Cyclic voltammetry studies on 4 V and 5 V plateaus of non-stoichiometric spinel Li_1+xMn_2-yO₄

BAI Ying(白 莹)^{1, 2}, WU Chuan(吴 川)^{1, 2}, WU Feng(吴 锋)^{1, 2}, WANG Guo-qing(王国庆)^{1, 2}

1. School of Chemical Engineering and Environment, Beijing Institute of Technology, Beijing 100081, China;

2. National Development Center for High Technology Green Materials, Beijing 100081, China

Received 23 August 2005; accepted 27 December 2005

Abstract: Li_1+xMn_2-yO₄ spinels with various Li/Mn ratios were synthesized by a solid-state reaction. By X-ray diffraction analysis, Li₂MnO₃ was detected as a second phase with increasing the Li/Mn ratio; and the role of Li₂MnO₃ in Li_1+xMn_2-yO₄ spinel was discussed. A slow scanning cyclic voltammetry(CV) at the rate of 0.1 mV/s was adopted to characterize the evolutions of 4 V and 5 V plateaus of Li_1+xMn_2-yO₄ spinels. An additional Li⁺ insertion in 4 V region was observed in both Li-lack and Li-rich spinels at 3.95 V, which is different from the general Li⁺ insertion with weak Li-Li interaction and strong Li-Li interaction; and this plateau disappeared in the subsequent cycles. The 4.4 V/3.8 V plateaus correspondent to Li⁺ insertion and extraction of Li₂MnO₃ were discussed, and these plateaus have a high reversibility with cycling. The 5 V plateau was found only in the Li-rich samples, and this plateau has a tendency to emerge at higher voltage region with increasing Li/Mn ratio.

Key words: lithium manganese oxides; spinel; LiMn₂O₄; Li₂MnO₃; cyclic voltammetry

1 Introduction

As one of the most promising cathode materials for lithium ion batteries, spinel LiMn₂O₄ has received much attention in recent years. This material has reversible capacities at both 3 V and 4 V plateaus[1]. However, the Li⁺ insertion and extraction at 3 V regions are two phase reactions representing the structural evolution of LiMn₂O₄ to Li₂Mn₂O₄, and Li₂Mn₂O₄ back to LiMn₂O₄. This evolution is correspondent to a phase transition between cubic symmetry and tetragonal symmetry, causing a Jahn-Teller effect in the lattice, and leading to a serious capacity fading with cycling[2]. Therefore, most of the studies on spinel LiMn₂O₄ were concerning with the structures and the electrochemical performances in 4 V regions. The typical 4 V plateaus of LiMn₂O₄ are 4.0 V and 4.1 V plateaus. Since the structures of spinel Li-Mn-O compounds are sensitive to the synthesis conditions, Li/Mn ratios, and cation or anion dopants, some other plateaus such as 3.3 V/3.95 V and 4.5 V plateaus have been observed in some LiMn₂O₄-based spinels[3-7].

More recently, many groups pay more and more attentions to the transition metal substituted spinel LiMn₂O₄, which exhibit high voltage plateaus at around 5 V[8-11]. Therefore, the studies on the structural and electrochemical properties of LiMn₂O₄ and its derivates at 5 V region are of interest. Actually, the 5 V plateaus were observed in the LiMn₂O₄ spinels synthesized by melt-impregnation method[12], electron-beam evapora- tion or radio-frequency sputtering[13].

In this study, a slow scanning cyclic voltammetry is employed to analysis the evolutions of 4 V and 5 V plateaus in a series of Li_1+xMn_2-yO₄ spinels with -0.03≤x≤0.27 and 2.12≥2-y≥1.75, namely, with the Li/Mn ratio from 0.46 to 0.72. The changes of spinel structure, average Mn valence, and Li⁺ insertion and extraction with increasing Li/Mn ratio are discussed.

2 Experimental

2.1 Material synthesis and characterization

Li_1+xMn_2-yO₄ spinels were synthesized by a solid-state reaction as described previously[14]. Lithium hydroxide (LiOH) and chemical manganese dioxide (MnO₂) served as the raw materials. On preparing Li-doped spinels, raw materials with various nominal Li/Mn ratios (0.45 to 0.70) were mixed and ground carefully, then preheated at 480 ℃ for 6 to 8 h. Subsequently, they were reground and pressed into pellets, sintered at 750 ℃ for 24 h before being cooled down for about 12 h in air. The Li_1+xMn_2-yO₄ spinels with various Li/Mn ratios were obtained by this way.

