heat dissipation rate and the enthalpy accumulation rate the total generated
heat was determined in dependence of discharge C-rate. Tests under adiabatic conditions, i.e. under negligible
heat loss, more accurately simulate a battery pack where several cells are closely packed and the neighboring
cells prevent the heat transfer to the ambient. The cell temperature was largely increasing at 1C rate over three
cycles by more than 40C rate before reaching the safety limit temperature of 75C.
This work presents also a short overview of some ECMs followed by a first implementation of an extended ECM
with a simplified thermal model in Matlab/Simulink/Simscape. The identification problem of the structure
and the parameters of an ECM are discussed in terms of the Current Interruption Technique (CIT).
In addition to the calorimetric measurements the distribution of the surface temperature was determined on the
pouch cell during charging and discharging using a thermographic camera system (FLIR, X6540sc) which
allows temperature measurements with spatial resolution. The resulting information from the IR images could be
correlated to the results of measurements with the calorimeter. It could be also used to identify temperature
gradients and hot spots on the surface of the cell.

1 Electrochemical-calorimetric measurements
Lithium ion batteries used in electric vehicles are getting more and more attention. Therefore it is important to
understand the heat generation and dissipation of single lithium ion cells and packs during operation, since this is
closely related to battery performance and safety issues.

1.1 Energy balance


As shown in Figure 1 the heat generation and heat dissipation in a single cell obeys an energy balance [1]:

!"#

"

$$

'()* !"#

$$

% &

"

% &

!1&

where
is the total heat generation rate,
the enthalpy accumulation rate,
the heat dissipation rate,
m the mass of the cell,
the measured heat capacity of the cell, h a heat transfer coefficient, A the battery area
exposed to the environment, T and T the cell surface and environment temperatures, respectively, ARC the
calorimeter constant [1]. The heat capacity of the cell and the calorimeter constant were determined to calculate
the total generated heat during cycling. For the calibration of the setup a dummy cell made of AlMgSi0.5 with the
cell dimensions was used.

enthalpy accumulation
heat dissipation

Figure1: Heat balance in a single cell arranged in ARC.


The heat effect of lithium-ion cells was measured using an Accelerating Rate Calorimeter (ARC) with a battery
cycler (internal or external) as seen Figure 2. The cycler was used for charging and discharging the cell. One or
more thermocouples are attached to the surface of the cell to measure its temperature during cycling. One 18650
cell and two pouch cells from different manufactures were tested to study the heat generation and dissipation of
single lithium ion cells. Chemistries and specifications of the cells are listed in Table 1.

Figure 2: ES ARC, EV+ ARC and Digatron MCT Cell Tester.


Table 1: Specifications of lithium-ion cells.
Cell

Chemistries (cathode//

LiMn2O4//

LiFePO4//

NMC//

anode)

graphite

graphite

graphite

Nominal capacity (Ah)

1.6

20

40

Nominal voltage (V)

3.7

3.3

3.7

Voltage window (V)

2.5 4.2

2.0 3.6

2.7 4.2

Dimensions(mm)

Cylindrical 1865 (DL)

Pouch 2271607.25

Pouch 22021510.7

1.2 Isoperibolic and adiabatic studies


In this study, commercial 18650 lithium ion cells (1.6Ah) with LiMn2O4 cathodes, 20Ah pouch cells with
LiFePO4 cathodes as well as 40Ah pouch cells with NMC cathodes (see Table 1) were cycled under isoperibolic
and adiabatic conditions in an Accelerating Rate Calorimeter (ARC).
Isoperibolic condition means constant environmental temperature. Under adiabatic conditions there is no heat
loss in the system and all the heat generated in the cell is used to heat the cell. This is important because in
reality the cell is surrounded with cells in battery pack and the heat produced in the cell can not be dissipated, so
the cell temperature increases continuously.
Figure 3 shows the experimental results for the cylindrical cell at C/2 charge/discharge rate. During isoperibolic
cycling the temperature rise of cylindrical cell is only about 2oC. Under adiabatic conditions the temperature rise
of the cell after one cycle is about 18oC.

