Synthesis of highly crystalline spinel LiMn2O4 by a soft chemical route and its electrochemical performance

Electrochimica Acta 52 (2007) 4525–4531

Synthesis of highly crystalline spinel LiMn2O4 by a soft chemical route and its electrochemical performance Jia-Yan Luo a , Xi-Li Li b , Yong-Yao Xia a,∗ a

Chemistry Department and Shanghai Key Laboratory of Molecular Catalysis and Innovative Materials, Fudan University, Shanghai 200433, China b Gejiu Superhoo Industrial Co., Ltd., Gejiu 661000, China Received 10 October 2006; received in revised form 19 December 2006; accepted 19 December 2006 Available online 21 January 2007

Abstract Highly crystalline spinel LiMn2 O4 was successfully synthesized by annealing lithiated MnO2 at a relative low temperature of 600 ◦ C, in which the lithiated MnO2 was prepared by chemical lithiation of the electrolytic manganese dioxide (EMD) and LiI. The LiI/MnO2 ratio and the annealing temperature were optimized to obtain the pure phase LiMn2 O4 . With the LiI/MnO2 molar ratio of 0.75, and annealing temperature of 600 ◦ C, the resulting compounds showed a high initial discharge capacity of 127 mAh g−1 at a current rate of 40 mAh g−1 . Moreover, it exhibited excellent cycling and high rate capability, maintaining 90% of its initial capacity after 100 charge–discharge cycles, at a discharge rate of 5 C, it kept more than 85% of the reversible capacity compared with that of 0.1 C. © 2007 Elsevier Ltd. All rights reserved. Keywords: LiMn2 O4 ; Lithiated manganese dioxide; Rechargeable lithium batteries; Cathode material; Highly crystalline

1. Introduction The cathode materials for lithium-ion batteries are usually oxides transition metal due to their high electrochemical potentials during the highly reversible lithium insertion/deinsertion. It has been demonstrated that spinel structure lithium manganese oxides are the most promising cathodes because they are much lower in cost, richer in natural sources, more permissible in an environmental standard [1–8]. The details of the crystallinity, crystal structure, the deviation from theoretical stoichimetry of elemental composition, the grain size and grain size distribution, all play important and eventually decisive roles in the electrochemical performance of the LiMn2 O4 spinel. Most of these parameters are decided by, or at least influenced by the synthesis process. LiMn2 O4 was typically obtained by reaction of a mixture of lithium salt (e.g., Li2 CO3 ) and manganese oxides at around 800 ◦ C in air for many hours. The high temperature solidstate synthesis process could achieve a high crystalline spinel LiMn2 O4 , but suffers from a problem in controlling the stoichimetry, especially causing an oxygen deficiency, which was

∗

Corresponding author. Tel.: +86 21 55664177; fax: +86 21 55664177. E-mail address: [email protected] (Y.-Y. Xia).

0013-4686/$ – see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2006.12.049

accompanied with a faster capacity fading during cycling [9–12]. On other hand, it has been demonstrated that the high crystallinity would improve the stability of crystallographic structure during charge–discharge cycles and thus enhance the cycling reversibility of the LiMn2 O4 [13–15]. In recent years, several low-temperature preparation techniques such as sol–gel precipitation [16–18], Pechini process [19,20], electrochemical process [21,22], aqueous reduction method [23] and hydrothermal process [24] template method [25] have been developed in which all the components can been homogeneously distributed to the atomic scale, thus allowing a reduction of heating temperature and sintering time. In spite of that, most of resulting materials show either low discharge capacity or large capacity fading due to the presence of impurity phases and low crystallinity, and further heating treatments involving temperature as high as those used in ceramic route are always needed to enhance their electrochemical performance. Therefore, it is of great importance, but a challenge, to explore an alternative method, which could at a relative low temperature prepare well-crystalline LiMn2 O4 spinel single phase having excellent electrochemical performance. Pistoia et al. [26,27] have reported that various MnO2 polymorphs can be lithiated with LiI solutions to give Lix MnO2 , which would transform to spinel upon heating. In the present work, we explored the optimal conditions to obtain highly crys-

