Eric McCalla

Experimental and Theoretical Considerations

The method used to synthesize hundreds of samples across the pseudo-ternary systems was to make combinatorial arrays of milligram-scale oxides. Throughout this section, the method used to make samples in the Li-Mn-Ni-O system will be described. The only variation needed to make samples in the Li-Co-Mn-O system is to replace the nickel starting solution with a cobalt solution (both were nitrates in these studies). This method, developed by Carey and Dahn [10], was closely based on that typically used for large scale samples made in a tank reactor. Carey [10] mixed a total of 10 μL of roughly 2 M lithium nitrate (Aldrich, 98 \(\%\)), manganese nitrate (Sigma-Aldrich, 97 \(\%\)) and nickel nitrate (Sigma-Aldrich, 97 \(\%\)) using a Cartesian Pixsys solution-processing robot shown in Fig. 2.1. Figure 2.2 (a) shows how these solutions were dispensed onto an alumina plate (Pi-Kem, 96 \(\%\)) coated with stearic acid (Aldrich, 96 \(\%\)) which served to bead the solutions. Carey then added ammonium bicarbonate (Alfa Aesar, 98 \(\%\)) in excess to cause co-precipitation of Li, Mn and Ni carbonates. After drying at 55 °C, the sample was made up of the mixed carbonate and any other products of the reaction (in Chap. 3, this will be shown to be primarily ammonium carbonate). Carey then heated the samples to 800 °C for 3 h in air to form the oxides. Silicon (100) wafers were then covered in a tacky mixture of Trilene-65 (a polymer mixture made by Lion Copolymer) and cyclohexane. The wafer was placed over the alumina plates and flipped in order to transfer the samples onto the silicon wafer. The final products were shown to be the expected spinel oxides by XRD.

Materials Near the Layered Boundary

The structural changes taking place during regular cooling of materials with metallic compositions near that of LiNi0.5Mn0.5O2 were studied after heating under various oxygen partial pressures and were related to features in the phase diagram determined previously. Materials made with 5 \(\%\) excess lithium (at point B9 in Fig. 9.1) were found to stay single phase even when the regular cooling rate of 5–10 °C/min was used. These single-phase layered materials showed good electrochemical performance with a reversible capacity of 180 mAh/g for the quenched sample and 170 mAh/g for the regular cooled sample when cycled up to 4.8 V. By contrast, under the same cycling conditions, samples without excess lithium (at point A9 in Fig. 9.1) made in 2 \(\%\) oxygen had a capacity of 140 mAh/g when quenched and only 90 mAh/g when regular cooled. The changes in the XRD pattern for this A9 sample were small with peak broadening only seen at high angle, consistent with the first sign of phase separation into layered–layered nano-composites. The dramatic loss in capacity seen in the sample made in 2 \(\%\) oxygen at point A9 can be attributed to nickel clustering on the lithium layer such that lithium islands form, many of which would be surrounded by the clustered nickel. This behavior was expected from the phase diagram where phase separation into a nickel rich and nickel poor phase was demonstrated in Chap. 6, and this clustering of nickel was also seen in a Monte Carlo simulation performed in this chapter.

Combinatorial Studies of Compositions Containing Layered Phases in the Li–Mn–Ni–O System

The phase diagrams obtained for quenched samples show that in both three-phase regions all samples contain some of the N and S phases, while during slow cooling they convert to structures containing at least some R and M phases. Though the thermodynamics of this are complex, a qualitative understanding can be achieved by considering the results of the Rietveld refinement. The R and M phases were collectively more ordered than the N and S phases (the N phase in particular showed considerable disorder in the hexagonal layers). This would suggest that the R and M phases have lower combined internal energies and lower combined entropies than the combination of the N and S phases. Since the structure of the three-phase regions at high temperature is driven by entropy, the N and S phases are present in all points in the regions. Upon slow cooling, a temperature is reached where internal energy becomes more important and there is still sufficient thermal energy for atomic transport such that the R and M phases begin to appear in all samples. Further study is required to have a better understanding of the thermodynamics and kinetics of these phase transformations.

Combinatorial Studies in the Li–Co–Mn–O System

The layered single-phase region in the Li–Co–Mn oxide pseudo-ternary system was explored by a solution-based combinatorial approach. The results showed that the layered region is a single composition line corresponding to cobalt being constrained to the 3+ oxidation state only. This composition line, joining LiCoO2 to Li2MnO3, was a solid solution over its entire length when samples were quenched from either 800 or 900 °C. Upon slow cooling, the structures phase separated near the center of the line with the maximum phase separation occurring over the range x = 0.2–0.8 in Li[Li\(_{(1-x)/3}\)Co\(_x\)Mn\(_{(2-2x)/3}\)]O2 when samples were cooled from 900 °C at a rate of 1 °C/min. These endpoints correspond to Co and Mn2Li domains with approximately 20 \(\%\) disorder on the transition metal layers. Such disorder was also found over a range of temperatures during cooling using a Monte Carlo simulation.

Investigations of Bulk Li–Mn–Ni–O Samples to Confirm the Combinatorial Studies

The key features in the Li–Mn–Ni–O pseudo-ternary phase diagrams, as determined using combinatorial samples, were confirmed with bulk samples synthesized under various conditions via two different synthesis routes. The four-phase equilibrium was observed and transformations were found to be reversible, thereby confirming that two three-phase regions exist and transform during slow cooling. The primary importance of this work is a better understanding of how the phase boundaries and coexisting regions transform when cooled at rates typically used commercially. It is also important to recognize that a small amount of transformation occurred even for samples quenched on a copper or steel plate. These changes cannot be avoided entirely without quenching in liquid nitrogen or avoiding the compositions where the transformations occur. Understanding the phase diagrams should have a significant impact on research focused on composite electrodes in the Li–Mn–Ni–O system, since this work identified the compositions and conditions required to obtain layered–layered nano-composites and layered–spinel composites.

