Crystal structural change during charge–discharge process of LiMn1.5Ni0.5O4 as cathode material for 5 V class lithium secondary battery

Solid State Ionics 176 (2005) 299 – 306 www.elsevier.com/locate/ssi

Crystal structural change during charge–discharge process of LiMn1.5Ni0.5O4 as cathode material for 5 V class lithium secondary battery Yasushi Idemoto*, Hiroshi Sekine, Koichi Ui, Nobuyuki Koura Department of Pure and Applied Chemistry, Faculty of Science and Technology, Tokyo University of Science, 2641 Yamazaki, Noda-shi, Chiba 278-8510, Japan Received 18 May 2004; received in revised form 31 August 2004; accepted 5 September 2004

Abstract We investigated the relation between the cycle performance and crystal structural change during the charge–discharge process of LiMn1.5Ni0.5O4 as a 5 V class cathode active material, which was prepared by changing the calcination temperature using the sol–gel method. The lithium content of Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) was controlled by electrochemical lithium extraction. The crystal structure was determined by Rietveld analysis using powder neutron diffraction. As a result, all samples consisted of three phases (space group: P4332) of different lattice constants and Ni valences. The main phase, which has the maximum percentage, was shifted to a phase with a lower lattice constant with the decreasing lithium content, and then finally Li1
x Mn1.5Ni0.5O4 (x=1.0) was almost oxidized to Ni4+ by a charging process. Furthermore, LiMn1.5Ni0.5O4,, by changing the synthesis temperature, was different for a few oxidation processes; the structure of the phase at Ni3+ was not stable based on the distortion of each phase and the Madelung energy. It was suggested that these factors should provide an effective cycle performance. D 2004 Elsevier B.V. All rights reserved. PACS: 84.60; 61.12; 61.66.F Keywords: Li secondary battery; Cathode; Crystal structure; Neutron diffraction; Lithium manganese nickel oxide

1. Introduction LiMn2O4 is a promising cathode material for the lithium secondary battery since it is inexpensive and environmentally friendly. LiMn2
x MxO4 has an ca. 5 V plateau when part of the Mn-site is substituted for the 3d-transition metals (M=Cr, Co, Ni, Fe, Cu) [1–8]. LiMn1.5Ni0.5O4 is easy to prepare, and has been shown to have a high capacity, and good cycle performance [3,4]. However, this material has a different cathode performance, which appears in the 4 V-region, charge–discharge behavior, and cycle performance, by changing the synthesis method [2,3]. The authors reported that LiMn1.5Ni0.5O4 synthesized by the sol–gel method has a good cycle performance with the increasing

* Corresponding author. Tel.: +81 4 7122 9493; fax: +81 4 7123 9890. E-mail address: [email protected] (Y. Idemoto). 0167-2738/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2004.09.003

oxygen content [9]. We also reported that the sample is assigned to an ordered form (space group P4332) based on a neutron diffraction analysis, and the crystal structure was stable, since it has a lower lattice energy than LiMn2O4, which is a spinel cubic (space group Fd-3m) [10,11]. The cycle performance is different from the calcined temperature at 650–700 8C. From the crystal structural analysis by Rietveld method using powder neutron diffraction data, LiMn1.5Ni0.5O4, which calcined at 650 8C, composed of mixing two space group P4332 phases. It is suggested that the oxidation states of Ni change in the sample, as these phases are different from each lattice parameter. Ni valence of sub-phase for LiMn1.5Ni0.5O4 calcined at 650 8C, which was calculated from bond valence sum, increases [11]. In recent years, Terada et al. [12] and McBreen et al. [13] reported that the oxidation reaction mechanism due to lithium deintercalation for LiMn1.5Ni0.5O4 had three distinct valence states, Ni2+, Ni3+, and Ni4+, using the Ni

300

Y. Idemoto et al. / Solid State Ionics 176 (2005) 299–306

K-edge XAFS technique. They reported that the lower cycle performance for LiMn1.5Ni0.5O4 is due to the coexistence of several phases in this potential region [13,14]. The average Ni2+–O, Ni3+–O, and Ni4+–O octahedral distances were found to be 0.199, 0.192, and 0.186 nm, respectively [12]. Thus, it is suggested that the crystal structural change of LiMn1.5Ni0.5O4, accompanying the intercalation–deintercalation of lithium, is important, and yet there are few reports about the stability of the (Mn,Ni)–O6 octahedral site, which is the frame of the spinel structure. In this study, LiMn1.5Ni0.5O4, as a 5 V cathode material was synthesized by the sol–gel method, in which the calcination temperature was changed, and the lithium content of Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) was controlled by an electrochemical treatment. We investigated the crystal structural change, which includes the (Mn,Ni)–O6 octahedral site, during the charge–discharge process by Rietveld analysis with powder neutron diffraction, and the relations between the structure and the cycle performance.

