Toward Positive Electrode Materials with High-Energy Density: Electrochemical and Structural Studies on LiCo x Mn2-x O4 with 0 ≤ x ≤ 1.

To obtain positive electrode materials with higher energy densities (Ws), we performed systematic structural and electrochemical analyses for LiCo x Mn2-x O4 (LCMO) with 0 ≤ x ≤ 1. X-ray diffraction measurements and Raman spectroscopy clarified that the samples with x ≤ 0.5 are in the single-phase of a spinel structure with the Fd3̅m space group, whereas the samples with x ≥ 0.75 are in a mixture of the spinel-phase and Li2MnO3 phase with the C2/m space group. The x-dependence of the discharge capacity (Q dis) indicated a broad maximum at x = 0.5, although the average operating voltage (E ave) monotonically increased with x. Thus, the W value obtained by Q dis × E ave reached the maximum (=627 mW h·g-1) at x = 0.5, which is greater than that for Li[Ni1/2Mn3/2]O4. Furthermore, the change in the lattice volume (ΔV) during charge and discharge reactions approached 0%, that is, zero-strain, at x = 1. Because ΔV for x = 0.5 was smaller than that for Li[Ni1/2Mn3/2]O4, the x = 0.5 sample is found to be an alternative positive electrode material for Li[Ni1/2Mn3/2]O4 with a high W.

Introduction

The bottom line of batteries in baseball is the harmonizing of a pitcher and a catcher. In the same way, the balancing of the electrochemical performances between a positive electrode and a negative electrode is crucial for secondary batteries. As previously reviewed for lithium-ion batteries (LIBs),[1,2] the energy densities (Ws) for positive electrode materials are significantly lower than those for negative electrode materials. This imbalanced situation still restricts the widespread applications of LIBs, such as in electric vehicles (EVs) and stationary energy storage systems (ESSs).[1] In other words, positive electrode materials with higher Ws are urgently required to realize more practical EVs and ESSs.

The W value for a positive electrode material is determined as followswhere Qrecha is the rechargeable capacity and Eave is the average operation voltage. Therefore, a positive electrode material with a high Qrecha or a high Eave or both provides a high W. Over the past decade, lithium–nickel–manganese spinel Li[Ni1/2Mn3/2]O4 has attracted much attention because it offers ∼135 mA h·g–1 of Qrecha and 4.5 V versus Li+/Li of Eave, resulting in more than 600 mW h·g–1 of W.[3−6] Lithium–cobalt–manganese spinel LiCoMn2–O4 (LCMO) with 0 ≤ x ≤ 1 is also promising from the viewpoint of Eave;[7−13] namely, the redox reaction of Co3+ ↔ Co4+ in LCMO is ∼0.3 V higher than that of Ni2+ ↔ Ni4+ in Li[Ni1/2Mn3/2]O4.[3,4,8,9,11]

Previous X-ray absorption near-edge structure (XANES)[8] analyses and magnetic measurements of electron paramagnetic resonance (EPR)[14,15] and susceptibility[16] indicated that Co ions in LCMO are in the trivalent state with a low-spin t2g6 (S = 0) configuration. The ideal electrochemical reaction of LCMO is thus represented aswhere the theoretical capacity (Qtheo) is calculated to be (148.23 – 3.21x) mA h·g–1. The current Qrecha for x = 1 is, however, significantly lower than Qtheo and is usually[7−11] limited to ∼100 or 116 mA h·g–1 at the maximum.[13] This comes from the presence of Li2MnO3 impurities in x = 1, which are electrochemically inactive.[9] By contrast, the Eave for LCMO almost linearly increases on increasing the amount of the Co3+ ↔ Co4+ redox reaction. Because W is the product obtained by the multiplication of Qrecha and Eave, as shown by eq , the opposite trend between Qrecha and Eave in LCMO is expected to indicate a maximum W value at a certain x composition. Despite the intensive studies on LCMO thus far,[7−16] a systematic study to explore the optimum x composition for the highest W has not been undertaken.

