Spinel LiMn₂O₄ Cathode Materials in Wide Voltage Window: Single-Crystalline versus Polycrystalline

Abstract

Single-crystal (SC) layered oxides as cathodes for Li-ion batteries have demonstrated better cycle stability than their polycrystalline (PC) counterparts due to the restrained intergranular cracking formation. However, there are rare reports on comparisons between single-crystal LiMn₂O₄ (SC-LMO) and polycrystalline LiMn₂O₄ (PC-LMO) spinel cathodes for Li-ion storage. In this work, the Li-ion storage properties of spinel LiMn₂O₄ single-crystalline and polycrystalline with similar particle sizes were investigated in a wide voltage window of 2–4.8 V vs. Li/Li⁺. The SC-LMO cathode exhibited a specific discharge capacity of 178 mA·h·g⁻¹, which was a bit larger than that of the PC-LMO cathode. This is mainly because the SC-LMO cathode showed much higher specific capacity in the 3 V region (Li-ion storage at octahedral sites with cubic to tetragonal phase transition) than the PC-LMO cathode. However, unlike layered-oxide cathodes, the PC-LMO cathode displayed better cycle stability than the SC-LMO cathode. Our studies for the first time demonstrate that the phase transition-induced Mn(II) ion dissolution in the 3 V region rather than cracking formation is the limiting factor for the cycle performance of spinel LiMn₂O₄ in the wide voltage window.

1. Introduction

Lithium-ion batteries (LIBs) have been the most successful power sources for portable electronics over the last four decades. Recently, LIBs are also emerging as promising energy storage device for electric vehicles (EVs). Nevertheless, the requirement of higher specific energy and improved safety for automotive LIBs is far from over due to the range and safety concerns of EVs [1,2,3]. A natural pathway to improve the energy densities of automotive LIBs is to increase the reversible capacities of the cathode materials. Compared to currently prevailing Co- and Ni-based layered-oxide cathode materials (such as LiCoO₂ and Ni-rich LiNi_1−x−yCo_xMn_yO₂), the Mn-based spinel LiMn₂O₄ cathode possesses unique characteristics of low toxicity, abundant source, superior safety, and environmental friendliness [4,5]. The 4 V electrochemistry of the LiMn₂O₄ cathode has been popularly explored in LIBs [4,5,6,7,8,9,10,11,12]. Unfortunately, practical specific capacities of spinel LiMn₂O₄ cathodes in the 4 V region are limited to ~80–120 mA·h·g⁻¹, which are lower than layered-oxide cathode materials. The low specific capacity of the spinel LiMn₂O₄ cathode originates from the lithiation/delithiation in spinel frameworks in the 4 V region occurring at Li-ion occupied tetrahedral (8a) sites only involving one-electron redox reaction. Interestingly, extending the discharge voltage of spinel LiMn₂O₄ cathode to 2 V vs. Li/Li⁺ can double its specific capacity. There is another voltage plateau at around 2.9 V vs. Li/Li⁺ (labeled as 3 V region) with extra Li-ion intercalation into the unoccupied octahedral sites (16c) sites of the spinel structure. However, upon discharging to the 3 V region, severe phase transition from cubic phase to tetragonal phase may appear because the reduction in average Mn valence state close to trivalence easily induces serious Jahn–Teller distortion and Mn(III) disproportionation reactions [13]. Both Jahn–Teller distortion and Mn(III) disproportionation reactions would lead to the severe capacity fading of spinel LiMn₂O₄ cathodes. Thus, most previous efforts in spinel LiMn₂O₄ cathodes were mainly focused on electrochemical studies in the 4 V region [4,5,6,7,8,9,10,11,12].

To suppress the capacity fading of cathodes, one effective strategy is to design a single-crystal morphology. Currently, single-crystal layered-oxide cathodes have demonstrated better structural integrity than their polycrystalline counterparts due to the reduced phase boundaries and limited surface reactivity [14,15,16,17,18,19,20,21,22,23]. Typically, the polycrystalline cathodes are composed of smaller primary particles agglomerated into large secondary particles. During cycles, the expansion/contraction of the small primary particles in nonuniform directions usually leads to cracking propagation in large secondary particles. On the one hand, the cracking propagation causes a severe loss of electrical contact between the active cathode material and the current collector. On the other hand, the formed cracking exacerbates the parasitic reactions between the electrolyte and freshly exposed particle surfaces [16]. Thus, cracking formation is regarded as one of the main reasons for the capacity loss and structural collapse in layered-oxide cathodes. It has been revealed that single-crystal morphology can avoid the formation of particle cracking because of the uniform lattice expansion/contraction and fewer phase boundaries. For example, Dahn et al. [22] demonstrated that LiNi_0.5Mn_0.3Co_0.5O₂ cathodes in pouch cells exhibited long-term stability up to 4700 cycles. Zhou et al. [23] reported that the serious oxygen release was effectively restrained within a single-crystal Li-rich Li_1.2Ni_0.2Mn_0.6O₂ electrode, whereas the polycrystalline sample displayed severe irreversible capacity during the initial cycle and capacity fading. In general, the research on single-crystalline and polycrystalline cathodes is still mainly focused on Ni-rich and Li-rich layered-oxide cathodes [14,17,18,19,23]. There are rare reports on comparisons between single-crystalline and polycrystalline LiMn₂O₄ spinel cathodes.