2. Experimental LiMn1.5Ni0.5O4 was prepared by the sol–gel method using the LiNO3 (99.9% Wako Pure Industries, Japan), Mn(NO 3 ) 2 d 9H 2 O (99.9% Wako Pure Industries), Ni(NO3)2d 6H2O (99.9% Wako Pure Industries) precursor and poly vinyl alcohol with subsequent decomposition at 150 8C. The resulting solids were heated at 600 8C for 24 h in O2, then calcined at 700 and 680 8C for 24 h in O2 [9,10]. The lithium content of the Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) was controlled by an electrochemical treatment. A three-electrode electrochemical cell composed of LiMn1.5Ni0.5O4 as the working electrode, a lithium metal anode as the counter electrode and lithium film as the reference electrode was used for controlling the lithium content. The cathodes were prepared by mixing the active material, a conducting material (acetylene black), and a binding material (polytetrafluoroethylene) at a weight ratio of 5:2:1 by pressing them onto an aluminum mesh current collector. The electrodes were dried at 150 8C in a vacuum atmosphere, and pressed at 40 MPa. The electrolyte was 1 M LiPF6 in an ethylene carbonate/diethyl carbonate (1:1) solution. All procedures were carried out in a dry box containing an argon atmosphere. First, these samples were examined by powder X-ray diffraction and the particle sizes of the samples were observed by SEM (JSM-5500, JEOL). Next, the crystal structures of Li1
x Mn1.5Ni0.5O4 were studied by powder neutron diffraction using HERMES [15]. The data were refined with the Rietveld technique using the Rietan-2000 program [16]. The bond length was calculated by the attached software ORFFE of RIETAN-2000. The Madelung energy was calculated by the attached software MADEL of FAT-RIETAN. The bond valence sum [17–19] was calculated by the VICS program [20], and the nuclear densities

were calculated by the maximum entropy method using the PRIMA program [21].

3. Results and discussion 3.1. Sample characterization and electrode characteristic The X-ray diffraction pattern of the LiMn1.5Ni0.5O4, prepared by the sol–gel method as good samples, indicated a cubic spinel phase. The metal composition of the samples determined by ICP was almost the same as the nominal composition [11]. Fig. 1 shows the X-ray diffraction pattern of the Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0), which controlled the amount of lithium. Each sample of Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) has a split (111) peak. This suggests the appearance of a new phase or phase transition. Since each peak is shifted in the direction of a higher angle with the increasing x, it suggests that LiMn1.5Ni0.5O4 is oxidized from Ni2+ (0.69 nm) to Ni4+ (0.48 nm) via Ni3+ (high spin: 0.60 nm, low spin: 0.56 nm) [22] with the decreasing Li content [23,24]. These details, that determined by Rietveld analysis using powder neutron diffraction, are described in Section 3.2. The lithium contents in the samples, calculated from the charge capacity, are shown in Table 1. As a result, it was controllable in the target lithium content.

Fig. 1. X-ray diffraction patterns of Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0). (a) calcined at 700 8C, (b) calcined at 680 8C. D: PTFE (polytetrafluoroethylene).

Y. Idemoto et al. / Solid State Ionics 176 (2005) 299–306

301

Table 1 Li content (1
x) calculated from charge capacity for Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0)

analysis using powder neutron diffraction, and the cycle performance.

