1. Introduction

Lithium manganese oxide (LMO) spinel has been considered as one of the most promising cathode materials for large power batteries and energy-storage systems in recent years. It exhibits good electrochemical performance and high safety at high current rate. Manganese is one of the most abundant elements in the Earth crust, and it is environmentally benign. However, this LMO material shows a significant capacity fading during long-term electrochemical cycling, particularly at elevated temperatures (55 ℃), mainly because of the following possible reasons: (1) the structural collapse due to Jahn-Teller distortion in the charge and discharge process, (2) enhanced manganese ion dissolution into the electrolyte, (3) the thermal decomposition of LiPF₆-based carbonate electrolytes [1-5].

A large number of efforts are carried out to resolve the above issues. First of all, synthesis methods are studied. The traditional methods mainly include solid state method and soft chemical method. For solid state method, repeated grinding and high temperature reaction for a long time are always needed to form spinel crystal. Soft chemical methods include Pechini method, solgel method, hydrothermal method and co-precipitation method etc. [6-12]. In these methods, the precursors are obtained from liquid solution, so all the components can be distributed very homogeneously at atomic-molecular level in these precursors. Soft chemical synthesis can be performed at comparatively low temperatures and within short periods. Solid state method is an easy preparation process, the obtained product shows high bulk density but poor electrochemical performance. In the synthesis process, some undesirable impure phases can be formed in the product because of the long-range diffusion of atoms. The product synthesized by soft chemical method shows good electrochemical performance but low density, and it is a complex process. In recent years, some new method are put forward such as solid-state combustion method [13], electrospinning method [14], spray pyrolysis method [15], solution combustion method [16] and mechanical alloying method [17].

Secondly, modified methods are adopted to improve the performance of LMO spinel including doping and coating. The investigated doping ions include cations and anions, the cationic doping elements mainly include Li, Mg, Ni, Zn, Cu, Al, Cr, Co, Ga, Gd, La, Ce, Nd, Sm, Tb, Sr, Dy, Mo, B, Fe, Ti, Sn [18-30], Er, Sc, Y [2] etc., while the reported anions mainly consist of F^- [31], Cl^-, Br^- [2], S^2- [32], BO₃^3-, PO₄^3-, SiO₃^2- [33, 34] etc. Sometimes, single ion doping cannot reach satisfactory effect, dual or multiple ions doping simultaneously have also been developed. Surface coating is an effective approach in which an artificial protective layer is used to prevent Mn ions from contacting with electrolyte, and restrain the Mn dissolution. The coating materials mainly include various oxides, fluorides, phosphates, lithium compounds, carbon materials, and polymers [2, 35].

Besides doping and coating, some new techniques are developed in LMO spinel such as controlling the nanoparticles granularity and spheroidizing of particles [2, 13, 16, 36]. Although a certain progress has been made in these researches, LMO spinel has not been put into large-scale commercial applications. The properties of LMO spinel need to be further improved.

In terms of choice for doping elements, few researchers select +5 valence cations. Hernan et al. investigated the electrochemical properties of vanadium doped LMO spinel in different voltage window [37]. The results show that it has a discharge capacity of 105 Ah kg^-1 within the range of 4.6-2.5 V. Herein, we synthesized a spherical Li-rich LMO spinel cathode material doped with +5 valence niobium ion and phosphate by an ion implanted method. The morphology of the prepared material is obtained by controlling the morphology of the manganese dioxide precursors. The properties of the material and synthesis mechanism were also investigated in detail in this work.

2. Results and discussion 2.1. XRD analysis on the raw materials, MnO₂ precursor, SP and IP

Figs. 1 and S1 (Supporting information) are the XRD patterns of the precursor mixtures prepared by different method and the raw material lithium hydroxide monohydrate and manganese dioxide precursor. Fig. S1a (Supporting information) suggests that the lithium hydroxide monohydrate is regular crystal structure with high crystallinity, while the precursor MnO₂ indicate a γ-MnO₂ structure (JCPDS card No. 30-820) with low crystallinity as shown in Fig. S1b (Supporting information) [36, 38].