In this work, we synthesized single-crystalline LiMn₂O₄ (SC-LMO) and polycrystalline LiMn₂O₄ (PC-LMO) cathode materials with similar particle sizes. The differences in electrochemical properties of SC-LMO and PC-LMO cathodes were investigated in a wide voltage window of 2–4.8 V vs. Li/Li⁺. The SC-LMO cathode exhibited higher Li-ion storage capacity than the PC cathode due to the large contribution at the 3 V voltage plateau. In contrast to what is reported on Ni-rich and Li-rich layered-oxide cathodes, PC-LMO showed superior cycling stability to SC-LMO.

2. Materials and Methods

2.1. Synthesis of SC-LMO and PC-LMO

The PC-LMO sample was synthesized using Mn₂O₃ microspheres as the template [24]. First, the MnCO₃ microspheres were prepared via a precipitation method [25]. NaHCO₃ (8.4 g) was dissolved in 700 mL of distilled water. Then, ethanol (70 mL) and the NaHCO₃ solution were added to the MnSO₄·H₂O solution (1.6 g in 700 mL distilled water) in sequence under stirring. The mixture was kept under stirring for 3 h at room temperature. After centrifugation, filtration, and drying, MnCO₃ powders were obtained. Then, the as-obtained MnCO₃ powders were calcined at 700 °C for 10 h in the air to synthesize the Mn₂O₃ sample. Finally, the Mn₂O₃ sample was thoroughly mixed with LiOH·H₂O in a molar ratio of Mn₂O₃:LiOH = 1:1.05 and heat-treated at 650 °C for 10 h in a muffle furnace.

SC-LMO was synthesized using a molten salt method. Stoichiometric amounts of LiOH·H₂O and commercial Mn₂O₃ (molar ratio 1.05:2) were thoroughly mixed as lithium and nickel precursors. Then, the mixed precursors were ground with a large excess of CsCl salt (melting point: 645 °C). The molar ratio between the CsCl flux and precursors (referred to as the R ratio) was 4. The mixtures of CsCl salt and precursors were put in alumina crucibles and annealed at 850 °C for 10 h. Afterward, the product was washed with deionized water to remove the residual CsCl and then filtered, before finally drying at 100 °C in an oven for 12 h.

2.2. Material Characterizations

Powder X-ray diffraction (XRD) patterns were collected using a BrukerD4 X-ray diffractometer (Bruker, Dresden, Germany) with λ (Cu Kα) = 0.154 nm (40 kV, 30 mA). All the obtained XRD patterns are shown with background (without any smoothing). The morphologies of the prepared materials were observed by scanning electron microscopy (Philip XL30, Dresden, Germany) at 15 kV. Elemental analysis was carried out on an inductively coupled plasma optical emission spectrometry (ICP-OES). Dissolution of Mn ions in the electrolyte was quantified using a two-electrode glass device with 10 mL of electrolyte after 10 cycles. The electrolyte was treated under vacuum-drying and then under a small amount of concentrated HNO₃, before finally diluting to a 50 mL solution using deionized water for ICP-OES tests.

2.3. Electrochemical Measurements

The working electrode was prepared by mixing the as-prepared active material (SC-LMO or PC-LMO), conductive agent (acetylene black), and binder (polyvinylidene fluoride, PVDF) with the help of N-methyl-2-pyrrolidone (NMP). The mass ratios of the LMO cathodes, acetylene black, and PVDF were 80 wt.%, 10 wt.%, and 10 wt.%, respectively. The slurry was coated onto an aluminum foil current collector and dried at 120 °C in a vacuum oven overnight. Then, the dried foil was cut into circular electrode with an area of 1.0 cm². The areal mass loading of cathode is in the range of 1.2–1.5 mg·cm⁻². Half-cell assembly was carried out in a glovebox with moisture and oxygen concentrations of 0.1 ppm. A solution of 1.0 M LiPF₆ in ethylene carbonate/ethyl methyl carbonate (EC/EMC, 3:7) solvents was used as electrolyte. Li metal was used as both the counter and reference electrodes for half-cell tests, and all voltages are reported versus the Li/Li⁺ redox couple. Cyclic voltammetry tests were performed using a CHI660 electrochemical workstation between 2.0 V and 5.0 V at 0.1 mV·s⁻¹. The galvanostatic charge/discharge measurement was carried out by a Land tester (CT2001A) between 2.0 and 4.8 V vs. Li/Li⁺. All electrochemical tests were performed at room temperature. The specific capacities were obtained as a function of the mass of active cathode materials not including the mass of electrolyte and the Li metal anode.