The real stoichiometry of lithium and manganese was determined on an ICP spectrometer (IRIS/AP, Thermao Jarrell Ash), while the oxygen stoichiometry was estimated from those of Li and Mn. Powder XRD patterns were recorded with a Rigaku B/max-2400 X-ray diffractometer with Cu K_a radiation at 40 kV and 120 mA.

2.2 Cyclic voltammetry investigation

The cyclic voltammograms of the Li_1+xMn_2-yO₄ spinels were investigated in typical Swagelok two-electrode cells, which adopt metal lithium foils as counter electrodes, and 1mol/L LiPF₆ in a 1∶1 (volume ratio) mixture of ethylene carbonate(EC) and dimethyl carbonate(DMC) as electrolyte. A sheet of Celgard^? 2300 was used as a separator. The cathodes were prepared by coating slurries of Li_1+xMn_2-yO₄ powder, carbon black and cyclopentanone-dissolved poly- vinylidene fluoride(PVDF) onto aluminum foils. Subsequently, the films were vacuum-dried at 55 ℃ for 24 h, compressed at 1 MPa between two stainless steel plates, and cut into sheets with an area of 0.5 cm². The resultant mass ratio of Li_1+xMn_2-yO₄, carbon black, and PVDF was 85∶10∶5. All the tested cells were assembled in an argon-filled glove box (MBRAUN).

After a 12 h aging, the as-assembled cells were investigated. The cyclic voltammograms were performed on a CHI 660A electrochemical workstation in the voltage range from 3.0 to 5.2 V at a scanning rate of 0.1 mV/s.

3 Results and discussion

3.1 Structural characterization

Fig.1 shows the X-ray diffraction patterns of Li_1+xMn_2-yO₄ spinels. Though all the samples show a spinel phase with the structure of  space group, the emergency of Li₂MnO₃ is observed with increasing Li/Mn ratio. According to the Li-Mn-O and Li₂MnO₃-MnO-l-MnO₂ trinary phase diagrams[2], Li-substituted spinel LiMn₂O₄ has a tendency to generate a stable rock-salt phased Li₂MnO₃. In this study, when the Li/Mn ratio is 0.61, a new diffraction peak implying rock-salt Li₂MnO₃ emerges around the (311) diffraction of LiMn₂O₄; and the shoulder of (400) diffraction shows asymmetry, as shown in Fig.2. When the Li/Mn ratio is increased to 0.67, the (400) shoulder grows evidently. As the Li/Mn ratio is further increased to 0.72, the diffraction peaks of Li₂MnO₃ are distinct. The second phase in a spinel cathode may have effects on the cation valences[15], and Li₂MnO₃ here in Li_1+xMn_2-yO₄ spinels doesn’t make an exception.

Fig.1 Powder XRD patterns of Li_1+xMn_2-yO₄ spinels synthesized by solid state reaction: (a) Sample H09, Li/Mn ratio=0.46; (b) Sample H10, Li/Mn ratio=0.50; (c) Sample H11, Li/Mn ratio=0.54; (d) Sample H12, Li/Mn ratio=0.61; (e) Sample H13, Li/Mn ratio=0.67; (f) Sample H14, Li/Mn ratio=0.72.

Fig.2 Evolution of (311) and (400) diffraction peaks of Li_1+xMn_2-yO₄ spinels, in which Li/Mn ratios are 0.54(a), 0.61(b), 0.67(c) and 0.72(d), respectively

The real stoichiometry and average Mn valences of the samples are listed in Table 1. The average Mn valences are calculated from the real stoichiometry of the samples based on charge balance. Obviously, the average Mn valences of the Li-rich samples are higher than 3.5, which is the critical of Jahn-Teller effect. The increase of the valence is due to two factors. On one hand, the excessive Li⁺ ions have inserted into 16d sites of MnO₆ octahedron to replace Mn³⁺ ions; therefore, some of the rest Mn³⁺ ions have to convert to Mn⁴⁺ ions in order to keep the charge balance. On the other hand, the growth of the second phase Li₂MnO₃, which has no Mn³⁺ ions but Mn⁴⁺ ions, increases the relative content of Mn⁴⁺ ions in manganese ion species, and upgrades the average Mn valence.