4526

J.-Y. Luo et al. / Electrochimica Acta 52 (2007) 4525–4531

talline LiMn2 O4 spinel and employed thermo-gravimeter (TG), X-ray diffraction (XRD) and scanning electronic microscope (SEM) to characterize its composition, structure and morphology. Moreover, the relationships among the morphology, crystal structure and electrochemical performance were discussed in detail. 2. Experimental Lithiated manganese oxide, Lix MnO2, was prepared by chemical lithiation of the electrolytic manganese oxide (EMD, Xiangtan Chemical Co.) and a LiI solution in acetonitrile. The preparation procedure can be simply described as follows: 0.1675 g LiI was dissolved in 25 mL of acetonitrile and then stirred at room temperature until the solution became clear. Different amounts of EMD (with molar ratios of LiI to EMD ranging from 0.5:1 to 1:1) were then added into the above solution. The suspension was heated to 70 ◦ C under reflux for 24 h, followed by centrifugation. The resulting product was washed with acetonitrile several times, dried under vacuum at 200 ◦ C for 1 day and then annealed at the temperature range of 200–800 ◦ C for 2 h at a heating rate of 2 ◦ C min−1 in air. For comparation, the LiMn2 O4 was prepared by heating a stoichiometric mixture of LiOH and EMD with a Li/Mn molar ratio of 0.5 at 600 ◦ C for 12 h in air. XRD measurements were performed using a Rigaku D/MAX-IIA X-ray diffractometer using Cu K␣ radiation. SEM images were obtained on Philip XL30 operated at 20 kV. The thermo-gravimetric (TG) analysis was carried out on a PerkinElmer TGA 7 thermal analyzer. A CR2016 coin type cell was fabricated to test its electrochemical performance with metallic lithium as counter electrode. The working electrode was prepared by compressing a mixture of active materials (LiMn2 O4 ), conductive material (acetylene black, AB) and binder (polytetrafluoroethylene, PTFE) in a weight ratio of LiMn2 O4 /AB/PTFE = 16:3:1 onto an aluminum grid at 10 MPa. The electrode was punched in disk form with a typical diameter of 12 mm and dried at 120 ◦ C for 12 h before assembly. Typical mass loading of the active material was about 10 mg cm−2 . The cell was assembled in a glovebox filled with pure argon. The electrolyte solution was 1 M LiPF6 /ethylene carbonate (EC)/dimethyl carbonate (DMC)/ethyl methyl carbonate (EMC) (1:1:1, v/v). The cell was galvanostatically cycled between 3 and 4.3 V versus Li/Li+ at various current densities at 25 ◦ C. Lithium insertion into LiMn2 O4 electrode was referred to as discharge and extraction as charge. The cell capacity was determined by only the weight of the positive active material. Cyclic voltammograms (CV) were characterized using a three-electrode cell; the metallic lithium was used both as a counter and reference electrodes. The experiments were performed using a Solartron Instrument Model 1287 electrochemical interface at a scanning rate of 0.1 mV s−1 . 3. Results and discussion Firstly, the formation mechanism for spinel was studied by thermo-gravimeter (TG), and differential thermal analy-

Fig. 1. TG and DTA curves for lithiated MnO2 recorded over the temperature range from ambient to 900 ◦ C at a heating rate of 10 ◦ C min−1 in air at 100 ml min−1 flow rate.

sis (DTA). The DTA and TG traces for the transformation of Lix Mn2 O4 from the lithiated Lix MnO2 in air are displayed in Fig. 1. The weight loss until 200 ◦ C was due to the loss of the adsorbed water. From the TG and DTA traces, and the transformation of the defected Lix Mn2 O4+y from Lix MnO2 (if x = 0.5) may occur from 200 to 500 ◦ C. The reaction can be described as follows: 2Li0.5 MnO2 (orthorhombic) + 0.5yO2 = LiMn2 O4+y (cubic) The transformation from a defect to a well-ordered spinel proceeds via a loss of oxygen was occurred at 500 ◦ C. The following reaction clearly corresponds to an exothermic peak on DTA curve: LiMn2 O4+y (defect) → LiMn2 O4 (well-ordered) + 0.5yO2 Weight loss over 800 ◦ C was also observed on TG curve. That was also accompanied by the formation of oxygen deficient spinel: LiMn2 O4 (cubic) → LiMn2 O4−y (tetragonal) + y/2O2 All above results suggest that optimum synthesis conditions for the transformation of well-ordered spinel from lithiated Lix MnO2 was occurred performed between 500 and 800 ◦ C. Fig. 2 gives the XRD pattern of EMD, lithiated MnO2 heattreated at different temperatures. Fig. 2a reveals that ␥-MnO2 is relative low degree of crystallinity with broad, low-density peaks, which is a common feather of commercial EMD materials widely used in primary batteries. Fig. 2b and c shows the XRD patterns of the lithiated EMD with Li/Mn = 0.75 after heat treatment at 200 and 400 ◦ C, respectively. Both curves are similar to that of the above mentioned pristine EMD, but most peaks continuously shift toward a lower diffraction angle and the intensity increases correspondingly. This feather confirms that lithium intercalation was occurred in the EMD in a topotactic manner without much distribution in its structure [27]. In addition, diffraction peaks from spinel LiMn2 O4 can also been found in Fig. 2c, suggesting that LiMn2 O4 was initially formed by heat treatment of lithiated EMD at 400 ◦ C. Upon further heating treatment at 600 and 800 ◦ C, all the diffraction peaks could be