Fig. 1 shows that the XRD patterns of SP comprise the patters of lithium hydroxide monohydrate and manganese dioxide. The characteristic peaks of lithium hydroxide monohydrate and manganese dioxide both appear on the diffraction pattern of SP. It demonstrates that the lithium hydroxide monohydrate and the manganese dioxide were just simply blended in physical form. However, the characteristic peaks of the lithium hydroxide monohydrate disappear in the XRD patterns of IP. It implies that lithium hydroxide monohydrate has been dissolved.

2.2. Thermal analysis

To further investigate the component of the mixture and synthesis process, TG/DSC is carried out from room temperature to 900 ℃. Fig. 2(a) is the TG-DSC curve of the pure lithium hydroxide monohydrate. A dehydration process can be found at 100 ℃ with a strong endothermic peak in DSC curve accompanied by a significant weight loss in TG curve. The solid lithium hydroxide transforms into liquid phase at 430-480 ℃ where there is a strong heat absorption phenomenon without weight change. Lithium hydroxide decompose at about 600 ℃ with weight loss. The whole thermal reaction process can be revealed by Eqs. (1)-(3):

(1)

(2)

(3)

In this experiment, 43.0% of weight loss is observed, and it is extremely close to the theoretical weight loss value of 42.8% calculated from Eqs. (1)-(3).

Fig. 2(b) is the curve of manganese dioxide precursor. The results show that there is a small amount of heat absorption along with continuous weight change. It should be related with the adsorbed water removal from the manganese dioxide. It can be inferred that the endothermic decomposition of the manganese dioxide happened near 555 ℃ because of the apparent sharp endothermic peak with a straight line loss in weight. The reaction is as follows:

(4)

Fig. 2(c) is the TG-DSC curve of the precursor SP obtained by the solid method. It can be found that there are five endothermic peaks corresponding to the dehydration of lithium hydroxide monohydrate, the melting of lithium hydroxide, the decomposition of lithium hydroxide, the decomposition of manganese dioxide and the forming of LMO spinel respectively at 100 ℃, 400 ℃, 500 ℃, 650 ℃, 830 ℃. It may exist the following reaction equations according to the raw material component of SP.

(5)

(6)

(7)

(8)

(9)

It is obvious that the endothermic peaks are consistent with that of the pure lithium hydroxide and manganese dioxide except the peak at 830 ℃, although some alterations to the temperature coordinates may be caused by the instrument. It also proves that SP is simply consisted of lithium hydroxide and manganese dioxide by physical mixing. The results are consistent with that of XRD analysis. The Eq. (9) is not so consistent with the TG curve at about 830 ℃, it indicates that the formation process of the LMO spinel is complex in the solid state method. The theoretical weight loss is 19.2% according to Eqs. (5)-(9), while the actual experimental result is 24.6%. It is reasonable because the manganese dioxide usually contains about 5 wt% water.

Fig. 2(d) is the TG-DSC curve of the precursor IP obtained by the ion implanted method. It shows that there are two endothermic peaks on the DSC curve at about 560 ℃ and 800 ℃ along with two significant weight change on TG curve. The two peaks are corresponding to the decomposition of manganese dioxide and the formation of LMO spinel. It can also be inferred that there is no lithium hydroxide in IP, because no peaks due to the dehydration of lithium hydroxide monohydrate and the melting of lithium hydroxide were found. The possible reaction mechanism is as follows, lithium hydroxide monohydrate is dissolved with the help of ethylene glycol or glycerol in the mixing process, when a large amount of heat is produced, and the lithium ions have been implanted into the pore of manganese dioxide under the action of the heat. Then the mixture is transformed into Li₂O·4MnO₂ compound after drying. So the forming process of LMO can be represented by Eq. (10).

(10)

The actual experimental weight loss is 5.0%. It is close to the theoretical value of 4.2% according to Eq. (10).

