A new approach to improve the electrochemical performance of ZnMn2O4 through a charge compensation mechanism using the substitution of Al3+ for Zn2+

ZnMn2O4 and Zn1−xAlxMn2O4 were synthesized by a spray drying process followed by an annealing treatment. Their structural and electrochemical characteristics were investigated by SEM, XRD, XPS, charge–discharge tests and EIS. XPS data indicate that the substitution of Al3+ for Zn2+ causes manganese to be in a mixed valence state by a charge compensation mechanism. Moreover, the presence of this charge compensation significantly improves the electrochemical performance of Zn1−xAlxMn2O4, such as increasing the initial coulombic efficiency, stabilizing the cycleability as well as improving the rate capability. The sample with 2% Al doping shows the best performance, with a first cycle coulombic efficiency of 69.6% and a reversible capacity of 597.7 mA h g−1 after 100 cycles. Even at the high current density of 1600 mA g−1, it still retained a capacity of 558 mA h g−1.


Introduction
Recently, rechargeable lithium ion batteries (LIBs), as energy storage devices, have been wildly applied in hybrid electric vehicles (HEVs), pure electric vehicles (EVs) and plug-in hybrid vehicles (PHEVs). However, in these cases, consumers want the vehicles to have a long cruising mileage and good safety, which requires LIBs with enhanced performance in terms of their energy density and safety. [1][2][3][4] Most commercial LIBs still use graphite as the anode material due to its low cost, stable capacity, and long cycle life. 5 However, the structure of graphite means that its theoretical capacity is only 372 mA h g À1 and the practical capacity is near to 360 mA h g À1 . Thus, the demand for next-generation LIBs with higher capacity has stimulated efforts to develop new materials 6,7 such as graphene, carbon nanotubes, silicon, tin oxide, etc.
These new materials can usually be classied into three groups in terms of their electrochemical mechanism. 8 The rst is the intercalation/de-intercalation type, including many carbonaceous materials like hard carbons, [9][10][11] carbon nanotubes [12][13][14][15] and graphene, 16 and some titanium oxides, such as TiO 2 (ref. 17) and LiTi 4 O 5 . 17 The second type is alloy/de-alloy mechanism. This type materials include such as silicon, 18 germanium, 19,20 tin, 21 antimony, 22 tin oxide 23 and SiO. 24 The third type is usually named as conversion mechanism mainly referring to some transition metal oxides such as manganese oxide, 25-27 cobalt oxide 28 and iron oxide. 29 These materials have the advantages of high capacity, high energy, low cost and environmentally compatibility but also exist disadvantages like low coulombic efficiency, unstable SEI formation, large potential hysteresis and poor cycle life.
ZnMn 2 O 4 belongs to the third type mechanism and not only has the same advantages, but also has an appropriated working potential above lithium, which could restrain the formation of lithium dendrites. However, ZnMn 2 O 4 has two obvious drawbacks. One is poor electrical conductivity. The other is instability during the charge/discharge process. ZnMn 2 O 4 can't be commercialized unless these two issues are solved.
So far, the investigations on the structural and electrochemical performances of the ZnMn 2 O 4 are inadequate and rather few works were reported. Deng et al. synthesized agglomerated pure spinel ZnMn 2 O 4 via calcination of an agglomerated Zn-Mn citrate complex precursor maintained a specic capacity of 650 mA h g À1 over 200 cycles. 30 Courtel et al. synthesized nanoparticles ZnMn 2 O 4 showed a capacity of 690 mA h g À1 aer 90 cycles by using co-precipitation method. 31 Feng et al. successfully synthesized ZnMn 2 O 4 with the particle size about 50 nm retained a capacity of 745 mA h g À1 aer 160 cycles through a rheological phase method. 32 The groups mentioned above usually focused on the preparation of nano-size ZnMn 2 O 4 . 33 It is well known that nano-size materials have many merits such as high surface area, short diffusion distance. However, the nano size materials usually need more binder as well as more conductive carbon in the formation of the electrode. The smaller the particle size is, the more binder amount needs.
It is well known that foreign metal ions doping can improve the electrochemical performances of the cathode materials. Generally, there are two kinds of substitution in terms of the charge balance. 34 One is the equivalent and the other is the nonequivalent substitution.
The equivalent-substitute ZnMn 2 O 4 , such as Cd-doped ZnMn 2 O 4 , 35 has been studied already. However, investigations of the nonequivalentsubstituted ZnMn 2 O 4 are rather limited.
In our work, we want to improve the electrochemical performances through the nonequivalent substitution. ZnMn 2 O 4 and Zn 1Àx Al x Mn 2 O 4 were synthesized by spray drying process following with annealing treatment. We also investigated the inuence of aluminum doping on the structural and electrochemical performances of ZnMn 2 O 4 .

