- Email: [email protected]
RGO cathode material with excellent voltage platform and cycle performance
Electrochimica Acta 225 (2017) 272–282
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Low-temperature synthesis of LiMnPO4/RGO cathode material with excellent voltage platform and cycle performance Xiaoning Fu, Kun Chang* , Bao Li, Hongwei Tang, Enbo Shangguan, Zhaorong Chang* Collaborative Innovation Center of Henan Province for Green Manufacturing of Fine Chemicals, Key Laboratory of Green Chemical Media and Reactions, Ministry of Education, School of Chemistry and Chemical Engineering, Henan Normal University, Xinxiang 453007, PR China
A R T I C L E I N F O
Article history: Received 24 October 2016 Received in revised form 16 December 2016 Accepted 25 December 2016 Available online 26 December 2016 Keywords: Lithium ion battery lithium manganese phosphate low temperature synthesis high voltage platform graphene oxide
A B S T R A C T
Pure and well-crystallized LiMnPO4/reduced graphene oxide (RGO) nanopowders are synthesized by adding a small amount glucose and graphene oxide simultaneously in dimethyl sulfoxide (DMSO)/H2O, under constant atmospheric pressure and at low-temperature (108 C). The scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images show that the addition of different amounts of graphene oxide can significantly affect the particle size and morphology of LiMnPO4/RGO composites. With small amounts of graphene oxide (1 and 3 wt.%), the small LiMnPO4 particles are wrapped in RGO, in a cocoon-like structure. This special morphology can be maintained after a rapid carbon coating treatment at high temperature. Electrochemical studies show that these cocoonlike C-LiMnPO4/G nanocomposites not only have a higher discharge specific capacity, but also show improved high voltage platform and high rate cycle performance. When the added graphene oxide is 3%, the specific capacity of C-LiMnPO4/G nanocomposite is 160.8 mAh g1 at 0.05 C, the discharge capacity in the area of more than 4.0 V is up to 115 mAh g1, accounting for 70% of the total discharge capacity. The proposed C-LiMnPO4/G nanocomposites also exhibit an outstanding high rate capability, where the discharge specific capacity at 1C can reach to 99.6 mAh g1 and after 1000 cycles at 5 C, it still has 83% of capacity retention. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction At present, olivine-type polyanionic compounds such as LiMPO4 (M = Fe, Mn, Ni and Co) have attracted increasing attention as powerful cathode materials for lithium ion batteries, due to their good structure and chemical stability [1,2]. Compared with LiFePO4, LiMnPO4 is a more interesting material because of its redox potential (4.1 V vs. Li/Li+) which shows 0.7 V higher than that of LiFePO4, and is still within the stable electrochemical window of currently used liquid electrolytes. Meanwhile, LiMnPO4 also has a much higher theoretical energy density than LiFePO4, 701 Wh kg1 = 171 mAh g1 4.1 V vs. 586 Wh kg1 = 170 mAh g1 3.45 V; this plays an important role for applications in electric cars. In addition, LiMnPO4 has other important properties; in fact it can be produced using abundant and low cost raw materials, it is environmental friendly, it has good structural stability, wide chemical compatibility, and it is safe to use; because of all this, it is
* Corresponding authors. E-mail addresses: [email protected] (K. Chang), [email protected] (Z. Chang) . http://dx.doi.org/10.1016/j.electacta.2016.12.161 0013-4686/© 2016 Elsevier Ltd. All rights reserved.