2.3. The morphology analysis

Fig. 3(a) is the SEM image of the MnO₂ precursor. It shows that the pristine submicron fibrous MnO₂ particles were gathered into micrometer spherical secondary particles. A large number of pores were formed between the fibrous particles, which would provide good channels for the lithium ion injection in the ion implantation method. The spherical particle diameter is of about 10 μm.

Fig. 3(b)-(f) are the SEM images of the prepared LMO. All the particles with spherical morphology were consisted of primary submicron crystalline grain with different size and shape. SLM is consisted of the octahedral and irregular tiny particles. It indicates that the material does not fully grow into octahedral particles during the high temperature reaction. The particles of LM and LLM are composed of octahedral grains with high crystallization, while those of LLMN and LLMNP are composed of grains with irregular shape and smooth surface. Besides, the octahedral crystal edges disappear in LLMN and LLMNP.

The results of Fig. 3 demonstrate that the synthesis method and doping have influence on the particle morphology of lithium manganese oxide spinel prepared. This effect will be further reflected from the microstructure and electrochemical performance.

2.4. XRD analysis of as prepared samples

Fig. 4 is the XRD patterns of the prepared samples in the range of 2θ from 10-90°. The diffraction peaks of the samples can be well indexed to a cubic spinel structure with Fd-3 m space group, which is consistent with JCPDS Card No: 35-0782. The diffraction peak intensity of SLM is lower than that of the other samples. There are two weak diffraction peaks at ca. 32.5° and 55.3° for SLM which corresponds to the strongest characteristic diffraction peak of Mn₂O₃ (PDF 78-0390), indicating the presence of a trace amount of Mn₂O₃ impurity phase in this material [39]. It cannot be noticed any strange reflection in the patterns of LM, LLM, LLMN, LLMNP. The (220) peak cannot be found in the XRD patterns for all the five samples. It indicates that the tetrahedral sites (8a) are only occupied by lithium atoms.

The ratios of the diffraction peak intensity due to different crystal planes and corresponding 2θ are showed in Table S1 (Supporting information). The ratios I₍₁₁₁₎/I₍₃₁₁₎ of sample SLM, LM, LLM, LLMN, LLMNP are 2.68, 2.69, 2.87, 2.03, 2.15 and 2.53, 3.16, 2.53, 2.34, 2.24 for I₍₁₁₁₎/I₍₄₀₀₎ respectively. The ratios of I₍₁₁₁₎/I₍₃₁₁₎ and I₍₁₁₁₎/I₍₄₀₀₎ for sample SLM, LM, LLM are larger than that of LLMN, LLMNP, it indicates that SLM, LM, LLM crystal particles grow well along the (111) crystal plane, so the particles exhibit an octahedral shape. The ratios of I₍₁₁₁₎/I₍₃₁₁₎ and I₍₁₁₁₎/I₍₄₀₀₎ for sample LLMN and LLMNP decrease, which implies that the crystal growth along the (311) and (400) surface is enhanced, while that along the (111) surface is restrained. So the crystalline grain presents arc shape with smooth surface without crystal edge. This result is consistent with the SEM results shown in Fig. 3.

The location and strength of the diffraction peaks and lattice parameters are also summarized in Table S1 (Supporting information). Afterlithium ions have replaced parts of themanganese ions in LLM, the lattice parameter decreases from 8.2417 Å(LM) to 8.2296 Å (LLM). The diffraction peaks are slightly shifted to larger angle as shown in Fig. S2a-c (Supporting information). It may be caused by the reason that the content of the Mn⁴⁺ is increased because of the replacement of the Mn with Li, while the radius of Mn⁴⁺ ion issmaller than that of Mn³⁺ ion [40]. The lattice parameter of LLMN (8.2363 Å) is increased compared to LLM, because Nb⁵⁺ radius is close to Mn³⁺, and the Mn³⁺ content is much increased after doping niobium. The lattice parameter of LLMNP (8.2394 Å) is increased compared to LLMN, for the radius of polyanion phosphoric acid root is larger than that of oxygen ionic. All in all, the lattice parameter is increased after doping with niobium and phosphate ions and the diffraction peaks move toward to smaller angle as shown in Fig. S2a-c (Supporting information).