Sample

3.2. Crystal structural change during the charge–discharge process

x=0.5 x=0.7 x=1.0

Li content, 1
x 700 8C, O2, 24 h

680 8C, O2, 24 h

0.509 0.309 0.018

0.499 0.304 0.058

The electrode characteristics of the LiMn1.5Ni0.5O4 were examined by CV and charge–discharge tests. The CV data showed that the redox reaction occurred only in the 5 V region (oxidation potential: ca. 4.65–4.75 V vs. LijLi+, reduction potential: ca. 4.55–4.65 V vs. LijLi+) with the deintercalation–intercalation of Li [11]. These results suggested that the obtained sample does not exist as Mn3+. The Mn valence of the samples was in good agreement with about 4.0 by chemical analysis, assuming that the Ni valence was 2.0 [11]. The sample, which was calcined at 700 8C, showed a good cycle performance with the maximum discharge capacity of 125.3 mAh/g, and the capacity after 100 cycles was 95.1% of the maximum capacity [11]. On the other hand, the sample, which was calcined at 680 8C, showed the maximum discharge capacity of 122.6 mAh/g, and the capacity after 100 cycles was 90.5% of the maximum capacity [11]. The oxygen content, the surface morphology by SEM, and the crystal structure determined by powder neutron diffraction of the samples were almost the same [11]. Accordingly, in this study, we investigated the stability of the host structure during the charge–discharge process, the oxidation reaction mechanism of nickel that contributed to the redox reaction, the relation between the crystal structural change, which was determined by Rietveld

The neutron diffraction intensity profile of LiMn1.5Ni0.5O4 was analyzed assuming the space group P4332 (cubic, Z=8) [10,11]. From these results, good agreement between the observed and calculated patterns was obtained and the good-fit indicator, the S value (=R wp/R e), was low. As an example of the samples that controlled the lithium content, the final refined parameters for Li0.5Mn1.5Ni0.5O4 (Phase1) (Phase2), which was calcined at 700 8C, are shown in Table 2. Fig. 2 illustrates the results of the Rietveld refinement patterns for Li0.5Mn1.5Ni0.5O4, which was calcined at 700 8C. The refinement was carried out by assuming a single phase, two phases, and three phases of space group P4332, and the best-fit indicator was obtained using the three-phase model. The lattice constant of each phase for Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) showed almost the same value. Accordingly, all samples were analyzed using a three-phase model composed of the three-space group P4332 phase. The range of 50.1b2hb 52.1 was excluded from the refinement of all the samples, since the peak of PTFE existed. For example, a refinement was carried out such that O1 (8c) is fully occupied, and the other site occupancies are variable in Phase1 for x=0.5. Only the Li site was contained the variables, and the other site occupancies are fixed in Phase2 for x=0.5, which is near the lattice constant of Phase1 for x=0.7. The Li content of Phase3 is fixed for the calculation value by the existing percentage of each phase and the Li content in Table 1, and the other site occupancies

Table 2 Final results of Rietveld refinements for Li0.5Mn1.5Ni0.5O4, which was synthesized at 700 8C, in space group P4332 (cubic) y

z

102B (nm2)

Site occupancy

(a) Phase1, a=0.8151(4) nm, V=0.5415(5) nm Ni1 4b 5/8 Mn1 4b =x(Ni1) Li 8c 1.015(7) Mn2 12d 1/8 Ni2 12d =x(Mn2) O1 8c 0.386(2) O2 24e 0.108(2)

5/8 =y(Ni1) 1.015(7) 0.379(5) =y(Mn2) 0.386(2) 0.123(2)

5/8 =z(Ni1) 1.015(7)
0.129(5) =z(Mn2) 0.386(2) 0.391(1)

0.3(7) =B(Ni1) 1.0(29) 0.2(9) =B(Mn2) 0.4(6) 0.5(5)

0.76(5) 0.24(5) 0.81(28) 0.94(3) 0.06(3) 1 0.95(6)

(b) Phase2, a=0.8098(9) nm, V=0.5311(11) nm 3 Ni1 4b 5/8 Mn1 4b =x(Ni1) Li 8c 1.026(83) Mn2 12d 1/8 Ni2 12d =x(Mn2) O1 8c 0.380(15) O2 24e 0.122(16)

5/8 =y(Ni1) 1.026(83) 0.371(23) =y(Mn2) 0.380(15) 0.114(8)

5/8 =z(Ni1) 1.026(83)
0.121(23) =z(Mn2) 0.380(15) 0.387(9)

0.45 =B(Ni1) 1 0.42 =B(Mn2) 0.67 0.42

0.69 0.31 0.28(98) 0.91 0.09 1 0.99

Atom

Site

x 3

B is isotropic thermal parameters. Numbers in parentheses are estimated standard deviations of the last significant digit, and those without deviations. R wp=4.07%, R p=3.21%, R exp=2.79%, S(=R wp/R exp)=1.45. Phase1: Phase2: Phase3 [a=0.7985(21) nm]=0.532: 0.274: 0.194.

