Investigation of cycle-life of lithium-ion batteries by means of electrochemical impedance spectroscopy

Transcription

1 Investigation of cycle-life of lithium-ion batteries by means of electrochemical impedance spectroscopy Jochen Bernhard Gerschler, Abderrezak Hammouche, Dirk Uwe Sauer 1 Introduction To date, Lithium-ion batteries are among the few systems showing sufficient promise for various applications, particularly, automotive and other transport applications. A key issue in the commercialization of this technology is the ability to estimate its long-term durability and lifecycle performance. Electrochemical impedance spectroscopy (EIS) is today a common method for the characterization of electrochemical processes and physical properties of batteries. Among its benefits is the small influence on the object under investigation, and that measurement can be made both at open circuit conditions or during battery operation. Moreover, processes with different time constants are easier to identify in the frequency domain. EIS can thus be used as a non-destructive in-situ analysing technique to identify the actual origin of cell ageing processes, which may be hardly recognized by traditional test methods, as well as to monitor the actual degree of ageing of the battery modules. This work shall give an insight into the use of EIS in cycle-life and calendar life tests with lithium-ion cells. Furthermore it points out the possibilities of EIS in lithium-ion cell's diagnostics. 2 Experimental: Cells and set-up Data and investigations on two different lithium-ion cells will be presented within this paper. page 1 / 12 1

2 The positive electrode of the lithium-ion cells of type A consists of a LiNi 0.80 Co 0.15 Al 0.05 O 2 - based composite with the following geometric parameters: wide (W) x long (L) x roughness factor (r/s) = 54 mm x 450 mm x 2 = mm 2. The anode is a rough sheet having the dimensions: W x L x r/s = 56 mm x 500 mm x 2 = mm 2. The separator is made of polyethylene imbedded in EC/DCE - 1M LiPF 6 electrolyte. The nominal capacity is 500 mah. For the measurements new and aged cells have been used. The aged cells have been aged by conducting 500 full discharge-/charge cycle with 1.5 to 1.7 C rate at 60 C. Calendar life tests have been made with lithium-ion battery B consisting of prismatic LiMn 2 O 4 -cells, which have a nominal capacity of 40 Ah, a nominal voltage of 3.6 V, a weight of 1.35 kg, a specific energy of 115 Wh/kg and an internal resistance of 2.5 mω. Figure 1 displays a photo of three cells of type A mounted in series to be tested at the same time. Adequate cell-holders with four electrical contacts (two for the current and two for the voltage) have been designed at the ISEA workshop for this purpose. Figure 1: Three Lithium-ion cells of type A tested in series with connections for separate impedance measurements on each cell Laboratory instrumentation, used for the presented experiments, has been developed at ISEA for high-precision impedance measurements on batteries. One of the main feature of the ISEA- page 2 / 12 2

3 multi-channel EISmeter is its ability to perform impedance measurements during battery charge and discharge, while the battery DC voltage is continuously drifting. The other technical characteristics of this equipment can be found, for instance, in [1] or [13]. 3 Analysis of cells' state by means of impedance spectroscopy 3.1 Typical impedance diagram A complete impedance spectrum that comprises the characteristics of all involved internal processes and allows for reliable modeling, needs to be measured over a sufficiently large frequency domain. Obviously, this spectrum will call for long measurement time if frequencies below 10 mhz are required in order to observe slow diffusion phenomena. Such a prolonged diagram shall be done only in absence of a DC current, otherwise the SOC value will vary significantly (especially under high DC currents) and the impedance diagram will thus correspond to a large distribution of SOC values and not to a single SOC value. Figure 2 shows impedance spectra plotted measured on a new cell and an aged cell of type A at 70% SOC in a frequency interval from 1500 Hz down to 0.5 mhz without DC bias current. In addition to the inductive behaviour observed at frequencies above ca. 300 Hz, both diagrams exhibit two high frequency capacitive loops (labeled I and II) followed by a low frequency diffusion characteristic. The shape of these impedance spectra suggests the equivalent electrical circuit shown also in Figure 2 to describe the dynamic behaviour of the cells. The inductive behaviour is considered to be related with the geometry of the cell and with the porosity of the plates. It is not changing significantly with the age of the battery cell. The intersection of the diagram with the real axis (labelled R ser ) corresponds approximately to the sum of internal ohmic resistances due to the electrolyte, current collectors and electrodes material bulk. The medium and the low frequency domains describe the behaviour of the different electrochemical phenomena involved in the reactions mechanism. Namely, loop I is typically associated with relaxation of electrical charges within the so-called solid-electrolyte inter-phase page 3 / 12 3