In this contribution, we report the results of Qrecha, Eave, and W as a function of x in LCMO and compare them with Li[Ni1/2Mn3/2]O4. Structural analyses using synchrotron radiation X-ray diffraction (XRD) measurements and Raman spectroscopy were also performed to clarify the relation between the electrochemical properties and crystal structures of LCMO. Consequently, we revealed that the x = 0.5 composition provides the maximum W value of 627 mW h·g–1, with a change in the lattice volume of ∼4%. These performances were found to be superior to those of Li[Ni1/2Mn3/2]O4.

Results and Discussion

Particle Morphology

For positive electrode materials with high Eave values, the particle size and morphology significantly affect their electrochemical properties,[17] probably due to the decomposition of electrolytes. To clarify changes in the particle size and morphology with x, Figure shows the scanning electron microscopy (SEM) images for the LCMO samples with (a) x = 0, (b) x = 0.5, and (c) x = 1. The x = 0 sample indicates a nonuniform particle shape with an average particle size (dave) of 2 μm. For the x = 0.5 and 1 samples, some particles show a truncated octahedral shape, although their dave values are similar to that of the x = 0 sample. As seen in Figure S1, particles for the x = 0.25 and 0.5 samples also show a truncated octahedral shape with dave of ∼2 μm. Thus, the substitution of Co ions for Mn ions mainly influences the particle morphology for LCMO.

Figure 1

SEM images on the 5 μm scale for the LiCoMn2–O4 samples with (a) x = 0, (b) x = 0.5, and (c) x = 1. One of the particles in the x = 1 sample (surrounded by the red dotted line) shows a truncated octahedron with a facet growth velocity ratio (α) of 2.25 < α < 2.5.

The Li[Ni1/2Mn3/2]O4 particles prepared by the two-step solid-state reaction technique exhibit an octahedral shape with smooth {111} facets, not a truncated octahedral shape.[17] The difference between the octahedral and truncated octahedral shapes originates from the proportion of the {001} facets to {111} facets. One simple method to quantitatively describe such a difference is given by the Wulff construction[18]where α is the facet growth velocity ratio and V001 and V111 are the growth velocities of the {001} and {111} facets, respectively. When α = 1, the particle indicates only the {001} facets, that is, cubic, whereas when α = 3, the particle indicates only the {111} facets, that is, regular octahedron.[18] As seen from the x = 1 particle surrounded by a red dotted line in Figure c, the surface area of one {001} facet is estimated to be 1.1 μm2, whereas that of one {111} facet is estimated to be 3.9 μm2. Thus, the ratio of the {001} facets to {111} facets is estimated to be ∼0.2, resulting in 2.25 < α < 2.5.

Electrochemistry

Figure shows the charge and discharge curves of the LCMO/Li cells with (a) x = 0, (b) x = 0.25, (c) x = 0.5, (d) x = 0.75, and (e) x = 1. For the x = 0 sample, the cell voltage rapidly increases to ∼4.0 V, then maintains an almost constant voltage at around 4.2 V up to a charge capacity (Qcha) of ∼80 mA h·g–1, and finally indicates another voltage plateau at around 5.2 V. The voltage plateau above 5.0 V is also reported for the charge curves for Li[LiMn2–]O4 with 0 ≤ x ≤ 1/3.[19,20] The Qcha and discharge capacity (Qdis) at the initial cycle are 126 and 119 mA h·g–1, respectively. For the x = 0.25 and 0.5 samples, the Qcha values below ∼4.0 V decrease compared with that for x = 0; and in turn, the Qcha values above ∼5.0 V increase with x. The Qcha and Qdis values at the initial cycle are not different between the x = 0.25 and 0.5 samples; that is, the Qcha and Qdis values are 147 and 138 mA h·g–1, respectively, for x = 0.25, and the Qcha and Qdis values are 146 and 140 mA h·g–1, respectively, for x = 0.5. However, as clearly seen from the intersecting voltages between the charge and discharge curves, the Eave value for x = 0.5 is expected to be greater than that for x = 0.25.

Figure 2

Charge and discharge curves of the LiCoMn2–O4/Li cells with (a) x = 0, (b) x = 0.25, (c) x = 0.5, (d) x = 0.75, and (e) x = 1. The cells were operated at a current of 0.6 mA (0.3 mA·cm–2) in the voltage range between 3.0 and 5.4 V at 25 °C.