302

Y. Idemoto et al. / Solid State Ionics 176 (2005) 299–306

Fig. 2. Rietveld refinement patterns for Li0.5Mn1.5Ni0.5O4, which was synthesized at 700 8C. Plus marks show observed neutron diffraction intensities, and a solid line represents calculated intensities. The vertical marks below the patterns indicate the positions of allowed Bragg reflections {upper: Phase1 [a=0.8151(4) nm], middle: Phase2 [a=0.8098(9) nm], bottom: Phase3 [a=0.7985(21) nm]}. The curves at the bottom are the difference between the observed and calculated intensities in the same scale.

were fixed in Phase3 for x=0.5, which is near the lattice constant of Phase1 for x=1.0. The isotropic thermal parameters of Phase1 are variables, while the Phase2 and Phase3 were fixed with values at Phase1 for x=0.7, and Phase1 for x=1.0, respectively. The refinement of x=0.7 and 1.0 was carried out by the same method. The metal composition, oxygen content and the nickel valence of each sample, which were calcined at 700 8C, calculated from the final results of the structure parameter obtained by Rietveld analysis are shown in Table 3, and those at 680 8C are shown in Table 4, respectively. The Ni valence was calculated from the Mn valence assumed to be 4. These results suggested that the three phases corresponded to the Ni valence of Ni2+ (ai0.816 nm), Ni3+ (ai0.809 nm), and Ni4+ (ai0.800 nm) in all cases.

Moreover, the results of the average metal composition and oxygen content were in good agreement with the chemical analysis. Ariyoshi et al. [24] reported the change in lattice parameter as function of charge capacity of LiMn1.5Ni0.5O4, which prepared by a solid state reaction (calcined at 700 8C in air) and determined by X-ray diffraction. They described the reaction at ca. 4.7 V as consisting of two cubic/cubic two-phase reactions, The three-lattice parameter of cubic phases in this region agrees with the present results. On the other hand, the present results show all samples (x=0.5, 0.7, 1.0) consisted of three cubic phases determined by neutron diffraction, but they reported two-phase reactions. For the reason, they determined by X-ray diffraction data and it is difficult to distinguish the each phase of the existence of low percentage.

Table 3 Composition and Ni valence of Li1
x Mn1.5Ni0.5O4, which was calcined at 700 8C, calculated from Rietveld refinement

Table 4 Composition and Ni valence of Li1
x Mn1.5Ni0.5O4, which was calcined at 680 8C, calculated from Rietveld refinement

Sample

Composition Li

Mn

Ni

Ni valencea

Sample

O

Li

Mn

Ni

O

[Phase1] [Phase2] [Phase3]

0.81 0.33 0.01 0.49

1.51 1.52 1.51 1.51

0.49 0.48 0.49 0.49

3.88 3.94 3.93 3.91

1.86 3.06 3.69 2.63

(b) x=0.7 a=0.8098(9) nm a=0.8007(27) nm a=0.8154(5) nm Average

[Phase1] [Phase2] [Phase3]

0.23 0.01 0.74 0.30

1.52 1.51 1.51 1.51

0.48 0.49 0.49 0.49

3.97 3.94 3.89 3.94

3.40 3.73 2.04 3.14

(c) x=1.0 a=0.8001(5) nm a=0.8085(9) nm a=0.8161(2) nm Average

[Phase1] [Phase2] [Phase3]

0.01 0.07 0.62 0.07

1.51 1.52 1.51 1.51

0.49 0.48 0.49 0.49

3.94 3.97 3.89 3.94

3.73 3.73 2.29 3.61

(a) x=0.5 a=0.8151(4) nm a=0.8098(9) nm a=0.7985(22) nm Average

[Phase1] [Phase2] [Phase3]

0.81 0.28 0.01 0.51

1.53 1.52 1.52 1.53

0.47 0.48 0.48 0.47

3.85 3.97 3.96 3.90

1.64 3.29 3.81 2.51

(a) x=0.5 a=0.8159(2) nm a=0.8087(9) nm a=0.7986(22) nm Average

(b) x=0.7 a=0.8083(9) nm a=0.8011(17) nm a=0.8158(6) nm Average

[Phase1] [Phase2] [Phase3]

0.26 0.01 0.73 0.31

1.59 1.51 1.52 1.55

0.41 0.49 0.48 0.45

3.97 3.94 3.84 3.93

3.22 3.73 1.81 3.00

(c) x=1.0 a=0.8013(15) nm a=0.8089(10) nm a=0.8163(19) nm Average

[Phase1] [Phase2] [Phase3]

0.01 0.23 0.60 0.08

1.52 1.52 1.53 1.52

0.48 0.48 0.47 0.48

3.97 3.97 3.84 3.96

3.85 3.40 2.04 3.67

a

The Ni valence calculated from Mn valence is 4.