considered the most promising cathode material for Li batteries [3–5]. Nevertheless, LiMnPO4 has a very poor electronic conductivity and Li+ diffusion coefficient, which make it an insulator material [6], with poor electrochemical activity [7]. To improve LiFePO4 performance, different methods were developed which can increase the ion diffusion coefficient and the electron conductivity; these include carbon coating on the surface of nanoparticles, bulk phase element doping, nanocrystallization of material particles [8– 13]. These improvements are mostly focused on the enhancement of the discharge capacity at low rates. Less investigation was performed, however, on the performances of the discharge of high voltage platform and of the high rate discharges; consequently some materials have a higher specific capacity at low rate, but the platform performance at high voltage is poor [14–19] .For examples, Chen et al. [20] synthesized CeO2-modified LiMnPO4/ C composites by solvothermal method at 190 C for 24 h, and then calcined at 550 C for 3 h under dynamic argon atmosphere, the discharge capacity of the prepared composites was about 60 mAh g1 above 4.0 V high voltage. Zhu et al. [21] used Mn (CH3COO)24H2O and LiH2PO4 as the raw material, synthesized LiMnPO4 in diethylene glycol solution keeping 3 h in the
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
temperature of 160 C. LiMnPO4/C composites were prepared using planetary ball-milling of the LiMnPO4 powders with 20 wt.% carbon black; they can deliver a discharge capacity of 129 mAh g1 at 0.05 C, but at more than 4.0 V high voltage discharge area, the discharge capacity of the material was very low, about less 30 mAh g1. Dai et al. [22] studied the impact of the rate properties of LMP, when doped with different proportions of vanadium. A series of LiMn1xVxPO4/C (0 < x < 0.075) samples were synthesized via an improved solid-state method. Results showed that the LiMn0.075V0.025PO4/C exhibited the highest discharge capacity of 108 mAh g1at 0.2 C for the first cycle; for over 4.0 V high voltage discharge area, however, the discharge capacity of the material was still too low, about less than 40 mAh g1. Guo et al. [23] synthesized the plate-like LiMnPO4 material by a solvothermal method. After a high temperature carbon treatment, at over 4.0 V high voltage discharge area, the discharge platform of C- LiMnPO4 was relatively good. Coating with too much carbon (weight ratio of LiMnPO4/glucose is 2:1) would greatly reduce the amount of active material and, hence, reduce the volume ratio of energy of the electrode. Meanwhile, the solvent thermal conditions at high temperature and high pressure were quite harsh. As it is well-known, graphene has good electrical conductivity and chemical and mechanical stability; literature showed that it can improve the electrochemical performance of LiMPO4 cathode materials [24–27]. Zong et al. [28] synthesized LiMnPO4-C/ Acetylene black (AB) and LiMnPO4-C/graphenes by solvothermal method. The samples had a capacity of 134 mAh g1 and 139 mAh g1 at 0.05 C, respectively. The retention of discharge capacity maintained 90% and 94% of the initial values after 40 cycles. When the discharge ratio was increased to 0.5 C and 1.0 C, the discharge capacities of LiMnPO4-C/Acetylene black were 92 mAh g1and 60 mAh g1, but that of LiMnPO4-C/graphenes still could be as high as 125 mAh g1and 119 mAh g1. These data confirm that the electric capacity and cycle performance of LiMnPO4 samples improved substantially with the addition of graphene. Despite this, however, the discharge voltage platform was still not ideal; in fact at over 4.0 V high voltage discharge area, the discharge capacity of the material was about less than 40 mAh g1, accounting for only 30% of the total capacity. To take advantage of the 4.1 V vs. Li/Li+ redox potential of LiMnPO4, further improvements in the performance of the materials are needed. These can be achieved by optimizing the size of the