Delithiation Mechanisms in Acid of Spinel LiMn_2-xM_xO₄ (M = Cr, Fe, Co, and Ni) Cathodes

Synthesis

Characterization

As pointed out in the introduction, the delithiation of spinel oxides with acid has been proposed to occur by a disproportionation of Mn³⁺ ions into Mn²⁺ and Mn⁴⁺ ions that is accompanied by the loss of soluble Li₂O and MnO into the solution while Mn⁴⁺ remains in the solid (Reaction 1).⁸ Using this mechanism and assuming only Mn³⁺ undergoes disproportionation, the expected (calculated) compositions after delithiation with acid are given in Table I for the LiMn_2-xNi_xO₄ (0 ≤ x ≤ 0.5) and LiMn_2-xM_xO₄ (0 ≤ x ≤ 1.0) (M = Cr, Fe, and Co) series. The oxidation states of the ions in the initial samples are indicated in column 2, the species dissolving into the solution are given in column 3, and the calculated compositions of the solid remaining after delithiation are given in column 4 of Table I. The compositions in column 4 are then normalized to obtain an oxygen content of 4 to be consistent with the spinel formula and are given in column 5. As the transition-metal dopant content in the initial samples increases, the amount of Mn³⁺ content decreases (column 2 in Table I).

Figures 2 through 5 compare the XRD patterns of the LiMn_2-xNi_xO₄ (0 ≤ x ≤ 0.5) samples, the LiMn_2-xCo_xO₄ (0 ≤ x ≤ 1.0) samples, the LiMn_2-xCr_xO₄ (0 ≤ x ≤ 1.0) samples, and the LiMn_2-xFe_xO₄ (0 ≤ x ≤ 1.0) samples, respectively, before and after delithiation. Their lattice parameters can be seen in Table II. All of the reflections in the samples before and after delithiation could be indexed based on the Fd-3m cubic spinel structure, indicating the absence of impurity phases, except for a few samples. A trace amount of Ni_1-xLi_xO impurity can be found in the x = 0.5 sample in the Ni series. In the case of LiMnCoO₄, a small amount of a Co-rich layered oxide phase is present. In the Fe series, when x ≥ 0.75, there are several additional peaks on the lower 2θ side of the existing peaks, which are most likely from an Fe-rich spinel impurity.^28,29 The initial LiMn_2-xCo_xO₄ samples in the Co series exhibit increased peak broadening with increasing Co content. The Co and Mn ions do not order in the spinel structure, so increasing the Co content increases the structural disorder and leads to peak broadening. However, the trend reverses upon going from the x = 0.75 sample to the x = 1.0 sample, as the 1:1 ratio of Mn:Co may provide a better structural order and crystallinity due to the lack of Mn³⁺ ions. This behavior was also seen by Amarillo et al.³⁰ Although another study has reported Cr ions in the 8a tetrahedral sites depending on the composition and synthesis method,³¹ this was not found to be the case in our study as the (111) peaks at ∼19° did not show a decrease in intensity and the (220) peaks at ∼31° did not exhibit an increase in intensity as would occur if an appreciable amount of Cr were in the tetrahedral sites.^29,32

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The peaks in the as-prepared Co and Cr samples shift to higher 2θ values as the dopant content increases due to the substitution of smaller Co³⁺ ions (0.61 Ǻ) and Cr³⁺ ions (0.615 Ǻ) for the larger Mn³⁺ ions (0.645 Ǻ). The peaks in the as-prepared Ni samples also shift to higher 2θ values as the Ni content increases since the substitution of Ni for Mn creates a Ni²⁺ ion (0.69 Ǻ) and a Mn⁴⁺ ion (0.53 Ǻ) at the expense of two Mn³⁺ ions (0.645 Ǻ), so the net effect is a decrease in lattice parameter. However, the peaks in the as-prepared Fe samples have relatively constant locations with increasing Fe content since Fe³⁺ ions (0.645 Ǻ) and Mn³⁺ ions (0.645 Ǻ) have identical size. Except for the materials where no delithiation occurs, the peaks in the delithiated samples all shift to higher 2θ values and the lattice parameter values given in Table II decrease due to the oxidation of the larger Mn³⁺ ions to smaller Mn⁴⁺ ions. As expected, the shift to higher 2θ values and the change in lattice parameters after delithiation decrease with increasing dopant content due to the decrease in Mn³⁺ content in the initial samples and amount of the disproportionation Reaction 1. For the highest-doped sample in each series, which contains little or no Mn³⁺, the reflections should not shift and the lattice parameters should not change after the delithiation treatment with the acid. This is true for the Ni and Cr series. However, LiMnCoO₄, which may contain a small amount of Mn³⁺ due to the Co-rich layered oxide impurity, shows a very small change in lattice parameter after the acid-delithiation treatment. The difficulty in keeping all Mn as Mn⁴⁺ in LiMnCoO₄ during the high-temperature synthesis may lead to the small amount of Mn³⁺, which would be accompanied by the presence of a trace amount of layered oxide impurity phase or oxygen deficiency. Similarly, LiMnFeO₄, which also contains an impurity phase, exhibits a slight peak shift and lattice parameter change upon delithiation.