Table 1 Real stoichiometry and general structural parameters of Li_1+xMn_2-yO₄ spinels

The evolutions of lattice parameter with various Li/Mn ratios for Li_1+xMn_2-yO₄ spinels are shown in Fig.3. It is evident that the lattice of Li_1+xMn_2-yO₄ shrinks with increasing the Li/Mn ratio. The higher the Li/Mn ratio is, the higher the Mn⁴⁺ content is. Thus, the Mn—O bonds in spinel lattice are enhanced, which results in smaller lattice parameter. This phenomenon is correspondent with the average Mn valences listed in Table 1.

Fig.3 Lattice parameter vs Li/Mn ratio plot for Li_1+xMn_2-yO₄ spinels

3.2 Cyclic voltammogram

3.2.1 Li⁺ extraction in 4.05 V to 4.15 V plateaus

The insertion and extraction of lithium ions were investigated via a slow cyclic voltammetry, as shown in Fig.4 and Fig.5. When concerning with the 4V plateau of spinel LiMn₂O₄, the split phenomena of the redox peaks are generally interpreted as two-step insertion and extraction. Namely, during the charge process, the Li⁺ ions with strong Li-Li interaction are removed from half of the tetrahedral positions at lower potential, and some of the Mn³⁺ ions are converted to Mn⁴⁺ ions. As the content of Mn⁴⁺ increases, the second half of Li⁺ ions with weak Li-Li interaction is extracted at higher potential. The two binary equilibrium systems during Li extraction and insertion are considered LiMn₂O₄- Li_0.5Mn₂O₄ and Li_0.5Mn₂O₄-l-MnO₂[16]. Since the two steps for Li⁺ insertion or extraction can not be separated distinctly, the reductive peaks may overlap with each other in some case, as well as the oxidative peaks.

Here we found the overlap of 4.05 V and 4.15 V plateaus with each other (see Fig.4). In the initial cycle, the two plateaus in 4 V regions can be identified distinctly when the Li/Mn ratio is in the range of 0.46-0.54. However, these plateaus are gradually combined to one plateau when the Li/Mn ratio exceeds 0.61. This phenomenon may be related to the increase of average Mn valence. With increasing the Li/Mn ratio, the average Mn valence increases as well. This results in a high proportion of Mn⁴⁺ in 16d sites, enhancing the Mn—O bonds, and weakening the Li-O interaction. Thus, the Li⁺ ions with weak Li-Li interaction can also be extracted out of the spinel lattice at a relative low voltage region, in company with those having strong Li-Li interaction.

3.2.2 Li⁺ insertion in 4.1 V to 3.8 V plateaus

Even though all the cells have gone through a 12 h aging, there are still kinetic polarizations found in the initial cycling. In the initial cycle, the second reductive peak that implies the insertion of Li⁺ ions into spinel lattice, which should be located around 3.90 V (see Fig.5), is located in 3.80-3.85 V (see Fig.4).

Unlike the common Li⁺ insertion in LiMn₂O₄ spinel, an additional Li⁺ insertion plateau at 3.95V is found in this study, as shown in Figs.4(a) and (c). This plateau is different from the 3.95 V Li⁺ extraction observed by Palacín et al[7], which corresponds to a 3.3 V Li⁺ insertion step. Since there is no corresponding extraction plateau for the 3.95 V plateau observed here, it is reasonable to consider this plateau as a third kind of insertion, which differs from the Li⁺ insertion with weak Li-Li interaction and with strong Li-Li interaction in the 8a sites. For the Li-lack spinel H09, Li⁺ ions occupy less than 1/2 of the tetrahedral 8a sites; the 3.95 V plateau may be correspondent to the Li⁺ insertion to fill the 8a vacancies, until the Li⁺ occupation achieves 1/2.

Fig.4(c) shows that the slight Li-rich sample H11 also has a 3.95 V plateau. Since the 16c octahedral shares faces with the 8a tetrahedral, the excess Li⁺ ions may be crowded into the neighboring vacant 16c octahedral. In the spinel LiMn₂O₄ phase, the tetrahedral 8a sites are of lower energy than the octahedral 16c sites. Therefore, in the 4 V plateau region, some of the Li⁺ ions extracted from octahedral 16c sites are preferable to insert tetrahedral 8a sites, rather than inserting back to 16c octahedral sites; and the 3.95 V plateau may imply this Li⁺ insertion process.

