Evolutionary algorithm simulation study of MnO2 nanoclusters

Abstract

Transition metal-oxides have garnered significant attention due to their different allotropic polymorphs with unique structural and electronic properties necessary for the creation of new and improved rechargeable battery systems. The increasing demand for rechargeable batteries with high energy densities has fuelled interest in the research, development and manufacturing of new battery systems capable of powering high-powered machinery as well as portable household appliances. These demands have led to the heavy reliance on renewable fossil Consequentially, these have resulted in the unavoidable environmental pollution, which has in turn led to dangerous climatic conditions and unwanted health threats to human beings hence the need for renewable energy. Manganese dioxide is one of the most promising materials in rechargeable lithium-ion batteries. Specifically, pyrolusite (β-MnO2) as the most stable and abundant polymorph of MnO2 is selected for this purpose. In this study, evolutionary algorithm and first principle methods were used to generate (MnO2)n=2-20 nanoclusters. Firstly, the β-MnO2 bulk structure was optimised using existing interatomic potentials (IP). The selected IP parameters successfully reproduced the β-MnO2 bulk structure to within 0.43 % of experimental data. The XRD patterns of the β-MnO2 bulk structure were compatible to experimental data displaying similar peak intensities, 2θ positioning and Miller indexes such as the (110), (101), (200) and (211) peaks. The verified IP parameters were then used in conjunction with the Buckingham potential and the 12-6 Lennard-Jones potential to generate subsets of stable nanoclusters using the Knowledge Led Master Code (KLMC) software. A combination of global search techniques and density functional theory (DFT) based codes such as FHI-aims, CASTEP and VASP were used to refine the energy ordering of the generated nanoclusters in an effort to determine the global minima of the nanoclusters for all the concerned atomic sizes i.e., n = 2 to n = 20. The predicted order of stability for the nanoclusters was found to be similar to those predicted for isostructural ZrO2, SiO2 and TiO2 nanoclusters. Furthermore, the larger stable nanoclusters adopted compact ring configurations as they grew larger in atomic size. Additionally, they displayed improved stability and electrical conductivity as seen by the shorter band-gap energy of 0.113 eV for the stable n20 nanocluster. Secondly, the effect of temperature changes on the stability of the nanoclusters was investigated using the NVE ensemble in CASTEP and the molecular dynamics code DMol3. The nanoclusters showed a preference towards compact circular bonding configurations at higher temperatures as seen by the decreased bond lengths and inward bending of the exterior terminal oxygen atoms. An indirect relationship were the binding energy decreased with increasing temperatures was observed showing that high temperatures weakened the bonds in the n3-01 nanocluster. The results also predicted that the nanoclusters have metallic characteristics with the density of states (DOS) curves being continuous at the Fermi level with minimal band-gaps between the valence and conduction bands indicating a good conductive nature. The XRD patterns for the most stable n3 nanoclusters revealed common peaks such as the (001), (101) and (200) indicative of the tetragonal phase signifying the stable rutile β-MnO2. Moreover, the stable nanoclusters showed a prevalence towards a cubic bonding configuration composed of two manganese atoms bonded to two oxygen atoms. The atomistic substitutional doping technique proved to be a better strategy as compared to the virtual crystal approximation (VCA) technique using Fe, Co and Ni. The doping was preferential on the central manganese atom with the highest coordination denoted as Mn2. Nickel was the most favourable dopant due to its better binding energy with smaller bond lengths. However, the cobalt-doped nanoclusters were shown to be the better electrical conductors even though they were not as stable as the nickel-doped nanoclusters. Dual doping with nickel and cobalt did not sufficiently improve the stability and conductivity of the smaller nanoclusters. However, the simultaneous doping using Fe, Co and Ni may succeed in improving the stability of the larger nanoclusters as compared to singular doping. Lastly, the electronic charge density differences of the Ni-doped n3-01 nanoclusters displayed considerable occurrences of covalent bonding. However, the weaker ionic bonding was observed with singular Fe-doping and Co-doping. The Ni-doped nanoclusters on the Mn2 atomic position with the highest coordination had the highest voltage potential of 3.038 V showing that nickel is indeed the most preferable dopant. Fe-doped nanoclusters had the lowest potential below the Co-doped nanoclusters. Furthermore, all the doped nanoclusters had an operational voltage range between 2.7 V and 3.9 V showing their efficacy for energy storage. The aims of the study were achieved hence the study can be considered a success. Recommendations are made to continue the work towards improving future rechargeable batteries.