Vrije Universiteit Brussel

Electrochemical impedance spectroscopy characterization and parameterization of lithiumnickel manganese cobalt oxide pouch cells: dependency analysis of temperature and state ofchargeGopalakrishnan, Rahul; Li, Yi; Smekens, Jelle; Barhoum, Ahmed; Van Assche, Guy; Omar,Noshin; Van Mierlo, JoeriPublished in:Ionics

DOI:10.1007/s11581-018-2595-2

Publication date:2019

Document Version:Proof

Link to publication

Citation for published version (APA):Gopalakrishnan, R., Li, Y., Smekens, J., Barhoum, A., Van Assche, G., Omar, N., & Van Mierlo, J. (2019).Electrochemical impedance spectroscopy characterization and parameterization of lithium nickel manganesecobalt oxide pouch cells: dependency analysis of temperature and state of charge. Ionics, 25(1), 111-123.https://doi.org/10.1007/s11581-018-2595-2

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portalTake down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Download date: 21. Nov. 2020

ORIGINAL PAPER

Electrochemical impedance spectroscopy characterizationand parameterization of lithium nickel manganese cobalt oxide pouchcells: dependency analysis of temperature and state of charge

Rahul Gopalakrishnan1,2& Yi Li1,2 & Jelle Smekens1,2 & Ahmed Barhoum3,4

& Guy Van Assche3& Noshin Omar1,2 &

Joeri Van Mierlo1,2

Received: 21 November 2017 /Revised: 21 April 2018 /Accepted: 8 May 2018# Springer-Verlag GmbH Germany, part of Springer Nature 2018

AbstractCharacterizing lithium-ion batteries is of prime importance as it helps in understanding the safety, working temperature, voltagerange, and power capabilities. Based on these results, we can then choose operating conditions which include safety protocols,application, and working environment. In this study, EIS studies of commercially available 20-Ah lithium-ion battery and a 28-Ah prototype cell with nickel manganese cobalt oxide (NMC)/graphite chemistry are used to determine the contribution oftemperature and state of charge (SoC) towards the electrochemical impedance spectroscopy. These cells are manufactured forelectric vehicle (EV) application. The electrode structure, particle size, stacking of the electrodes, and other entities for both thecells are provided to compare the similarities and differences between both the cells. Equivalent circuit modeling is used toanalyze and comprehend the variation in impedance spectrum obtained for both the cells. It is observed that the ohmic resistancevaries with both temperature and SoC and the variation with temperature is more significant for the prototype cell. The prototypecell showed better charge-transfer characteristics at lower temperatures when compared to the commercial cell.

Keywords State of charge . Separators . Electrolytes . Electrochemical impedance spectroscopy . Electric vehicle

Introduction

Lithium-ion batteries (LIBs) are becoming the main energystorage devices in the communications, transportation, andrenewable energy sectors [1]. However, the highest energystorage possible for LIBs is insufficient for the long-termneeds of society, for example, extended-range electric vehicles

(EVs), hybrid electric vehicles (HEVs), and other portabledevices [2]. Nowadays, the tendency to have EVs increasesdrastically especially in the developed countries since the car-bon dioxide emission becomes a serious issue in this decade.To meet up the increasing energy demand, LIBs technologyscientists are still working on refining the cycling stability,looking for larger voltage window stable electrolytes, higherenergy density positive electrode (cathode) materials, andhigher capacity negative electrode (anode) materials.

Electrochemical impedance spectroscopy has always beenan effective, nondestructive technique which could analyze/characterize lithium-ion batteries. The impedance characteris-tic is directly linked to the power capabilities of the battery anddecides the voltage drop observed in a battery when current isapplied [3]. In [4], the authors have demonstrated the used ofimpedance spectroscopy to estimate the battery temperature. Itwas proved that a simple equivalent circuit could simulate thecharge-discharge behavior of LIB [5]. Equivalent circuitmodels have been demonstrated by many authors as the bestchoice for EV applications compared to other physics-basedmodels which require many parameters for developing a

* Rahul GopalakrishnanRahul.gopalakrishnan@vub.be

1 Mobility, Logistic and Automotive Technology Research Center(MOBI), Department of Electrical Engineering and EnergyTechnology (ETEC), Elsene, Belgium

2 Flanders Make, 3001 Leuven, Heverlee, Belgium3 Department of Materials and Chemistry, Vrije Universiteit Brussel

(VUB), Pleinlaan 2, 1050 Brussels, Belgium4 Chemistry Department, Faculty of Science, Helwan University,

Helwan, Cairo 11795, Egypt

Ionicshttps://doi.org/10.1007/s11581-018-2595-2

model [6]. Drawbacks do exist with this kind of modeling as itis difficult to use this technique in the design process of thecell but these are the kind of models preferred in battery man-agement applications [7].

EIS can be performed at different levels either at electrodelevel [8] or at pouch cell level [3]. It can also be used tocharacterize beyond lithium-ion technologies like lithium-sulfur and lithium-air [9, 10]. Methods have also been pro-posed to measure the impedance in real time for pouch cells,using a noise or signal injected; it also uses an algorithm that isused to estimate the impedance parameters from the model[11]. Fast Fourier transform is generally used to detect thesignals [12]. This is done to estimate the state of health ofthe battery [13]. The impedance studies are generally per-formed to analyze the influence of parameters with respectto temperature, SoC, and current rate as well as to quantifystate of health (SOH) and aging effect at different cycling/storage conditions. There have been several works related totemperature and SOC studies in LIB [14–16].