nanoparticle and their structural morphology, and by optimizing the preparation of LiMnPO4/ graphene composites. An effective improvement of the electrical conductivity of the particle will enhance the material’s discharge platform, taking full advantage of its high energy density. In our previous work, our team had synthesized sphere-like LiFePO4/C nanocomposites with particle sizes of about 100– 300 nm via a solution-phase method. The synthesized LiFePO4/C sample showed excellent rate and cycle performances [29,30]. Based on our previous works, we added glucose as additive in dimethyl sulfoxide solvent; the glycoside formed during the reaction acted as a surfactant and led to the successful preparation of hollow spindle LiMnPO4 particles [31]. The discharge capacity was 161.8 mAh g1mAh g1 at 0.05 C. To further improve the high voltage platform performance, the high rate performance and the electrical conductivity between the particles, we also added different mass fraction of graphene oxide in the solution, together with the glucose. The results showed that nano LiMnPO4/RGO was synthesized after the reaction. With the graphene oxide ratios of 1%, 3%, the materials have a cocoon-like morphology, with the surface of the particles completely wrapped up by soft graphene fold layer. Nano C-LiMnPO4/G composite was then produced after a rapid high temperature carbon treatment. This kind of composite with special morphological structure had excellent conductive network
273
structure, produced by the close contact of LiMnPO4 particles with amorphous carbon and graphene layers; this will greatly improve the high voltage platform performance and the high rate cycle performance of the material. This proposed method is based on a simple and facile procedure, using cheap raw materials and under mild reaction conditions (no high temperatures or pressure need to be employed). This method can also be applied in the preparation of other olivine structure for lithium ion battery materials, it has therefore great application prospects. 2. Experimental section 2.1. Synthesis of graphene oxide Graphene oxide (GO) was prepared by an improved Hummers method [32]: 1.0 g graphite were added to 100 ml concentrated H2SO4 (98%). 4.5 g of KMnO4 and 0.5 g of NaNO3 were slowly added into the above solution, which was magnetically stirred in an ice bath for 10 min. The mixture continued to be stirred for 3 days at room temperature. Subsequently, water was added dropwise to the mixture under vigorous stirring, and the stirring was continued until the solution was cooled to room temperature. 15 ml of H2O2 solution (30 wt.%) and 200 ml of water were added dropwise under vigorous stirring for another 2 h. Then, the resulting graphite oxide suspension was washed 3 times with HCl (0.5 M), and washed repeatedly with distilled water, until the solution pH keep a constant value of about 5.0. The sol was scattered to H2O again. After ultrasound treatment, the complete delamination of graphite oxide was achieved. 2.2. Synthesis of samples The synthesizing procedure of sample LiMnPO4/RGO is shown in Fig.S1 (See Supplementary materials); the process is based on our previous work to prepare LiMnPO4 [31]. Deionized water and DMSO (VH2O = VDMSO = 25 ml) were mixed in a three neck round bottom flask. 5% glucose was slowly added to the above solution. The DMSO-H2O solution with glucose was vigorously stirred and heated to 108 C for 2 h, then cooled to room temperature. This is named Solution A. Solution B was obtained by dissolving 0.05 mol of manganese acetate tetrahydrate (Mn(CH3COO)24H2O, 99%) and 3.42 ml of phosphoric acid (H3PO4, 85%) into the solution A under constant stirring, for 1 h at 50 C. Then, 6.3545 g of lithium hydroxide (LiOHH2O, 99%) were slowly added into the Solution B. The final concentration of lithium hydroxide was 1.5 mol L1. The reaction solution was heated at 108 C for 2 h under nitrogen atmosphere, cooled to room temperature, and stored overnight. The LiMnPO4 material (named LMP) was obtained by precipitation; the precipitate was washed several times with deionized water and ethanol and dried overnight in an oven at 60 C. To facilitate the measurement and guarantee the repeatability of experiments, experiments according to the theoretical production of LiMnPO4 were performed adding different amounts of GO to the reaction system; more specifically, solution with GO concentrations of 1 wt.