Conclusion

Figure 4.9: Normalized cobaltL2,3 white line in- tensity and L3/L2 ratio from two different crys- tallites of Li1−xCoO2.

valence in the anode, and are not Li⁺ ions as is implied by the name “lithium-ion battery”

[37]. The present experimental work shows, as expected, that the lithium in the cathode LiCoO2 is ionic Li⁺. In conjunction with the previous theoretical work [54, 55], however, we see that the chemistry of the cathode material originates primarily with electron transfers from lithium to the oxygen ions. Perhaps the name “lithium-lithium oxide” battery better conveys the chemistry of the full cell.

Chapter 5 Electronic Structure and Charge Compensation in LiNi

0.8

Co

0.2

O

2

5.1 Introduction

The compound LiNi0.8Co0.2O2 is a promising cathode for lithium batteries. It has the same layered structure as LiCoO₂ but with a substitution of nickel atoms at cobalt sites.

In LiNi0.8Co0.2O2 the nickel and cobalt ions are both octahedrally coordinated by oxygen ions and occupy the same crystallographic sites. This material is an improvement over LiNiO2 because it combines a high cycling capacity with safe operating conditions even in the delithiated charge state. LiNi0.8Co0.2O2 remains as a single phase during deep lithium deintercalation, allowing reversible cycling from Li1.0 to Li0.4 [70]. Also, the LiNi0.8Co0.2O2

cathode is less susceptible to cationic disorder (TM ions occupying lithium sites) than the LiNiO2 system. LiNi0.8Co0.2O2 has a slightly lower voltage vs. lithium when compared to LiCoO2 and consequently, electrochemical cells are less prone to electrolyte decomposition.

Finally, the reduction of cobalt also allows for a cheaper, more environmentally friendly cathode.

During electrochemical discharge, lithium ions and their associated electrons are inter- calated into the TMO2 host. There have recently been a number of investigations into how the intercalated lithium affects the electronic structure of the host [10, 48, 57, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81]. The results of these studies have led to recent debate over the amount of charge transferred to the oxygen and TM ions. Analyses of the TM core edges using x-ray absorption spectroscopy (XAS), magnetic susceptibility, electronic conductivity, and thermoelectric measurements, have found that upon lithium deintercalation the TM ion is oxidized to the tetravalent state [10, 70, 71, 77, 78]. Studies of the LiNi1−yCoyO2 system suggest that during discharge, the nickel ions are immediately oxidized to the tetravalent state while the cobalt oxidation occurs only in the heavily delithiated state [10, 70, 71]. In LiNi0.8Co0.2O2 the preferential oxidation of nickel over cobalt is attributed to the position of the Fermi level, which lies within the energy gap between the Ni²⁺/Ni³⁺ and Ni³⁺/Ni⁴⁺

redox couples. It is argued that the Fermi level drops into the Ni³⁺/Ni⁴⁺ band during

the early stages of delithiation and the nickel is oxidized. The t2g band of cobalt appears at a lower energy, requiring the removal of up to 80 percent of the lithium atoms before the Fermi level is lowered sufficiently for cobalt oxidation [70]. Although there is general agreement that the cobalt ion participates little in the redox processes that occur during cycling, the behavior of the nickel ion is still not well understood. The notion of a trivalent nickel ion in the fully lithiated material has been disputed by recent results of Montoro et al., which suggest that in LiNi0.5Co0.5O2 the nickel ions are divalent and are partially oxidized to a trivalent state during delithiation [48].

Absorption studies using hard X rays have been used to investigate the electronic struc- ture around the TM ions. However, these studies are unable to measure electronic transi- tions with low absorption energies like those contributing to the oxygenK-edge. Therefore, these measurements are relatively insensitive to the atomic states of oxygen. More indirect experiments, such as magnetic measurements, can be difficult to interpret, and may be misleading because changes in the spin density, or the number of unpaired electrons about an ion, are not necessarily correlated with changes in the charge density.

Soft x-ray absorption and transmission EELS are well suited for studying the electronic structure near the oxygen and TM ions at different states of lithiation. Recent studies of the oxygen K-edge and nickel and cobalt L2,3-edges have demonstrated that the oxygen ion is responsible for accommodating much of the lithium 2s electron during intercalation [57, 79, 80, 81]. These studies suggest a nearly constant net charge density in the vicinity the transition metal ion, with significant changes in occupancy of the oxygen 2p band. The details of the redox energies of the individual transition metal ions appear to be subordinate to changes occurring in the oxygen valence.

First-principles calculations by Wolverton and Zunger have found that in LiCoO2the net charge density about the cobalt ion is unaffected by lithium concentration. This behavior is attributed to a rehybridization of the Co-O bond, which produces invariant energy levels and electronic stability over a wide range of lithium concentration [69]. Other first-principles calculations of LiCoO2have suggested that much of the lithium 2selectron is donated to the oxygen, allowing the cobalt valence to remain predominately in the trivalent state [54, 82, 64]. Theseab-initiostudies add insight into the experimental results for the ordered LiCoO₂ and LiNiO2systems. However, calculations on a disordered distribution of nickel and cobalt atoms are not practical owing to the complexity of the unit cell of Li(Ni,Co)O₂. Here we

report results from an experimental investigation using transmission EELS to elucidate changes in the electronic structure of Li1−xNi0.8Co0.2O2 with the removal of lithium.