4 "SEI"; a thin passivating layer usually formed on top of the negative electrode surface, but also likely on top of the positive electrode surface [3]. Its main feature is its little sensitivity to the electrode potential (i.e., battery SOC). Loop II is related with the parallel combination of the charge-transfer resistance and the double-layer capacitance at the positive electrode. This loop is expected to be influenced by the SOC. Since both semi-circles, I and II, appear slightly depressed, the capacitive component is better described by the so-called constant-phase element (CPE). The low frequency part of the diagram describes the diffusion process in the bulk of the positive active material. The shape of the impedance curve usually comprises a more or less tilted part prolonged by a quite vertical branch, depending on the type of diffusion (cylindrical, spherical...) and its boundary conditions (transmissive, reflective...). Fitting of the curves was limited to high and medium frequency domain since the spectra should not be extended to very low frequencies, in order to prevent large changes in SOC during measurement of a given impedance diagram while a DC bias current is added. Also, fitting is much more precise in this case. The resistive parameters for the high and medium frequency domain of new and aged cells in the above mentioned conditions achieved by fitting the impedance curve to the equivalent circuit diagram where fitting simple capacitors are used instead of CPE elements (diagrams shown in Figure 2), are given in figure 3. The parameters of the equivalent electrical circuit have been numerically determined with the toolbox MEISP. MEISP performs a non-linear complex least square fitting algorithm. More information and details about MEISP are given in [12]. page 4 / 12 4

5 -0,03 fresh aged -0,02 II Z" (Ω) -0,01 R ser I I II 0,00 0,01 0,02 0,03 0,04 0,05 0,06 0,07 0,08 Z' (Ω) L R ser CPE SEI CPE ct Z diff R SEI R ct Figure 2. Impedance spectra of a new (fresh) and an aged Li-ion cells of type A measured at SOC=70% and room temperature and their analogies in the equivalent electrical circuit Resistance (Ω), Inductance (mh) fresh L R_ser R_SEI R_ct aged Figure 3: Model parameters for a new (fresh) and aged cell achieved from fitting the impedance spectra shown in Figure 2 to the equivalent circuit diagram from the same figure. page 5 / 12 5

6 3.2 Impedance diagrams during continuous charge and discharge Usually, battery performance is investigated over a wide range of state of charges in order to identify the SOC interval where the battery is efficiently usable. A simple way to assess the characteristic electrical parameters is by performing impedance measurements when the battery is on rest (i.e., at I dc = 0 A). But, the parameters thus obtained may not correspond to those parameters which occur under load. In fact, during battery charging or discharging, the electrical parameters change continuously as a result of the progressive change in the kinetics of the electrochemical processes which depend on the battery internal conditions (active mass utilisation, concentration gradients, etc...). Therefore, impedance measurements under dc current flow are more meaningful. However, in order to avoid to interfere with changes in the impedance due to changes in the SOC during such measurements. However, this limits the minimum frequency, because the lower the frequencies are the longer is the measurement period and therefore the change in the state of charge is a DC bias current is applied. Figure 4 presents the impedance spectra obtained on a new cell and an aged cell of type A during discharge at C/5 rate. With a minimal frequency of 10 mhz, the duration for the measurement of each spectrum is 25 min, during which SOC will have changed by 8.3%. Qualitatively, the shape of the curves is maintained, but their magnitude has grown by approximately 30% due to accelerated ageing. -80 Fresh cell during discharge at -C/5 DOD_04.2% -80 Aged cell during discharge at -C/5 DOD_04.2 % -60 DOD_20.8% -60 DOD_20.8 % -40 DOD DOD_45.7% -40 DOD DOD_45.7 % Z" (mω) DOD_70.6% DOD_95.5% Z" (mw) DOD_70.6 % DOD_87.2 % Z' (mω) Z' (mw) Figure 4: Impedance diagrams plotted during discharge at -C/5 on a new (fresh) cell (left hand graph) and an aged cell (right hand graph) of type A Figure 5 shows the parameters for the elements of the equivalent circuit diagram (figure 2) as a function of the depth of discharge. The depth of discharge is an expression for the state of charge, but is defined by DOD = 100% - SOC. The main observations are listed below. page 6 / 12 6