%, 3 wt.%, 5 wt.% and 7 wt.% were prepared. The corresponding samples obtained by liquid phase reaction were denoted as LMP/RGO (1 wt.%), LMP/RGO (3 wt.%), LMP/RGO (5 wt. %), and LMP/RGO (7 wt.%). The C-LiMnPO4/G was synthesized via the following steps; samples LMP, LMP/RGO(1 wt.%), LMP/RGO(3 wt.%), LMP/RGO(5 wt. %) and LMP/RGO(7 wt.%) were milled with 10% of glucose. The milled powder was collected in boat shape alumina crucible and then heated at 650 C for 4 h in an atmosphere of (N2:H2 = 95:5) in a tube furnace, and then cooled to room temperature. The resulting product was named as C-LMP, C-LMP/G(-1), C-LMP/G(-3), C-LMP/G
274
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
(-5), C-LMP/G(-7). According to the TGA analysis in Fig. S2, it can be seen that the carbon weight loss of sample C-LMP/G(-3), C-LMP, and LMP/RGO (3 wt.%) is 4.8%, 3.4%, and 1.4%, respectively. It is therefore calculated that the carbon content is 0.47% for LMP/RGO (1 wt.%), 2.35% for LMP/RGO (5 wt.%), and 3.29% for LMP/RGO (7 wt. %); after coating carbon, the carbon content is 3.87% for C-LMP/ RGO (-1), 5.75% for C-LMP/RGO (-5), and 6.69% for C-LMP/RGO (-7), respectively. 2.3. Characterization of samples Phase analysis and cell parameter determination of all samples were performed by X-ray diffraction (XRD) using a D8 X diffractometer (Germany, Bruker) with Cu-Ka radiation. The particle morphology of all samples was analyzed by the Scanning electron microscopy (SEM, S4800, Japan) and transmission electron microscopy (TEM, JEM-2010, JEOL). Raman spectra of CLFP/G cathodes were recorded on a Raman spectrometer (HR800) using laser excitation energy of 532 nm. 2.4. Electrochemical measurements The electrodes were prepared using the following procedure: the LMP-based sample, acetylene black and polyvinylidene fluoride (PVDF) were mixed at a mass ratio of 7:2:1. The suspensions were prepared after adding a certain amount of
analytical grade N-methyl-2-pyrrolidinone (NMP) (Sinopharm Chemical Reagent Co., Ltd.) to the mixture, and grinding for 60 min with a pulp refiner. Then, the suspensions were uniformly coated on an aluminum foil with an applicator. The solvents were then evaporated by drying the fabricated electrodes at 80 C overnight in a vacuum oven. The dried electrodes were pressed using a twin roller to ensure a close contact between the electrode materials and the current collector. The pressed electrodes were cut into circular discs of 12 mm diameter and then dried again at 120 C overnight in a vacuum oven. The LMP-based material loading on the electrode was about 2.5 mg. For half cells, lithium metal was used as negative electrode, polypropylene microporous membrane as the separator, while 1 mol L1 LiPF6 (EC: DEC: DMC is 1: 1: 1, V/V/V) was used as the electrolyte. The coin cells (CR2025) were assembled in an Ar-filled box (MBraun, Germany). The cells were tested using a land CT2001A battery tester (Wuhan Jinnuo Electronics Co. Ltd., China) at room temperature, at rates of 0.05 C– 10 C. In the electrochemical stability experiments, the cells were cycled between 2.50 V and 4.50 V vs. Li/Li+. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were performed using an electrochemical work-station (Shanghai Chenhua Instrument Company, China). The CV test scan rates were 0.2 mV s1, 0.5 mV s1, 0.8 mV s1 and 1.0 mV s1 over a cell voltage from 2.5 V to 4.5 V. EIS was performed over a frequency range of 0.05 Hz–10 kHz.
Fig. 1. (a, b) XRD patterns of LiMnPO4/RGO and C-LMP/G; (c, d) XRD spectrums of GO before and after reaction.