Characterization
The phase composition and structure of the sample were performed using an X-ray diffractometer equipped with Cu Ka radiation (XRD, Rint 1000, Rigaku, Japan) over the 2q range of 10-90 . The morphologies and sizes of the samples were directly examined by scan electron microscopy (SEM). The electronic states of Mn in the samples were determined X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi).
CR2032 coin-tape cell was used to investigate the electrochemical performances of the synthesized materials. The active material was mixed in slurry containing 10 wt% of Super P as a conductive agent and 10 wt% NaCMC as a binder. The homogeneous slurries were cast onto copper foils to obtain electrode laminate which were dried at 110 C in the vacuum drying oven all over the night. The working cell was assembled in a glove box lled with dry argon. A lithium disk (F 15 Â 1 mm) was used as a negative electrode (counter electrode and reference electrode). A Celgard 2400 porous polypropylene lm was served as a separator. The electrolyte was 1 M LiPF 6 dissolved in a compound of diethyl carbonate/ethylene carbonate (DEC/EC, 1 : 1 by volume). The galvanostatic charge-discharge cycling and cyclic voltammetry were tested at room temperature (RT $ 25 C) by means of a computer-controlled battery evaluation system (LAND CT 2001, Wuhan, China). And the electrochemical impedance spectroscopy (EIS) measurements were performed via PARSTAT 2273 electrochemical workstation system (Princeton Applied Research, AMETEK, America) over frequency range of 100 kHz to 10 mHz with amplitude of 10 mV.

