Correlation between electrochemical characteristics and thermal stability of advanced lithium-ion batteries in abuse tests—short-circuit tests

Correlation between electrochemical characteristics and thermal stability of advanced lithium-ion batteries in abuse tests—short-circuit tests

Electrochimica Acta 49 (2004) 1803–1812 Correlation between electrochemical characteristics and thermal stability of advanced lithium-ion batteries i...

169KB Sizes 0 Downloads 1 Views

Electrochimica Acta 49 (2004) 1803–1812

Correlation between electrochemical characteristics and thermal stability of advanced lithium-ion batteries in abuse tests—short-circuit tests Mao-Sung Wu∗ , Pin-Chi Julia Chiang, Jung-Cheng Lin, Yih-Song Jan Materials Research Laboratories, Industrial Technology Research Institute, 195-5 Chung Hsing Road, Section 4 Bldg 77, Chutung, Hsinchu 310, Taiwan Received 26 September 2003; received in revised form 2 December 2003; accepted 8 December 2003

Abstract The shutdown function of a separator is an important factor in the safety of advanced lithium-ion batteries (ALB). When a separator without proper shutdown function is used, battery safety would depend on the thermal stability of electrode materials. Results show that thermal stability of a battery, contributed from both the anode and cathode, decreases noticeably after cycling. DSC shows that exothermicity from SEI decomposition and the reaction of the lithiated graphite and electrolyte around 140 ◦ C increases as cycle number increase; main reason is the gradual thickening of passivation film, observed through three-electrode ac impedance measurements. DSC also shows a similar trend of exothermicity for Lix CoO2 cathode. The lesser the amount of lithium (x-value) in Lix CoO2 , the larger the exothermicity and the lower the decomposition temperature. Using a three-electrode system to observe the changes of open-circuit potential in Lix CoO2 cathode, thermal instability is a consequence of decreased lithium content as cycle increases. © 2003 Elsevier Ltd. All rights reserved. Keywords: Advanced lithium-ion batteries; Thermal stability; Differential scanning calorimetry; AC impedance; Short-circuit test

1. Introduction With booming development in portable electronics, demand on batteries of higher energy density and efficiency has become more pressing. Among the vast selections of rechargeable batteries, lithium-ion battery especially ALB (with laminated aluminum exterior) receives most attention because of its high energy density, high working voltage, and low self-discharge rate. However, a major weakness, safety, has yet to be solved. Studies on this topic are underway all over the world [1–7]. Tobishima et al. [1,2] released a safety test report on commercial lithium-ion batteries for cellular phones: a direct relationship between battery safety and its testing specifications. Take overcharge for example, a battery with 1 C charging only resulted in cell bulging due to gas build-up, but with larger than 2 C the battery ignited or exploded. Tobishima et al. [3] and Dan et al. [4] studied the safety issues of AA-sized lithium-ion battery in which lithium metal was used as anode. In general, safety and battery designs are closely related. Kitoh and Nemoto [5] used ∗

Corresponding author. Tel.: +886-3-5913100; fax: +886-3-5820442. E-mail address: ms [email protected] (M.-S. Wu).

0013-4686/$ – see front matter © 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2003.12.012

a micro-porous separator, a temperature fuse, a current fuse, and a safety vent in the design, and successfully passed many safety tests including nail penetration, external short-circuit testing, overcharge, and heating. Generally, thermal stability of the battery components is a decisive factor in its safety performance. Differential scanning calorimetry (DSC) and accelerating rate calorimetry (ARC) techniques are commonly used to determine the thermal stability of electrode materials in electrolyte or under inert conditions [8–17]. Du Pasquier et al. [8] studied the reactivity of carbon anodes in plastic lithium-ion batteries by DSC and found that first reaction at ca. 120–140 ◦ C to be the transformation of passivation layer products into lithium carbonate, and lithiated carbon reacts with molten binder via dehydrofluorination only at T > 300 ◦ C. Maleki et al. [9] also used DSC and ARC to evaluate the thermal stability of several lithiated negative electrode (NE) and found that exothermic heat generation of lithiated NEs, in the absence of electrolyte, is attributed to the reaction of PVDF with lithiated carbon (Lix C6 ). Further, Richard and Dahn [11] used ARC to measure the thermal stability of a lithiated MCMB material in electrolyte under adiabatic conditions. They also developed a mathematical model to calculated