For the x = 0.75 sample, the Qcha value at around 4.0 V is decreased to ∼25 mA h·g–1; however, the whole Qcha (=134 mA h·g–1) and Qdis (=127 mA h·g–1) values at the initial cycle are slightly decreased compared with those for x = 0.25 and 0.5. For the x = 1 sample, the Qcha and Qdis values at the initial cycle are further decreased to 112 and 108 mA h·g–1, respectively, in exchange for the almost disappearance of the Qcha value at around 4.0 V. Such Qcha and Qdis values are similar to the reported values for x = 1.[7−13] According to the previous XANES analyses[8] and EPR measurements,[14,15] the electrochemical reaction at around 4.0 V is attributed to the redox reaction of Mn3+ ↔ Mn4+, whereas that at around 5.0 V is attributed to the redox reaction of Co3+ ↔ Co4+. The small amount of Qdis at around 4.0 V in x = 1 indicates that the Mn3+ ions still exist in the sample, although its ideal chemical formula is described as Li[Co3+Mn4+]O4. Figure S2 shows the charge and discharge curves for the x = 1 sample prepared at 1000 and 1100 °C. As the maximum heating temperature increases, the Qdis value decreases; for example, Qdis = 72 mA h·g–1 for 1100 °C. This is because the amount of the Li2MnO3 impurity increases, as later shown in Figures c and 6c.

Figure 4

Results of the Rietveld analyses for the pristine LiCoMn2–O4 samples with (a) x = 0, (b) x = 0.5, and (c) x = 1. Enlarged XRD patterns are also shown in the insets to clarify the presence of the Li2MnO3 phase (indicated by *). The 220 diffraction in x = 1 indicates the presence of Co ions at the tetrahedral 8a site.

Figure 6

Results of the Rietveld analyses for the fully charged LiCoMn2–O4 samples with (a) x = 0, (b) x = 0.5, and (c) x = 1. All samples were prepared using the electrochemical reaction charging up to 5.4 V. Enlarged XRD patterns are also shown in the insets to clarify the presence of the Li2MnO3 phase (indicated by *). The 220 diffraction in x = 1 indicates the presence of Co ions at the tetrahedral 8a site.

Figure shows the x-dependence of (a) Qdis at the initial cycle, (b) Eave of the initial discharge curve, and (c) W obtained by Qdis × Eave. Results of Li[Ni1/2Mn3/2]O4 with the P4332 space group[5] are also shown for comparison. The x-dependence of Qdis indicates a broad maximum at around x = 0.5, whereas Eave almost monotonically increases from 4.166 V at x = 0 to 4.996 V at x = 1. The observed Qdis is lower than the calculated Qtheo over the whole x range. Consequently, W gradually increases from 480 mW h·g–1 at x = 0, then reaches the maximum (=627 mW h·g–1) at x = 0.5, and finally decreases to 516 mW h·g–1 at x = 1. The maximum W value for x = 0.5 is slightly greater than that for Li[Ni1/2Mn3/2]O4 (=607 mW h·g–1), and as far as we are concerned, the x-dependence of W in LCMO has been clarified for the first time because previous studies on LCMO were focused on the Qdis values.[7−16]

Figure 3

Results of the electrochemical measurements for the LiCoMn2–O4 samples with 0 ≤ x ≤ 1: (a) discharge capacity at the initial cycle (Qdis), (b) average voltage (Eave) of the initial discharge curve, and (c) energy density (W) obtained by Qdis × Eave. The solid line indicates the Qtheo for LCMO calculated by (148.23 – 3.21x) mA h·g–1. The results for Li[Ni1/2Mn3/2]O4 with the P4332 space group are taken from ref (5).