Combinatorial Studies of the Spinel and Rocksalt Regions in the Li–Mn–Ni–O System

The entire spinel and rocksalt solid-solution regions of the Li–Mn–Ni oxide pseudo-ternary system were determined at 800 °C when regular cooled in air and when quenched or regular cooled in oxygen. All samples discussed here either contained one or two phases; and no evidence for nonequilibrium behaviour was seen. Two-phase fits in the coexistence regions were used to determine lattice parameters to allow the drawing of tie-lines. The lever rule was used to determine the boundaries and showed excellent agreement with visually identified phases. Milligram-scale combinatorial samples can therefore be used to obtain a high degree of precision in the mapping of pseudo-ternary phase diagrams.

Optimization of the Synthesis of Combinatorial Samples

Mechanisms for lithium loss during and after the formation of combinatorial samples of lithium nickel oxide were identified. During synthesis in air, the main source of lithium loss was the decomposition of lithium carbonate that failed to react with the nickel oxide structure. The fact that the formation of LiNiO2 was hindered significantly in air was attributed to the presence of constituents in air other than oxygen and nitrogen, the likeliest candidates being carbon dioxide and water vapor. The second mechanism for lithium loss was the thermal decomposition of Li\(_x\)Ni\(_{2-x}\)O2. TGA was used to confirm that both lithium oxide and oxygen were lost when the samples were heated in either air or oxygen. In both cases, the loss of lithium from the samples was attributed to the conversion of lithium oxide to lithium peroxide vapor. Synthesizing the samples in dry, carbon dioxide free air would therefore result in lithium content very close to that seen in oxygen, the only difference being attributed to a slight increase in the rate of decomposition of lithium nickel oxide as seen in the TGA. Combinatorial samples of Li\(_x\)Ni\(_{2-x}\)O2 cannot be made with x > 0.77 at 700 °C in air. In a flow of oxygen, excess lithium was used to react a sufficient amount of lithium into the material to form Li0.95Ni1.05O2 at 800 °C on an alumina substrate. These conditions allow the simultaneous synthesis of the layered and spinel structures in the Li–Mn–Ni–O system.

Layered Materials with Metal Site Vacancies

Layered Li–Mn–Ni–O materials were analyzed in a region of the phase diagram where a strange bump was seen in the boundary of the layered region. These structures were found to contain a significant amount of metal site vacancies. The maximum vacancy content was found to result in highly ordered monoclinic structures where manganese occupies two of the \(\sqrt{3} \times \sqrt{3}\) superlattices on the transition metal layers while the third was randomly filled with nickel, lithium and vacancies. The resulting ordering predicted by a Monte Carlo simulation was consistent with the sharp ordering peaks seen in the XRD patterns. The vacancy concentrations were confirmed by Rietveld refinement, density measurements and redox titration; all of which were in good agreement. The role of the vacancies during electrochemical cycling, if there is one, remains unclear. The material with the greatest possible vacancy concentration, Li[Ni\(_{1/6}\Box_{1/6}\)Mn\(_{2/3}\)]O2, showed electrochemical behavior consistent with lithium-rich layered materials, namely high irreversible capacity associated with the high voltage plateau and voltage fade associated with conversion to spinel. However, this material was not, in fact, lithium-rich given that the vacancies result in lithium occupying only 50 % of metal sites. The vacancy results also demonstrated that there were roughly 1 % vacancies in a stoichiometric lithium-rich material lying along the line from Li2MnO3 and LiNi0.5Mn0.5O2. The Monte Carlo simulation suggested that this allows Ni\(^{3+}\) to substitute for Mn\(^{4+}\) on two of the superlattices. This has never been recognized before and a complete understanding of the starting material is crucial to fully understand the complex electrochemical behavior of the lithium-rich positive electrode materials. The exact shape of the other side of the bump (to the left of Li[Ni\(_{1/6}\Box_{1/6}\)Mn\(_{2/3}\)]O2 in the Gibbs triangles) has not been determined. It is also unclear as to why the bump would be so sharp on both sides and this is worthwhile for further study.

Conclusions and Future Works

Figure 10.1 shows the Li–Co–Mn–Ni–O pseudo-quaternary system with the Li–Co–Mn–O and Li–Mn–Ni–O faces shown, as determined with combinatorial samples quenched from 800 °C. Some approximations were made to join the two faces since the Li-Co-Mn-O face was synthesized in air while the nickel containing samples were made in oxygen. Nonetheless, the two faces join quite well. The pyramid strongly suggests that both single-phase regions of importance for battery materials, spinel and layered, extend into the pyramid and form relatively large three dimensional shapes. On the Li–Co–Mn–O system, the layered region is restricted to a single line showing that cobalt is always synthesized in the 3+ state as it is in LiCoO2. By contrast, nickel can be in the 2+ state as in NiO rocksalt or the 3+ state as in layered LiNiO2 such that a much larger and more complex layered region exists on the Li–Mn–Ni–O face. The spinel-layered co-existence region is also simpler in the Li–Co–Mn–O triangle with all tie-lines connecting to either the cobalt spinel, Co3O4, or the manganese layered material, Li2MnO3, while in the Li–Mn–Ni–O system there are 2 three-phase regions. The differences between the roles of cobalt and nickel should prove significant in upcoming combinatorial work in the Li–Co–Mn–Ni–O pseudo-quaternary system that is of extreme interest for battery materials as it includes commercial materials such as Li[Ni\(_{1/3}\)Mn\(_{1/3}\)Co\(_{1/3}\)]O2 [93] as well as promising spinel-layered core-shell materials [88] and lithium-rich layered materials [3].