Ni valencea

Composition

a

The Ni valence calculated from Mn valence is 4.

Y. Idemoto et al. / Solid State Ionics 176 (2005) 299–306

Fig. 3 shows the relation between the lattice parameter, a, and the removed lithium content, x, of Li1
x Mn1.5Ni0.5O4, which was calcined at 700 and 680 8C. The numerical value in the figure shows the existing percentage of each phase. Each sample has three phases with different lattice constants. The main phase [Phase1], which has the maximum percentage, was shifted to the phase with a small lattice constant due to the decreasing lithium content. These results suggested that Li1
x Mn1.5Ni0.5O4 is gradually oxidized from Ni2+ with a larger ion radius, to Ni4+ via Ni3+ with a smaller one by the charging process, since the change in the Ni valence for each lattice constant in Tables 3 and 4. The existing percentage of each corresponding lattice constant is slightly different from the samples, which were calcined at 700 and 680 8C. The oxidation process from Ni2+ to Ni4+ via Ni3+ was different, and the sample, which was calcined at 680 8C, had the high Ni3+ ratio of nearly x=0.7. This suggested that the electrochemical characteristics only slightly change with the synthetic methods. The relations between the bond lengths of Ni,Mn (4b)– O2, Mn,Ni (12d)–O and Li content in the main phase of Li1
x Mn1.5Ni0.5O4 (x=0, 0.5, 0.7, 1.0) are shown in Table 5.

303

Here, the error bar is large, since these samples are analyzed for multi-phases. The bond length of Ni,Mn (4b)–O2, which is mainly occupied by Ni, decreased with the increasing x. The reason is that the Ni valence increased with increasing x, and agreed with the above consideration. On the other hand, the distortion of the Mn,Ni (12d) octahedral site, which has a high multiplicity, increased with the increasing x, that is, the Ni valence of the main phase shifted from Ni2+ to Ni4+ via Ni3+. The distortion of x=0.7 and 1.0, which were calcined at 680 8C, is greater than that of the samples, which were calcined at 700 8C. The quadratic elongation k, the bond angle variance r 2 [20], and the bond valence sum (B.V.S.) [19] in the Ni,Mn(4b)–O octahedral site, and Mn,Ni(12d)–O octahedral site for Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) are shown in Tables 6 and 7, respectively. Here, the B.V.S. of Ni was calculated by assuming Mn4+, since the occupancy of nickel for the Ni,Mn(4b)–O octahedral site is high, and considering the site occupancy of Ni and Mn. The B.V.S. of Mn was calculated from the Ni valence corresponding to a lattice constant, since the multiplicity of Mn for the Mn,Ni(12d)–O octahedral site is high, and considering the site occupancy of Mn and Ni.

Fig. 3. Relation between the lattice parameter, a, and the Li content, x, of Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0). (a) 700 8C, O2, 24 h; (b) 680 8C, O2, 24 h. The numerical value in the figure shows the existing percentage of each phase. (o: x=0, D: x=0.5, 5: x=0.7, R : x=1.0).

304

Y. Idemoto et al. / Solid State Ionics 176 (2005) 299–306

Table 5 Bond lengths (nm) of main phase for of Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) (a) Ni,Mn(4b)–O Sample x=0 x=0.5 x=0.7 x=1.0

Bond lengths (nm) 700 8C, O2, 24 h

680 8C, O2, 24 h

0.203(1) 0.203(2) 0.197(20) 0.188(20)

0.204(1) 0.203(1) 0.188(9) 0.186(10)

(b) Mn,Ni(12d)–O Sample

Bond length (nm) 700 8C, O2, 24 h

x=0 x=0.5 x=0.7 x=1.0 a b

680 8C, O2, 24 h

Mn,Ni(12d)–O

Mn,Ni(12d)–Oa

Mn,Ni(12d)–Ob

Mn,Ni(12d)–O

Mn,Ni(12d)–Oa

Mn,Ni(12d)–Ob

0.193(1) 0.192(5) 0.196(87) 0.191(44)

0.190(1) 0.194(5) 0.192(88) 0.203(57)

0.192(1) 0.191(1) 0.187(18) 0.192(17)

0.191(1) 0.190(4) 0.210(37) 0.205(55)

0.191(1) 0.190(1) 0.183(8) 0.184(8)

0.190(1) 0.194(4) 0.204(34) 0.200(54)

(z
1/2, 1/2
x, y). ( y, z, x).