1804

M.-S. Wu et al. / Electrochimica Acta 49 (2004) 1803–1812

self-heating rate profiles [12]. Maleki et al. [13] and Zhang et al. [14] indicated that heat generation from decomposition of positive electrode material and reaction with electrolyte initiates thermal runaway in a Li-ion cell, under thermal or abusive conditions. Xia et al. [15] and MacNeil et al. [16] also used ARC to study the reaction between Lix CoO2 and electrolyte which was found to be autocatalytic. For a lithium-ion polymer battery, thermal stability of cathode materials in solid polymer electrolyte was studied and found a significant exothermic reaction of both LiNiO2 and LiCoO2 in contact with polymer electrolyte [17]. Most of the researches above illustrate test results and thermal stability between electrodes and electrolyte. However, under abusive conditions, safety behavior of commercial lithium-ion battery can be much more complicated. Recently, Tobishima et al. [2] and Leising et al. [18] have analyzed explicitly an overcharging testing of lithium-ion battery, whereas we focused on short-circuit testing of commercial ALB. In general, electric short can be classified into internal or external. An external short occurs by touching the conducting tabs through a direct metal media, whereas an internal short is more complicated and usually can be imitated by a soft-nail penetration test. Yet rarely there is a detailed report on short-circuit testing for lithium ion batteries (specifically the commercial ones), especially the effects by electrochemical characterization and thermal stability of electrodes after cycling. This paper, therefore, is to discuss the phenomena in short-circuit tests. Consequences of electrode thermal stability, separator and cycle number are also taken into consideration. Finally, a three-electrode system is used to investigate the electrochemical characteristics of electrodes. Relations between electrochemical characteristics and thermal stability of electrodes are also studied.

2. Experimental Three kinds of lithium-ion batteries (ALB type) studied differ in separators: PP (polypropylene, Celgard 2500, 25 ␮m), PE (polyethylene, Tonen, 25 ␮m) and PP/PE/PP (Celgard 2325 trilayer membrane, 25 ␮m). Batteries had laminated aluminum foil exterior, sized 3.8 mm × 35 mm × 62 mm, capacity 750 mAh, and weight 16 g. Cell balance (capacity ratio of anode to cathode) is about 1.05. The content of the lithium cobalt oxide electrodes (cathode) was 90 wt.% LiCoO2 (10 ␮m in diameter, Nippon Chemical), 7 wt.% KS6 (Timcal SA) and 3 wt.% PVDF (polyvinylidene fluoride, Kuraha Chemical) binder. Powder were mixed with solvent NMP (N-methyl-2-pyrrolidone, Mitsubishi Chemical) to form a slurry. The slurry was then coated onto aluminum foil (20 ␮m in thickness) and dried at 140 ◦ C. Then the electrodes were pressed by a roller machine to a resultant thickness of 150 ␮m. MCMB electrodes (anodes), composed of 92 wt.% MCMB (Osaka Gas, 25 ␮m in diameter) with 8 wt.% PVDF binder and NMP, were mixed and coated onto copper foil (15 ␮m in thickness) and went through the same

