Enhancing the durable performance of LiMn2O4 at high-rate and elevated temperature by nickel-magnesium dual doping

Various nickel and magnesium dual-doped LiNixMg0.08Mn1.92−xO4 (x ≤ 0.15) were synthesized via a modified solid-state combustion method. All as-prepared samples show typical spinel phase with a well-defined polyhedron morphology. The Ni-Mg dual-doping obviously decreases the lattice parameter that gives rise to the lattice contraction. Owing to the synergistic merits of metal ions co-doping, the optimized LiNi0.03Mg0.08Mn1.89O4 delivers high initial capacity of 115.9 and 92.9 mAh·g−1, whilst retains 77.1 and 69.7 mAh·g−1 after 1000 cycles at 1 C and high current rate of 20 C, respectively. Even at 10 C and 55 °C, the LiNi0.03Mg0.08Mn1.89O4 also has a discharge capacity of 92.2 mAh·g−1 and endures 500 cycles long-term life. Such excellent results are contributed to the fast Li+ diffusion and robust structure stability. The anatomical analysis of the 1000 long-cycled LiNi0.03Mg0.08Mn1.89O4 electrode further demonstrates the stable spinel structure via the mitigation of Jahn-Teller effect. Hence, the Ni-Mg co-doping can be a potential strategy to improve the high-rate capability and long cycle properties of cathode materials.

delivered a capacity of about 120.0 mAh·g −1 at 1 C and remained 90.6 mAh·g −1 at a high current rate of 10 C. Even at an elevated temperature of 55 °C, the LiNi 0.03 Mg 0.02 Mn 1.95 O 4 also obtained the high capacity of 90.0 mAh·g −1 at a higher rate of 10 C. In addition to the Ni-Mg co-doping strategy, other metals such as Ni-Mo 20 , Mg-Si 21 , and Ni-Ti co-doping 22 were also employed to improve the cycling property and structure stability of spinel LiMn 2 O 4 cathode materials. For these dual-doped LiMn 2 O 4 cathode materials, the traditional preparation methods including high-temperature solid-state reaction, sol-gel method, microwave irradiation and so on need long reaction time, high temperature and troublesome pre-treatment. Compared to high-temperature solid-state reaction, the solid-state combustion method has the advantages of time-saving and energy efficient and avoids oxygen deficiency. Additionally, different synthesis method lead spinel LiMn 2 O 4 have unique various morphology, so indicating various electrochemical performance. The high-rate capability and durable cycling performance is closely related to its kinetic properties, such as lithium ion diffusion coefficient and activation energy. Therefore, introducing an feasible method to synthesize the LiNi x Mg 0.08 Mn 1.92−x O 4 cathode materials facilely is a great challenge, whilst considerable attention should focus on structure, morphology and detailed high-rate and durable elevated temperature properties.
In this work, Ni-Mg co-doped LiNi x Mg 0.08 Mn 1.92−x O 4 (x ≤ 0.15) cathodes with polyhedron morphology were prepared by a facile solid-state combustion method. The effects of the Ni-Mg co-doping on the structure, morphology, high-rate and long cycle performance as well as kinetic properties of the LiNi x Mg 0.08 Mn 1.92−x O 4 cathode materials were investigated detailedly. Furthermore, the structure characterization of long-cycled electrode materials was performed to further determine the stability and Li-ion kinetics. The resultant optimal Ni-Mg co-doped LiMn 2 O 4 sample presented excellent high-rate capability, long cycling stability and high temperature performance.

Experiment Section
Preparation of materials. A series of LiNi x Mg 0.08 Mn 1.92−x O 4 (x ≤ 0.15) products were synthesized by the solid-state combustion method using citric acid as a fuel. Firstly, with a total mass of reaction mixture is 6.0 g, the lithium carbonate (AR, Aladin), manganese carbonate (AR, Aladin), nickel acetate and magnesium acetate (AR, Aladin) were weighed according to the stoichiometric ratio of 1:(1.92 − x):x :0.08 (Li:Mn:Ni:Mg). Then adding 0.3 g citric acid into a polytetrafluoroethylene jar and using the ethanol as medium. Secondly, the mixture was ball-milled for 10 h by planetary. Thirdly, the mixture was dried at 80 °C in an oven. Thirdly, the as-obtained powder was calcined in a muffle furnace at 500 °C for 1 h. The pre-product was obtained after naturally cooling. Immediately, the pre-product was calcined again at 650 °C for 6 h, then cooled to room temperature and ground to obtain the ultimate LiNi x Mg 0.08 Mn 1.92−x O 4 (x ≦ 0.15) cathode materials.