Figure 3: Isoperibolic and adiabatic cycling of cylindrical cell.


In Figure 4 under isoperibolic conditions the temperature rise of pouch cell is about 3oC at 30oC and 2oC at 50oC.
Obviously the environmental temperature did not largely influence the battery thermal behavior.

Figure 4: Isoperibolic cycling of pouch cell.

2 ECM modeling for Li-Ion batteries using CITTechnique


In battery modeling the equivalent circuit model (ECM) is commonly used for two purposes: to predict battery
perfomance and to provide state of charge (SOC) estimation in battery management systems [2]. ECM model
describes the pure electrical behavior of the battery in terms of voltages, currents, resistances and capacitances.
Using the theory of passive electrical the resulting equations are a low dimensional system of ordinary differential equations which is in general nonlinear, since the system parameter depends on temperature T and state of
charge SOC. Due to these dependencies an extended ECM model is used together with an additional differential
equation for the SOC and an simplified temperature model in form of a linear inhomogenous differential equation. These models are able to capture the essential electrical phenomenas with good accurary with less complexity. They consist only of passive electrical elements and are easy to incorporate in BMS for real time application to ensure safe and reliable operation conditions. An overview of ECMs can be found in [3,4].
Figure 5 shows the general equivalent circuit model (ECM). It contains two voltage sources. The first one
describes the open circuit voltage of the battery, the second one describes parasitic voltage. Furthermore it
contains several parallel resistors and capacitors which represent polarizations like electrochemical or concentration polarization mechanism in the battery. The resistor R0 describes the ohmic resistance of the battery and
the Rp describe additional parasitic effects. This can be used to model thermal runaway with an ECM, where Rp
becomes of low impedance for rising temperatures which can result in a shorting circuit in the battery. All
parameters of this circuit depend on the SOC and temperature T.

Figure 5: General equivalent circuit model of an electrochemical cell.

2.1 Mathematical description and implementation


In the sequel the parasitic effects will be neglected. Using the passive electrical network theory the general
equivalent circuit model can be described as a system of n+2 nonlinear ordinary differential equations (2)-(4)
and one algebraic equation (5):
,?@
?A

23*

>6

)- */

BC D C

!T

/
5,
*4 6

>3*

)-

1*- /
5
123* *4 6

TE &

DC

Q,

1*1

)8 *8

!"

"9 &

*8

:;<= >

/
5,
*- 6

i=1, , n

(2)

(3)
(4)

56 HI

L/ > .

(5)

where Ri, Ci, Ui, Ii, i=1,n denotes the corresponding resistors, capacitors, voltages and currents of each RCbranch. UOC is the open circuit voltage, IL the load current and UL the terminal voltage, R0 is the ohmic resistance
of the battery. T, Ta are the temperature and ambient temperature, RT, CT the convection resistance and heat
capacitance respectively, Q the power dissipated inside the cell and SOC the dimensionless state of charge and
CQ the total charge of the battery. This system of equations can be rewritten in more general form using the state
space approach:
M

N!O, P, Q&, O!0&

T!O, P, Q&

OI

(6)
(7)