For aging studies EIS has been used as part of checkupprocedure to quantify the variation in impedance with respectto the impedance value of a new cell [17, 18]. This can provideinsights on which cycling condition (SoC, Current rate, tem-perature) has more effect on the lifetime of the batteries. Sucha study was performed in this European project on NMC cells[19] and several other works like [20–23]. The impedancemeasurement is generally done at the cycling/storage temper-ature or at ambient temperature (25 °C) for all SoC ranges orseveral chosen ranges (high, middle, and low SoC).

This work is aimed at determining the contribution of tem-perature, state of charge (SoC) towards the electrochemicalimpedance spectroscopy. Two NMC cells with different stoi-chiometry and designs are taken into consideration to observethis dependency. This investigation is done at the beginning oflife of both the cells (fresh cells) and aims to discuss thedifferences in the impedance behavior of the two and theircorrelation to electrode structure, cell design, etc.

Experimental

Materials

Two generations of fresh or uncycled NMC pouch cells havebeen selected, a commercial EIG 20 Ah and Leclanche 28 Ahwere tested. The commercial cell is designated as generation 1(G1) cell and the Leclanche cell is designated as generation 2(G2) cell. These high-power cells were part of the Batteries2020 European project [19] for HEV/EV purpose. The maindifferences between the two cells were the capacity, cathodestoichiometry, separators, and the electrode-separator stackingtopology. The capacities of the G1 and G2 cells were 20 and28 Ah at C/3 discharge rate at 25 °C. The NMC used in G1

cells was of 442 stoichiometry (Ni:Mn:Co ratio) whereas thecathode in G2 cell was NMC 622. A polymer-based separatorwas used in the G1 cell and the G2 cell consisted of a ceramic-based separator. The two cells were opened inside a gloveboxfilled with argon. G1 cell is the commercially available NMC/graphite cell where the electrodes are stacked in S- or Z-foldseparator topology. G2 cell is arranged in stacking topology(see Fig. 1).

In the G1 cell, the separator separates the two-sided cath-ode and anode sheets, the separator is continuous. In the G2cell, the electrodes are arranged like stacks of several bi-cellswhich form the pouch cell and each bi-cell is separated by asheet of separator to avoid short circuit. It can be observed thatthe G2 cell is heavier (645 g) than the G1 cell, owing to morenumber of cathode and anode sheets when compared with theG1 cell (428 g). The voltage window for the operation of cellswas chosen as per the manufacturer’s recommendation; forboth the cells, the end of discharge voltage (EODV) was3 V, but the end of charge voltage (EOCV) was 4.15 V where-as for G2 went up to 4.2 V. For G1 cell, the electrolyte wasLiFP6 in EC:DMC whereas G2 cell is manufactured with apatented electrolyte consisting of LiPF6 salt but the solventsand other additives cannot be disclosed. The energy densityfor both the cells is almost the same (174 Wh kg−1for G1 and176 Wh kg−1 for G2). The parameters for G1 and G2 cells aresummarized in Table 1.

Characterization

The particle size, morphology, and surface texture were exam-ined using a Phenom Pro X desktop SEM with a high-brightness CeB6 electron source and a BSED detector. Thetest samples were mounted on specimen stubs with double-sided adhesive tape. These were mounted in the charge reduc-tion sample holder (a holder with a pin hole for permitting airto enter the vacuum chamber) and examined by SEM. Anaccelerating voltage of 10 kV was selected.

The cells were tested using an EIS enabled MPG-205Biologic Tester inside a climate chamber (CTS), to controlthe temperature at which the cells are tested. The tester canprovide a maximum current of 5A and can perform EIS mea-surement from 10 μHz to 20 kHz. EC lab (11.10 version)

Fig. 1 Electrode-separator stacking topology in G1 and G2 cells

Ionics

software was used to analyze the results of the tests performed.The cells were characterized through two tests, discharge ca-pacity and impedance tests from low temperatures to highertemperatures (5, 15, 25, 35, 45 °C). The purpose of the capac-ity test is to obtain the available discharge capacity (constantcurrent − constant voltage mode) for both the cells at differenttemperatures at 5 A charge-discharge current. This capacity isvital as this is the reference used in the tester settings to dis-charge the cells to different SoC levels (ΔSoC) beforeperforming the impedance tests. Electrochemical impedancespectroscopy (EIS) measurements are more often performedunder potentiostatic control (PEIS). The impedance tests wereperformed after applying a 30-min pause/rest after the cell hasreached a SoC level. Similar procedure is also followed whenthe cell is completely charged (EOCV).This is performed tohelp the cell relax [24–26]. The amplitude should be set basedon good signal to noise ratio (SNR), and conserving the line-arity condition, this plays a major role in the accuracy of theresult [27]. The amplitude of the signal was chosen as 3 mVfor G1 and 8mV for G2 cell after several trials. The conditionsand parameters employed to perform the characterizations areshown in Table 2.