2.5. The chemical composition analysis

The XPS was carried out to determine the average oxidation state of manganese. Fig. 5 represents the Mn 2p XPS spectra for as prepared samples, in comparison with those for MnO₂ and Mn₂O₃. The Mn 2p_3/2 peak of MnO₂ isobservedat 642.6 eV and that of Mn₂O₃ at 641.6 eV. As shown in Fig. 5, the Mn 2p_3/2 peaks of as prepared samples are observed between those of MnO₂ and Mn₂O₃. The Mn 2p_3/2 peak position of LLM, LLMN and LLMNP shifts toward higher binding energies, it indicates that the average oxidation state of manganese in the three samples is increased. The average Mn valence and the relative ratio of Mn³⁺ and Mn⁴⁺ are calculated based on the results of XPS and summarized in Table S2 (Supporting information).

The contents of Li, Mn, Nb and P were measured using ICP-AES. The O content was calculated based on the ICP-AES data and the average Mn valence. The chemical composition of the as-prepared samples was shown inTable S2 (Supporting information). The actual chemical formulas are very close to the design formulas.

2.6. Electrochemical performance analysis

Fig. 6 is the initial charge and discharge curves of SLM, LM, LLM, LLMN and LLMNP at 0.2 ℃ (1 ℃ = 148 mA g^-1) rate. The discharge specific capacities are 113 mAh g^-1, 123 mAh g^-1, 117 mAh g^-1, 120 mAh g^-1 and 119 mAh g^-1 respectively. The specific discharge capacity of the sample synthesized by solid method is lower than that by the ion implantation method. This is mainly related to the different crystallization of the product prepared by the two methods. The sample synthesized by the solid method has a poor crystallization and contains impurities, while the samples synthesized by the ion implantation method have a good crystallization.

Fig. 6 also shows that the doping has great influence on the capacity. In the process of charging and discharging, the occurred redox reaction is mainly: Mn³⁺ → Mn⁴⁺. Thus, the capacity is determined by Mn³⁺ content [41]. Compared with LM, excess lithium ions were doped in LLM, so that part of Mn³⁺ would become Mn⁴⁺ and the average manganese valence would rise from 3.530 to 3.640 as shown in Table S2 in Supporting information. The specific discharge capacity of LLM is 117 mAh g^-1, while that of normal spinel LM is 123 mAh g^-1. The specific discharge capacity of LLMN recovers to 120 mAh g^-1 for the Mn³⁺ content is increased after doping with pentavalent niobium ion. When doping phosphoric acid root, Mn³⁺ concentration is slightly decreased, thus the discharge capacity of LLMNP is declined to 119 mAh g^-1.

The two voltage plateaus in 4 V region are obvious in sample SLM and LM, but not so evident in sample LLM. It is found that the voltage jump disappeared in sample LLMN and LLMNP and the two voltage plateaus do not exist. This suggests that the two-phase structure is transformed to an one-phase structure in the charging and discharging process [15].

Fig. 7 is the cycle performance of the prepared samples in room temperature (25 ℃) and high temperature (55 ℃) respectively at 1 ℃ rate. The capacity retention rates of SLM, LM, LLM, LLMN and LLMNP are 87.5%, 87.6%, 95.4%, 94.4% and 97.0% respectively after 50 cycles at 25 ℃ and 83.7%, 84.1%, 90.3%, 92.1% and 92.0% at 55 ℃. It indicates that doping can improve the cycle characteristics at both room temperature and high temperature. The sample LLMN and LLMNP doped with niobium have better cycle characteristics than LLM doped with single lithium at high temperature. The coulombic efficiency of these samples is showed in Fig. S3 (Supporting information) during the process of cycling. It also indicates that the sample LLMN and LLMNP have a higher coulombic efficiency than the other samples.