processing steps as the cathodes. The thickness of MCMB electrode was 135 ␮m. Batteries were assembled in dry room. Manufacturing process as follows: dry cathode and anode at 120 ◦ C for 3 h in a vacuum chamber, cut cathode and anode into the appropriate size for winding with separator, insert the roll into a laminated aluminum foil package. Inject 3.2 g of electrolyte into the pack and seal it off. Electrolyte was 1 M LiPF6 (lithium hexafluorophosphate, Tomiyama Pure Chemical) with ethylene carbonate/propylene carbonate/diethylene carbonate (EC/PC/DEC). The water content of electrolyte measured via Car Fischer titration in an argon-filled glove box (water content <5 ppm and oxygen content <5 ppm) is lesser than 10 ppm. Batteries were cycled between 4.2 and 3.0 V by means of a charge–discharge unit (Maccor model series 4000). Procedure consisted of a constant-current step at 750 mA followed by constant-voltage at 4.2 V until the current tapered to 20 mA. Discharge current was 750 mA. Unless otherwise stated, before any testing, all batteries were fully charged. Soft-nail test (internal short-circuit) was performed by sticking a stainless steel nail (diameter 3 mm, length 21 mm sharpened down to a point) into center of the battery to about half depth (about 2 mm). External short-circuit was carried out by using a copper circuit (total resistance: <30 m). Current changes in short-circuit tests were recorded with a current shunt device. K-type thermocouples were used to measure the external surface temperatures of batteries. All tests were conducted in an explosion-proof room at ambient temperature (about 25 ◦ C). To monitor voltage and impedance changes in anode or cathode, a lithium reference electrode was placed in between. A lithium chip was pressed onto one end of a fine copper wire to make the reference electrode. Three-electrode impedance measurements were taken by means of a potentiostat–galvanostat (Schlumberger SI 1286) and a frequency response analyzer (Schlumberger SI 1255). Scanning frequencies ranged from 50 kHz to 0.01 Hz, perturbation amplitude 10 mV. In order to understand separator’s role here, two copper foils replaced anode and cathode as in an ordinary battery. Test cell was put into an oven heating from room temperature to 200 ◦ C at 5 ◦ C min−1 . During the process, an impedance spectroscope (Hioki 3555) at 1 kHz was measuring changes of cell impedance. After cycling, batteries were opened in a dry glove box where samples of anodes and cathodes by scraping off sample from current collector were sealed in standard aluminum DSC pans. DSC scans were carried out on Perkin-Elmer apparatus at a heating rate of 20 ◦ C min−1 from 50 to 300 ◦ C.

3. Results and discussion 3.1. Internal short-circuit testing A battery can be fresh or has already been used under abusive condition, therefore batteries that had been cycled

M.-S. Wu et al. / Electrochimica Acta 49 (2004) 1803–1812 Table 1 Soft-nail test results for advanced lithium-ion batteries with different separators Cycles

10 200

Separator PE

PP

PP/PE/PP

+ +

+ −

+ +

“+” means all test samples passed; “−‘’ means only one out of five passed.

10 or 200 times were tested in a soft-nail penetration (internal short-circuit) test. Batteries with three different separators (PE, PP, and PP/PE/PP) were chosen for tests (five batteries with each separator type). Capacities were approximately 750 mAh for 10-cycle batteries and <650 mAh for 200-cycle batteries. Test results in Table 1 show that all met the passing requirement of the nail penetration test except for one cell with the PP separator after 200 cycles (four out of five samples). Generally in a nail penetration test, an instantaneous internal short would result the moment the nail is stuck into battery. Enormous heat is produced from current flow (double layer discharge and electrochemical reaction) in the circuit by the metal nail and electrodes. Contact area varies according to depth of penetration. The shallower the depth, the smaller the contact area and therefore the greater the local current density and heat production. Thermal runaway is likely to take place as local heat generation induces electrolyte and electrode materials to decompose. On the other hand, if a battery is fully penetrated, the increased contact area would lower the local current density, and consequently all cells would pass the test irregardless of the type of separator or extent of cycling. Fig. 1 shows the relationship between voltage and exterior temperature of a cell with a PE separator during a penetration safety test. Clearly there was a voltage drop from 4.2 to 0 V, instantaneously, as

1805

the nail penetrates through (when internal short occur), and temperature rose to a maximum of 105 ◦ C. There was no smoking, fire or explosion, only bulging of the battery. Similar results are obtained for batteries with PP/PE/PP separator. However, with cycling beyond 200 cycles, the majority of the batteries with PP separators caught fire and exploded, Fig. 2 shows its relation between voltage and temperature. 3.2. External short-circuit testing Another type of short is external short-circuit, by current collector contact. Test results show that all conditions pass this test. In this format of testing, instant contact current could reach as high as 25 A, stay at around 11 A for about 70 s, and then finally drop down to <1 A as shown in Fig. 3. The maximum temperature reaches 110 ◦ C. The results were identical for all three kinds of separators. External short-circuit clearly is safer than the soft-nail penetration described earlier, because contact area between current collectors is larger than metal nail contact where current density would therefore be smaller. In addition, external short-circuit takes place outside the battery, and unless internal temperature exceeds electrode material decomposition temperatures, the short should be safer. This phenomenon is applicable to all performed shorting experiments, independent of the separator or cycle number. 3.3. Separator materials There are a number of factors in battery safety, and the separator plays a very important role. Under certain abusive conditions (short-circuit or overcharge), the temperature of the cell increases and the shutdown of the separator helps prevent thermal runaway. Shutdown usually occurs near the melting temperature of polymer which closes the separator pores, converting the ionically-conductive porous