J.-Y. Luo et al. / Electrochimica Acta 52 (2007) 4525–4531

4527

Table 1 The relationship between the reaction product compositions and the synthesis conditions

Fig. 2. Powder XRD Patterns of (a) pristine EMD; the reaction product after heating treatment of lithiated MnO2 with LiI:MnO2 = 0.75:1 (b) at 200 ◦ C, (c) at 400 ◦ C, (d) at 600 ◦ C and (e) at 800 ◦ C; the reaction product after heating treatment of lithiated MnO2 at 600 ◦ C (f) with LiI:MnO2 = 0.5:1 and (g) with LiI:MnO2 = 1:1, (h) solid-state reaction sample at 600 ◦ C for 12 h in air.

assigned to the well-known spinel phase (PDF no. 350782) with an Fd-3m space group. The spinel structure can be described as ideally consisting of a cubic-packing arrangement of oxygen ion at the 32e sites, the Li+ ions occupy the tetrahedral 8a sites, and the Mn3+ and Mn4+ ions are at the octahedral 16d sites [7,8]. Furthermore, the relative intensity of those reflections representing spinel structure was significantly increased from 600 to 800 ◦ C, which was directly linked to the better crystallinity or the larger crystal grain size. Fig. 2f and g shows that the XRD patterns of the lithiated EMD with Li/Mn = 0.5 and Li/Mn = 1.0 after heat treatment at 600 ◦ C, respectively. The relative intensity of those reflections representing spinel structure in Fig. 2f and g was much lower compared with that in Fig. 2c, which was heated at the same temperature. Moreover, Mn2 O3 and Li2 MnO3 diffraction peaks can be seen clearly from Fig. 2f and g, respectively, which means, that the optimizing ratio of LiI to MnO2 was 0.75:1 and either excess or less this ratio will lead to impurities. Fig. 2h shows the XRD patterns of solid-state sample with heating treatment at 600◦ C for 12 h in air, which was similar to Fig. 2f. The XRD peaks represented

Sample no.

Synthesis conditions

Compositions

A B

LiI/MnO2 = 0.75, 200 ◦ C, 2 h LiI/MnO2 = 0.75, 400 ◦ C, 2 h

C D E F G

LiI/MnO2 = 0.75, 600 ◦ C, 2 h LiI/MnO2 = 0.75, 800 ◦ C, 2 h LiI/MnO2 = 0.50, 600 ◦ C, 2 h LiI/MnO2 = 1.0, 600 ◦ C, 2 h LiOH/MnO2 = 0.50, 600 ◦ C, 12 h

Intermediate compound Intermediate compound + LiMn2 O4 LiMn2 O4 LiMn2 O4 LiMn2 O4 + Mn2 O3 LiMn2 O4 + Li2 MnO3 LiMn2 O4 + Mn2 O3 + Li2 O