In spinel structure, 32 oxide ions form a face centered cubic lattice, and the lithium and manganese ions occupy 8a tetrahedral sites and 16d octahedral sites, respectively [35]. When doping excess lithium ions, part of the lithium ions would replace the manganese ions in 16d octahedral sites, which leads to the content of Mn⁴⁺ ions increased to keep the charge balance. Because the radius of Mn⁴⁺ ion is smaller than that of Mn³⁺ ion, the lattice constant would decrease, and the manganese oxide covalent bond strength would also be increased, so the crystal structure became more stable. Doretta Capsoni reported that the capability of a lattice to maintain the undistorted structure is determined by the ratio r = [Mn⁴⁺]/[Mn³⁺] [42]. Jahn-Teller transition can be suppressed when r≥ 1.18. As shown in Table S2 (Supporting information), the r values of LLM, LLMN and LLMNP reach 1.78, 1.56 and 1.63. After doping with niobium and phosphorus, although the crystal structure constant slightly increases, but the r value is greater than 1.18, so LLMN and LLMNP still have good cycle characteristics. Luo [20] reported that the Mn valence in spinel cathodes needs to be >3.6 to offer good cycle ability at 25 ℃ and 55 ℃. The average valences of manganese are 3.522, 3.530, 3.640, 3.610 and 3.625 as shown in Table S2 (Supporting information). The experimental results are consistent with the conclusion.

The cycling features of LLMN and LLMNP are better than that of LLM at high temperature. The reasons may be associated with the morphology of circular arc and smooth surface due to the niobium dopant, which can reduce the contact area between electrode and electrolyte due to the small specific surface area as shown in Table S3 (Supporting information). It is well known that the capacity retention of LMO spinel would typically deteriorate at elevated temperature which is related to the manganese dissolution [31].

Fig. 7 indicates that the sample LLMNP with phosphate ion dopant exhibits the best cycle performance. In this case, the Mn-O distance is increased due to the phosphate doping, thus the polarizing effect exerted by Mn³⁺ over O^2- is decreased, and the positive charge on the Mn atom is increased as well. It produces a beneficial effect that the average valence of manganese would increase when increasing the a-lattice parameter. This effect can be used strategically to diminish the influence of the undesired Jahn-Teller effect on the electrochemical behavior of the LiMn₂O₄ cathode material through phosphate doping [43]. The electrochemical performance of the optimal sample and some published results of Li-rich spinels are showed in Table S4 (Supporting information) [44-47].

Fig. 8 represents the rate capability of the as-prepared samples at different rates ranging from 0.2 C to 10 C and the recovery ability after the charging and discharging at high rate. It shows that the discharge capacities decrease with the increasing chargedischarge rate. The specific discharge capacity of SLM, LM, LLM, LLMN and LLMNP are 113 mAh g^-1, 123 mAh g^-1, 117 mAh g^-1, 120 mAh g^-1 and 119 mAh g^-1 at 0.2 C rate, 106.6 mAh g^-1, 116.2 mAh g^-1, 109.2 mAh g^-1, 115.4 mAh g^-1 and 118.7 mAh g^-1 at 1 C rate, 102.5 mAh g^-1, 112.4 mAh g^-1, 105.9 mAh g^-1, 111.6 mAh g^-1 and 114.6 mAh g^-1 at 5 C rate, which then decrease to 96 mAh g^-1, 105 mAh g^-1, 98.5 mAh g^-1, 108.3 mAh g^-1and 112.8 mAh g^-1 at 10 C rate respectively. When the charge-discharge rates reduce from 10 C to 0.2 C, the specific discharge capacity of these samples could be recovered with 2.9%, 2.2%, 2.7%, 1.0%, 0.5% capacity loss. The two samples doped with niobium have more superior rates characteristics than those of the other samples. Doping the phosphate root further improves the rate characteristics. LLMNP has the best cycle characteristics at high rates.