with f : xup f n+2 and g : xup g , where the manifolds x n+2, u 3, p


2n+2 are suitable chosen. Furthermore u = (IL, Ta, Q) T 3, x = (U1,, Un, SOC, T) T n+2, y = UL and
p = (R1,, Rn, C1,, Cn, Uoc) T 2n+2, p = p (SOC, T) denotes the input, state-space, output and parameter
vector respectively. In this description the parasitic effects are neglected, all other constants not mentioned in
parameter vector p are assumed to be constant.
The Matlab /Simulink /Simscape environment is an ideal tool to implement this system of differential and
algebraic equations. For simulation purposes an ECM with two relaxation mechanisms is implemented. The
corresponding electrical network in Simulink can be seen in Figure 6. For the passive electrical elements a
library is programmed using Simscape and lookup-tables are used in approximation of the nonlinear functions
f and g. The main difficulty is that the nonlinear functions f and g are in general not completely known due to
the dependence of the passive electrical elements of the ECM on the SOC and the temperature T. This problem
can be solved by using lookup-tables for the passive elements of the ECM. Lookup-tables are able to approximate this functions via interpolation and extrapolation in the simulation task. These tables can be filled via data
from the experimental Current Interruption Technique (CIT), where for known SOC and temperature T, charge
and discharge curves of real batteries are measured and from these curves the polarization mechanism, relaxation times, numerical values for restistances and capacitances can be extracted [2, 5].

2.2 The current interruption technique (CIT)


The current interruption technique (CIT) is a frequently used tool to describe the unknown function f, g via
interpolation and extrapolation technique using lookup-tables in the ECM. In the CIT the cell is charged and
discharged stepwise by interrupting the current at a certain voltage or after a certain time followed by a defined
relaxation time. Lookup-tables will be used to model the nonlinear behavior of ECM to extract system parameter
of ECM. This lookup-tables can be measured via CIT as described in Figure 7. This pulse provided important
information about the open-circuit voltage and the circuit dynamics at the given SOC. The ideal pulse data
would have reached steady-state voltage each time before a new pulse begins [2].
In this work the charge and discharge current was I=10A (C/2 rate) for the pouch (20 Ah) cell. After each current
interruption, the cell was left to relax for 60 min at open circuit (see Figure 8). The instantaneous voltage drop
(Ui) was measured after current interruption, and the cell voltage drop during relaxation (Urel) was also recorded.
The instantaneous voltage drop (Ui) is mainly due to the ohmic resistance of the cell and partially due to the
concentration polarization. The relaxation voltage drop (Urel) is due to concentration polarization in the liquid
electrolyte and in the solid electrode materials, which is a very slow process. The instantaneous impedance (Ri)
and the relaxation impedance (Rrel) were estimated using Eqs. (8) [6].
H

,U

HV

,W

(8)

Figure 6 : Simulink/SimscapeTM implementation of the ECM with two polarizations [7].

Figure 7: Pulse discharge tests [2].

Figure 8: Measured DC-impedance (current interruption technique).


In Figure 9 a closer look at a single pulse and cell impedance at different points of depth of discharge (DOD)
(DOD=1-SOC) are shown. The instantaneous impedance was almost identical at the beginning of discharge and
a sudden increase was noticed at DOD 0.98, which means the ohmic resistance of the cell increases at the end of
discharge. A slight increase with a slight decrease in the the relaxation impedance was noticed at DOD 0.900.98. This may be due to a phase change in the cathode material or to structural transformation in the anode
electrode [6]. There was also a sudden increase in the the relaxation impedance at DOD 0.98. This may be due to
the concentration polarization, which usually increases significantly with DOD [6].

a)

b)

Figure 9: a) One pulse from the test and b) Impedance as a function of DOD.

2.3 Simulation of a Pouch Cell using ECM


For the Simulink/ Simscape implementation of the model first simulations of a pouch cell with the dimensions 0.220.2150.01 m at ambient temperature Ta=25C were performed for the drive profiles Artemis Road
and Artemis Urban. The simulated battery is first charged to SOC=1 over charging time 1200 s and then the
profile is performed in the simulation using the ode15s solver from Matlab and the battery is completely
discharged to SOC=0. The simulation results are shown in the Figure 10. For this simulation lookup-tables for
the system parameter R0, R1, R2, C1, C2, Uoc are determined with the CIT-technique.

Figure 10: Simulation of Artemis Road (Top) and Artemis Urban (Bottom) in Simulink/ Simscape.