LiMn2 O4 spinel structure, however, the relative intensity was much lower compared with that in Fig. 2c and Mn2 O3 impurities can be seen more clearly indicating the incompleteness of solid-state reaction at 600 ◦ C for 12 h. The relationship between the reaction product compositions and the synthesis conditions are summarized in Table 1 and hereafter the samples prepared under various synthesis conditions are denoted as the samples A–G. Fig. 3 presents the morphology of the products prepared by the low-temperature, soft-chemistry method and in comparison with the sample of solid-state reaction. As can be seen from Fig. 3, samples A–C, were composed of nanosized particles of 60–100 nm with grains stack together. The particle size increases with increasing of the annealing temperature from 200 to 600 ◦ C. As shown in Fig. 3d, sample D has much larger size particles with planar-lamination shape with 300–400 nm in size. As compared with samples A and B, samples C and D have a well-defined crystalline structure and the crystal shape indicates that the LiMn2 O4 particle is close to single crystal, especially for sample D. However, much lower crystallinity can be seen clearly in Fig. 3e for the solid-state reaction sample G. The SEM results are consistent with the XRD patterns in Fig. 2. The low-temperature, soft-chemistry method is a promising way to prepare highly crystalline nanostructured LiMn2 O4 spinel pure phase at a relative low temperature of 600 ◦ C. Fresh electrodes composed of as-synthesized samples A–D possessing a similar amount of active masses were prepared and characterized by cyclic voltammograms (CV) with the potential range between 3.0 and 4.5 V versus Li/Li+ at a scan rate of 0.1 mV s−1 . As can be seen from Fig. 4A, only one pair of redox peaks around 3.5 V was observed. This feature was anticipated for an intermediate compound with an intergrown structure. Fig. 4B shows two pairs of prominently separated redox peaks which located at around the potentials of (3.24 and 3.87 V) and (4.09 and 4.12 V), respectively, that reflected the coexistence of the intermediate compound and spinel phase. The typical CV curves of well defined spinel LiMn2 O4 can be seen in Fig. 4C and D. The two sets of close anodic peaks around 4.0 and 4.1 V represent the reversible electrochemical deinsertion/insertion of Li+ from the tetrahedral sites of LiMn2 O4 , which occur in two stages. The first peak at about 4.0 V is ascribed to the removal of Li+ from half of the tetrahedral site in which Li–Li interactions exist, whereas the second peaks at around 4.1 V is attributed to

4528

J.-Y. Luo et al. / Electrochimica Acta 52 (2007) 4525–4531

Fig. 3. SEM photographs of: (a) sample A, (b) sample B, (c) sample C, (d) sample D and (e) sample G.

the removal of Li+ from the other tetrahedral sites where Li+ do not interact. Compared with CV data of samples C and D, the peaks of sample C are sharper, indicating the higher activity of the nanostructure LiMn2 O4 . The specific capacity of the products was determined by galvanostatic charge–discharge measurements at a current density of 40 mA g−1 between cutoff voltage of 3.0 and 4.3 V. From Fig. 5, sample A only showed one voltage plateaus at the cutoff voltage, which is well consistent with the CV results. Sample B has two prominently separated voltage plateaus again reflecting the coexistence of the intermediate compound and spinel phase. The discharge curves of the samples C and D both present two plateaus, one at 4.1 V and the other at 3.95 V. The appearance of the former plateau may because of the coexistence of ␭MnO2 –LiMn2 O4 , while the latter is due to the coexistence of two pseudophases in the form of Li0.5 Mn2 O4 –LiMn2 O4 . Sample C, however, displayed higher discharge capacity than that of sample D at the same current density. A slight increasing of capacity of sample C may contribute to that the Li+ –Li+ Coulomb repulsion which is lower at the surface of nanomaterial thus enhancing local capacity as there is no neighboring Li+ outside the particle. The above results indicate that both crystallinity and grain

size played important roles in the electrochemical performance of electrode materials. As can been seen in Fig. 6, samples E and F deliver discharge capacity of 95 and 74 mAh g−1 , respectively, both are much lower than that of sample C at the same current density especially compared with sample F. This is in accordance with the previous results [26] that the presence of by-products of Mn2 O3 and/or Li2 MnO3 in the spinel limited the available capacity. So based on our experiment results, the optimizing ratio of LiI to MnO2 is 0.75:1 and the optimizing annealing temperature of the lithiated samples is 600 ◦ C. At these optimizing conditions, highly crystalline nanostructured LiMn2 O4 spinel with homogeneous crystal grains and high purity could be obtained. This highly crystalline nanostructured LiMn2 O4 spinel delivered a high initial discharge capacity of 127 mAh g−1 at a current density of 40 mA g−1 between 3.0 and 4.3 V, which doubled that of the solid-state reaction sample G, as shown in Fig. 6. The cycling performance of the samples was given in Fig. 7. The cells were cycled between 3.0 and 4.3 V at a current density of 40 mA g−1 at room temperature. The discharge capacity fading of sample A is much faster than that of the others which was due to residual water observed by TG. The residual water would