Results and discussion

Surface morphology and particle size

Surface morphology (topography) is the most important phys-ical property of the cells. Particle size measurement is routine-ly carried out across a wide range of batteries and is often acritical parameter during manufacture. Particle size has a di-rect influence on battery properties such as internal resis-tances, reaction kinetics, surface current densities, lower deg-radation possibilities, etc. The SEM images of the cathode and

anode electrodes of G1 and G2 cells show a uniform, crack-free, and stable coating throughout the electrode surface.Generally, the commercial electrodes are optimized for highactive material (e.g., NMC and graphite) and low conductingaid and binder contents (e.g., 90:5:5 ratio) to maximize ener-getic densities. As shown in Fig. 2, the surface of the G1cathode electrode is characterized by higher numbers(content) of NMC particles and low binder contents than thoseof G2 cells. The NMC particles of G1 cathode have smallerparticle size distribution than those of G2. For both NMCelectrodes, the NMC particles are made up by small primarycrystallites (500–1000 nm) with little-sized pores which prob-ably also provide electrochemically active surface area. It canbe clearly observed that the particle distribution of G1 cathodeis smaller and narrower than that of G2. The surface topogra-phy of G1 cathode is also more uniform when compared tothat of G2 cathode.

Figure 3 shows the surface topography and particle size ofthe anode electrode of G1 and G2 cells; no damaging of softgraphite particles was observed. The surface if G1 anode con-tains sheet-like agglomerates whereas the surface of G2 anodeis made of sphere-like agglomerates. The binder andconducting aids (e.g., carbon black) which are interconnectedby the particle to particle contacts provide a continuous ionicand electronic path within the electrode layer from the sub-strate to the surface.

ImageJ software was used to analyze the particle size dis-tribution from the SEM images shown previously for the an-odes and cathodes [28]. The particle size distribution in theanodes and cathodes from G1 and G2 cells is shown in Table3.

There was also a difference in the separators used in thesecells. The G2 cell uses a patented ceramic separator fromLeclanche whereas G1 cell uses state-of-the-art polymer sep-arator. Figure 4a shows the SEM image of the ceramic

Table 1 Components andspecifications of G1 and G2 cells(datasheet/manufacturer/postmortem)

Parameters/components G1 cell G2 cell

Cell dimensions (L ×W × T) (mm) 216 × 130 × 7.2 162.5 × 173.5 × 12

Cathode NMC 442 NMC 622

Anode Graphite Graphite

Separator Polymer Ceramic

Electrolyte LiPF6 in EC:DMC LiPF6Capacity (C/3; 25 °C) 20 Ah 28 Ah

Voltage range 3–4.15 V 3–4.2 V

Average voltage 3.66 V 3.68 V

Number of anodes 20 34

Number of cathodes 19 33

Weight 428 g 645 g

Energy density 174 Wh kg−1 176 Wh kg−1

Ionics

separator used in the G2 cell. Ceramic separator is consideredto increase the safety of the cell by reducing the shrinkage andincreasing thermal stability [29, 30]. This ceramic separator isdenser (1.13 g/cm3) and porous (55%) with a low Gurleynumber (19 s) [manufacturer]. Gurley number is an indicationof the electrical resistance of the separator and a low numberindicates good electrical performance; for polymer-based sep-arators, the Gurley number is around 30–40 s [31]. The thick-ness of the separators should not be more than 40 μm for EV/PHEVapplications [32, 33]. The ceramic separator used in G2

cell is around 35 μm compared to the 25-μm-thick polymerseparator used in G1 cell.

Charge-discharge characteristics

The charge-discharge curves at 5 A for G1 and G2 cells at25 °C are shown in Fig. 4b; a difference of 8.7 Ah can beobserved from the discharge curve between the two cells. Thedischarge values for the G1 and G2 cells at other temperatures

Table 2 Testing conditionsParameter Capacity test EIS

C-rate (charge/discharge rate) (A) 5 5

Number of cycles 3 1

Temperatures 5, 15, 25, 35, 45 °C 5, 15, 25, 35, 45 °C

Frequency NA 5 mHz–5 kHz

Pause time after reaching EOCV 30 min 30 min

Pause before EIS NA 30 min

Number of frequencies per decade NA 10

Amplitude (Va) NA G1 cell (3 mV); G2 cell (8 mV)

Δ SoC 100% (complete discharge) Every 10% capacity

NA Not Applicable

Fig. 2 SEM images of G1 and G2 NMC cathode. Particle size distribution (a, b) and surface topography (c) of G1 cell cathode. Particle size distribution(d, e) of G2 cell cathode and surface topography (f) of G2 cell cathode

Ionics

are listed in Table 4. These values are the reference valuestreated as available capacity at that temperature for a cell.