Fig. 1. Voltage and temperature behavior of a battery with a PE separator during a soft-nail safety test. Battery was cycled for 200 cycles.

1806

M.-S. Wu et al. / Electrochimica Acta 49 (2004) 1803–1812

Fig. 2. Voltage and temperature behavior of a battery with a PP separator during a soft-nail safety test. The battery underwent thermal runaway. Battery was cycled for 200 cycles.

polymer film into a non-porous insulating layer between electrodes [19]. At this temperature there is a significant increase in cell impedance and current passage is restricted. This blocks further electrochemical reactions thereby shutting down the cell before an explosion can occur [19]. Fig. 4 shows curves for impedance versus temperature on cells of different separators. The pores in PE and PP/PE/PP close at around 130 ◦ C to terminate any further ionic transport hence the impedance increase. As the temperature continues to rise, separator shrinkage short-circuits the whole system internally and the impedance drop. PE has a low shutdown temperature (130 ◦ C) and PP provides mechanical stability at and above 130 ◦ C; this is why the PP/PE/PP separator

has a much wider shutdown window than PE alone. Consequently, a multi-layer separator is safer than a single layer for practical use. However, the impedance increase of PP separator at 160 ◦ C may not be large enough for shutdown as compared with PE and PP/PE/PP separators. AC impedance results show a close relationship between separator shutdown and temperature. But in case of non-uniform temperature distribution, shutdown would be ineffective (only some pores are closed so electrochemical reactions could continue), so temperature distribution therefore is as important as other factors. Accordingly, in the nail test, the location of penetration (depth) not only determines the heat production but also heat transfer within a battery. If

Fig. 3. Current and temperature behavior of a battery undergoing an external short-circuit test after more than 200 cycles.

M.-S. Wu et al. / Electrochimica Acta 49 (2004) 1803–1812

1807

Fig. 4. Impedance of different separators vs. temperature.

the heat produced is not dissipated fast enough, temperature difference in the battery could be large enough to cause failure of the shutdown of separator. In tests with penetration depths <20% of battery thickness barely any pass, regardless its cycle number or separator type. Generally speaking, temperature differences for large batteries (electric vehicles) are larger than in smaller batteries, therefore, the possibility of separator malfunction becomes more of a safety issue. 3.4. Anode material Undoubtedly there are many factors in anode stability; Edström et al. [10] indicated that thermal stability of solid

electrolyte interface (SEI)-layer formed on graphite, mesocarbon microbeads (MCMB) and carbon-black anodes is dependent upon the type of lithium salt in electrolyte. Richard and Dahn [11] demonstrated the self-heating of MCMB is influenced by: (i) the initial lithium content; (ii) the electrolyte; (iii) the surface area; and (iv) the initial heating temperature. Du Pasquier et al. [8] also studied the effects of several factors on anode thermal stability, such as the degree of intercalation, the carbon specific area, the cycle number, the reaction of polymer binders, and the type of carbon. Here, we focus on the relation between changes in electrochemical characteristics and thermal stability of anode for commercial lithium-ion batteries (ALB) after cycling.

Fig. 5. The ac impedance of a battery at different cycle number.

1808

M.-S. Wu et al. / Electrochimica Acta 49 (2004) 1803–1812

Fig. 6. The ac impedance of a three-electrode battery.