J.-Y. Luo et al. / Electrochimica Acta 52 (2007) 4525–4531

4529

Fig. 4. Cycling voltammogram over the potential 3.0–4.5 V of: samples A–D.

strongly impair the cycle performance and/or during cycling, the unit cell of the incomplete grown intermediate compound crystallites will distort, which lead to rapid capacity decreasing. For spinels, sample C has the highest initial capacity and the best capacity retention. It can been seen from Fig. 4 that

sample C shows an average capacity loss of 0.13 mAh g−1 per cycle, in other words, after 100 charge–discharge cycles, the discharge capacity of sample C maintain 90% of its initial capacity, which was much higher than most of the reported results for nanosized LiMn2 O4 [20–25]. The highly crystalline nanos-

Fig. 5. Typical discharge curves of samples A–D between 3.0 and 4.3 V at a current rate of 40 mAh g−1 .

Fig. 6. . Typical discharge curves of samples E–G, between 3.0 and 4.3 V at a current rate of 40 mAh g−1 .

4530

J.-Y. Luo et al. / Electrochimica Acta 52 (2007) 4525–4531

Fig. 7. Variation of discharge capacity vs. cycle number for samples A–D between 3.0 and 4.3 V at a current rate of 40 mAh g−1 .

0.1 C then, after some five cycles, the rate was increased in stages to 5 C. Specific capacities of around 129 mAh g−1 was obtained at the rate of 0.1 C for sample C then reduced to 125 mAh g−1 at 0.2 C, 122 mAh g−1 at 0.5 C, 120 mAh g−1 at 1 C, 119 mAh g−1 at 2 C, 117 mAh g−1 at 3 C, 114 mAh g−1 at 4 C and finally, 112 mAh g−1 at 5 C. In particular, more than 85% of the reversible capacity for sample C at 0.1 C rate can be discharged at 5 C rate compared to 60% for the commercial LiMn2 O4 . This rate capability is much higher than the rate capability for the commercial spinel LiMn2 O4 , as well as most of the reported results for the spinels synthesized by a sol–gel precipitation, Pechini process, electrochemical process and template method [16–25]. The low-temperature, soft-chemistry method we discussed in this report, we provide a promising way to synthesize high rate spinel LiMn2 O4 , which is vital for future EV application. 4. Conclusion

tructured LiMn2 O4 spinel pure phase prepared by this method exhibited excellent long-term cycling performance which could hardly be achieved though most of the other methods. Compared with the discharge curves of samples C and D, as the annealing temperature increased from 600 to 800 ◦ C, the degradation of cycling stability was observed because higher temperature (800 ◦ C) treatment was very likely to create an oxygen deficiency [10], which was widely acknowledged to accelerate the capacity fading of the spinel LiMn2 O4 . This is in a good agreement with the systematic studies on the relationship between the capacity fading and the oxygen deficiency reported by Xia and Yoshio [8,9]. Besides the cycling performance, the rate capability especially the high rate discharge performance is another important factor for the application of Mn-based spinel cathodes in EV/HEV power sources. Rates of up to 5 C (where C corresponds to complete discharge in 1 h) of sample C and the commercial LiMn2 O4 have been investigated and the results are shown in Fig. 8. The cell was first cycled at

We proposed a promising way to prepare well-crystalline nanostructured LiMn2 O4 spinel pure phase at a relative low temperature of 600 ◦ C through a low-temperature, soft-chemistry method. Both the LiI/MnO2 ratio and the annealing temperature played critical roles in the formation of preferable spinel LiMn2 O4 . Electrochemical measurement showed that the assynthesized LiMn2 O4 spinel delivered a high initial discharge capacity of 127 mAh g−1 at a current rate of 40 mAh g−1 , it exhibited excellent cycling ability and high rate capability, maintaining 90% of its initial capacity after 100 charge–discharge cycles, even discharge at 5 C, it kept more than 85% of the reversible capacity at 0.1 C discharge rate. Moreover, the soft chemical method described in the present work could be used to develop other nanostructured LiMn2 O4 such as nanorod, nanowire, nanobelt, nanotube, urchin-like nanostructure, hollow spheres and so on. Acknowledgments This work was partially supported by the National Natural Science Foundation of China (No. 20633040) and the program of New Century Excellent Talent in University of China (2005). References [1] [2] [3] [4] [5] [6] [7] [8] [9]

Fig. 8. Variation in discharge capacity vs. cycle number for sample C and commercial LiMn2 O4 cycled at 0.1, 0.2, 0.5, 1, 2, 3, 4 and 5 C, between 3.0 and 4.3 V.