Determination of equivalent circuit

The Nyquist plot for a lithium-ion battery generally consists oftwo semi-circles which represent the cathode and anode, re-spectively. The anode semi-circle is seen at higher frequencieswhereas the cathode semi-circle is seen at middle frequency[34]. These semi-circles are dependent generally on thecharge-transfer reactions occurring at the anode and cathode,depending on the electrode material (uniform coating/mixture,morphology) and surface characteristics (potential distribu-tion, roughness, etc.) [35]. At higher frequencies, an inductivebehavior is seen which is due to the connection cables which

are used to connect the cells to the tester [35]. At the lowerfrequencies, we can observe a tail or a straight line whichrepresents the diffusion process at the anode and cathode inthe lithium-ion battery. Normally, an equivalent circuit is cho-sen to fit the impedance spectra to extract the parameters. Thecircuit is chosen based on simplicity with better representationof the internal parameters of the battery and ease to use inbattery management systems [36, 37]. As mentioned before,we expect two semi-circles representing the anode and cath-ode but this is not what is observed in commercial high-powercells; the impedances of anode and cathode electrodes are notdistinguished clearly. This could be attributed to the chemicaleffects occurring at almost the same time constants on both theelectrodes [14, 38]. Only one semi-circle is seen in most of theSoC levels and temperatures except at certain low temperature

Fig. 3 SEM images of G1 and G2 graphite anodes. Particle size distribution (a, b) and surface topography (c) of G1 cell anode. Particle size distribution(d, e) of G2 cell anode and surface topography (f) of G2 cell anode

Table 3 Particle size distributionin anodes and cathodes from G1and G2 cells

Materials Mean particle/agglomeratesize (μm)

Min size (μm) Max size (μm)

G1 anode 2.6 1.5 4.6

G2 anode 19.5 11.7 33.4

G1 cathode 7.2 4.6 10.1

G2 cathode 12.1 4.85 21.4

Ionics

conditions when coupled with low SoC levels. Hence, we canconsider the semi-circle to represent the charge-transfer pro-cesses occurring in the electrolyte, solid-electrolyte interphase(SEI), and the active materials in anode and cathode [16]. Themodel to this study is shown in Fig. 4c, which is also used in[39] the top circuit; L1 represents the inductance which comesfrom the cables and connections to the cell. R1 represents theohmic resistance (RΩ) which is the value obtained from thepoint where the impedance spectra cut the real axis and it is theonly parameter which can be obtained directly from the im-pedance spectrum. It represents the conductivity of the elec-trolyte; therefore, the resistance offered for lithium-ion trans-port [16]. R2 represents charge-transfer resistance (RCT) forthe processes occurring on both anode and cathode electrodessimilarly, Q2 represents the capacitances for anode and cath-ode electrodes (CDL). The capacitance is modeled through a

constant phase element (CPE) with a dispersion constant (thesemi-circle is modeled by a constant phase element as itscenter-point is not on the x-axis of the impedance spectrum[15]). Q3 represents theWarburg impedance which is the imped-ance at lower frequencies related to mass transport modeledthrough CPE as the slope is not 45° [21]. The pros and cons ofCPE have been already explained in various research works[40–42]. The bottom circuit was used at lower SoC levels (>20%SoC) at 5 °Cbecause therewere two semi-circle formations.Similar approach of using a different equivalent circuit for lowtemperature conditions has been recorded in [43]. Even thoughthe resistances and capacitances were split (for anode and cath-ode charge-transfer resistance processes) to represent both thesemi-circles [44], the values were very small, and hence whenthe parameters were plotted, a single value for resistance andcapacitance representing both the electrodes were obtained bysumming up the values. In [16], it has been proved that oneRC circuit is best for large simulations to simulate charge-transfer resistances for different factors like state of charge, stateof health, and temperature and an addition RC circuit can repre-sent range of diffusion process. Here, we use RQ circuits insteadof RC circuits. Two minimalization algorithms were used toanalyse the data and extract the parameters, namely Randomizeand Levenberg-Maquardt. Randomize is used first to providessuitable initial values followed by Levenberg-Maquardt, whichwill help in further minimization.

The variation of the characteristics of the impedance circuitwas investigated as a function of temperature, state of charge,

Fig. 4 a SEM image of the ceramic separator used in G2 cell. b Charge/discharge curves for G1 and G2 cells at 25 °C at 5 A current. c Equivalentcircuits used for modeling G1 and G2 cells. The top circuit used for

modeling the cells at all the temperatures except for low SoC levels at5 °C, where the bottom circuit was used

Table 4 Discharge capacities for cell A and cell B at differenttemperatures

Temperature (°C) G1 cell (Ah) G2 cell (Ah)

5 18.159 26.822

15 20.096 28.897

25 20.775 29.538

35 21.726 30.163

45 22.222 30.596

Ionics

and between the cells itself. The tests were performed at thefollowing conditions:

& Temperature 5, 15, 25, 35, 45 °C& SoC range 100–0%

Temperature dependency

It is of prime importance to investigate the performance of thecell at lower and higher temperatures to check in which cli-matic region the cells can be used. Figure 5 shows how the

Fig. 5 Temperature effect on G1 and G2 cells at various SoCs. For G1 cell at a 80% SoC, b 50% SoC, and c 20% SoC and for G2 cell at d 80% SoC, e50% SoC and f 20% SoC

Fig. 6 SoC dependence onG1 andG2 cells at various temperatures. For G1 cell at a 5 °C,b 25 °C, and c 45 °C and for G2 cell at d 5 °C, e 25 °C, and f 45 °C

Ionics

cells G1 and G2 are influenced by the temperature at roomtemperature (25 °C), low temperature (5 °C), and high tem-perature (45 °C). Here, three different SoCs are chosen toanalyze the cells’ performance, 80, 50, and 20% SoC, whichrepresent high, mid, and low SoC range. The influence oftemperature is strong on G2 cell compared with G1 cell.