Results and discussion
XRD patterns of the Zn 1Àx Al x Mn 2 O 4 materials are shown in Fig. 1. All diffraction peaks can be indexed well to tetragonal ZnMn 2 O 4 (JCPDS card no. 24-1133; space group: I4 1 /amd). No additional peaks are observed, indicating that the samples are pure. The lattice parameters calculated by JADE are shown in Fig. 2. The lattice parameters of the Zn 1Àx Al x Mn 2 O 4 a/b are gradually decreased from 5.722Å to 5.6965Å when x is increased from 0 to 10%. In the case of lattice parameter c, it is initially increased from 9.236Å to 9.2793Å when x is increased from 0 to 2%, then decreased to 5.6965Å when x is increased 10%. This variation of lattice parameters should be ascribed to the substitution of Al 3+ to Zn 2+ since the radius of the Al 3+ is    smaller than that of the Zn 2+ (Al 3+ : 0.535Å, Zn 2+ : 0.74Å). These results also indicate that Al 3+ has been successfully introduced into the ZnMn 2 O 4 lattice. Fig. 3 shows the SEM photographs of ZnMn 2 O 4 and Zn 1Àx -Al x Mn 2 O 4 samples. All samples have an aggregated morphology of some irregular primary particles whose size is around 150 nm. There is no change in terms of the particle size and shape aer the substitution of Al 3+ for Zn 2+ . The EDX images of the ZAMO2 are shown in Fig. 4. It can be seen that the sample is composed of zinc, manganese, oxygen and aluminum and those elements are in uniform distribution. It further proves that the aluminum ions were successfully introduced into the ZnMn 2 O 4 lattice.
The cyclic voltammograms of Zn 1Àx Al x Mn 2 O 4 electrode are provided in Fig. 5. All cells were operated at a scan of 0.1 mV s À1 in the voltage range of 0.01-3.0 V versus Li/Li + . In the case of the ZnMn 2 O 4 , there are three cathodic peaks and two anodic peaks in the rst cycle. According to the discussion in literatures, the rst peak located at 1.2 V is attributed to the reduction of Mn 3+ to Mn 2+ . 36 The second one located at 0.8 V is ascribed to the electrolyte decomposition along with the formation of solidelectrolyte interface (SEI) on the surface of the electrode. Both of those two peaks disappear in the following cycles. The sharp and intense peak at 0.13 V is attributed to reduction of Mn 2+ and Zn 2+ to metallic Mn and Zn nanoparticles, respectively. 37 Simultaneously, the Li-Zn alloy is also formed at this low potential. There are two peaks in the rst charge process, one is at 1.3 V, and the other is at 1.5 V, corresponding to the oxidation of metallic Mn and Zn nanoparticles to MnO and ZnO. 38 In the second and the third cycles, there is only one peak located at 0.5 V in the discharge process. They are resulted from the reduction of MnO and ZnO to Mn and Zn. The similarity of the subsequent CV curves indicates the high electrochemical reversibility of the sample. The cyclic voltammograms of Zn 1Àx Al x Mn 2 O 4 electrodes are shown in Fig. 5b-e. They are almost the same as the cyclic voltammograms of ZnMn 2 O 4 , implying that the Al 3+ doping has no effect on the chargedischarge mechanism of ZnMn 2 O 4 . The rst discharge-charge curves of ZnMn 2 O 4 and Zn 1Àx -Al x Mn 2 O 4 are shown in Fig. 6 [43][44][45][46] The cyclic performances of ZnMn 2 O 4 and Zn 1Àx Al x Mn 2 O 4 are depicted in Fig. 7. All half-cells were operated at a current density of 100 mA h g À1 in the voltage range of 0.01-3.0 V in room temperature, using Li metal as an anode. The specic discharge capacity of ZnMn 2 O 4 is gradually decreased till the 40th cycle and then increases till the 120th cycle. This abnormal increase in the discharge capacity between the 40th and 120th cycle have been ascribed to two factors. One is the formation of a polymer organic layer on the electrode during the cycling which can reserve Li reversibly. 47,48 The other is the particles cracked during the cycling so that the contact area between the electrode and electrolyte solution increases. 49 This phenomenon is not always good. When the nanostructure completely crumbled, the active materials will be separated from the conductive agent. This is bad for charge and ion transmission. As expected, following the rise, there is a quick decline in the capacity.
The discharge/charge curves of Zn 1Àx Al x Mn 2 O 4 electrodes are shown in Fig. 7b. It is obvious that the uctuation of capacity disappears gradually aer the doping with Al and the cycleability tends to be steady. ZAMO2 has a reversible capacity of 597.7 mA h g À1 .
The rate capabilities of ve samples are shown in Fig. 8, in the picture, Zn 1Àx Al x Mn 2 O 4 exhibits a much higher capacity at   Paper a current density of 1600 mA g À1 . The capacities of ZAMO2 and ZMO are 557.8 mA h g À1 and 379.8 mA h g À1 . When the current density goes back to 100 mA g À1 , the capacity of ZAMO2 is also higher than ZMO. It indicates that ZAMO2 is more appropriate for charge/discharge at high current density.
In order to clarify the mechanism of this improved performances aer the introduction of Al, the X-ray Photoelectron Spectroscopy (XPS) and Electrochemical Impedance Spectroscopy (EIS) were measured and the results were provided in Fig. 9 and 10. Fig. 9 shows the characteristic peaks of Mn 2p , Mn 3s and O 1s . The binding energy of Mn 2p 3/2 , multiplet splitting of the Mn 3s level, the binding energy of O 1s and the atomic ratio of Mn 3+ / Mn 2+ are list in Table 1. The binding energy of Mn 2p 3/2 for the samples of Zn 1Àx Al x Mn 2 O 4 slightly shis from 641.1 eV to 641.2 eV, while the binding energy of O 1s shis from 529.8 eV to 529.9 eV, when x increases from 0 to 0.1. These results implied that the substitution of Al for Zn increases the interaction between Mn and O. As to the DE 3s of Mn, it increases from 4.5 eV to 5.1 eV when x is from 0 to 2%. According to previous research, 50 the higher DE 3s value indicated a lower valence of Mn. It is well known that Zn in ZnMn 2 O 4 exhibits +2 and Mn is +3, aer the substitution of Al 3+ for Zn 2+ , the valence of Mn should be lower so that the sample displays electric neutrality. Furthermore, the atomic ratio of Mn 3+ /Mn 2+ calculated through XPS analysis decreases from 6.5 to 4.5 when x is from 0 to 2%. Those results prove that some Mn 3+ in Zn 1Àx Al x Mn 2 O 4 is partly transformed to Mn 2+ in order to balance the Al 3+ doping. Fig. 10 shows the Nyquist plots of Zn 1Àx Al x Mn 2 O 4 before cycling. The depressed semicircles in the high frequency region are related to the ohmic resistance (R e ) and the charge transfer resistance (R ct ) through the electrode/electrolyte interface. [51][52][53] Low resistance is good for the charge and ion diffusion. From the gure, we can easily nd that ZAMO2 has the least diameter of the semicircles. Fig. 11 shows the Nyquist plots of Zn 1Àx Al x Mn 2 O 4 aer 10 cycles. The curves are almost lines with a slope of 45 degrees. This indicates that the batteries are controlled by the diffusion step of Li + in solid phase.
According to the result of XPS and EIS, we believed that the introduction of Al makes Mn into a mixed valence state which can shorten the band gap width of manganese oxide and result in materials with higher conductivity. SEM shows that the particle size and morphology have no change aer the doping, so that this improvement is all result from the charge compensation aer the nonequivalent substitution.

Conclusion
In summary, the nonequivalent substitution of ZnMn 2 O 4 was successfully realized by spray drying process following with annealing treatment. Results of XRD, EDX and XPS indicate that the materials are pure and the aluminum ions are successfully introduced into the ZnMn 2 O 4 lattice. Meanwhile, the valance of Mn in Zn 1Àx Al x Mn 2 O 4 is in a mixed state because of charge compensation. ZAMO2 exhibits excellent stability, which maintains a high reversible of 597.7 mA h g À1 at the current density of 100 mA g À1 aer 100 cycles. Even at high current density of 1600 mA g À1 , the reversible capacity is still kept at 557.8 mA h g À1 , which is higher than the capacity of theoretical commercial graphite. The electrochemical results demonstrated that the nonequivalent-substitution is a successful try for development of advanced anode material for highperformance LIBs.

Conflicts of interest
There are no conicts to declare.