3 Thermography
Accelerating Rate Calorimeters (ARC) are reliable and very useful for spot temperature measurement, but for
scanning large areas or components, its easy to miss critical effects. A thermal imaging camera can scan entire
components at once never missing any overheating hazards. In our work, a thermal camera was used to investigate the heat effects and the local temperature distribution during cycling in more details.
Every object emits radiation in the infrared region. Infrared radiation (IR) is not detectable by the human eye, but
an IR thermography camera can convert it to a visual image that depicts thermal variations across an object [8].
From Stefan-Bolzmann law, the radiated energy from a blackbody is
X

Y" Z !X/

&,

(9)

where is the Stefan-Bolzmanns constant (5.6710-8 W/m2K4). Since there is no perfect blackbody, the emissivity was used to calculate the temperature of a normal object [8]:
]

X ^_ X^^ ,

(10)

where Wbb is the emitted energy from a blackbody and Wobj is the emitted energy of a normal object at the same
temperature. So emissivity is a number between 0 and 1. Thus, the emissive power of a normal object is [8]
X

]Y" Z !X/

).

(11)

As shown in Figure 11, the radiation that impings on the IR camera lens comes from three different sources [8].
The camera receives radiation from the target object, puls radiation from its surroundings that has been reflected
9

onto the objects surface. Both of these radiation components become attenuated when they pass through the
atmosphere. Since the atmosphere absorbs parts of the radiation, it will also emit radiation. [8]

Figure 11 : The radiation received by the IR camera [8].


The total radiation power receiced by the camera can be written as [8]
X

] b X ^_

!1

]& b XV

c$

!1

b& X9

(12)

where is the object emissivity, is the transmission through the atmosphere, Trefl is the reflected ambient
temperature, and Tatm is the temperature of the atmosphere. To calculate the correct target object temperature, IR
camera software requires inputs for the emissivity of the object, atmosphere transmission and temperature, and
the temperature of the ambient surroundings. These factor can be measured, assumed, or found from look-up
tables [8].
The thermal imaging setup for a prismatic format of the lithium iron phosphat pouch cell was performed using
FLIR X6540sc IR camera (Figure 12). The camera has a maximal resolution of 640 by 512 pixels and the wavelength range is between 3.5 and 5 m. The camera is connected to a computer to save the thermal imaging using
software (FLIR Research IR, FLIR Systems). The cell was oriented in a vertical configuration and connected to
the Digatron MCT Cell Tester. The aluminum packaging of the cells provides a too low emissivity (=0.09) for
accurate IR imaging. Therefore, the surface of the cell was coated with graphite that has an emissivity of approximately 1.0 to ensure a uniform emissivity. A high emissivity surface effectively eliminates reflections from
the object being imaged [9].

Figure 12: Thermal imaging setup.


The cell was thermally imaged at charge/discharge rate C/2 at room temperature. Figure 13 shows the thermal
images taken at the beginning and the end of the discharge. The cell was fully discharged from 100% to 0%
SOC. In Figure 13, a rectangle outlines a separate area (Ar1), where the cathode, separator, and anode are physically located. The thermal imaging was almost symmetrical with the maximum temperature at the center of the
cell area (Ar1) and minimum temperature at corner. At the end of the discharge the cell has the temperature
difference of 1.7oC (Tmax=27.8oC, Tmin=26.1oC), compared to the beginning of the discharge where it is 0.6oC
(Tmax=21.4oC, Tmin=20.8oC). So the cell shows relative good temperature uniformity during discharge at dis10

charge rate C/2. The heterogeneity of the temperature distribution on the surface of the cell with increasing
discharge rate will be studied.

Figure 13: Thermal imaging at the beginning and the end of discharge
The local temperature on the surface of the cell was captured, as shown in Figure 14. It is evident that in horizontal and vertical direction no obvious temperature heterogeneity was observed.