Materials characterization.
The crystalline phase of the samples was identified by powder X-ray diffraction (XRD, Bruker Company) using Cu Kα radiation (λ = 0.15406 nm) over the 2θ range of 10°-70°. Morphological and particle size was examined by scanning electron microscopy (SEM, QUANTA-200 America FEI Company) and transmission electron microscopy (TEM, JEM-2100, Japan Electronics Corporation). X-ray photoelectron spectroscopy (XPS, Thermo fisher Scientific) analysis was performed by using Al Kα (1486.6 eV) radiation. The cycled electrodes were disassembled, washed with NMP and dried, further characterized by the XRD, SEM and TEM tests.
Cell assemble and electrochemical measurement. The electrochemical performance of as-synthesized LiNi x Mg 0.08 Mn 1.92−x O 4 samples was evaluated in CR2032 type coin cells using lithium metal as the anode and reference electrode. The working electrodes were fabricated by mixing active materials, carbon black and polyvinylidene fluoride (PVDF) binder in 1-methyl-3-pyrrolidone (NMP) solvent with a mass ratio of 8:1:1. The electrolyte was 1 M LiPF 6 that dissolved in ethylene carbonate (EC), dimethyl carbonate (DMC) and methyl ethyl carbonate (EMC) at a volume ratio 1:1:1. The electrochemical cells were assembled in a high-purity argon atmosphere (<1 ppm of O 2 and H 2 O). The electrode activities were performed at various current rate (1 C is defined as 148.0 mAh·g −1 ) and voltage range from 3.0 to 4.5 V (vs. Li + /Li) by using Land CT2001A system (Wuhan Jinnuo Electronics). The cyclic voltammogram (CV) measurements at a scan rate of 0.05 mV·s −1 and the electrochemical impedance spectroscopy (EIS) tests in the frequency range of 0.1 Hz to 100 kHz were performed on an electrochemical workstation (Shanghai Chenhua Instrument Co., Ltd.). showing that the Ni-Mg co-doping doesn't change the pristine spinel structure. The amplified pattern of (400) peaks present a slight movement towards the larger angle for the Ni-Mg co-doped samples, which indirectly interprets the decrease of the unit cell volume of the co-doped samples (Fig. 1b). As shown in Fig. 1(b), the lattice parameters of the LiNi x Mg 0.08 Mn 1.92−x O 4 samples display the decrease trend with the increased Ni 2+ content. Generally, the Mn 4+ exhibits an ionic radius of 0.53 Å, while the Mn 3+ shows two ionic radius of 0.58 Å and 0.645 Å in low spin state and high spin state, respectively 23 . In this regard, the high-spin state Mn 3+ ions (0.645 Å) is considered to be substituted due to the similar ionic radius of Ni 2+ (r = 0.69 Å) and Mg 2+ (0.65 Å), to balance the valence electrons in this structure, the low-spin state trivalent manganese ion would change to tetravalent manganese ion. The above two reasons lead to the decreased lattice constant in the doped samples. It has been confirmed that the Ni-Mg co-doping is attributed to the cell volume contraction, which is due to the the Ni-O bond (0.1915 nm) is shorter than that the Mn-O bond (0.1937 nm), and the of the Mg-O bonding energy is stronger than the Mn-O boding energy 14 . Moreover, the FWHM of (400) peaks gradually decreased with the increasing of Ni 2+ amount, which demonstrate the improved crystalline quality.   