Crystal Structure

In this section, we examined the crystal structures before and after the charge reaction to understand the electrochemical reaction scheme of LCMO. Figure shows the results for the Rietveld analyses for the pristine (a) x = 0, (b) x = 0.5, and (c) x = 1 samples. The crystal structure for x = 0 is assigned as a spinel structure with the Fd3̅m space group, in which Li+ and Mn3+/Mn4+ ions occupy tetrahedral 8a and octahedral 16d sites, respectively. Structural parameters, such as the cubic lattice parameter ac [=8.2225(1) Å] and oxygen positional parameter u [=0.264(1)], are consistent with the previous results for LiMn2O4 (Table ).[19−21]

Table 1

Structural Parameters for the Pristine LiCoMn2–O4 Samples with x = 0, 0.25, 0.5, 0.75, and 1 Determined by the Rietveld Analyses

				atomic coordination
x in LiCoxMn2–xO4	atom	Wyckoff position	occupancy	x	y	z	Biso (Å2)
SG: Fd3̅m, ac= 8.2226(1) Å, Rwp= 5.78%, and S = 0.513
x = 0	Li	8a	1.00	0.125	0.125	0.125	0.86(1)
	Mn	16d	1.00	0.5	0.5	0.5	0.61(1)
	O	32e	1.00	0.264(1)	0.264(1)	0.264(1)	0.87(1)
SG: Fd3̅m, ac= 8.1595(1) Å, Rwp= 5.21%, and S = 0.439
x = 0.25	Li	8a	1.00	0.125	0.125	0.125	0.67(1)
	Co2	16d	0.125	0.5	0.5	0.5	0.45(1)
	Mn	16d	0.875	0.5	0.5	0.5	0.45(1)
	O	32e	1.00	0.264(1)	0.264(1)	0.264(1)	0.52(1)
SG: Fd3̅m, ac= 8.1300(1) Å, Rwp= 7.39%, and S = 0.647
x = 0.5	Li	8a	1.00	0.125	0.125	0.125	1.2(1)
	Co	16d	0.25	0.5	0.5	0.5	0.37(1)
	Mn	16d	0.75	0.5	0.5	0.5	0.37(1)
	O	32e	1.00	0.264(1)	0.264(1)	0.264(1)	0.50(1)
SG: Fd3̅m, ac= 8.0791(1) Å, Rwp= 6.99%, and S = 0.582
x = 0.75	Li	8a	0.970(1)	0.125	0.125	0.125	0.76(1)
	Co1	8a	0.030(1)	0.125	0.125	0.125	0.76(1)
	Co2	16d	0.371(1)	0.5	0.5	0.5	0.26(1)
	Mn	16d	0.629(1)	0.5	0.5	0.5	0.26(1)
	O	32e	1.00	0.264(1)	0.264(1)	0.264(1)	0.37(1)
SG: Fd3̅m, ac= 8.0589(1) Å, Rwp= 6.42%, and S = 0.532.
x = 1	Li	8a	0.887(1)	0.125	0.125	0.125	0.43(1)
	Co1	8a	0.113(1)	0.125	0.125	0.125	0.43(1)
	Co2	16d	0.5(1)	0.5	0.5	0.5	0.22(1)
	Mn	16d	0.5(1)	0.5	0.5	0.5	0.22(1)
	O	32e	1.00	0.264(1)	0.264(1)	0.264(1)	0.52(1)

Before describing the result for x = 0.5, we wish to mention the result for x = 1. The majority of the x = 1 sample is in the Fd3̅m space group, as in the case for x = 0. However, as seen in the inset of Figure c, the diffraction lines indicated by * are clearly observed in the vicinity of the diffraction line around 2θ = 10° (the 111 diffraction line of the spinel structure). These diffraction lines originate from the Li2MnO3 phase with the monoclinic structure, whose weight fraction is determined to be 6.23%. Coexistence of the Li2MnO3 phase is also reported in previous structural analyses on x = 1.[9] Moreover, the intensity of the 220 diffraction line at around 2θ = 17° is slightly larger than that for x = 0 [compare Figure a,c], indicating that a small amount of metal (Co) ions exists in the tetrahedral 8a site. As a result, the crystal structure for the x = 1 sample is assigned as a mixture of the spinel phase with the Fd3̅m space group and the Li2MnO3 impurity with the C2/m space group. Because the amount of Co ions at the 8a site is determined to be 0.113(1), the actual formula for x = 1 can be represented as Li0.887Co0.113[CoMn]O4. Note that the mixture of 93.77 wt % Li0.887Co0.113[CoMn]O4 and 6.23 wt % Li2MnO3 is consistent with the Li/Co/Mn (=1/1/1) composition of the starting material: mol % of Li0.887Co0.113[CoMn]O4 and Li2MnO3 are calculated to be 89.3 and 10.7, respectively, providing the composition of Li/Co/Mn = 1.01/0.99/1.00.