Figure 14 : Temperature Comparison in different direction on the surface of the cell at a C/2 discharge rate

11

4 Conclusion and outlook


Different lithium-ion cells were cycled at different ambient temperature and different charge/discharge rates to
evaluate their performance and measure their capacity, impedance etc. Experimental results for isoperibolic and
adiabatic cycling were presented. The measured data were then applied to the energy balance getting the total
heat generated in the cell during cycling. An endothermic effect was noticed at the beginning of charge, followed
with a exothermic effect. This is due to the deintercalation of Li ions from positive electrode at lower SOC and at
higher SOC ohmic and polarization impedance increase significantly and is greater than the entropic heat effect
(endothermic during charge and exothermic during discharge). The total cell impedance during discharge was
defined by using CIT. Its change is well matched with the heat effect of the cell at the end of the discharge.
We used infrared thermal imaging for only a single discharge rate, but later it will be conducted at a range of
discharge rates to incorporate non-uniform temperature distribution on the surface of the cell with different
discharge rate.
Using the Simulink/ Simscape implementation ECM model offers a quick and easy possibility to simulate
the electrical-thermal dynamical behavior of Li-ion batteries based on look-up tables coming from experimental
measurement. It is planned to measure more lookup-tables for different types of Li-ion batteries and to perform
corresponding simulations. Also parasitic effects will be taken into account for simulations of Thermal runaway
with ECM models in the near future. Furthermore it is planned to extend the Simulink/ Simscape implementation for whole battery stacks and packs. Additional the Simulink/ Simscape implementation will be extended in that way that an optimization of the parameter of the model in comparison to a specific experimental
measurement with arbritary load profile will be possible due to an Least Squares fit.

Bibliography
[1] J. S. Hong, H. Maleki, S. Al Hallaj, L. Redey, J. R. Selman, Electrochemical-Calorimetric Studies of Lithium-Ion Cells,
J.Electrochem. Soc., 145 (1998)1489-1501.
[2] R. Jackey, M. Saginaw, P. Sanghvi, J. Gazzarri, T. Huria, M. Ceraolo, Battery Model Parameter Estimation Using a
Layered Technique: An Example Using a Lithium Iron Phosphate Cell, SAE International, 2013-01-1547, (2013).
[3] H. He, R. Xiong, J. Fan, Evaluation of Lithium-Ion Battery Equivalent Circuit Models for State of Charge Estimation by
an Experiemtal Approach. Energies 4, 582-596, (2011).
[4] X. Hu , S. Li, H. Peng, A Comperative Study of Equivalent Circuit Models for Li-ion Batteries. Journal of Power Sourcest
198, 359-367, (2012).
[5] T. Huria, M. Ceraolo, J. Gazzari, R. Jackey, High Fidelity Electrical Model with Thermal Dependence for
Characterization and Simulation of High Power Lithium Battery Cells. IEEE International Electric Vehicle Conference
2012, (2012).
[6] S. Al Hallaj, J. Prakash, J. R. Selman, Characterization of Commerical Li-ion Batteries using Electrochemicalcalorimetric Measurements, Journal of Power Sources, 87 (2000) 186-194.
[7] http://www.mathworks.de/de/help/physmod/simscape/examples/lithium-battery-cell-two-rc-branch-equivalentcircuit.html?searchHighlight=lithium%20battery.
[8] FLIR: The Ultimate Infrared Handbook for R&D Professionals.
[9] Matthew A. Keyser, Ahmad Pesaran, Mark Mihalic, Ji-Sang Yu, Soo-Ryung Kim, Mohamed Alamgir, Daniel Rivers,
Thermal Characterization of Advanced Lithium-Ion Polymer Cells, Third Advanced Automotive Battery Conference, June
2003.

Acknowledgement
This R&D project is partially funded by the Federal Ministry for Education and Research (BMBF) within the
framework IKT 2020 Research for Innovations under the grant 16N12515 and is supervised by the Project
Management Agency VDIVDEIT.

12