The B.V.S. of Ni for the samples, which was calcined at 700 8C, is different from each phase in Table 6(a). Based on these results, it is well supported that the Ni valence increased with the decreasing lattice constant. The same tendency was also obtained in each phase as shown in Table 6(b) and (c). Furthermore, the values of k and r 2 of the samples, which were calcined at 700 8C, decreased with the decreasing lattice constant in Table 6(a), hence the distortion of the Ni,Mn(4b)–O octahedral site decreased. These results were similar to that of the bond lengths in Table 5, and they considered changing the Ni valence from Ni2+ to Ni4+ via Ni3+, which have a small ionic radius. The values of k and r 2 in each phase in Table 6(b) and (c) showed the same tendency. On the other hand, the values of k and r 2 for the samples, which were calcined at 680

Table 6 Distortion of octahedral-site, k, r 2, bond valence sum (B.V.S.) for Ni,Mn(4b)–O of Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) Sample a (nm)

k

r

2

a

B.V.S. (Ni)

700 8C

680 8C

700 8C

680 8C

700 8C

680 8C

(a) x=0.5 0.816 [Phase1] 1.0172 0.809 [Phase2] 1.0048 0.800 [Phase3] 1.0047

1.0217 1.0169 1.0103

66.024 17.586 16.514

83.367 58.394 36.414

1.760 3.444 4.989

2.007 4.586 5.416

(b) x=0.7 0.809 [Phase1] 1.0132 0.800 [Phase2] 1.0203 0.816 [Phase3] 1.0172 (c) x=1.0 0.800 [Phase1] 1.0041 0.809 [Phase2] 1.0131 0.816 [Phase3] 1.0172 a

1.0162 1.0046 1.0217

1.0104 1.0147 1.0217

49.776 72.218 65.848

14.493 49.317 65.984

56.227 16.167 83.302

36.000 53.266 83.302

The B. V. S.(Ni) calculated from Mn valence is 4.

3.034 4.938 1.748

4.718 2.988 1.736

4.413 4.839 2.016

5.340 2.821 2.004

8C, were greater than those of the samples, which were calcined at 700 8C, that is, the distortion of the lattice for the 680 8C samples was greater than that for the 700 8C samples. The results have the same tendency in Table 6(b) and (c); they were similar to that of the bond lengths in Table 5. The B.V.S. of Mn for the Mn,Ni(12d) site decreased in Table 7, since the site occupancy of Ni increased by changing the Ni valence. The value of k for the Mn,Ni(12d)–O octahedral site, that is the distortion from the bond length, increased with the increasing Ni valence, and the decreasing lattice constant of Phase1 (700 8C samples) in Table 7. On the other hand, the value of r 2 for the Mn,Ni(12d)–O octahedral site has a maximum at ai0.809 nm, that is, the bond angle variance was Table 7 Distortion of octahedral-site, k, r 2, bond valence sum (B.V.S.) for Mn,Ni(12d)–O of Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) Sample a (nm)

r2

k

B.V.S. (Mn)a

700 8C 680 8C 700 8C 680 8C

700 8C 680 8C

(a) x=0.5 0.816 [Phase1] 1.0079 0.809 [Phase2] 1.0056 0.800 [Phase3] 1.0138

1.0094 1.0430 1.0132

28.953 18.641 33.786

34.757 3.849 113.14 3.369 34.590 3.310

3.910 3.518 3.480

(b) x=0.7 0.809 [Phase1] 1.0108 0.800 [Phase2] 1.0182 0.816 [Phase3] 1.0079

1.0109 1.0140 1.0092

37.824 47.748 28.971

43.644 3.952 34.209 3.823 34.194 3.835

3.369 3.259 3.918

(c) x=1.0 0.800 [Phase1] 1.0115 0.809 [Phase2] 1.0103 0.816 [Phase3] 1.0079

1.0132 1.0158 1.0092

28.572 36.326 28.952

34.824 3.228 46.069 3.904 34.194 3.819

3.439 3.837 3.900

a The B.V.S. (Mn) calculated from Ni valence corresponding to a lattice constant.