7 The ohmic resistance (R ser ) (high frequency resistance) does not change appreciably with SOC both during charge and discharge. It is equally almost insensitive with respect to ageing effects in this case. Three things are remarkable with regard to this result: 1. Eventhough the capacity of the cell declined by approx. 20% no changes in the high frequency resistance can be observed. Therefore the high frequency resistance is disqualified as an indicator for the capacity of a cell. 2. The high frequency resistance is not revealing a relevant signal for the state of charge indication. 3. The ageing of this specific cell is not due to changes in the electrolyte resistance of contact resistances between the active masses and the current collectors. This is an important information for the battery manufacturers as they can learn easily, where they have to check for problems in the cell design or electrochemistry. The resistance (R SEI ) associated with the solid-electrolyte inter-phase (loop I) shows no significant change with SOC, but it doubled due to ageing. This is a clear indication, that ageing is affected significantly by a growth of the solid electrolyte interface. On the one hand a growth of the thickness increases the resistance for the Li ions diffusion through the interface and on the other hand the growth of the SEI is always related with a consumption of Li and therefore a reduction of the capacity of the cell. The charge-transfer resistance (R ct ) doubles as well due to ageing. The variations of R ct with SOC generally present a U-shape, i.e., R ct increases both at high and low SOC values and it shows a minimum at intermediate SOCs. For SOC<20%, R ct is considerably high. A similar behaviour has been observed in the literature [2] for LiNi 0.85 Co 0.15 O 2 -based cathodes. It was related with the variation of the crystallographic c-axis of LiNi 0.85 Co 0.15 O 2 hexagonal lattice structure, which goes through a maximum at intermediate SOC values and rolls off at both ends. The increase in the charge-transfer resistance due to ageing indicates that the available surface for the reaction is reduced. This could be caused by structural changes of the active masses. The diffusion impedance (not shown in figure 5) remains almost constant in the % SOC range. Also, the accelerated ageing did influence the diffusion impedance significantly. This shows that the porosity of the active mass has not changed significantly. The relative contributions of the four resistive terms to the total polarisation resistance (measured down to 10 mhz) are unequal. For SOC>20%, their ratios are roughly: R ser (50%), R SEI (6%), R ct (16%) and R diff (28%) for new cells, and page 7 / 12 7

8 R ser (38%), R SEI (8%), R ct (30%) and R diff (24%), for aged cell, revealing that the major increase due to ageing stems from the charge-transfer resistance. Rser, Rct, RSEI (Ohm) Fresh cell during discharge at -C/5 R_ser R_ct R_SEI Rser, Rct, RSEI (Ohm) Aged cell during discharge at -C/5 R_ser R_ct R_SEI DOD (%) DOD (%) Figure 5: Electric Parameters' variation during discharge at -C/5 for a new cell (left plot) and for an aged cell (right plot) The results presented and analysed above can be summarised as follows: The effect of ageing is evident both on the charge-transfer resistance and on the Solid-Electrolyte-Interface (SEI) resistance, but it has minor influence on the ohmic resistance. The impedance magnitude grows by approximately 40%. The major increase is observed for the charge-transfer resistance. A growth of the SEI is the cause of an increase of R SEI, the increase of R ct is caused by an alteration of the interface properties. Figure 6 describes schematically the changing of structure in lithiumion cells during cycles. Conductive particles Conductive particles Current collector Active mass Binder Current collector Active mass Binder new cell aged cell Figure 6: Structure of an electrode on a new cell and on an aged cell. SEI growth page 8 / 12 8

9 The SEI growth and formation leads to an impedance rise at the anode. It can considered as an ageing effect, which occurs mainly due to cycling and storage at higher temperatures. Impedance rise can be directly related to power fade. Moreover SEI growth is only a partial process in the degradation of the electrodes. So in parallel to the described processes, corrosion of lithium in the active carbon takes place. This process also causes a capacity and power fade of lithium cell. Other ageing effects are more detailed described in [5]. The authors of [6] and [7] focus on the structure, formation and the degradation processes on the SEI. All things considered this example shows, in how far the impedance can give information about the actual state of a lithium-ion cell. By the measurement of the impedance it becomes possible to reveal statements about the internal structure of the cell and about its capability. The relation between impedance rise and power fade is shown by measurements for instance in the works of [8-11]. 3.4 Impedance diagrams during calendar life testing Ageing approach In addition to ageing mechanisms in lithium-ion cells as a result of cycling, storage of lithiumion cells under high temperature conditions causes an accelerated ageing as well. In order to analyse these mechanisms a calendar life test has been made with the battery of type B, which have been stored at 55 C for 24 weeks. A check-up with an analysis of the alterations in the capacity and the resistance have been made after storage duration s of 6, 12, 18 and 24 weeks. The check-up included also an impedance measurement Measurement results The impedance spectra of the lithium-ion cell measured after storage duration s of 6, 12, 18 and 24 weeks are presented in figure 7. Due to a special arrangement and the cell design it was possible to measure the impedance of the positive and the negative electrode separately. page 9 / 12 9