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
3. Results and discussion 3.1. Structure and morphology characterization Fig. 1(a, b) shows the XRD patterns of the LMP/RGO and C-LMP/ G materials obtained by adding different content of graphene oxide. Comparing with the standard LiMnPO4 pattern (PDF reference number 33-0804), it can be seen that all samples are well crystallized, regardless of the graphene oxide content. All diffraction peaks could readily be indexed to the pure phase of LiMnPO4 with an orthorhombic olivine-type structure, without other impurity peaks. These data confirmed that the pure phase LiMnPO4 can be directly prepared using the liquid phase method; the addition of graphene oxide does not change the crystal structure of the sample When a certain amount of glucose is blended with samples, followed by a rapid high temperature carbon treatment, the obtained samples C-LMP/G(-1), C-LMP/G(-3), C-LMP/G(-5), C-LMP/ G(-7) still showed to be pure phases, with an orthorhombic olivinetype structure. As no other impurity peaks were detected, it is likely that carbon originated from the glucose thermolysis in the composite is amorphous.
275
The crystal cell parameters, crystal cell volume, the grain size, FWHM of samples were calculated using the Scherrer equation, and they listed in Table S1(See Supplementary materials). It can be seen that the cell volumes of C-LMP/G(-1), C-LMP/G(-3), C-LMP/G (-5), C-LMP/G(-7) were larger than that of C-LMP. It can be seen that the introduction of graphene oxide could increase the cell volume of LiMnPO4 particles, which is beneficial for the deintercalation of lithium ions. [33]. From the values of FWHM of the (111), (020) and (311) crystal faces it can be seen that, with the increase of graphene oxide, the FWHM of the characteristic peaks first decreased and then increased again; for the grain size, on the contrary, the opposite behavior was observed, i.e. an increase followed by a decrease. The FWHM of the main crystal plane diffraction peaks (111, 020 and 311) reached a minimum for sample C-LMP/G(-3), while the corresponding grain size is the largest. This not only shows that CLMP/G (-3) had the highest crystallinity, but also that the grain orientation growth was preferential along the vertical crystal direction, especially along the vertical direction of the (020) crystal face. This lattice orientation can favor the deintercalation of lithium ions, beneficial to the improvement of the performance [34,35]. To examine the possible changes of the graphene oxide due to the reaction in liquid phase, we took a certain quantity of graphene
Fig. 2. SEM micrographs of (a) LMP; (b) LMP/GO (1 wt.%); (c) LMP/GO (3 wt.%); (d) LMP/GO (5 wt.%);(e) LMP/GO (7 wt.%); (f) C-LMP/G(-1); (g) C-LMP/G(-3).
276
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
oxide, and made it react for 2 hours in the same liquid phase system. XRD spectrums of GO before and after the reaction are shown in Fig. 1(c, d). It can be seen that after the liquid phase reaction, the characteristic peak of graphene oxide for 2u = 11 disappeared; different strong diffraction peaks, however, appeared for 2u = 26 , 43 and 46 , which correspond to the characteristic peaks of graphite (PDF#26-1079). This shows that graphene oxide can be reduced to RGO by glucose in DMSO-glucose-GO-H2O reaction system. Therefore, in the above reaction medium, GO was gradually reduced to RGO and wrapped on the surface of LiMnPO4 particles, generating LMP/RGO composites with olivine phase. 3.2. Study of the morphology Fig. 2 shows SEM images of LMP/RGO samples with different content of graphene oxide and of samples C-LMP/G(-1), C-LMP/G (-3). It can be seen that the different amounts of graphene oxide affected both the particle size and the morphology of the material. As seen in Fig. 2(a), the morphology of the LMP sample without graphene oxide is spindle-like, with a mean length of 700–900 nm and width of 200–300 nm. The particle size distribution was very uniform, with no agglomeration occurring. Under the lower contents of graphene oxide (1 wt.% and 3 wt.%), the LMP particles