The surface chemical compositions and distribution of the as-prepared materials were further determined by XPS, as shown in Fig. 3(a). The Mn2p 3/2 binding energy is about at 642.5 eV, demonstrating that the Mn valance state in the co-doped spinel LiMn 2 O 4 is the Mn 3+ and Mn 4+ . On the basis of the XPS data, Fig. 3 Fig. 3(b), the r values of all samples are bigger than 1.18, which demonstrates that the Ni-Mg co-doping can actively limit Jahn-Teller effects of the LiMn 2 O 4 . In addition, the increase in the Mn 4+ content and the average oxidation state of Mn can enhance the structure stability and reduced the dissolution of Mn, hence improving the high rate capacity of the LiMn 2 O 4 cathodes. Figure 4(a) depicts the initial charge/discharge curves of the LiNi x Mg 0.08 Mn 1.92−x O 4 samples at 1 C between of 3.0 and 4.5 V at 25 °C. Seen that two well-defined voltage plateaus at 3.9-4.3 V can be observed for all samples, corresponding to a representative two-step intercalation/de-intercalation process of LiMn 2 O 4 . With the increased Ni 2+ content from x = 0 to 0.03, the charge voltage platforms are gradually descend, whilst the discharge platforms are elevated, however, when the Ni 2+ content increases from x = 0.05 to 0.15, the opposite can be true, which implying that the polarization of the electrode is increased. Figure 4(b) is the corresponding cycling performance of the LiNi x Mg 0.08 Mn 1.92−x O 4 (x ≤ 0.15) samples at 1 C and 25 °C. As shown, the cycling stability of the Ni-Mg dual-doped materials firstly rise with the increased Ni 2+ content from x = 0 to 0.03. When the Ni 2+ content gradually increases, the capacity decreases by degrees. Moreover, the introduction of Ni 2+ and Mg 2+ in the spinel structure weakens the first capability to some extent. This unfortunate results is attributed to the reduction of electrochemically active trivalent manganese ions. According to the previous reports 25 , the traditional Li + diffusion in the spinel structure is along the zigzag that hop from the 8a position to 16c site, providing that the next 8a site is vacant. When the Ni 2+ content increases, the 8a site will be replaced by increased Ni 2+ ions, so the Li + diffusion pathway is blocked by Ni 2+ ions. As a result, the initial capacity of the materials is greatly reduced when the nickel ions are excessive. Among all samples, the optimal LiNi 0.03 Mg 0.08 Mn 1.89 O 4 presents a initial capacity of 115.9 mAh·g −1 with an excellent capacity retention of 67% after 1,000 cycles. Figure 4(c) shows the discharge capacities cycled sequentially from 0.5 C to 10 C. As shown, the discharge specific capacities of all samples show a downward trend as the increased discharge rate. This is mainly because that the de-intercalation/intercalation process of Li + ions is hindered at the high rate 26 . Noted that the LiNi 0.03 Mg 0.08 Mn 1.89 O 4 exhibited a good rate performance than other samples at higher rate, which is attributed to the addition of Mg 2+ ions that enhance the ionic conductivity by lowering local Li + ions reaction energy barrier barriers 25 .

Results and Discussions
Additionally, Fig. 4  www.nature.com/scientificreports www.nature.com/scientificreports/ rate and 55 °C, as shown in Fig. 4(e). The discharge capacity of LiNi 0.03 Mg 0.08 Mn 1.89 O 4 sample displays smaller downward trend as the discharge rate increases. The initial capacity is 106.8 mAh·g −1 , 103.2 mAh·g −1 and 92.2 mAh·g −1 at 1, 5 and 10 C, respectively. Even at 10 C after 500 cycles, the 45.0 mAh·g −1 can be maintained. The above results demonstrate the improvement effect of Ni 2+ and Mg 2+ dual-doped on the discharge capacity at high rate and temperature.