Batteries with PP separators could not pass soft-nail tests after 200 cycles; it then can be concluded that separators like PE and PP/PE/PP pass safely, independent of cycle number, because they have a lower shutdown temperature (130 ◦ C). PP, on the other hand, has a very close relation with cycle number. It is, therefore, necessary to study the changes and effects of the electrodes during cycling. Generally, the anode after cycling would accumulate a so-called passivation film—a resultant product of anode material and electrolyte. The film thickens as cycle number increases, and the thickening is believed to be the cause of danger. When cycled batteries are examined, many gray-white spots are seen on the anodes of older batteries but very little on fresh ones. EIS is often used to analyze passivation films in batteries

and Fig. 5 shows EIS at different cycle numbers for complete cells. The hemisphere at high frequencies shows that surface film resistance is composed of smaller hemispheres, each from a parallel resistance and capacitance component of the passivation films. At low frequencies the hemisphere resembles charge transfer, and the linear Warburg impedance is due to the solid-state lithium ion diffusion [20]. Clearly the impedance of passivation film increases as cycle number increases, and this is likely the reason for failing the nail test. These data have contribution from both anode and cathode in a two-electrode system, and results are difficult to interpret. An alternative is to use a three-electrode system. A three-electrode system, in principle, with lithium metal as a reference electrode, is capable of separating individual

Fig. 7. DSC results for anode, after 10 cycles and 200 cycles.

M.-S. Wu et al. / Electrochimica Acta 49 (2004) 1803–1812

1809

Fig. 8. DSC result for cathode, after 10 cycles and 200 cycles.

contribution from the positive and negative electrode. Fig. 6 is the ac impedance for a three-electrode system after 10 cycles. Semicircle of cathode in high frequencies is small, referring to a low passivation film resistance; comparatively, the semicircle of charge transfer in low frequencies is large. Fig. 6 also shows a pronounced semicircle in high frequencies for anode, as a result of high passivation film resistance from the irreversible deposition of lithium ions onto the surface of MCMB; the situation gets worse as cycle number increases. The results show that passivation film resistance in anode contributes to the hemisphere at high frequencies. To ensure the reliability of readings from a three-electrode system, ac impedance scan of an ordinary two-electrode system is compared, and the nearly overlapping curves prove its applicability and reliability.

All the discussion so far is based on the electrochemical changes in anode and cathode after cycling, but in fact such changes very likely have altered the electrode thermal stability already. The cycled anode was studied further by differential scanning calorimeter (DSC), and the results are shown in Fig. 7. There is an apparent decomposition peak formed at about 140 ◦ C, whose amplitude increases with increasing cycle number. After 10 cycles, anode has a heat release of 53.8 and 95.5 J g−1 after 200 cycles. This 140 ◦ C peak was identified as the decomposition of passivation layer formed during electrochemical reduction of carbon by lithium and the reaction of the lithiated graphite and electrolyte [8,13,21,22]. Prior ac impedance results show that passivation film thickens as cycle number increases, therefore, an increasing heat released in DSC is predictable. In real battery abuse tests, if

Fig. 9. Open-circuit potential of the cathode in a fully-charged three-electrode battery.

1810

M.-S. Wu et al. / Electrochimica Acta 49 (2004) 1803–1812

the separator does not stop the temperature from rising (by pore closing) before 140 ◦ C, the exothermic decomposition of passivation film may lead to thermal runway. With very little passivation film, the amount of heat generated may be inadequate to cause danger, but with thicker films there is a higher risk; this is exactly why a battery with more than 200 cycles failed to pass the safety test. 3.5. Cathode material The passive film on cathode, after cycling, does not cause as much dangerous effects as the film on the anode. However, changes in material may still influence thermal stability of the battery. Similar to anode, there are many aspects to cathode thermal stability. Dahn et al. [23] showed that, at or above some critical temperature, positive materials such as Lix NiO2 , Lix CoO2 , or Lix MnO2 are not stable and liberate oxygen into electrolyte which, when heated above its flash point, reacts violently. Amount of oxygen released by heating increases with decrease in the x-value. Maleki et al. [13] believed that oxygen liberation rate can be dependent upon the crystal structure and particle size of the delithiated positive electrode materials. Li et al. [24] have demonstrated that the thermal stability of LiNiO2 depends on the individual particle size, not agglomeration size, hence large particles are thermally more stable. Fig. 8 is the DSC result for cathode after 10 and 200 cycles. An apparent decomposition peak of Lix CoO2 around 230 ◦ C is the main source of heat release, coincides with the result by Zhang et al. [14], Xia et al. [15] and MacNeil et al. [16] used accelerating rate calorimeter (ARC) to study the reaction between Lix CoO2 and electrolyte and it was found to be autocatalytic. Experiments show an increased peak as cycle number increases. The heat released after 10 cycle is