The cyclic voltammograms of the prepared samples are shown in Fig. S4 (Supporting information) at a cutoff voltage between 3.0 V and 4.35 V with a scanning rate of 0.2 mV s^-1. The CV curve of the undoped samples SLM and LM shows two sharp anodic and cathodic peaks, it is corresponding to the two voltage plateaus of the charge and discharge curve as shown in Fig. 6. There are two binary equilibrium systems (LiMn₂O₄-Li_0.5Mn₂O₄ and Li_0.5Mn₂O₄-λMnO₂) for normal spinel LiMn₂O₄ during Li ion intercalation and deintercalation with a 100 mV voltage difference. The first pair redox of the peaks (4.15 V/4.05 V) is attributed to the intercalation/ de-intercalation of lithium ion into the first half of the tetrahedral sites, where Li⁺ ions do not interact. The second pair of the peaks (about 4.05 V/3.95 V) is attributed to the removal/addition of Li ion from/into the other half of tetrahedral sites inwhich Li/Li⁺ interaction occurs. It would increase the repulsion between the lithium ions, resulting in a small increase in the free energy in the system as reflected by the observed potential dropping or rising [7, 10]. It can be inferred that the change of structure is rapid in the process of charging and discharging for the un-doped samples SLM and LM, it would lead to final structure damage and deteriorate the cyclic performance. Fig. S4 (Supporting information) also shows that the sample LM has better reversibility than SLM.

The CV of the Li-rich spinel LLM shows two gentle redox peaks. The first pair redox of the peaks (4.15 V/4.05 V) becomes slight bulge for the CVs of LLMN while the two peaks disappear and become one peak for the CVs of LLMNP. It indicates that the structure changes gradually in the process of charging and discharging due to the adding of dopant. It helps to improve the cycle characteristics of LLM, LLMN and LLMNP.

Fig. S4 (Supporting information) also shows that the sample LLMNP has the largest peak current compared with that of other samples at the same scanning rate, so it can be inferred that the sample LLMNP has the highest Li⁺ ion diffusion coefficient which accounts for an excellent rate capability.

3. Conclusion

Spherical LMO spinel was prepared by an ion implanted method assisted by polyalcohol. Compared with the solid phase method, lithium ion can be immersed into the pore of manganese dioxide at a low temperature by the ion implantation method. The materials prepared by the ion implantation method have a high discharge capacity and excellent crystallization.

The LMO material was doped with lithium, niobium, and phosphate simultaneously. The results show that the rich lithium doping can improve the cycle stability at both room temperature and high temperature. Niobium doping can enhance the capacity of Li-rich LMO spinel. The crystal growth along the (111) and (400) planes is reinforced. The crystal grain exhibits circular shape and smooth morphology, so the BET specific surface area is decreased. Phosphate-doped LMO spinel exhibits good high-rate capacity and structure stability. The prepared Li_1.09Mn_1.87Nb_0.031O_3.99(PO₄)_0.021 has a discharge capacity of 119 mAh g^-1 at 0.2 C rate and 112.8 mAh g^-1 at 10 C rate. The retention rates of the discharge capacity reach 98% at 1 C rate after 50 cycles in 25 ℃ and 94% in 55 ℃.

4. Experimental

All the chemical reagents are of analytically pure grade. Reagents and materials used for the electrochemical performance evaluation are specialized for lithium ion battery.

4.1. Synthesis of spherical MnO₂ precursor

To prepare spherical MnO₂, a redox reaction was adopted in a liquid phase using sodium permanganate solution and manganese chloride as reactant. Firstly, the manganese chloride was dissolved in deionized water with the concentration of 1.5 mol L^-1 and the sodium permanganate solution was diluted to 1.5 mol L^-1. Then some H₂SO₄ solution with pH = 5 was added into a reaction kettle which was placed in a 40 ℃ water bath pot. After that, the sodium permanganate and manganese chloride solution were added into the reaction kettle continuously with a feeding speed of 8 mL min^-1 by constant flow pump under the condition of stirring. Keep stirring for half an hour after the end of feeding, the mixture was aged for ten hours. Finally, the spherical MnO₂ precursor was obtained by filtering, washing and drying. The Mn wt% was measured by inductively coupled plasma atomic emission spectroscopy.