Figure 5a–c represents G1 cell’s impedance curves at80, 50, and 20% SoCs respectively at 5, 25, and 45 °Cand Fig. 5d–f represents the same for G2 cell. It can beclearly seen that R1 or ohmic resistance for G1 cell iscloser together in magnitude with a change in tempera-ture (~ 2 mOhms) whereas the G2 cell shows a widerrange of ohmic resistance. A general trend can be ob-served from the impedance spectrum that the increase intemperature decreases the resistance value and the de-crease in temperature increases the resistance. This isalso observed in [14, 16, 22]. The R2 or charge-transfer resistance increases with decreasing the temper-ature which suggests that the chemical processes pro-ceed in a slower path because of low thermal energyresulting in low kinetic energy [14]. Generally, thecharge-transfer reactions are very fast for commercial

cells [38], one can observe two distinct semi-circles atlow SoC and low temperature for both the cells and itis quite evident for G2 cell (20% SoC; 5 °C), and thevalues for the two distinct processes have been summedup together in order to compare the values with otherplots. This effect of increase in the charge-transfer re-sistance values was observed to be reversible as men-tioned in [16].

State of charge dependency

State of charge of the cells plays a big role in thecharge-transfer processes in the cells [14, 16, 45]. Thesize of the semi-circle increases with decreasing SoCwhich could be the decrease in the kinetics of the cellwhich can also been seen in [16, 46]. For clarity pur-pose, we choose four SoC’s: 80, 60, 40, and 20%. Itcan be observed from Fig. 6 that the G1 commercialcells perform better in terms of comparatively less in-crease in charge-transfer resistance when compared toG2 cells which means that G1 cells can perform at abigger SoC window compared to G2 cells. The ohmic

0 10 20 30 40 50

Temperature (0C)

1.3

1.4

1.5

1.6

1.7

1.8

1.9

R1

resi

stan

ce (

mO

hm

s)

R1(G1)

80% SoC60% SoC50% SoC40% SoC20% SoC

0 10 20 30 40 50

Temperature (0C)

2

2.5

3

3.5

4

4.5

R1

resi

stan

ce (

mO

hm

s)

R1(G2)

80% SoC60% SoC50% SoC40% SoC20% SoC

0 10 20 30 40 50

Temperature (0C)

0

2

4

6

8

10

R2

resi

stan

ce (

mO

hm

s)

R2(G1)

80% SoC60% SoC50% SoC40% SoC20% SOC

0 10 20 30 40 50

Temperature (0C)

0

2

4

6

8

10

R2

resi

stan

ce (

mO

hm

s)

R2(G2)

80% SoC60% SoC50% SoC40% SoC20% SoC

(a) (b)

(c) (d)

Fig. 7 Dependence of state of charge on R1 and R2 resistances at different temperatures for G1 (a, c) and G2 cells (b, d)

Ionics

resistance also changes with variation in SoC as previ-ously studied by [14, 22, 47]. It can also be seen that ata lower temperature and lower SoC, two semi-circlesbecome more visible as seen from the Fig. 6a, d whichcould suggest that two distinct charge-transfer processesmight be occuring at different time constants and thevalues of resistances have been summed up. Figure 7compiles all the parameters which were extracted usingthe equivalent circuit model. From Fig. 7a, b, we cansee that spread between the SoC’s at different tempera-tures is between 1.4 to 1.9 mOhms for G1 cell between80 and 20% SoC whereas it was between 2.25 and4.25 mOhms for G2 cell for R1 resistance (ohmic resis-tance). This clearly suggests that G1 is the betterperforming cell. The R1 values for both the cells runinto steady state above 25 °C. But in the case of R2

(charge-transfer resistance) (Fig. 7c, d), we see similarbehavior but the resistance values for G1 cells aresmaller than those for G2. At higher temperatures of35 and 45 °C, the charge-transfer resistance is ~1 mOhm for SoC which ranges up to 20% SoC. TheR2 values for G1 cell run into steady state above 35 °Cbut the value keeps decreasing for G2 cell. Comparablebehavior has also been reported in [48] for Kokam 40-Ah NMC cells.

Design contribution

Figure 8 compares G1 and G2 cells at different temper-atures and SoC conditions. The frequency for the onsetof diffusion tail and the ohmic resistance of both cellswere checked. The difference in charge-transfer

Fig. 8 Comparison of G1 and G2 cells at various temperatures and SoCs. When the temperature is 5 °C a 80% SoC, b 50% SoC, and c 20% SoC; whenthe temperature is 25 °C d 80% SoC, e 50% SoC, and f 20% SoC and at 45 °C g 80% SoC, h 50% SoC, and i 20% SoC

Ionics

resistance for both the cells at different temperature andat different SoC’s can be attributed to the surface reac-tion kinetics and particle morphology, basically the elec-trode structure.

The G2 cell has a higher ohmic resistance compared to theG1 cells which could be due to the thickness of the ceramicseparator (35 μm). Increase in thickness of the separator alsoenhances the tortuosity in the separator, leading to longer dif-fusion pathway for lithium-ions (mass transfer). Moreover,electrolyte wettability issues on the separator were reportedby the manufacturer which could have also lead to the de-crease in performance of the G2 cell. These could be thereasons for the G2 cell showing lesser performances comparedto G1 cell.

