First-Principles Investigation On The Structures And Properties Of LiMnO₂ System

In today's rapidly developing, floating information society, thecommercial Li-ion batteries which use rhombohedral LiCoO₂(space group R3 m, hereafter denoted as r-LiCoO2) as the cathodematerial are almost the standard electrical sources for theindispensable portable products such as mobile phone, notebookPC, electronic entertainment equipments and so on. Nevertheless,in order to satisfy people's increasing demands, it should manageto decrease the cost, increase the output and improve theperformance of the Li-ion batteries. In this aspect, r-LiCoO2 is themain bottleneck. Therefore, exploiting new cathode materials isalways a research hotspot in the field of Li-ion batteries andLiMnO₂ is regarded as one of the most promising substitutes.In order to understand the states and properties of the materialin nature, and direct its synthesis and improvement, LiMnO2system has been investigated by first-principles calculations withultra-soft pseudo-potentials plane-wave method based on thedensity functional theory. Three different crystal structures, i.e.rhombohedral, monoclinic and orthorhombic symmetries as well asthree different chemical compositions, i.e. pure, Li-deintercalatedand Al-doped phases have been considered.Among these three crystal structures mentioned above,rhombohedral LiMnO2 ( R3 m, r-LiMnO2) is isostructural tor-LiCoO2 and thereby attracts most expectation. Unfortunately,pure r-LiMnO2 is only an imaginative figment to solid-statechemists until now. The research results of this work reveal thatthis is mainly because that the Mn3+ ion in r-LiMnO2 adoptslow-spin but not high-spin state, i.e. its electronic configuration is40t 2 g eg but not t 23 g e1g. This unusual low-spin state of Mn3+ ion ismetastable and is stabilized only in a strong enough oxygenoctahedron ligand field caused by short enough Mn-O bonds.According to this finding, three possible approaches that cancompress the Mn-O bond effectively and thereby obtain r-LiMnO2have been predicted theoretically, which are high-pressureexperiment under low temperature, thin-film growth with preferredorientation and doping with Co, Ni, Cr or 4d, 5d elements.Following the proposed synthesis approaches deduced from thecalculations, Cr-doped r-LiMnO2 has been obtained successfullythrough hydrothermal reaction, and the magnetic measurements onit verify that the the Mn3+ ion is indeed in low-spin state.Similar to r-LiCoO2, r-LiMnO2 has fully ionized Li ions andstrongly hybridized Mn-3d and O-2p states, which can ensure thismaterial has good electrochemical properties. Theoretical analysesalso found that the unusual low-spin Mn3+ ion makes r-LiMnO2have some unique and useful properties compared to the othermanganites. Firstly, there is not strong Jahn-Teller (JT) effect in thestructure any longer, which ensures a good electrochemical cyclingperformance for this material. Secondly, its excited spectra for theMn ions will not be identical to the standard ones for the othermanganites but will similar to the materials containing Mn4+ ions.With this conclusion, the so-called "conflicts" observed in theexperiments reported by some literatures can be understood verywell. Most significatively, r-LiMnO2 containing low-spin Mn3+ ionis a possible half-metal, which not only has a relative highelectronic conductivity and thereby can overcome the disadvantagesuch as security problem and unsuitability for power suppliesfound in r-LiCoO2, but only can be applied extensively in the risinghigh-tech field of spintronics.Therotical studies on r-LiMnO2 compared its JT-distortedform, monoclinic LiMnO2 (C2/m, m-LiMnO2) found that whenMn3+ ion is in the high-spin state, the JT transition temperature Tt is1910 K, far above the melting point of LiMnO2, which reveals thatin the high-spin state JT distortion always exists and onlym-LiMnO2 can be synthesized;While when Mn3+ ion adoptslow-spin state, Tt is 4 K, which means JT distortion will disappearat the temperature above 4 K, then low-spin r-LiMnO2 can beobtained. With JT distortion, the two degenerated eg levels havebeen split, which decreases the system's total energy and makem-LiMnO2 an insulator, consistent well with the experiments. Atthe same time, JT distortion has compressed the structureperpendicularly to the triangular layers but relaxed in the layers,which directs useful ways to suppress the JT distortion.With