3. Results

The crystallographic structure and crystallinity of the SC-LMO and PC-LMO were characterized by XRD patterns. The diffraction peaks (Figure 1a) of these two samples could be well indexed to the cubic structure in a space group $\mathrm{Fd}\overline{3}\mathrm{m}$ (standard card, PDF#53-1237) [13], where Li, O and Mn ions occupy the 8a tetrahedral, 32e octahedral, and 16d octahedral sites (Figure 1b). The Li-ion can deintercalate from the 8a tetrahedral sites, and the 8a tetrahedral sites together with 16c octahedral sites form a 3D pathway for Li-ion diffusion. The spinel structure has a three-dimensional network of channels for fast Li-ion conduction. Moreover, the Li-ion can intercalate into the empty 16c octahedral sites in an over-lithiation state, which may cause the transition of crystal structure from cubic LiMn₂O₄ into tetragonal Li₂Mn₂O₄ [26]. Notably, the SC-LMO showed relatively stronger diffraction peaks than PC-LMO, indicating a higher degree of crystallinity in SC-LMO than PC-LMO.

The morphology of the as-synthesized LiMn₂O₄ samples was examined by SEM. As exhibited in Figure 2a, the morphology of PC-LiMn₂O₄ was microspherical with a rough surface. The diameter of the microsphere was in the range of 2.0–2.5 μm. The enlarged SEM image indicated that the spherical particle was densely composed of numerous small primary particles with sizes in the range of 100–200 nm (Figure 2b). In contrast to the spherical-shaped polycrystalline particles consisting of small grains, SC-LMO had a relatively regular polyhedral shape with a smooth surface and clear edges. SC-LMO had an average particle size of ~2.0 μm.

The electrochemical properties of SC-LMO and PC-LMO samples were evaluated on the basis of cyclic voltammograms (CV) and galvanostatic charge–discharge (GCD) measurements. As shown in the CV curves of both SC-LMO and PC-LMO (Figure 3a), there were two pairs of redox peaks in the 4 V region, corresponding to a two-stage Li-ion intercalation/deintercalation process at the 8a tetrahedral sites with LiMn₂O₄/Li_0.5Mn₂O₄ and Li_0.5Mn₂O₄/Mn₂O₄ transition [27,28]. Another pair of redox peaks in the 3 V region resulted from the further lithiation at the octahedral vacancies, forming over-lithiated Li₂Mn₂O₄ with the phase transformation of cubic to tetragonal phases (LiMn₂O₄/Li₂Mn₂O₄) [13,29]. Interestingly, in the CV curve of the SC-LMO cathode, the current of the cathodic peak associated with the lithiation of LiMn₂O₄ to Li₂MnO₄ in the 3 V region was slightly larger than that associated with the lithiation of Mn₂O₄ to Li_0.5Mn₂ O₄ and Li_0.5Mn₂O₄ to LiMn₂O₄ in the 4 V region. On the contrary, PC-LMO presented a smaller current of the cathodic peak in the 4 V region than in the 3 V region.

Figure 3b,c show the GCD profiles of SC-LMO and PC-LMO cathodes at 50 mA·g⁻¹ in a wide voltage range of 2.0–4.8 V vs. Li/Li⁺. During the first charging process, both SC-LMO and PC-LMO cathodes exhibited two distinct voltage plateaus at ~4.1/4.2 V vs. Li/Li⁺. The first discharge capacities of SC-LMO and PC-LMO cathodes at a current density of 50 mA·g⁻¹ were 178 and 172 mA·h·g⁻¹, respectively. Clearly, the SC-LMO had a much higher specific capacity (88 mA·h·g⁻¹) than PC-LMO (56 mA·h·g⁻¹) in the 3 V region, which is consistent with the CV tests. This indicates that extra Li-ion storage at the 16c octahedral vacancies in spinel LiMn₂O₄ [13] was more easily activated in single-crystalline particles than in polycrystalline secondary particles.

