Multi-scale modelling of pristine and transition metal doped spinel LiMn2o4 cathode materials

Abstract

Lithium-manganese-oxide with a spinel structure has attracted enormous attention as a future cathode material for lithium-ion batteries. LiMn2O4 is affordable due to the abundance of manganese in South Africa (80% of the world reserves) and provides high-rate capability. However, its commercialization is currently hindered by structural instabilities linked to its Mn3+ component. During cycling, the increasing amount of Mn3+ (≥ 50%) onsets the Jahn-Teller effect, which destabilizes the spinel structure, leading to a loss of capacity. Hence, in this work, there emanates a need to introduce Ni and Co as dopants to reduce the concentration of Mn3+ and also stabilize the transition metal and oxygen bond in the structure. To incorporate this effectively, the Buckingham potentials for the interactions that arise from doping LiMn2O4 with Co and Ni were derived using a technique inspired by machine learning methods. The Buckingham potential parameters were derived with guidance from potential energy surfaces from electronic structure calculations. The potential reduced the fitted structure properties and elastic constant to an acceptable percentage difference of less than 7.1%. The potentials were utilized to investigate the microstructural changes and Li+ transport properties of LiMn2O4 influenced by the introduction of Co and Ni on the Mn3+ sites in the structure. The developed potentials were able to determine the thermodynamic properties aligned with experimental observations such as melting point and the crucial high-temperature behaviour of LiCo2O4 and LiNi2O4 in the test of their veracity. The melting point of LiNi2O4 was found to be lower than that of LiCo2O4, which could be attributed to the observed lower melting point of Ni when compared to that of Co. Moreover, the potentials were further employed to perform simulated synthesis of LiCo2O4 and LiNi2O4 nanospherical structures. X-ray diffraction (XRD) patterns and atomic structure snapshots showed the dominance of the LiM2O4 (M=Co, Ni) co-existing with the M3O4 and Li2MO3 high-temperature impurity phases, which is in line with the findings of Mouhib and co-workers. Moreover, the average bond distances of the Co-O (~1.92 Å) and Ni-O (~1.91 Å) interactions were found to be smaller than that of the Mn-O (~1.923 Å) interaction, in line with the experimental values. Therefore, the M-O (M=Co, Ni) bond will be stronger than the Mn-O bond, which will improve the structural stability of LiMn2O4. Since a shorter bond length will result in higher electron affinity between M-O (M=Co, Ni) as compared to Mn-O. The subsequent step was to employ the derived interatomic potentials to define the interactions within the Co- and Ni-doped LiMn2O4 spinel. Monitoring of the impact of these dopants demonstrated that the introduction of Co and Ni in LiMn2O4 did not compromise the cubic spinel structure. Therefore, in line with our findings in LiCo2O4 and LiNi2O4 spinel systems, the bond length between the dopants (Co and Ni) and oxygen (< 1.923 Å) was also found to be shorter than the bond length of manganese and oxygen (~1.923 Å). This also supports the view that the partial substitution of manganese with Co and/or with Ni could result in improved structural stability, which will improve capacity retention. Furthermore, Co and Ni were only observed in the 16d octahedral sites and were not found in the 8a tetrahedral sites on the microstructure, unlike in the case of pure LiMn2O4 phases, where some Mn atoms migrate into the tetrahedral sites. Therefore, Co and Ni do not disrupt Li transport in the 8a tetrahedral sites, which affects capacity retention and rate capabilities. Furthermore, one extremely crucial aspect is the impact of the dopants on the ionic diffusion of the spinel material. When comparing the effects of the two dopants, Ni was found to improve the diffusion of lithium ions in the spinel structure compared to Co. Since, the Ni-doped LiMn2O4 spinel structure (1.6 x 10-12 m2/s) was found to have a higher value for diffusion coefficient at 300 K as compared to the Co-doped LiMn2O4 spinel structure (0.61 x 10-12 m2/s). Moreover, the ionic conductivity of Li also demonstrated the same behaviour, Ni-doped LiMn2O4 spinel nanomaterial was found to have a value of 0.94 x 10-6 S cm-1, while the Co-doped LiMn2O4 spinel nanomaterial had a value of 0.23 x 10-6 S cm-1. The ultimate exploration in this work was the morphological impact on the ionic diffusion and conductivity of the LiNi2O4 spinel cathode material. This sparked particular interest in further exploration since the Ni-dopant demonstrated enhanced diffusivity. As such, the role of pore size in LiNi2O4 nanoporous structures on transport properties (i.e., ionic diffusion) and microstructural changes was probed. Increasing the pore size was found to also increase the surface area and the diffusion of lithium ions in the structure. The surface areas of the 0.30, 0.21, and 0.15 nm nanoporous materials were found to be 22.16, 15.50, and 12.05 Å2, respectively. The same was also noted in the calculated ionic conductivities of Li+, which were found to be ~2.34 x 10-8, ~7.86 x 10-8, and ~1.50 x 10-7 S/cm for the nanoporous structures with pore diameters of 0.15, 0.21, and 0.30 nm, respectively. Therefore, a nanoporous spinel structure could enhance rate capabilities, and fast charge and discharge rates could be achieved. The overall stability of Li-Mn-O can be enhanced by the partial substitution of Mn with Co and Ni. Additionally, the transport of Li+ ions is not disrupted by Co and Ni at dopant concentrations of ≤ 0.02. Moreover, Ni was found to have superior influence over the migration of Li+ ions in the structure. Therefore, the charge and discharge processes can be enhanced by single-doping LiMn2O4 spinel with Ni. The rate of charge and discharge can also be further enhanced by the design of nanoporous materials with larger pore diameters (≥ 0.21 nm), which are crucial for large-scale high-energy and high-power density applications.