The Li2MnO3 impurity is not observed in the x = 0.5 sample [see Figure b]. The crystal structure for x = 0.5 is thus assigned as the single-phase of the spinel structure with the Fd3̅m space group. Figure S3 shows the results for the Rietveld analyses for the pristine (a) x = 0.25 and (b) x = 0.75 samples. The situation for x = 0.25 is similar to that for x = 0.5; the sample crystallized into a single-phase of the spinel structure. By contrast, the Li2MnO3 impurity was observed in the XRD pattern for the x = 0.75 sample [see diffraction lines indicated by * in Figure S3b]. The weight fraction of the Li2MnO3 phase in x = 0.75 is determined to be 1.67%, and the actual formula for x = 0.75 can be represented as Li0.970Co0.030[CoMn]O4. As seen from Figure S4, the amount of the Li2MnO3 phase in LCMO rapidly increases at x ≥ 0.75.

We next examined the distributions of the Li2MnO3 phase in the x = 1 sample by Raman spectroscopy. Figure S5 shows the Raman spectra for the LCMO samples with 0 ≤ x ≤ 1. According to a factor group analysis,[22] five Raman active modes of A1g + Eg + 3F2g are predicted for the spinel structure with the Fd3̅m space group. The Raman spectrum for x = 0 shows a major Raman band at around 623 cm–1 and three minor Raman bands at 572, 477, and 365 cm–1. The major Raman band is assigned as the A1g mode, which corresponds to a symmetric vibration between Mn3+/Mn4+ and O2– ions.[16,23] The major Raman band at around 623 cm–1 splits into two or three Raman bands with x, probably due to the change in the proportion of Mn3+/Mn4+ ions.[16] As seen in Figure a, the x = 1 sample indicates three major Raman bands at around 652, 575, and 538 cm–1 and three minor Raman bands at around 475, 382, and 179 cm–1. Although these Raman bands are still unassigned, the Raman spectrum for x = 1 significantly differs from that for the single-phase of Li2MnO3 [Figure b]. This enables the LiCoMnO4 sample to be distinguished into Li0.887Co0.113[CoMn]O4 and Li2MnO3 phases. As provided in Figure c, the Li2MnO3 phase segregates from the Li0.887Co0.113[CoMn]O4 phase, not coexisting in the Li0.887Co0.113[CoMn]O4 phase.

Figure 5

Raman spectra for the (a) LiCoMn2–O4 sample with x = 1 and (b) Li2MnO3. (c) Two-dimensional Raman mapping for the x = 1 sample in the region of width 100 μm and height 25 μm. The regions of Li1−δCo1+δMnO4 and Li2MnO3 phases are shown by red and blue, respectively.

Figure shows the results of the Rietveld analyses for the fully charged LCMO samples with (a) x = 0, (b) x = 0.5, and (c) x =1. Figure S6 shows the results of the fully charged LCMO samples with (a) x = 0.25 and (b) x = 0.75. Here “fully” means the lithium cells were charged up to 5.4 V before the XRD measurements. The y values were calculated to be y = 0.15 for x = 0, y ≈ 0 for x = 0.25, y ≈ 0 for x = 0.5, y = 0.08 for x = 0.75, and y = 0.228 for x = 1, by ignoring the capacities consumed for electrolyte decompositions. As seen from Figures and S6, all LCMO samples maintain a spinel structure with the Fd3̅m space group, as in the cases for the initial (pristine) state. Structural parameters, such as ac and u, are summarized in Table .