Fig.4 Initial cyclic voltammograms of Li_1+xMn_2-yO₄ spinels at scanning rate of 0.1 mV/s in voltage range from 3.0 to 5.2 V

In a word, the 3.95 V plateau may represent a further Li⁺ insertion in tetrahedral 8a sites of the Li_1+xMn_2-yO₄ spinels, which have some phase defections. After the initial cycle, the spinel phases are activated to more perfect structures, and the 3.95 V plateau is no longer existent in the subsequent cycles, see Figs.5(a) and (c).

3.2.3 Additional 4.4 V insertion and 3.8 V extraction plateaus

With increasing the Li/Mn ratio, a pair of additional redox plateaus is observed at 4.4/3.8 V more and more evidently, as shown in Fig.5. These plateaus are highly reversible, but have much less capacities than other plateaus.

According to previous studies[3], the 4.5 V plateau has been considered as a result of the co-occupation of lithium and manganese ions in the tetrahedral 8a sites. However, from the analysis of powder X-ray diffraction and Rietveld refinement[3], GAO et al hypothesized oxygen defect as the inducement of the additional reduction process. Subsequently, Amatucci et al[5] and Tabuchi et al[6] reported respectively that the 4.5 and 3.3 V plateaus are relative to low average Mn valence, which leads to Jahn-Teller effect during cooling process. More recently, Palacín et al[7] detected that the origin of these peaks is the formation of double hexagonal-type layers in the spinel.

Fig.5 Second and third cyclic voltammograms of Li_1+xMn_2-yO₄ spinels at scanning rate of 0.1 mV/s in voltage range from 3.0 to 5.2 V

In this paper, the 4.4/3.8 V plateaus are also relative to phase transition. However, they are not correspondent to the transition of spinel phase, but that of the second phase, Li₂MnO₃. Since the manganese ions in rock-salt phased Li₂MnO₃ are at a high oxidation state, namely, Mn(Ⅳ), it is difficult for the spinels to get a charge compensation, which allows Li⁺ ions to be extracted out of the lattice. Kalyani et al[17] reported that only after several cycles, the typical discharge-charge curves of Li₂MnO₃ could be obtained. Hence, there is no plateau for Li⁺ extraction from the lattice of Li₂MnO₃ observed in the initial cycle, see Fig.4.

In the 4 V insertion plateau region, Li⁺ insertion of Li₂MnO₃ is prior to that of Li_1+xMn_2-yO₄ spinels, for the average Mn valence of the former is higher than that of the later, and has a priority to be reduced at 4.4 V, whereas the reduction of Li_1+xMn_2-yO₄ is around 4 V, see Figs.3(d), (e) and (f). In the subsequent cycles, the extraction plateau correspondent to the 4.4 V insertion plateau emerges at 3.8 V, and this pair of plateaus show a high reversibility, see Figs.4(d), (e) and (f). In addition, since the Li⁺ insertion of Li₂MnO₃ is located at 4.4 V, the effect of Li₂MnO₃ on the charge-discharge capacity is in dependence on the cut-off charging voltage of the cathode.

3.2.4 Li⁺ insertion and extraction at 5 V plateaus

Reversible 5 V plateaus are observed in Li-rich samples, see Figs.4(c)-(f) and Figs.5(c)-(f). When studying the capacity fading of 4 V Li/LiMn₂O₄ batteries, XIA et al[12] had found a 5 V plateau during cycling. They took this plateau for electrolyte decomposition or Mn(Ⅳ) oxidation to higher valence. In this paper, on one hand, the electrolyte adopted in the batteries is 1mol/L LiPF₆-EC/DMC (1∶1), which is stable to 5.2 V; on the other hand, samples H09 and H10, which are cycled under the same conditions as the other spinels, have no  5 V plateau. Therefore, the 5 V plateau here is not implying the decomposition of electrolyte, but the evolution of spinel structure with increasing Li/Mn ratio.

In the study of thin-film LiMn₂O₄ electrodes, Thackeray[13] discussed the 5 V plateaus observed with a small charge-discharge current of 0.2 mA. There were two kinds of spinel LiMn₂O₄ in the thin-film electrodes, which were described as Li_8a[Mn₂]_16dO₄ and Li_8b[Mn₂]_16cO₄. The presence of manganese ions on both sets of octahedral sites destabilizes the tetrahedral sites, and displaces some of the tetrahedral-site Li⁺ into neighboring octahedral sites. The residual Li⁺ ions in 16d octahedral sites must go through the energetically unfavorable 8b tetrahedral sites to be extracted out, and requiring more energy.