Li-deintercalation, the Fermi level EF of m-LiMnO2shifts down below the d z2 level from above it, revealing vividlythat the high-spin Mn3+ ion has been oxidized to be Mn4+ ion.Accordingly, the strong JT effect disappears and a phase transitionto rhombohedral symmetry is driven. Therefore, repeatingcharge/discharge, i.e. Li-deintercalation/Li-intercalation processeswill make the material's structure change ceaselessly between therhombohedral and monoclinic symmetries. Consequently, thestructure collapses easily and the capacity of the material losesseverely.With Al-doping, the calculated Mulliken population analysesand partial density of states of Mn-3d and O-2p both reveal that asingle Al dopant in m-LiMnO2 has stabilized its sixnearest-neighbor Mn ions in their respective octahedral sites, whichcan hinder the migration of Mn ions into the interlayer Li sitesduring the Li intercalation-deintercalation and therefore improvingboth the structural stability and the electrochemical performance ofthe material. It is also found out that the Al doping can decrease theJT distortion, increase the intercalation voltage and induce negativeformation energy.Results of theoretical calcuations on the magnetic structures ofm-LiMnO2 show that antiferromagnetic (AF) state favors longerMn-Mn distance and less JT distortion than ferromagnetic (FM)state does, while these two magnetic states have comparable totalenergies. Additionally, it is found that their lattice parameterscompared to the experimental ones follow Vegard law, whichsuggests that the experimental structure is a "solid solution" of thetheoretical AF and FM ones. Based on these findings, a complexmagnetic ground state of coexistent, competitive AF and FMinteractions is suggested for m-LiMnO2. Compared to theconventional viewpoint of geometric frustration, this competitivemechanism can supply more reasonable and more thoroughexplaination on the complex magnetic behaviors of m-LiMnO2,which displays in turn short AF ordering, spin glass and weak FMordering when the temperature decreases experimentally.The same as m-LiMnO2, the JT distortion in orthorhombicLiMnO2 (Pmnn, o-LiMnO2) is also driven by high-spin Mn3+ ionand the complex magnetic properties of o-LiMnO2 is also the resultof coexistent, competitive AF and FM interactions. Differently, theoccupied eg level in o-LiMnO2 is d x2 ? y2 but not d z2, and becauseof the asymmetric distortion in the MnO6 octahedrons, there is abigger splitting in the d x2 ? y2 sub-band than between the d x2 ? y2and d z2 sub-bands.After Li-deintercalation, some d x2 ? y2 states still remain belowEF, which suggests that the JT active high-spin Mn3+ ions have notbeen oxidized thoroughly to JT inactive Mn4+ ions. Analysis on theMn-O bondlengths in the MnO6 octahedrons confirms that the JTdistortion in the structure has not been eliminated with theLi-deintercalation. Therefore, the traditional formal redox notation,i.e. Mn3+/Mn4+ couple can not be used to describe the realelectrochemical cycling process in o-LiMnO2, which is similar tor-LiCoO2 but different from m-LiMnO2. Interestingly, the partiallyoxidized Mn ions make the Li-deintercalated structure ofo-LiMnO2 displays half-metallic character in the electronicstructures. As a result, a possible new approach to find materialsfor spintronics named "electrochemical Li-deintercalation" hasbeen exploited and a possible new applying field for o-LiMnO2 hasbeen discovered.Utilizing the advantage of theoretical research, this work hasdecomposed the Li-deintercalation effect into two parts that can notbe distinguished in experiments: chemical effect and structuraleffect. Results show that the chemical effect in o-LiMnO2 hasstrong local character, which does harm to the Li-deintercalationand may be the main reason for the bad electrochemical activity ofthis material. One the other hand, the changes induced by thestructural effect open channels for the migration of Mn ions fromtheir own octahedral sites to the interlayer Li sites and helps to thestructural transition from orthorhombic phase into a spinel-like one,which has more activity in electrochemistry but has only halftheoretical capacity compared to o-LiMnO2. Thus, the commonexperimental fact that the capacity of o-LiMnO2 increases but notdecreases rapidly with the first some electrochemical cyclingnumbers and soon levels off with a capacity less than halftheoretical value can be understood in nature.