Table II gives the experimentally obtained compositions based on the Li, Cr, Mn, Fe, Co, and Ni contents determined by ICP-OES analysis for the LiMn_2-xNi_xO₄ (0 ≤ x ≤ 0.5) and LiMn_2-xM_xO₄ (0 ≤ x ≤ 1.0) (M = Co, Cr, and Fe) series. As seen in Table II, the experimentally determined lithium contents in the samples match well with the values calculated based on the disproportionation reaction of Mn³⁺ (Reaction 1). The experimental compositions of the transition-metal dopants in column 5 are all slightly lower than the expected values in column 4, except the samples where no delithiation occurs. To investigate this issue, the dilute sulfuric acid solvent was also tested using the ICP, and it was found that 0.03–0.04 moles of Cr, Fe, Co, and Ni were dissolved in addition to the expected amounts of Li and Mn. This explains the discrepancy between the actual and expected transition-metal contents found in Table II. This can be understood by the fact that as Li₂O and MnO are dissolved from the solid, the transition-metal dopants left on the surface may slightly dissolve into the sulfuric acid medium during the extended reaction time. This dissolution, however, does not cause any additional delithiation.

There are a few other discrepancies in Table II as well. Although no lithium is expected to be extracted from LiMnCoO₄, the small amount of lithium extracted from it is due to the following. As pointed out by Andriiko et al.,³³ LiMnCoO₄ lies just past a phase boundary (LiMn_1.1Co_0.9O₄ was the homogenous limit) and two phases are formed: one is the desired spinel phase, while the other is a Co-rich layered oxide phase. From our XRD pattern, we can see that our LiMnCoO₄ sample shows a small amount of layered oxide impurity phase. Because Co is preferentially present in the layered oxide impurity phase, the spinel phase is slightly richer in Mn and deficient in Co compared to the nominal composition of LiMnCoO₄. This results in a small amount of Mn³⁺ in the spinel phase and a consequent extraction of a small amount of lithium from the spinel phase. A similar reason also causes the delithiated x = 0.75 and 1.00 samples in the Fe series to contain much less Li than expected. As mentioned above in the XRD discussion, those two samples have significant amounts of an Fe-rich spinel impurity phase: one closer to Fe₃O₄ and the other closer to LiMn₂O₄. The LiMn₂O₄-like phase would experience delithiation due to its Mn³⁺ content.

These data presented in this investigation with four series of samples clearly establish that acid delithiation of spinel oxides occurs by a disproportionation of Mn³⁺ into Mn⁴⁺ and Mn²⁺ that is accompanied by a leaching out of corresponding amounts of Li₂O and MnO into solution according to Reaction 1 proposed by Hunter.⁸ The data also establish that in spinel solid solutions consisting of Li, Mn, and either Cr, Fe, Co, or Ni, the amount of lithium extracted with acid is proportional to the Mn³⁺ content in the initial sample. Thus, the amount of lithium extracted with acid could be used to reveal quantitatively the Mn³⁺ content in such complex spinel oxides.

The experimental composition data presented in Table II reveal that M³⁺ = Cr³⁺, Fe³⁺, and Co³⁺ ions in spinel oxides do not disproportionate into M⁴⁺ and M²⁺. For example, if Co³⁺ ions also disproportionate similar to Mn³⁺ ions, then one would expect to obtain spinel-like λ-MnCoO₄, but that is not the case. The same statements can be made for Cr³⁺ and Fe³⁺ ions as well. However, lithium extraction with acid has been reported in the literature with layered oxides such as LiCoO₂ and LiNi_1-xCo_xO₂.^9–12 This could be largely related to the ion exchange of Li⁺ ions by H⁺ ions in acid. However, such ion exchange of Li⁺ ions by H⁺ ions does not occur in these spinel oxides as illustrated by the compositional analysis data in Table II, which shows final compositions very close to the expected values. If such an ion exchange occurred, then the calculated and observed lithium contents and compositions in Table II would not match with each other. For example, if an exchange of Li⁺ ions by H⁺ ions occurs in acid, then the LiMn_1.5Ni_0.5O₄ composition would not remain identical (Table II) before and after treating with the acid. Because ICP analysis cannot provide compositional data for H and O, one could argue that protonation may be occurring with a corresponding increase in oxygen content, but the experimentally determined oxygen contents of 4.0 ± 0.03 for the initial materials and delithiated samples further confirm that no ion exchange has occurred during the acid delithiation process of these spinel oxides. It is possible that the protons could not be accommodated or stabilized in the 8a tetrahedral sites of the spinel lattice, while they may be readily accommodated or stabilized in the octahedral sites of the lithium layer in layered oxides.