Table 2

Structural Parameters for the Fully Charged LiCoMn2–O4 Samples with x = 0, 0.25, 0.5, 0.75, and 1 Determined by the Rietveld Analyses

				atomic coordination
x in LiyCoxMn2–xO4	atom	Wyckoff position	occupancy	x	y	z	Biso (Å2)
SG: Fd3̅m, ac= 8.0585(1) Å, Rwp= 5.46%, and S = 0.489
x = 0	Li	8a	0.15	0.125	0.125	0.125	0.86(1)
	Mn	16d	1.00	0.5	0.5	0.5	0.61(1)
	O	32e	1.00	0.262(1)	0.262(1)	0.262(1)	0.87(1)
SG: Fd3̅m, ac= 8.0525(1) Å, Rwp= 5.90%, and S = 0.539
x = 0.25	Co	16d	0.125	0.5	0.5	0.5	0.45(1)
	Mn	16d	0.875	0.5	0.5	0.5	0.45(1)
	O	32e	1.00	0.263(1)	0.263(1)	0.263(1)	0.52(1)
SG: Fd3̅m, ac= 8.0196(1) Å, Rwp= 7.70%, and S = 0.692
x = 0.5	Co	16d	0.25	0.5	0.5	0.5	0.37(1)
	Mn	16d	0.75	0.5	0.5	0.5	0.37(1)
	O	32e	1.00	0.263(1)	0.263(1)	0.263(1)	0.50(1)
SG: Fd3̅m, ac= 8.0109(1) Å, Rwp= 11.49%, and S = 1.04
x = 0.75	Li	8a	0.008	0.125	0.125	0.125	0.76(1)
	Co1	8a	0.030	0.125	0.125	0.125	0.76(1)
	Co2	16d	0.371	0.5	0.5	0.5	0.26(1)
	Mn	16d	0.629	0.5	0.5	0.5	0.26(1)
	O	32e	1.00	0.263(1)	0.263(1)	0.263(1)	0.37(1)
SG: Fd3̅m, ac= 8.0081(1) Å, Rwp= 6.02%, and S = 0.535
x = 1	Li	8a	0.228	0.125	0.125	0.125	0.43(1)
	Co1	8a	0.113	0.125	0.125	0.125	0.43(1)
	Co2	16d	0.5	0.5	0.5	0.5	0.22(1)
	Mn	16d	0.5	0.5	0.5	0.5	0.22(1)
	O	32e	1.00	0.263(1)	0.263(1)	0.263(1)	0.52(1)

Δac and ΔV for LCMO

Figure a shows the ac values before and after the fully charged reaction as a function of x in LiCoMn2–O4. The ac value before the charge reaction (pristine sample) monotonically decreases from 8.2225(1) Å at x = 0 to 8.0589(1) Å at x = 1. This linear decrease in ac agrees with the change in the ionic radius (r) from Mn3+ ions with CN = 6 (rMn = 0.58 Å) to Co3+ ions (rCo = 0.55 Å) with CN = 6, where CN is the coordination number.[24] The ac value after the fully charged reaction also indicates a linear decrease in ac with x, however, its slope (ac/x) is much smaller than that for the LCMO samples before the charge reaction. That is, the ac value slightly decreases from 8.0581(1) Å at x = 0 to 8.0083(1) Å at x = 1.

Figure 7

(a) Cubic lattice parameters (acs) for the pristine LiCoMn2–O4 samples and fully charged LiCoMn2–O4 samples. The solid line in (a) indicates the calculated ac value from . (b) Change in Δac and change in the lattice volume (ΔV) during charge and discharge reactions as a function of x in LiCoMn2–O4. The ΔV value for Li[Ni1/2Mn3/2]O4 was calculated from the data in ref (4).

Using these ac values for LCMO, the change in ac (Δac) and the change in the lattice volume (ΔV) are determined. As seen in Figure b, both Δac and ΔV values decrease with increasing x; for instance, Δac = −1.99% and ΔV = −5.88% for x = 0, Δac = −1.36% and ΔV = −4.01% for x = 0.5, and Δac = −0.63% and ΔV = −1.87% for x = 1. Here, the Δac and ΔV values for Li[Ni1/2Mn3/2]O4 are calculated to be −1.98 and −5.83%, respectively, using the reported ac values before (=8.167 Å) and after (=8.005 Å) the fully charged reaction.[4] Therefore, the Δac and ΔV values for x = 0.5 are smaller than those for Li[Ni1/2Mn3/2]O4, although the W value for x = 0.5 is greater than that for Li[Ni1/2Mn3/2]O4.