Figures 9 and 10 show the Bode plots of both the cells.Figure 9 shows the variation of magnitude of the cells againstfrequency in logarithmic scale whereas Fig. 10 shows thevariation of phases during occurring in the cell against thefrequency. These plots further assess the cells in terms of thetime constants (τ). Each semi-circle has a single time constant.It can be seen from Fig. 9 that the flat zones are going from flatat higher temperature to almost a slope at low temperature.

This could confirm that there are two processes occurringhaving different time constants leading to two semi-circle ob-servations. At higher temperatures probably, the kinetics arefaster and both process occur at comparable time constants.When comparing G1 and G2 cells, G1 has more defined timeconstants for the processes. From Fig. 10, we can see that at alltemperatures, we can see an inductive effect at higher frequen-cies where the curve tails upwards which is due to currentconduction, tabs, wires, etc. and it is generally independentof temperature. The intersection of the real axis (region wherethe curve becomes flat) shifts towards lower values with de-crease in temperature which is clearly evident for 5 °C plotand in G2 cell is that with increase in temperature, the induc-tive tail seems to change from inductive towards resistiveeffect at very high frequencies when compared with the G1cell.

Characterizing the electrodes used in G2 cells separatelyagainst lithium reference with the ceramic separator can yieldsome clarification in this regard. The charge-transfer resis-tance of G2 cell is observed to be higher than that of G1 cellat all temperatures except 5 °C where the charge-transfer char-acteristics of the G2 cell were better than those of the G1 cell

Fig. 9 Comparison of Bode plots (log (Z) vs log (frequency)) for G1 and G2 cells at 50% SoC at different temperatures

Ionics

(Fig. 8a–c); looking at the results, we could conclude that theG2 cell can perform better than the G1 cell at lower tempera-tures hence, widening the operation window but further half-cell characterization would be needed to confirm thisconclusion.

Conclusion

Electrochemical impedance spectroscopy was used to mea-sure the impedance of commercial and a prototype cell. Thegeneration 1 (G1—commercial cell) and generation 2 (G2––prototype cell) cells which were chosen were different basedon various entities including active material stoichiometry,active material particle size, stacking of the electrodes, elec-trode architecture, separator, and electrolyte. The impedancespectrum of these two cells was studied as a function of stateof charge and temperature. Two equivalent circuits, one forlow temperature and low SoC second for all other conditions,were used to extract parameters from two different kinds ofcells. CPEs were used to model the double-layer capacitanceas well as the diffusion process as the semi-circles obtained forboth the cells were depressed and the diffusion tail was not45°. The G1 cells were found to have lower ohmic resistancescompared to G2 cells at all the temperature and SoC condi-tions which could be related to electrolyte solvents and maybe

the separator (uptake of electrolyte). The ohmic resistancevaries with both temperature and SoC and the variation withtemperature is more significant for the G2 cell. The overallreason for the diminished performance of G2 cells could bedue to the morphology of the electrodes and the separator usedin the cell. The EIS modeling technique is very crude as it isquite difficult to relate/interpret each entity from the cell to aparameter in the EIS spectra because one or more entitiescould act simultaneously and affect a parameter. This is themajor limitation of this technique when used for pouch cellscharacterization. The further research with would be to devel-op a half-cell impedance models with the cathodes and anodesagainst lithium to check the influence of the separator andelectrolyte. This could give a better insight as to which com-ponent influences the full cell impedance behavior.

Acknowledgements We acknowledge the support to our research teamfrom BFlanders Make.^

References

1. Opitz A, Badami P, Shen L, Vignarooban K, Kannan AM (2017)Can Li-ion batteries be the panacea for automotive applications?Renew Sust Energ Rev 68:685–692. https://doi.org/10.1016/j.rser.2016.10.019

Fig. 10 Comparison of Bode plots (phase vs log (frequency)) for G1 and G2 cells at 50% SoC at different temperatures

Ionics

2. Nykvist B, Nilsson M (2015) Rapidly falling costs of battery packsfor electric vehicles. Nat Clim Chang 5:329–332. https://doi.org/10.1038/nclimate2564

3. Waag W, Fleischer C, Sauer DU (2013) On-line estimation oflithium-ion battery impedance parameters using a novel varied-parameters approach. J Power Sources 237:260–269. https://doi.org/10.1016/j.jpowsour.2013.03.034

4. Beelen HPGJ, Raijmakers LHJ, Donkers MCF, Notten PHL,Bergveld HJ (2016) A comparison and accuracy analysis ofimpedance-based temperature estimation methods for Li-ion batte-ries. Appl Energy 175:128–140. https://doi.org/10.1016/j.apenergy.2016.04.103

5. Liaw BY, Jungst RG, Nagasubramanian G, Case HL, Doughty DH(2005) Modeling capacity fade in lithium-ion cells. J PowerSources 140:157–161. https://doi.org/10.1016/j.jpowsour.2004.08.017

6. Franco AA (2013) Multiscale modelling and numerical simulationof rechargeable lithium ion batteries: concepts, methods and chal-lenges. RSC Adv 3:13027. https://doi.org/10.1039/c3ra23502e

7. Fotouhi A, Auger DJ, Propp K, Longo S, Wild M (2016) A reviewon electric vehicle battery modelling: from lithium-ion toward lith-ium-sulphur. Renew Sust Energ Rev 56:1008–1021. https://doi.org/10.1016/j.rser.2015.12.009