Y. Idemoto et al. / Solid State Ionics 176 (2005) 299–306

maximum. These results suggested that the phase of Ni3+ remarkably increased the distortion for the Mn,Ni(12d)–O octahedral site. The distortion of the samples, which were calcined at 680 8C, was overall greater than that of the 700 8C samples, and corresponded well to the tendency of the bond length in Table 5. Fig. 4 shows the bird’s eye view of the nuclear densities on the (110) plane for Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) in the main phase using the maximum entropy method. The minimum and the maximum values of the nuclear densities

305

are different from the samples, so a maximum value is unified by 1,137,410 fm/nm3. The value of each sample shows the percent of the unified maximum value. The nuclear densities of the Mn,Ni(12d) sites decreased with the decreasing Li content; this corresponds with the decreasing Ni content of this site. The nuclear density of the oxygen around the Mn,Ni(12d) site furthest from the Ni,Mn(4b) site was extremely high for the phase with ai0.809 nm in Fig. 4(c) and (g), that is, the Ni3+ phase. It is suggested that bonding around Ni,Mn(4b) becomes weak, since Ni2+ of the

Fig. 4. Bird’s eye view of nuclear densities on the (110) plane for Li1
x Mn1.5Ni0.5O4 (x=0.5, 0.7, 1.0) calcined at 700 8C (a)–(d), and 680 8C (e)–(h) [main phase]. The maximum is 1137410 (fm/nm3). (a) x=0, ai0.816 nm; up to 69% of the maximum, (b) x=0.5, ai0.816 nm; up to 79% of the maximum, (c) x=0.7, ai0.809 nm; up to 49% of the maximum, (d) x=1.0, ai0.800 nm; up to 35% of the maximum, (e) x=0, ai0.816 nm; up to 100% of the maximum, (f) x=0.5, ai0.816 nm; up to 93% of the maximum, (g) x=0.7, ai0.809 nm; up to 52% of the maximum, (h) x=1.0, ai0.800 nm; up to 23% of the maximum.

306

Y. Idemoto et al. / Solid State Ionics 176 (2005) 299–306

Table 8 Results of the Madelung energy (E M) for Li1
x Mn1.5Ni0.5O4 (x=0, 0.5, 0.7, 1.0) Sample [a (nm)]

x=0 x=0.5 x=0.5 x=0.7 x=0.7 x=1.0 x=1.0

[0.816] [0.809] [0.809] [0.800] [0.800] [0.809]

for which the oxidation–reduction reaction from Ni3+ and the reduction reaction from Ni4+ to Ni3+ easily occur.

Madelung energy (kJ/mol) 700 8C, O2, 24 h

680 8C, O2, 24 h


8356.9
8260.8
8096.6
7867.5
8473.3
8484.4
7735.1


8338.8
8226.7
8196.7
8209.3
8498.2
8642.3
8193.7

initial stage becomes Ni3+, which is a Jahn-Teller ion, during the charge process. In addition, the nuclear density of Li decreases with increasing x. The Madelung energy, which was calculated from the Rietveld analysis result, is shown in Table 8. It is suggested that the structure of the Ni3+(ai0.809 nm) phase is less stable than that of Ni2+(ai0.816 nm), since the Madelung energy of Ni3+ (ai0.809 nm) increased. On the other hand, the structure of the Ni4+ phase (ai0.800 nm) was more stable than those of Ni2+ and Ni3+. Accordingly, it is suggested that the LiMn1.5Ni0.5O4 samples will have a good conservation performance. The absolute value of the Madelung energy for the structure with space group P4332 is greater than that of the structure with space group Fd-3m [25] before and after the charge–discharge process. Consequently, the crystal structure with space group P4332 is more stable, and it should provide an excellent cycle performance.