[10] [11] [12]

J. Kim, A. Manthiram, Nature 390 (1997) 265. J.M. Tarascon, M. Armand, Nature 414 (2001) 359. T. Tanaka, K. Ohta, N. Arai, J. Power Sources 2 (2001) 97. R. Hamlen, G. Au, M. Brundage, M. Hendricks, E. Plichta, S. Slane, J. Barbarello, J. Power Sources 22 (2002) 95. R.A. Marsh, S. Vukson, S. Surampudi, B.V. Ratnakumar, M.C. Smart, M. Manzo, P.J. Dalton, J. Power Sources 25 (2001) 97. B. Deng, H. Nakamura, M. Yoshio, J. Power Sources 141 (2005) 116. Y. Xia, Y. Zhou, M. Yoshio, J. Electrochem. Soc. 144 (1997) 2593. Y. Xia, M. Yoshio, J. Electrochem. Soc. 143 (1996) 825. Y. Xia, T. Sakai, T. Fujieda, X. Yang, X. Sun, Z. Ma, J. McBreen, M. Yoshio, J. Electrochem. Soc. 148 (2001) 723. J.M. Tarascon, D. Guyomard, Electrochim. Acta 38 (1993) 1221. J.C. Hunter, J. Solid State Chem. 39 (1981) 142. T. Ohzuku, M. Kitagawa, T. Hirai, J. Electrochem. Soc. 137 (1990) 769.

J.-Y. Luo et al. / Electrochimica Acta 52 (2007) 4525–4531 [13] K. Kanamura, K. Dokko, T. Kaizawa, J. Elctrochem. Soc. 152 (2005) A391. [14] A. Blyr, C. Sigala, G. Amatucci, D. Guymard, Y. Chabre, J.M. Tarascon, J. Electrochem. Soc. 145 (1998) 194. [15] Y. Gao, J.R. Dahn, J. Electrochem. Soc. 143 (1996) 100. [16] S. Bach, M. Henry, N. Baffier, J. Livage, J. Solid State Chem. 88 (1990) 325. [17] J.P. Pereira-Ramos, J. Power Sources 54 (1995) 120. [18] J.H. Choy, D.H. Kim, C.W. Kwon, S.J. Hwang, Y.I. Kim, J. Power Sources 77 (1999) 1. [19] P. Barboux, J.M. Tarascon, F.K. Shokoohi, J. Solid State Chem. 94 (1991) 185. [20] W. Liu, G.C. Farrington, F. Chaput, B. Dunn, J. Electrochem. Soc. 143 (1996) 879.

4531

[21] L.I. Hill, R. Portal, A. Verbaere, D. Guyomard, Electrochem. Solid-State Lett. 4 (2001) A180. [22] L.I. Hill, R. Portal, A.L.G.L. Salle, A. Verbaere, D. Guyomard, Electrochem. Solid-State Lett. 4 (2001) D1. [23] V.G. Kumar, J.S. Gnanaraj, S.B. David, D.M. Pickup, E.H. van-Eck, A. Gedanken, D. Aurbach, Chem. Mater. 15 (2003) 4211. [24] Y. Liang, S. Bao, B. He, W. Zhou, H. Li, J. Electrochem. Soc. 152 (2005) A2030. [25] X. Li, F. Cheng, B. Guo, J. Chen, J. Phys. Chem. B 109 (2005) 14017. [26] G. Pistoia, A. Antonini, D. Zane, M. Pasquali, J. Power Sources 56 (1995) 37. [27] M.M. Thackray, M.H. Rossouw, R.J. Gummow, D.C. Liles, K. Pearce, A. De-Kock, W.I.F. David, S. Hull, Electrochim. Acta 38 (1993) 1259.