8. Andre D, Meiler M, Steiner K, Walz H, Soczka-Guth T, Sauer DU(2011) Characterization of high-power lithium-ion batteries by elec-trochemical impedance spectroscopy. II: modelling. J PowerSources 196:5349–5356. https://doi.org/10.1016/j.jpowsour.2010.07.071

9. Canas NA, Hirose K, Pascucci B, Wagner N, Friedrich KA,Hiesgen R (2013) Investigations of lithium-sulfur batteries usingelectrochemical impedance spectroscopy. Electrochim Acta 97:42–51. https://doi.org/10.1016/j.electacta.2013.02.101

10. Højberg J,McCloskey BD, Hjelm J, Vegge T, Johansen K, Norby Pet al (2015) An electrochemical impedance spectroscopy investiga-tion of the overpotentials in Li-O2 batteries. ACS Appl MaterInterfaces 7:4039–4047. https://doi.org/10.1021/am5083254

11. Lohmann N,Weßkamp P, Haußmann P, Melbert J, Musch T (2015)Electrochemical impedance spectroscopy for lithium-ion cells: testequipment and procedures for aging and fast characterization intime and frequency domain. J Power Sources 273:613–623.https://doi.org/10.1016/j.jpowsour.2014.09.132

12. Christophersen JP, Motloch CG, Morrison JL, Donnellan IB, WHMorrison (2008) Impedance noise identification for state-of-healthprognostics. 43rd Power Sources Conference, Philadelphia

13. Eddahech A, Briat O, Bertrand N, Delétage J-Y, Vinassa J-M(2012) Behavior and state-of-health monitoring of Li-ion batteriesusing impedance spectroscopy and recurrent neural networks. Int JElectr Power Energy Syst 42:487–494. https://doi.org/10.1016/j.ijepes.2012.04.050

14. Andre D, Meiler M, Steiner K, Wimmer C, Soczka-Guth T, SauerDU (2011) Characterization of high-power lithium-ion batteries byelectrochemical impedance spectroscopy. I. Experimental investi-gation. J Power Sources 196:5334–5341. https://doi.org/10.1016/j.jpowsour.2010.12.102

15. Samadani E, Farhad S, Scott W, Mastali M, Gimenez LE, FowlerM, Fraser RA (2015) Empirical modeling of lithium-ion batteriesbased on electrochemical impedance spectroscopy tests.Electrochim Acta 160:169–177. https://doi.org/10.1016/j.electacta.2015.02.021

16. Westerhoff U, Kurbach K, Lienesch F, Kurrat M (2016) Analysis oflithium-ion battery models based on electrochemical impedancespectroscopy. Energy Technol 4:1620–1630. https://doi.org/10.1002/ente.201600154

17. USABC (1996) Electric vehicle Battery Test Procedures Manual.Revision 2, report, January 1, 1996; Idaho Falls, Idaho. (https://www.digital.library.unt.edu/ark:/67531/metadc666152/: accessed

September 24, 2017), University of North Texas Libraries, DigitalLibrary, digital.library.unt.edu; crediting UNT LibrariesGovernment Documents Department

18. BSI (2014) Electrically propelled road vehicles—test specificationfor lithium-ion traction battery packs and systems. BSI Stand Publ2012

19. Batteries2020.eu | Towards competitive european batteries n.d.http://www.batteries2020.eu/ (accessed 15 Sept 2017).

20. Ecker M, Nieto N, Käbitz S, Schmalstieg J, Blanke H,Warnecke A,Sauer DU (2014) Calendar and cycle life study of Li(NiMnCo)O2-based 18650 lithium-ion batteries. J Power Sources 248:839–851.https://doi.org/10.1016/j.jpowsour.2013.09.143

21. ISO 12405-3 (2014) Electrically Propelled Road Vehicles- TestSpecification for Lithium-Ion Traction Battery Packs andSystems- Part 3: Safety Performance Requirements

22. Stroe DI, Swierczynski M, Stan AI, Knap V, Teodorescu R,Andreasen SJ (2014) Diagnosis of lithium-ion batteries state-of-health based on electrochemical impedance spectroscopy tech-nique. Energy Convers Congr Expo (ECCE), IEEE 2014:4576–4582. https://doi.org/10.1109/ECCE.2014.6954027.

23. Arunachala R,Makinejad K, Athlekar S, Jossen A, Garche J (2013)Cycle life characterisation of large format lithium-ion cells. WorldElectr Veh Symp Exhib EVS 2014 2014:1–9. https://doi.org/10.1109/EVS.2013.6914865.

24. Barai A, Chouchelamane GH, Guo Y, McGordon A, Jennings P(2015) A study on the impact of lithium-ion cell relaxation onelectrochemical impedance spectroscopy. J Power Sources 280:74–80. https://doi.org/10.1016/j.jpowsour.2015.01.097

25. Fuller TF (1994) Relaxation phenomena in lithium-ion-insertioncells. J Electrochem Soc 141:982. https://doi.org/10.1149/1.2054868

26. Reichert M, Andre D, Rösmann A, Janssen P, Bremes HG, SauerDU, Passerini S, Winter M (2013) Influence of relaxation time onthe lifetime of commercial lithium-ion cells. J Power Sources 239:45–53. https://doi.org/10.1016/j.jpowsour.2013.03.053