It should be noted that the Δac and ΔV values for x = 1 are significantly smaller compared with those for other positive electrode materials, such as LiCoO2 (ΔV ≈ +2.5% at the half-charged state)[25] and LiFePO4 (ΔV ≈ −6.9%).[26] The ΔV value of −1.87% for x = 1 can be regarded as a “zero-strain” lithium insertion material because one of the zero-strain lithium insertion materials, Li[CrTi]O4, indicates a ΔV of +0.7%.[27,28] Thus far, this statement has not been explicitly expressed, although Alcántara et al.[10] reported the change in ac for LiCoMnO4.

Zero-strain lithium insertion materials such as Li[CrTi]O4 and Li[Li1/3Ti5/3]O4 (LTO)[29−31] are used as negative electrode materials, whereas the x = 1 sample is used as a positive electrode material. XRD and Raman spectroscopy clarified that the zero-strain reaction scheme of LTO is achieved by a change in u on proceeding with the discharge reaction, that is, local structural changes in the LiO6 and TiO6 environments.[30,31] To compare with such a zero-strain reaction scheme, u values during various charge states were determined by the Rietveld analyses for x = 1. As shown in Figure a, u for LiCoMnO4 maintains a constant value (∼0.263) up to the fully charged state, suggesting that the zero-strain reaction scheme for x = 1 is different from that for LTO. Considering the ac value (∼8.00 Å) at the fully charged state, one can understand the zero-strain reaction scheme for x = 1. That is, the minimum ac value for face-centered cubic (FCC) consisting of only O2– ions is calculated to be ∼7.8 Å, using the relations and rO = 1.38 Å (CN = 4) [see Figure b]. Hence, the zero-strain character for x = 1 is achieved by a rigid framework structure of O2– with FCC packing. In other words, negative electrode materials with zero-strain are due to reversible changes in the local structures, whereas positive electrode materials with zero-strain are due to invariance in the local structures.

Figure 8

(a) Change in the oxygen positional parameter (u) as a function of y in LiCoMnO4. (b) Schematics of zero-strain reaction scheme for LiCoMnO4. The ac value (≈8.0 Å) for LiCoMnO4 is close to the minimum lattice parameter () for the FCC consisting of O2– ions.

Finally, we wish to describe the strategy for positive electrode materials with more high-energy density. Although the x = 0.5 sample indicates the maximum W value among the various spinel oxides, its W value is still lower than layered oxides; for instance, layered oxides comprising Li/Ni/Mn/O[2,32] exhibit more than 300 mA h·g–1 of Qdis and ∼3.5 V of Eave, resulting in more than 1000 mW h·g–1 of W. The present findings confirm that the redox reaction of Co3+ ↔ Co4+ is effective to increase Eave. Thus, layered oxides comprising Li/Co/Mn/O would exhibit greater W values, although the combination of Co and Mn ions in the layered structure is reported to be thermodynamically unstable.[33] Trials for preparing layered Li/Co/Mn/O oxides are underway in our laboratory.

Conclusions

To pair harmoniously with the high W value negative electrode materials, explorations for positive electrode materials with high W values have been performed in a series of LCMO spinels with 0 ≤ x ≤ 1. The maximum Qdis value for LCMO was exhibited at x = 0.5, whereas Eave increased monotonically with x. Therefore, the maximum W value (=627 mW h·g–1) was obtained at the x = 0.5 composition. The W value for x = 0.5 was slightly greater than that for Li[Ni1/2Mn3/2]O4, which has attracted much attention as a positive electrode material because of its high W value. XRD measurements using synchrotron radiation clarified another advantage for the x = 0.5 sample; that is, the ΔV value for x = 0.5 was about 2% smaller than that for Li[Ni1/2Mn3/2]O4. Thus, the x = 0.5 sample is regarded as a next-generation positive electrode material with a high W value and a long cycle-life. Concerning the ΔV values, the x = 1 sample showed the minimum ΔV value (=–2%) among the various positive electrode materials and can be thought of as a zero-strain lithium insertion material, similar to LTO and Li[CrTi]O4. The zero-strain character for x = 1 is due to the invariance in the local structures, which is different from those for LTO and Li[CrTi]O4.