4. Conclusions We clarify that the Ni-doped spinel oxide, LiMn1.5Ni0.5O4, with a 5 V class cathode material is oxidized from Ni2+ (ai0.816 nm) to Ni4+ (ai0.800 nm) via Ni3+ (ai0.809 nm) with the deintercalation of Li based on a crystal structure analysis. The oxidization process of LiMn1.5Ni0.5O4 is slightly different from the synthesis temperature. The Ni,Mn(4b) octahedral site, which has a high Ni, shrunk during the deintercalation of Li, and the Mn,Ni(12d) octahedral site, which has a high multiplicity, was the most distorted for the Ni3+ phase. The crystal structure became unstable for the Ni3+ phase and stable for the Ni4+ phase based on the results of the nuclear density and Madelung energy. From these results, it is important for improvement of the cycle performance of this material to investigate the crystal structure and reaction mechanism,

Acknowledgements We are indebted to Dr. K. Ooyama (Tohoku University) for measurement of the powder neutron diffraction at HERMES (JRR-3 M). This work was partially supported by Grants-in-Aid for Scientific Research, Ministry of Education, Culture, Sports, Science and Technology, Japan. References [1] C. Sigala, D. Guyomard, A. Verbaere, Y. Piffard, M. Tournoux, Solid State Ionics 81 (1995) 167. [2] Q. Zhong, A. Bonakdarpour, M. Zhang, Y. Gao, J.R. Dahn, J. Electroanal. Chem. 144 (1997) 205. [3] T. Ohzuku, S. Takeda, M. Iwanaga, J. Power Sources 81–82 (1999) 90. [4] T. Ohzuku, K. Ariyoshi, S. Yamamoto, J. Ceram. Soc. Jpn. 110 (2002) 501. [5] B. Ammundsen, D.J. Jones, J. Roziere, F. Villain, J. Phys. Chem., B 102 (1998) 7939. [6] T. Zheng, J.R. Dahn, Phys. Rev., B 56 (1997) 3800. [7] Y.-K. Sun, D.-W. Kim, Y.-M. Choi, J. Power Sources 79 (1999) 231. [8] H. Kawai, M. Nagata, M. Tabuchi, H. Tukamoto, A.R. West, Electrochim. Acta 45 (1999) 315. [9] Y. Idemoto, H. Narai, N. Koura, Electrochemistry 70 (2002) 587. [10] Y. Idemoto, H. Narai, N. Koura, J. Power Sources 119 (2003) 125. [11] Y. Idemoto, H. Sekine, K. Ui, N. Koura, Electrochemistry, in press. [12] Y. Terada, K. Yasaka, F. Nishikawa, T. Konishi, M. Yosho, I. Nakai, J. Solid State Chem. 156 (2001) 286. [13] J. McBreen, S. Mukerjee, X.Q. Yang, X. Sun, Y. Ein-Eli, in: S. Surampudi, R.A. Marsh (Eds.), Lithium Batteries, The Electrochemical Society Proceedings Series, Pennington, NJ, 1999, p. 308, PV 98-16. [14] S. Mukerjee, R.C. Urian, X.Q. Yang, J. McBreen, Y. Ein-Eli, in: A. Landgrebe, R.J. Klingler (Eds.), Interfaces, Phenomena, and Nanostructures in Lithium Batteries Work Shop, The Electrochemical Society Proceedings Series, Pennington, NJ, 2001, p. 262, PV 2000-36. [15] K. Ohoyama, T. Kanouchi, K. Nemoto, M. Ohashi, T. Kajitani, Y. Yamaguchi, Jpn. J. Appl. Phys. 37 (1998) 3319. [16] F. Izumi, T. Ikeda, Mat. Sci. Forum 321 (2000) 198. [17] I.D. Brown, R.D. Shannon, Acta Crystallogr., Sect. A 29 (1973) 266. [18] I.D. Brown, D. Altermatt, Acta Crystallogr., Sect. B 41 (1985) 244. [19] N.E. Brese, M. O’Keeffe, Acta Crystallogr., Sect. B 47 (1991) 192. [20] K. Robinson, G.V. Gibbs, P.H. Ribbe, Science 172 (1971) 567. [21] F. Izumi, R.A. Dilanian, in: Recent Research Developments in Physics, vol. 3, Transworld Research Network, Trivandrum, 2002, p. 699. [22] R.D. Shanon, Acta Crystallogr., A 32 (1976) 751. [23] I. Takahashi, M. Saitou, M. Sano, M. Hujita, K. Kihune, Extended Abstracts; The 43rd Battery Symposium in Japan, 2002, p. 166. [24] K. Ariyoshi, Y. Iwakoshi, N. Nakayama, T. Ohzuku, J. Electrochem. Soc. 151 (2004) A296. [25] Y. Idemoto, K. Horiko, K. Ui, N. Koura, Electrochemistry, in press.