Here the samples H11, H12, H13 and H14 are all Li-rich spinels. The excess Li⁺ ions substitute some Mn³⁺ ions in the octahedral 16d sites. At a high voltage around 5 V, the Li⁺ ions begin to extract from spinel lattice, and form a 5 V plateau. When the Li/Mn ratio increases from 0.54 V to 0.72 V, the relative capacity of 5 V plateau to 4 V plateau is increasing as well. Furthermore, the extraction of Li⁺ in 16d sites has a tendency to a higher voltage with increasing Li/Mn ratio. An electrolyte stable to the voltage higher than 5.2 V is needed to further study the 5 V region activities. The further Li⁺ extraction at 5 V plateau is inevitable to induce an increase of average Mn valence. According to BATES et al[18], the 5 V plateau might be correspondent to the converting of Mn⁴⁺ to Mn⁵⁺. Actually, this plateau may not only be assigned to Mn⁴⁺→Mn⁵⁺, but also the converting of other Mn species, including the converting of the residual half of Mn³⁺ in 16d sites to Mn⁴⁺. Namely, it should be considered as an average effect of the oxidations of all the Mn species. This is unlike nickel substitute spinel LiNi_0.5Mn_1.5O₄, whose 5V plateau is resulted from Ni²⁺/Ni⁴⁺ redox[19, 20]. For samples H12, H13 and H14, which have a second phase Li₂MnO₃, the 5 V plateaus might also comprise the Li⁺ migrations in Li₂MnO₃.

4 Conclusions

1) Li_1+xMn_2-yO₄ spinels with various Li/Mn ratios were synthesized and characterized. Li₂MnO₃ was detected as a second phase by X-ray diffraction with increasing Li/Mn ratio. When the Li/Mn ratio exceeds 0.67, the diffraction peaks of Li₂MnO₃ are distinct.

2) Cyclic voltammetry was adopted to characterize the evolutions of 4 V and 5 V plateaus of Li_1+xMn_2-yO₄ spinels. An additional Li⁺ insertion plateau located at 3.95 V was observed in both Li-lack and Li-rich spinels. This plateau is different from the 3.95 V Li⁺ extraction observed by Palacín et al, and may represent a further Li⁺ insertion in tetrahedral 8a sites of the Li_1+xMn_2-yO₄ spinels, which have some phase defections. After the initial cycle, the spinel phases are activated to more perfect structures, and the 3.95 V plateau is no longer existent in the subsequent cycles. The 4.4/3.8 V plateaus correspondent to Li⁺ insertion and extraction of Li₂MnO₃ were detected, and these plateaus have a high reversibility in the following cycles. The presence of 5 V plateaus implies the existence of Li⁺ ions in octahedral 16d sites, which need more energy to be removed from the lattice. The further Li⁺ extraction at 5 V plateau is inevitable to induce an increase of average Mn valence, and has a tendency to emerge at higher voltage region with increasing Li/Mn ratio.

Acknowledgement

The authors gratefully thank Prof. CHEN Li-quan and Prof. HUANG Xue-jie for useful discussion.

References

[1] Thackeray m, Manganese oxides for lithium batteries[J]. Prog Solid St Chem, 1997, 25: 1-71.

[2] Thackeray m m, de Kock a, Rossouw m h, Liles d, Bittihn r, Hoge d. Spinel electrodes from the Li-Mn-O system for rechargeable lithium battery applications[J]. J Electrochem Soc, 1992, 139: 363-366.

[3] Tarascon j m, McKinnon w r, Coowar f, Bowmer t n, Amatucci g, Guyomard d. Synthesis conditions and oxygen stoichiometry effects on Li insertion into the spinel LiMn₂O₄[J]. J Electrochem Soc, 1994, 141: 1421-1432.

[4] Gao y, Dahn j r. Synthesis and characterization of Li_1+xMn_2–xO₄ for Li-ion battery applications[J]. J Electrochem Soc, 1996, 143: 100-114.