10 week positive plate -Z" (Ohm) negative plate 0,000 0,001 0,002 0,003 0,004 0,005 cell Z' (Ohm) Figure 7: Impedance spectra of both electrodes and the cell of battery type B measured during the check-up within the accelerated ageing test at elevated temperature of 55 C The main observations are: The shape of the curves does not change significantly during the storage. This shows that the SEI- and the charge-transfer resistant are not influenced by the storage No significant impedance effects are observed on the negative electrode at high frequencies. Therefore, the negative electrode is characterised mainly by the internal high frequency resistance. The impedance of the cell is dominated by the impedance of the positive electrode. The negative electrode is not showing significant changes in the impedance during the ageing test. The intersection point of the curves with the Z ' -axis moves to more positive values with proceeding ageing, this shows a growth of the ohmic resistance. From the findings listed above, it can be concluded, that a contact problem between positive active mass and current collector on the positive electrode exists. Indeed a post mortem analysis revealed a delamination of active mass as shown in figure 8 schematically. The phenomenon of page 10 / 12 10

11 contact loss inside the electrode as cause for ageing is detailed analysed in [4], an overview about the general effects of delamination on the electrode/electrolyte-interface at the anode is given in [5]. conductive particles current collector Binder Active mass Delamination Figure 8: Structure of a positive electrode after storage at 55 C. A delamination of active mass from the collector can be observed. 4 Conclusion Impedance spectroscopy as a non-destructive in-situ method of measurement enables a detailed investigation of the status and of the processes in lithium-ion cells. EIS is a workable solution to identify ageing processes and ageing mechanisms in lithium-ion cells. A detailed analysis of the impedance spectra allows to identify the reasons for ageing and the location within the cell. This is true especially if the electrodes can be characterised independently by using a reference electrode. Impedance spectroscopy is cheap, fast and non-destructive. It can be applied continuously during ageing tests and therefore gives information on the internal processes not only by physico-chemical post-mortem tear down analysis at the end of the test. So EIS can help to improve and accelerated the development of lithium-ion cells and to identify good and bad operation conditions. 5 Authors Dipl.-Ing. Jochen Bernhard Gerschler, Research Engineer since July 2005 with focus on battery management systems for nickel metal hydride and lithium-ion batteries, Electrochemical Energy Conversion and Storage Systems Group, Institute for Power Electronics and Electrical page 11 / 12 11

12 Drives (ISEA) at RWTH Aachen University, Jägerstrasse 17/19, D Aachen / Germany, Tel: +49 (0) , batteries@isea.rwth-aachen.de, Dr. Abderrezak Hammouche, worked for more than 5 years at the Institute for Power Electronics and Electrical Drives (ISEA) at RWTH Aachen University with a focus on material of lithium-ion batteries, algorithms and monitoring concepts for state of function detection and developed procedure for the use of impedance spectroscopy for characterisation of batteries. Since January 2006 he is with VB Autobatterie GmbH & Co. KGaA, Am Leineufer 51, D Hannover, Germany, Tel. +49 (0) , Abderrezak.Hammouche@jci.com Prof. Dr. Dirk Uwe Sauer, responsible for the Electrochemical Energy Conversion and Storage Systems Group, working in the field of battery and fuel cell characterisation, modelling, monitoring and system integration, Institute for Power Electronics and Electrical Drives (ISEA) at RWTH Aachen University, Jägerstrasse 17/19, D Aachen / Germany, Tel: +49 (0) , batteries@isea.rwth-aachen.de 6 Literature [1] E. Karden: "Using low-frequency impedance spectroscopy for characterization, monitoring and modeling of industrial batteries". Dissertation, RWTH Aachen, 2001, Shaker Verlag, ISBN [2] FY 2000, Progress Report for the Advanced Technology Development Program, R.A. Sutula, U.S. Department of Energy, Office of Advanced Automotive Technologies, December 2000 [3] E. Peled, in: J. P. Gabano: "Lithium Batteries", 1983, pp [4] Y. Wang, X. Guo, S. Greenbaum et al.; Electrochem, Solid State Letter 4 (2001), A68-A70 [5] J. Vetter et. al.; "Ageing mechanisms in lithium-ion batteries"; J. Power Sources, 147 (2005) [6] S. S. Choi, H. S. Lim; J. Power Sources,102 (2002) [7] R. Imhof, P. Novak; J. Electrochem. Soc., 145 (1998) [8] K. Amine, C. H. Chen et. al.; J. Power Sources, (2001) [9] J. Y. Song, H. H. Lee; J. Power Sources, 111 (2002) [10] C. Wang et. al; J. Electroanal. Chem., 497 (2001) [11] K. Amine et. al.; Proc. Electrochem. Soc , 2001, [12] MEISP, March Korea: Kumho Petrochemical R & D Center, 2002 [13] H. Blanke, T. Sanders, M. Kiel, T. Baumhöfer, B. Fricke, D. U. Sauer, EISMeter The art of Impedance Spectroscope for Batteries and Fuel Cells, Symposium Impedanzspektroskopie, Essen, 2006, Book of Proceedings page 12 / 12 12