about 60 and 102 J g−1 after 200 cycles. This phenomenon demonstrates a change in the Lix CoO2 electrode. In order to understand the mechanism involved in thermal stability deterioration, a lithium metal reference was used to observe the open-circuit potential (OCP) of the Lix CoO2 electrode during each fully-charged step. Fig. 9 shows an increase in OCP with cycle number, and there is a close relation to the lithium content (x) as well. One can see, from Fig. 10, that the smaller the x-value, the higher the OCP, in other words, the lithium concentration in Lix CoO2 decreases with increased cycling. One hypothesis is that as lithium ions slowly deposit onto the MCMB anode surface during cycling to form a passivation film or other lithium compounds, this irreversible loss not only diminishes the anode thermal stability, but also the cathode’s. From the point of view of material science, safety characteristics depend intrinsically on the stoichiometry of metal oxide electrodes and carbon electrodes. According to the thermal gravity analysis (TGA) by Dahn et al. [23], in the absence of electrolyte, small x in Lix CoO2 would induce more decomposition to release oxygen. From DSC data (Fig. 11), the Lix CoO2 electrode releases more heat as its OCP increases (x-value decrease) and the thermal stability deteriorates rapidly. From relation in Fig. 11, a decreasing trend can be seen in decomposition temperature with increasing potential. All results lead to the same conclusion that, cycled LiCoO2 electrode has a decreased decomposition temperature and hence decreased safety; this might as well be another important contribution to why cycled batteries with PP separator could not pass nail safety tests. 3.6. Effects of different upper voltages on battery safety Both the cathode and anode of a battery change drastically after cycling, yet different upper voltages during charging

Fig. 10. Relationship between open-circuit potential and lithium content (x) in a Lix CoO2 electrode.

M.-S. Wu et al. / Electrochimica Acta 49 (2004) 1803–1812

1811

Fig. 11. Behavior of the enthalpy and decomposition temperature of a battery at different OCP for the Lix CoO2 electrode.

Table 2 Soft-nail test results for advanced lithium-ion batteries with different upper voltages Cycles

10 200

Upper voltage (V) 3.8

4.0

4.2

4.4

5 5

5 5

5 2

1 1

Values refer to the number of samples passed safety; five samples are tested for each upper voltage.

generate different safety results (soft-nail testing). Results are shown in Table 2. Batteries after 200 cycles fail at 4.2 V, while it requires at least 4.4 V to have the same results on 10-cycle batteries. Plausibly 4.4 V is already in the state of overcharge for a standard lithium-ion battery where deposition of lithium ions is thick on anode, as well that thermal stability of cathode is reduced by loss of lithium, the battery fail the nail test. So, in case of an inefficient shutdown separator, overcharge may decrease a battery’s safety. Although Lix CoO2 can be charged to a very low x-value and still obtain a high cell capacity, results suggested that this would lead to a significant penalty in overall cell safety [14]. It is, therefore, a necessity to increase a battery’s cycle life performance and safety by having well-designed electrodes and charging parameters to reduce irreversible loss of lithium ions.

short-circuit, because the former (soft-nail test) induces an enormous heat instantaneously and locally to cause thermal runaway of the electrolyte and electrode materials. Experiments show that all separators provide a distinct shutdown except PP. This might not be as important in fresh batteries (cycle number about 10), but for batteries more than 200 cycles, usage of PP may result in fire because thermal stability of electrodes have changed after cycling, while the separator has no preventing mechanism. When a three-electrode system is analyzed with ac impedance, anode is found to be accumulating a passivation film on cycling; and this leads to a decreasing stability of anode, as shown from the DSC results. The exothermic amount is about 53.8 J g−1 for 10 cycles, and 95.5 J g−1 for 200 cycles. Changes in the cathode OCP reveals a decreasing amount of lithium content after cycling, and a decreasing thermal stability. DSC data for a 10-cycle cathode is about 60 J g−1 , whereas the value reaches as high as 102 J g−1 for 200-cycle. On the other hand, decomposition temperature for Lix CoO2 has a pronounced decrease with decreasing lithium ion content (x-value). It is, therefore, believed that the main reason for failing nail penetration safety test is the irreversible loss of lithium-ion during cycling. Lastly, the decline of thermal stability on overcharge is due to the deposition of lithium ions from cathode onto the anode surface as lithium compounds.

