Li-rich cathode materials are potential candidates for next generation Li-ion batteries. However, they exhibit large voltage hysteresis on the 1 st charge/discharge cycle involving a substantial (up to 1V) loss of voltage and therefore energy density. For Na cathodes, e.g. Na0.75[Li0.25Mn0.75]O2, voltage hysteresis can be explained by formation of molecular O2 trapped in voids within the particles. Here we show that this is also the case for Li1.2Ni0.13Co0.13Mn0.54O2. RIXS and 17 O MAS NMR show that molecular O2, rather than O2 2-, forms within the particles on oxidation of O 2at 4.6 V vs Li + /Li on charge. These O2 molecules are reduced back to O 2on discharge but at the lower voltage of 3.75 V explaining the voltage hysteresis in Li-rich cathodes. 17 O MAS NMR indicates a quantity of bulk O2 consistent with the O-redox charge capacity minus the small quantity of O2 loss from the surface. The implication is that O2, trapped in the bulk and lost from the surface, can explain O-redox.

The energy density of Li-ion batteries can be improved by storing charge at high voltages through the oxidation of oxide ions in the cathode material. However, oxidising O 2triggers irreversible structural rearrangements in the bulk and an associated loss of the high voltage plateau replacing it with a lower discharge voltage, as well as a loss of O2 accompanied by densification at the surface. Here we consider various models for O-redox proposed in the literature before describing a single unified model involving O 2oxidation to form O2, which is trapped in the bulk with the balance evolving from the surface. The model extends the O2 formation and evolution at the surface, which is well-known and well-characterised, into the electrode particle bulk as caged O2 that can be reversibly reduced and oxidised. This converged understanding allows us to propose practical strategies for avoiding O-redox-induced instability offering potential routes towards more reversible high energy density Li-ion cathodes.Since the discovery of 'anomalous' extra capacity to store charge in 3d transition metal oxide Li-rich cathode materials in the early 2000s, 1-4 there has been intense research interest seeking to understand the origin of the effect. [5][6][7][8][9] Over the years, these compounds have grown in number extending to include materials based on 4d and 5d transition metal oxides. [10][11][12][13] In the case of a conventional Li transition metal oxide intercalation cathode, Li + ions are extracted on charging, with charge-compensation by oxidation of the transition metal ion, the process is reversed on discharge, e.g. Li1-xMn2O4 (0 < x < 1). In contrast, the Li-rich cathodes, such as Li1.2Ni0.2Mn0.6O2, Li1.3Nb0.3Mn0.4O3, Li2RuO3 and Li2Ir0.5Sn0.5O3, extend the capacity to store charge by oxidation of the O 2ions. 10,[13][14][15][16] The ability of anions to undergo redox reactions is not without precedent, for example the S 2-/S2 2reaction in sulphides is well known but the phenomenon was not recognised in oxides until more recently. 17 The oxidation of O 2in cathode materials is typically accompanied by a high voltage plateau (usually ~4.5 V vs Li + /Li for 3d cathodes) on charge followed by an S-shaped discharge profile, Fig. 1. Early models posited that the charging plateau was associated completely with the irreversible loss of oxygen from the lattice (O-loss), which alongside extraction of Li + gives rise to the net loss of Li2O. 2,4 Online mass spectrometry showed that O2 gas was released from the surface of the material. 18 Later work showed that there was an insufficient degree of reduction observed of the transition metal ions on re-lithiation to explain the large discharge capacity 19 and quantitative studies also revealed an insufficient amount of evolved O2 to account for the charging capacity associated with the plateau. 14,20 Consequently, the idea was developed that reversible oxidation and reduction of O 2ions in the bulk compensate for the extraction and reinsertion of Li + beyond the limit of TM redox. 5,...