In the voltage window of 2.0 and 4.8 V vs. Li/Li⁺, the SC-LMO cathode delivered a discharge capacity of 114 mA·h·g⁻¹ with ~67% of its original capacity after 50 cycles, which was much better than the PC-LMO cathode with a capacity retention of ~48% after 50 cycles. Better cycling performance was achieved in the polycrystalline cathode than in the single-crystalline cathode. This is quite different from previous Ni-rich NMC and Li-rich cathodes [14,18,19,21].

To further understand the relationship between the cycle stability and crystallinity in the LiMn₂O₄ cathode, we carried out post-cycling structural characterizations of SC-LMO and PC-LMO electrodes using SEM, XRD, and ICP analysis. Figure 4a,b compare the morphology evolution of PC-LMO and SC-LMO cathodes after 10 cycles. The PC-LMO particles presented micro-cracks (intergranular fracture), whereas the SC-LMO particles showed no sign of grain boundaries after cycles. Regarding the cycle performance of polycrystalline LiMn₂O₄ cathodes, there was a slight gradual increase in discharge capacity during the 5th–20th cycles. In contrast, the increase in discharge capacity was not observed for the single-crystalline LiMn₂O₄ cathode (free of cracks after 10 cycles). Recent work from Jürgen Janek et al. [30] reported that the cracking of secondary polycrystalline particles leads to liquid electrolyte infiltration in the cathode active materials, lowering the charge-transfer resistance and increasing the Li-ion diffusion coefficient by more than one order of magnitude. This was also the case in our polycrystalline LiMn₂O₄ cathode. The slight increase in discharge capacities of PC-LMO might have resulted from the gradual electrolyte wetting due to the cracking formation.

Furthermore, post-XRD characterization (Figure 4c) suggested that more impurity phases (tetragonal Li₂Mn₂O₄ phase and orthorhombic LiMnO₂ phase) [31] appeared in the single-crystalline LiMn₂O₄ electrode than in the polycrystalline LiMn₂O₄ electrode after 10 cycles. Results from elemental analysis by ICP show that the dissolved Mn-ion amount in the electrolyte from the SC-LMO cathode was 1.5 times higher than that from PC-LMO (Figure 4d). The detected Mn ions were dissoluble Mn(II) ions, resulting from the disproportion reaction of Mn(III) before dissolving into the electrolyte. The loss of Mn at the LiMn₂O₄ cathode not only caused a gradual loss of active cathode, but also deteriorated the stable cathode–electrolyte interface, thereby leading to severe capacity decay during cycles [32]. It is well known that Mn(II) dissolution is also associated with the atomic arrangement of Mn-ions on the crystal facets of LiMn₂O₄. A large portion of Mn(II) ions exist at the {110} facet of LiMn₂O₄, leading to the {110} facet being more vulnerable to the Mn dissolution [33]. Compared to densely packed particles in polycrystalline LiMn₂O₄, it is probable that {110} facets are more exposed in single-crystalline LiMn₂O₄. This may be another reason for the more severe Mn(II) ion dissolution and worse cycling stability of single-crystalline LiMn₂O₄. Accurate crystal facet/orientation analysis is required for further characterization based on transmission electron microscopy (TEM) with a focused ion beam (FIB) in the future. Moreover, the high number of Mn(III) ions produced from SC-LMO would induce Jahn–Teller distortion due to the valence electronic configuration of t_2g³e_g¹, which may lead to the structural degradation of spinel phases [34]. Consequently, in spite of cracking formation, the PC-LMO cathode had better cycle stability than the SC-LMO cathode due to the low capacity contribution in the 3 V region with cubic/tetragonal phase transition. In general, these studies indicate that the phase transition-induced Mn(II) ion dissolution in the 3V region instead of cracking formation is the limiting factor for the cycle performance of the spinel LiMn₂O₄ cathode operating in a wide voltage window.

4. Conclusions

In summary, we investigated the electrochemical properties of single-crystal LiMn₂O₄ particles and polycrystalline LiMn₂O₄ particles with similar sizes as cathodes for Li-ion storage. Our studies suggest that, while single-crystalline LiMn₂O₄ cathodes might have the advantage of enhancing specific capacity, polycrystalline LiMn₂O₄ cathodes have less cubic–tetragonal phase transition in the 3 V region, thus presenting better cycle stability. Unlike layered Ni-rich cathodes, the single-crystalline design does not work well for stable spinel LiMn₂O₄ cathodes. Post-cycling structural characterizations confirmed that the limiting factor for the cycle performance of the spinel LiMn₂O₄ cathode in the wide voltage window was Mn(II) ion dissolution rather than intergranular cracking formation.