Furthermore, the Ni^2+/3+ redox energy is close to that of Mn^3+/4+ energy in oxides, so one may anticipate the following equilibrium under some conditions:

The fact that no lithium could be extracted from LiMn_1.5Ni_0.5O₄ with acid (Table II) reveals that Reaction 2 remains completely towards the left without the formation of any Mn³⁺ in the initial and delithiated samples. In other words, the Ni^2+/3+ and Mn^3+/4+ bands do not overlap in these spinel oxides as depicted in Figure 6, which is consistent with the neutron scattering and computational calculation data found in the literature^34–36. The compositional analysis data in Table II are also consistent with a lying of the Mn^3+/4+ energy well above that of the Co^3+/4+ energy as depicted in Figure 6.

Zoom In Zoom Out Reset image size

Extraction of lithium from four series of spinel oxides, LiMn_2-xNi_xO₄ (0 ≤ x ≤ 0.5) and LiMn_2-xM_xO₄ (0 ≤ x ≤ 1.0) (M = Cr, Fe, and Co), with acid has been investigated systematically with an aim to establish the lithium extraction mechanism. XRD and chemical-compositional analysis data establish that the lithium extraction from spinel oxides occurs by a disproportionation of Mn³⁺ ions into Mn⁴⁺ and Mn²⁺ ions, with the latter being leached out into the solution, as was proposed by Hunter in 1981. The amount of lithium extracted is proportional to the amount of Mn³⁺ in the initial samples. The chemical-composition analysis of the delithiated samples in the four series of samples also reveal that other M³⁺ (M = Cr, Fe, and Co) ions in spinel oxides do not disproportionate into M⁴⁺ and M²⁺ ions unlike the Mn³⁺ ions. Furthermore, the data indirectly establish that the Mn^3+/4+ band does not overlap with the Ni^2+/3+ band, so in mixed oxides the Mn⁴⁺ + Ni²⁺ ↔ Mn³⁺ + Ni³⁺ equilibrium (Reaction 2) lies completely towards the left and no Mn³⁺ is formed. Moreover, ion exchange of Li⁺ ions by H⁺ ions does not occur in spinel oxides as the H⁺ ions may not be stabilized in tetrahedral sites of the spinel framework, while such ion exchange is known to occur readily in layered LiMO₂ oxides, as the H⁺ ions may be readily stabilized in octahedral sites of the lithium layer in layered oxides.

Since only Mn³⁺ disproportionates into Mn⁴⁺ and Mn²⁺ while the other transition-metal ions do not, the study suggests that Mn dissolution could become much more pronounced than other transition-metal-ion dissolutions in mixed-metal oxide cathodes in the presence of trace amounts (ppm levels) of acid (e.g., HF) in the electrolyte when the cells are operated within the electrolyte stability region of <4.3 V vs. Li/Li⁺. However, at higher operating voltages (>4.3 V vs. Li/Li⁺), the cathode surface could become unstable in contact with the organic liquid electrolyte, which could lead to the dissolution of other metal-ions as well, even when no Mn³⁺ ions are present in the samples. Moreover, the chemical instability of cobalt-rich compositions due to an overlap of the Co^3+/4+:3d band with the top of the O²⁻:2p band (Figure 4) could further exacerbate the metal-ion dissolution since the oxidation of O²⁻ ions at higher operating voltages of >4.3 V in the Co-containing samples¹⁶ could promote the chemical reactivity and side reactions of the cathode with the electrolyte. Although the present work has been carried out at a much higher concentration of acid than would be encountered in an actual lithium-ion cell, the results clearly provide a trend and could provide insights in designing high-performance cathodes for lithium-ion batteries.

Zoom In Zoom Out Reset image size

This work was supported by the National Science Foundation Materials Interdisciplinary Research Team (MIRT) grant DMR-1122603 and Welch Foundation grant F-1254.