27. Yokoshima T, Mukoyama D, Nara H, Maeda S, Nakazawa K,Momma T, Osaka T (2017) Impedance measurements ofkilowatt-class lithium ion battery modules/cubicles in energy stor-age systems by square-current electrochemical impedance spectros-copy. Electrochim Acta 246:800–811. https://doi.org/10.1016/j.electacta.2017.05.076

28. Rasband WS, Image J (1997-2016) U.S. national institutes ofhealth. Bethesda, Maryland. https://imagej.nih.gov/ij/

29. Orendorff CJ (2012) The role of separators in lithium-ion cell safe-ty. Electrochem Soc Interface 21(2):61–65

30. Yu L, Jin Y, Lin YS (2016) Ceramic coated polypropylene separa-tors for lithium-ion batteries with improved safety: effects of highmelting point organic binder. RSCAdv 6:40002–40009. https://doi.org/10.1039/C6RA04522G

31. Ronald E (2016) Overview and Progress of United StatesAdvanced Battery Consortium (USABC) Activity.https://www.energy.gov/sites/prod/files/2016/06/f32/es097_elder_2016_o_web.pdf. Accessed 30 Oct 2017

32. Deimede V, Elmasides C (2015) Separators for lithium-ion batte-ries: a review on the production processes and recent developments.Energy Technol 3:453–468. https://doi.org/10.1002/ente.201402215

33. Arora P, Zhang Z (2004) Battery separators. Chem Rev 104:4419–4462. https://doi.org/10.1021/cr020738u

34. Zhuang Q-C, Qiu X-Y, Xu S-D, Qiang Y-H, Su S-G (2012)Diagnosis of electrochemical impedance spectroscopy in lithium-ion batteries. Lithium Ion Batter - New Dev:189–226. https://doi.org/10.5772/26749.

35. Orazem ME, Tribollet B (2017) Electrochemical impedance spec-troscopy, 2nd ed. John Wiley, New York

Ionics

36. He H, Xiong R, Fan J (2011) Evaluation of lithium-ion batteryequivalent circuit models for state of charge estimation by an ex-perimental approach. Energies 4:582–598. https://doi.org/10.3390/en4040582

37. He H, Xiong R, Guo H, Li S (2012) Comparison study on thebattery models used for the energy management of batteries inelectric vehicles. Energy Convers Manag 64:113–121. https://doi.org/10.1016/j.enconman.2012.04.014

38. Fleischer C, Waag W, Heyn HM, Sauer DU (2014) On-line adap-tive battery impedance parameter and state estimation consideringphysical principles in reduced order equivalent circuit batterymodels part 2. Parameter and state estimation. J Power Sources262:457–482. https://doi.org/10.1016/j.jpowsour.2014.03.046

39. Do DV, Forgez C, El Kadri Benkara K, Friedrich G (2009)Impedance observer for a Li-ion battery using Kalman filter.IEEE Trans Veh Technol 58:3930–3937. https://doi.org/10.1109/TVT.2009.2028572

40. Lukács Z (1999) Evaluation of model and dispersion parametersand their effects on the formation of constant-phase elements inequivalent circuits. J Electroanal Chem 464:68–75. https://doi.org/10.1016/S0022-0728(98)00471-9

41. Emmanuel B (2008) Constant phase elements, depressed arcs andanalytic continuation: a critique. J Electroanal Chem 624:14–20.https://doi.org/10.1016/j.jelechem.2008.07.014

42. Shoar Abouzari MR, Berkemeier F, Schmitz G, Wilmer D (2009)On the physical interpretation of constant phase elements. SolidState Ionics 180:922–927. https://doi.org/10.1016/j.ssi.2009.04.002

43. Olofsson Y, Groot J, Katrašnik T, Tavčar G (2014) Impedancespectroscopy characterisation of automotive NMC/graphite Li-ioncells aged with realistic PHEV load profile. IEEE Int Electr VehConf IEVC 2014 2015:1–6. https://doi.org/10.1109/IEVC.2014.7056095

44. Muenzel V, Hollenkamp AF, Bhatt AI, de Hoog J, Brazil M,Thomas DA,Mareels I (2015) A comparative testing study of com-mercial 18650-format lithium-ion battery cells. J Electrochem Soc162:A1592–A1600. https://doi.org/10.1149/2.0721508jes

45. Seaman A, Dao TS, McPhee J (2014) A survey of mathematics-based equivalent-circuit and electrochemical battery models for hy-brid and electric vehicle simulation. J Power Sources 256:410–423.https://doi.org/10.1016/j.jpowsour.2014.01.057Review

46. Osaka T, Momma T, Mukoyama D, Nara H (2012) Proposal ofnovel equivalent circuit for electrochemical impedance analysis ofcommercially available lithium ion battery. J Power Sources 205:483–486. https://doi.org/10.1016/j.jpowsour.2012.01.070

47. Karden E, Buller S, De Doncker RW (2002) A frequency-domainapproach to dynamical modeling of electrochemical power sources.Electrochim Acta 47:2347–2356. https://doi.org/10.1016/S0013-4686(02)00091-9

48. WaagW, Käbitz S, Sauer DU (2013) Experimental investigation ofthe lithium-ion battery impedance characteristic at various condi-tions and aging states and its influence on the application. ApplEnergy 102:885–897. https://doi.org/10.1016/j.apenergy.2012.09.030

Ionics