4.2. Synthesis of LiMn₂O₄ sample by a solid state reaction method

A stoichiometric mixture of lithium hydroxide monohydrate (LiOH·H₂O) and spherical MnO₂ according to the formula LiMn₂O₄ was thoroughly mixed by grinding in agate mortar for two hours. The mixture (denoted by SP) then sintered at 800 ℃ for 6 h in air, followed by slow cooling to ambient temperature. The LiMn₂O₄ sample synthesized by this solid state reaction was marked as SLM.

4.3. Synthesis of LiMn₂O₄, Li_1.1Mn_1.9O₄, Li_1.1Mn_1.87Nb_0.03O₄, Li_1.1Mn_1.87Nb_0.03O_3.98(PO₄)_0.02 samples by an ion implanted method

Lithium hydroxide monohydrate (LiOH·H₂O) and spherical MnO₂ were mixed according to the stoichiometric ratio in LiMn₂O₄, then 20 wt% ethylene glycol or glycerol was added into the mixture as soluble initiator, lithium hydroxide monohydrate was dehydrated and dissolved with the help of ethylene glycol or glycerol. Meanwhile, lithium ions were implanted into the pore of manganese dioxide under the action of solution heat. Then, a mixed paste was attained. After drying at 80 ℃ for 20 h the mixed paste turned into mixture powder (denoted by IP). The sample IP was calcined at 800 ℃ for 6 h in air atmosphere. The final sample LiMn₂O₄ was marked as LM. Li_1.1Mn_1.9O₄ (denoted as LLM) was synthesized by the same method. Using niobium pentoxide, diammonium phosphate as dopant, Li_1.1Mn_1.87Nb_0.03O₄ (denoted by LLMN) and Li_1.1Mn_1.87Nb_0.03O_3.98(PO₄)_0.02 (denoted by LLMNP) were synthesized by the same method.

4.4. Characterization of the as-prepared samples

The phase identification of the prepared samples was carried out by X-ray diffraction (XRD) using a Multiflex X-ray powder diffractometer. X-ray profiles were measured between 10 and 90° (2θ angle) with a monochromatic Cu radiation source. The morphologies of the samples were examined with a JEOLJSM-5600LV scanning electron-microscope (SEM). Thermal analysis was carried out with a TGA/DSC (Setaram Labsys) instrument at a heating rate of 10 ℃ min^-1. The BET specific surface area was measured by N₂ sorption using surface analyzer Gemini 2360. Elemental analysis was carried out using inductively coupled plasma atomic emission spectroscopy (ICP-AES) (PROFILE SPEC, USA) to determine the chemical composition of the prepared samples. The present XPS data were collected using an Imaging Xray Photoelectron spectrometer (Axis Ultra, UK) with a monochromatized Al Kα source.

The electrochemical properties of the samples were tested in CR2032 coin-type cells with metallic lithium as anode electrode. Cathode slurries were prepared by mixing 90 wt% spinel LMO, 4 wt % conducting carbon black, and 6 wt% PVDF binder in NMP solvent. Keep stirring for two hours by magnetic force, these slurries were then cast onto an aluminum foil and dried at 120 ℃ in a vacuum oven. Then the foil was punched into circular electrode. The cathode was separated from the Li anode by a layer of celgard 2300 membrane soaked with the electrolyte of 1 mol/L LiPF₆ in a 1:1(v/ v) mixture of ethylene carbonate (EC) and dimethyl carbonate (DMC). The cells were assembled in an argon-filled glove box. The charge and discharge curves, cycle performances and rate characteristics of the cells were evaluated on a Land cell tester. Cyclic voltammograms (CVs) were measured with CHI660C type electrochemical workstation.

Acknowledgments

This work was supported by a grant from the National High Technology Research and Development Program of China (863 Program) (No. 2008AA11A102).

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cclet.2017.03.035.