Experimental Section

Synthesis and Characterization

Powder samples of LCMO with x = 0, 0.25, 0.375, 0.5, 0.75, and 1 were prepared using a two-step solid-state reaction technique, as reported previously,[4,5,19] to obtain highly crystallized LCMO samples. Regent grade LiOH·H2O (Wako Pure Chemical Industries, Ltd.), Co3O4 (Kojundo Chemical Laboratory Co., Ltd.), and MnO2 (Kojundo Chemical Laboratory Co., Ltd.) were mixed with a pestle and mortar and then pressed into a pellet of diameter 23 mm and thickness ∼5 mm. The pellet was first sintered at 900 °C for 12 h under flowing oxygen and then heated at 700 °C for 24 h, 600 °C for 24 h, and 500 °C for 48 h, successively, without cooling to room temperature between each temperature step. The heating and cooling rates were 200 and 60 °C·h–1, respectively. The effects of heating temperature were investigated only for x = 1; we set the first sintering temperature to 1000 °C or 1100 °C. A powder sample of Li2MnO3 was also synthesized. The reaction mixture of LiOH·H2O and MnO2 was heated at 900 °C for 12 h under flowing oxygen.

The obtained LCMO samples with 0 ≤ x ≤ 1 were characterized by powder XRD measurements equipped with Fe Kα radiation (D8 ADVANCE, Bruker AXS, Inc.), SEM (S-3600N, Hitachi High-Technologies Co., Ltd.), and Raman spectroscopy (NRS-3300, Jasco Co. Ltd.). XRD measurements were conducted in the 2θ range between 15 and 140° at a step-scan rate of 0.014°. Raman spectroscopy was performed using an excitation wavelength of 532 nm, supplied by a diode-pumped solid-state laser. Before the measurements, the sharp Raman shift of crystalline silicon was calibrated as 520 cm–1. The laser power and duration for taking one Raman spectrum were 1 mW and 720 s, respectively. For the x = 1 sample, a two-dimensional mapping for the Raman spectra was recorded using the excitation wavelength of 532 nm, to obtain information about distributions of Li1−δCoδ[CoMn]O4 and Li2MnO3 phases (inVia Raman microscope, Renishaw plc.). Here, δ is the amount of Co ions at the tetrahedral 8a (Li) site. The area for the Raman mapping was 100 × 25 μm. The laser power and duration for taking one spot of the Raman spectrum were 0.5 mW and 20 s, respectively.

Electrochemical Measurements

Electrochemical properties for the LCMO samples were examined in a nonaqueous lithium cell. The mixed electrode consisted of 88 wt % LCMO, 6 wt % conducting carbon, and 6 wt % polyvinylidene binder, as reported previously.[5,19] The surface area of the electrode was ∼2.00 cm2 (ϕ 16 mm). A stainless steel plate (ϕ 19 mm) was pressed onto the lithium metal and used as the counter electrode. The electrolyte was 1 M LiPF6 dissolved in ethylene carbonate (EC)/dimethyl carbonate (DMC) (EC/DMC = 3/7 by volume) solution (KISHIDA Chemical Co. Ltd.). After fabricating the lithium cells in an argon-filled glovebox, the cells were operated at a current of 0.6 mA (≈0.3 mA·cm–2) in the voltage range between 3.0 and 5.4 V. This applied current corresponds to ∼2 C rate. The temperature for the electrochemical measurements was 25 °C.

Synchrotron Radiation Study

To clarify changes in the crystal structures in the charged states, XRD measurements were also performed at the synchrotron radiation facility, Aichi Synchrotron Radiation Center. All of the charged LCMO samples were prepared using electrochemical reactions and packed into borosilicate capillary tubes with a diameter of 0.3 mm (W. Müller Glas Technik) in the argon-filled glovebox. The XRD patterns were recorded in the 2θ range between 5 and 95° using the two-dimensional detector (PILATUS 100 K, DECTRIS Ltd.) of the BL5S2 beamline. The wavelength of X-ray was determined to be 0.779547(3) Å from the XRD measurement of a silicon standard (NIST 640d). Rietveld analyses were carried out using the RIETAN-FP software,[34] and the schematics of the crystal structures were drawn by the VESTA software.[35] The weight fraction, that is, the mol fraction of the LCMO and Li2MnO3 phases were determined by the multiphase mode in the RIETAN-FP program.