To further study the structural stability of the materials, Fig. 5(a,b) shows the contrastive XRD patterns after 1,000 cycles at 1 C and 25 °C. The two electrodes have the similar diffraction patterns before cycle and after 1,000 cycles, indicating an integrated spinel structure of the LiMn 2 O 4 with a Fd3m space group. Especially, compared with the cycled LiMg 0.08 Mn 1.92 O 4 sample, the LiNi 0.03 Mg 0.08 Mn 1.89 O 4 sample shows relatively higher peak intensity and narrower FWHM, implying that the co-doped cathode maintain a good crystallinity and enhanced cycling performance. The composition of the LiNi 0.03 Mg 0.08 Mn 1.89 O 4 electrode after 1000 cycles was also further analyzed by XPS. The Mn2p 3/2 spectra is shown in Fig. 5(c) 27 . The content of Mn 4+ peak was determined after 1000 cycles to be greater than those of the fresh electrode (as shown in Fig. 3), indicating the Mn 3+ ions have dissolved during the charge-discharge process, so the later discharge capacity is relatively lower.  Fig. 5(d,e). After 1,000 cycles of 1 C, both the two cathodes still maintain inherent polyhedral morphology like the fresh cathodes.  No other significant structure transformation or particles damage is observed, except there is small amount of PVDF and carbon black on the particle surface. This result could explain that why the LiMg 0.08 Mn 1.92 O 4 and LiNi 0.03 Mg 0.08 Mn 1.89 O 4 cathodes deliver the similar electrochemical performance at low current rate of 1 C, as shown in the above Fig. 4b. In order to detect any structure modifications after the long-cycled electrochemical measurement, the LiNi 0.03 Mg 0.08 Mn 1.89 O 4 sample was further detected using HRTEM. As shown in Fig. 5(f), the crystalline planes (111) of LiMn 2 O 4 (JCPDS NO.35-0782) with corresponding distance of 0.472 nm can be confirmed, no other peaks were detected after 1000 cycles. This results are in accordance with the XRD analysis. Figure 6 shows the CV curves of LiMg 0  Fig. 6(a) have two pairs of redox peaks corresponding to two-step de-intercalation/intercalation of Li + ions. After 1000 cycles, peak currents are decreased significantly, indicating a relatively poor cycling stability of the LiMg 0.08 Mn 1.92 O 4 sample. As shown in Fig. 6(b), the corresponding peak symmetry of the LiNi 0.03 Mg 0.08 Mn 1.89 O 4 changed relatively little. These suggest that the addition of Ni 2+ and Mg 2+ can improve the reversibility of lithium ions.
The lithium ion diffusion coefficient (D Li+ ) can be calculated according to the following equation 28 : where i p is the value of peak current (mA), n is the electron transfer number (n ≈ 1 for spinel LiMn 2 O 4 ), C Li+ is the bulk concentration of Li + (given as 0.02378 mol·cm −3 for spinel LiMn 2 O 4 ), D Li+ stands for the Li + diffusion coefficient (cm 2 ·s −1 ) and v represents the scan rate (mV·s −1 ). As seen from Fig. 6(c,d), the peak current increases with the increased scan rate, and the peak current and peak area of the LiMg 0  Fig. 4(c), indicating the fast Li + ions diffusion rate. Figure 7(a,b) presents the Nyquist plots of LiMg 0.08 Mn 1.92 O 4 and LiNi 0.03 Mg 0.08 Mn 1.89 O 4 electrodes, respectively. An equivalent circuit model was used to fit the impedance signal (as seen the insets in Fig. 7a,b). This circuit included ohmic resistance of electrolyte (R s ), charge transfer resistance (R ct ), double layer capacitance (CPE), and Warburg impedance (W) 29 . The R ct values of LiNi 0.03 Mg 0.08 Mn 1.89 O 4 are 146.9 Ω and 64.9 Ω before and after 1000 cycles, respectively. By contrast, the LiMg 0.08 Mn 1.92 O 4 presents the higher R ct values of 177.6 Ω and 77.6 Ω, respectively. These results indicate a faster lithium ions diffusion rate in the Ni-Mg co-doped samples. To further explore the energy among the Li + ions diffusion, the activation energy (E a ) was tested by impedance method. www.nature.com/scientificreports www.nature.com/scientificreports/ Figure 7(c,d) shows the Nyquist plots for each electrode at different temperatures. So the activation energy will be calculated by the following equations: where i 0 stands for the exchange current, R represents the gas constant (8.314 J·mol −1 ·K −1 ), T (K) is the absolute temperature, n is the number of the electron transfer (n ≈ 1 for spinel LiMn 2 O 4 ), F is the Faraday constant (96484.5 C·mol −1 ), A is a temperature coefficient. Combined with Eqs 2 and 3, the equation of E a can be expressed: E a = -Rkln10, where k is the slope of the fitting line, namely k is the (logi o )/(1⁄T). As shown in Fig. 7(c,d)