[5] Amatucci g g, Schmutz c n, Blyr a, Sigala c, Gozdz a s, Larcher d, Tarascon j m. Materials’ effects on the elevated and room temperature performance of C/LiMn₂O₄ Li-ion batteries[J]. J Power Sources, 1997, 69: 11-25.

[6] Tabuchi m, Masquelier c, Kobayashi h, Kanno r, Kobayashi y, Akai t, Maki t, Kageyama h, Nakamura o. Characterization of Li_1-δMn_2-2δO₄ defect spinel materials by their phase transition, magnetic and electrochemical properties[J]. J Power Sources, 1997, 68: 623-628.

[7] Palacín m r, Chabre y, Dupont l, Hervieu m, Strobel p, Rousse g, Masquelier c, Anne m, Amatucci g g, Tarasconc j m. On the origin of the 3.3 and 4.5 V steps observed in LiMn₂O₄-based spinels[J]. J Electrochem Soc, 2000, 147: 845.

[8] Eftekhari a. Electrochemical performance and cyclability of LiFe_0.5Mn_1.5O₄ as a 5 V cathode material for lithium batteries[J]. J Power Sources, 2003, 124: 182-190.

[9] Hong k j, Sun y k. Synthesis and electrochemical characteristics of LiCr_xNi_0.5-xMn_1.5O₄ spinel as 5 V cathode materials for lithium secondary batteries[J]. J Power Sources, 2002, 109: 427-430.

[10] Alcántara r, Jaraba m, Lavela p, Tirado j l. Optimizing preparation conditions for 5 V electrode performance, and structural changes in Li_1-xNi_0.5Mn_1.5O₄ spinel[J]. Electrochimica Acta, 2002, 47: 1829-1835.

[11] Ohzuku t, Ariyoshi k, Takeda s, Sakai y. Synthesis and characterization of 5 V insertion material of Li[Fe_yMn_2-y]O₄ for lithium-ion batteries[J]. Electrochimica Acta, 2001, 46: 2327-2336.

[12] Xia y y, Zhou y, Yoshio m. Capacity fading on cycling of 4 V Li/LiMn₂O₄ cells[J]. J Electrochem Soc, 1997, 144: 2593-2600.

[13] Thackeray m m. A comment on the structure of thin-film LiMn₂O₄ electrodes[J]. J Electrochem Soc, 1997, 144: L100-L102.

[14] Wu c, Wang z, Wu f, Chen l, Huang x. Spectroscopic studies on cation-doped spinel LiMn₂O₄ for ithium ion batteries[J]. Solid State Ionics, 2001, 144: 277-285.

[15] Wu c, Wu f, Chen l, Huang x. X-ray diffraction and X-ray photoelectron spectroscopy analysis of Cr-doped spinel LiMn₂O₄ for lithium ion batteries[J]. Solid State Ionics, 2002, 152-153: 335-339.

[16] Thackeray m, Johnson p j, de Picciotto l a. Electrochemical extraction of lithium from LiMn₂O₄[J]. Mater Res Bull, 1984, 19: 179-187.

[17] Kalyani p, Chitra s, Mohan t, Gopukumar s. Lithium metal rechargeable cells using Li₂MnO₃ as the positive electrode[J]. J Power Sources, 1999, 80: 103-106.

[18] Bates j b, Lubben d, Dudney n j, Zuhr r a, HART F X. 5-volt and 4.6V plateaus in LiMn₂O₄ thin films[A]. BATES J B. Thin Film Solid Ionic Devices and Materials, The Electrochemical Society Proceedings Series[C]. Pennington, NJ: The Electrochemical Society, 1995. 215-226.

[19] Arunkumar t a, Manthiram a. Influence of chromium doping on the electrochemical performance of the 5 V spinel cathode LiMn_1.5N_i0.5O₄[J]. Electrochimica Acta, 2005, 50: 5568-5572.

[20] Park s h, Sun y k. Synthesis and electrochemical properties of 5V spinel LiNi_0.5Mn_1.5O₄ cathode materials prepared by ultrasonic spray pyrolysis method[J]. Electrochimica Acta ,2004, 50: 431-434.

Foundation item: Project(2002CB211800) supported by the National Basic Research Program of China

Corresponding author: WU Chuan; Tel: +86-10-68912508; Fax: +86-10-68451429; E-mail: chuanwu@bit.edu.cn

(Edited by YUAN Sai-qian)