References 4. Conclusions Safety in soft-nail penetration test of commercialized lithium-ion battery with laminated aluminum package not only is related to the shutdown function of separator but also to the cycle number. For short-circuit safety testing, an internal short is much more dangerous than an external

[1] S. Tobishma, J. Yamaki, J. Power Sources 81–82 (1999) 882. [2] S. Tobishma, K. Takei, Y. Sakurai, J. Yamaki, J. Power Sources 90 (2000) 188. [3] S. Tobishma, Y. Sakurai, J. Yamaki, J. Power Sources 68 (1997) 455. [4] P. Dan, E. Mengeritski, Y. Geronov, D. Aurbach, I. Weisman, J. Power Sources 54 (1995) 143. [5] K. Kitoh, H. Nemoto, J. Power Sources 81–82 (1999) 887.

1812

M.-S. Wu et al. / Electrochimica Acta 49 (2004) 1803–1812

[6] J. Cho, Electrochim. Acta 48 (2003) 2807. [7] S. Tobishma, K. Hayashi, K. Saito, T. Shodai, J. Yamaki, Electrochim. Acta 42 (1997) 119. [8] A. Du Pasquier, F. Disma, T. Bowmer, A.S. Gozdz, G. Amatucci, J.-M. Tarascon, J. Electrochem. Soc. 145 (1998) 472. [9] H. Maleki, G. Deng, I. Kerzhner-Haller, A. Anani, J. Howard, J. Electrochem. Soc. 147 (2000) 4470. [10] K. Edström, A.M. Andersson, A. Bishop, L. Fransson, J. Lindgren, A. Hussenius, J. Power Sources 97–98 (2001) 87. [11] M.N. Richard, J.R. Dahn, J. Electrochem. Soc. 146 (1999) 2068. [12] M.N. Richard, J.R. Dahn, J. Electrochem. Soc. 146 (1999) 2078. [13] H. Maleki, G. Deng, A. Anani, J. Howard, J. Electrochem. Soc. 146 (1999) 3224. [14] Z. Zhang, D. Fouchard, J.R. Rea, J. Power Sources 70 (1998) 16. [15] Y. Xia, T. Fujieda, K. Tatsumi, P.P. Prosini, T. Sakai, J. Power Sources 92 (2001) 234.

[16] D.D. MacNeil, L. Christensen, J. Landucci, J.M. Paulsen, J.R. Dahn, J. Electrochem. Soc. 147 (2000) 970. [17] D.D. MacNeil, T.D. Hatchard, J.R. Dahn, J. Electrochem. Soc. 148 (2001) A663. [18] R.A. Leising, M.J. Palazzo, E.S. Takeuchi, K.J. Takeuchi, J. Electrochem. Soc. 148 (2001) A838. [19] G. Venugopal, J. Moore, J. Howard, S. Pendalwar, J. Power Sources 77 (1999) 34. [20] D. Aurbach, J. Power Sources 89 (2000) 206. [21] Ph. Biensan, B. Simon, J.P. Pérès, A. de Guibert, M. Broussely, J.M. Bodet, F. Perton, J. Power Sources 81–82 (1999) 906. [22] J. Yamaki, H. Takatsuji, T. Kawamura, M. Egashira, Solid State Ion. 148 (2002) 241. [23] J.R. Dahn, E.W. Fuller, M. Obrovac, U. von Sacken, Solid State Ion. 69 (1994) 265. [24] W. Li, J.C. Currie, J. Wolstenholme, J. Power Sources 68 (1997) 565.