maintained the spindle morphology and wrapped with soft graphene oxide fold layers. During the stirring solution phased system, the composites resembled to form a silkworm cocoon-like structure, where the graphene layers wrapped LMP particles with 300–500 nm, as show in Fig. 2(b, c-1, c-2). During the formation process of the final products, the Mn2+ ions can be adsorbed on the surface of graphene oxide due to the containing large amount of oxygen functional groups, and then reacted with Li+ ions and PO43 ions in the solution. The LiMnPO4 nanocrystals were in-situ anchored on the surface of graphene oxide layers and finally grown into nanoparticles. Due to the less amount of GO, the particle morphology of LMP/RGO (1 wt. %, 3 wt.%) sample did not change but maintained a spindle shape, with the particles just becoming smaller. The surface of the particles is wrapped up by reduced graphene oxide layer, it is similar to cocoons. Fig. 2(d-1–d-3) are SEM Fig. of LMP/RGO(5 wt.%) and at different magnifications. As it can be seen, when the amount of graphene oxide was increased to 5 wt.%, the particle of LMP/RGO(5 wt.%) has a clear change in morphology, gradually becoming diamond cube-like, and it is mixed with RGO. In the images with higher magnification of Fig. 2(d-2 and d-3), it can be seen clearly that, for the diamond cubic particles, the local surface is smooth, and clearly not wrapped by RGO. This phenomenon shows that when the amount of graphene oxide is
Fig. 3. The TEM of (a) LMP/RGO (1 wt.%); (b) LMP/RGO(3 wt.%); (c) LMP/RGO(5 wt.%) and (a-1) C-LMP/G(-1); (b-1) C-LMP/G(-3); (c-1) C-LMP/G(-5) and) (a-2, b-2, c-2 the local amplification Fig. 3(a-1, b-1, c-1).
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
more than 5 wt.%, the ability of GO to wrap around crystal grain decreases. According to Fig. 2(e), when the amount of graphene oxide is to 7 wt.% (sample LMP/GO (7 wt.%)) the particles gradually became a larger rhombus block structure, with a very smooth surface. It also can be seen that the RGO aggregates together, not package crystal particles. These images show that when the amount of GO is to 7 wt.%, the wrapping ability of LiMnPO4 grains almost disappears. With the increase of GO content, the solution became acidic because of many oxygen functional group in GO. Thus, with the increase of GO content, the pH value in reaction system would be decreased, leading to the aggregation the GO, which decreased the ability of coating LiMnPO4 particles. In addition, when GO content increased, partial of added glucose as reductant and surfactant would take part into the GO reduction reaction, decreasing the effect of its surfactant. The interfacial tension of the LiMnPO4 particles would also affected during the growing process, resulting in the final structure and morphology. Based on these results, samples LMP/RGO(1 wt.%) and LMP/RGO (3 wt.%) underwent a short high temperature carbon treatment; the morphology of the obtained samples C-LMP/G(-1) and C-LMP/ G(-3) is shown in Fig. 2(f, g). The images show that the morphology of the composites did not change significantly; indeed particles still maintained the original morphology of silkworm cocoon. The size of the particles also did not show significant changes. The close contact of LiMnPO4 particles with amorphous carbon and the reduced graphene layers formed an excellent conductive network, which led to the deintercalation of lithium ions and enhanced electron transportation. Fig. 3 shows the TEM images of LMP/RGO (1 wt.%), LMP/RGO (3 wt.%), LMP/RGO (5 wt.%) before and after high temperature carbon treatment. It can be seen that LMP/RGO (1 wt.%) and LMP/ RGO (3 wt.%) have a spindle crystal structure, wrapped up by RGO, similar to the shape of silkworm cocoon. After a rapid high
277
temperature carbon treatment, the composites maintain the original morphology of silkworm cocoon. Graphene and amorphous carbon form a dense cladding layer on the surface of LiMnPO4 particles. The carbon layer thickness is about 10 nm. LMP/RGO (5 wt.%) is like a diamond cube with a mean length of 600–800 nm. Some RGO layers are distributed around the LMP/ RGO (5 wt.%) particles. The originally smooth surface was coated by the amorphous carbon after a carbon treatment, the carbon layer thickness being approximate 10 nm. Obviously, the larger size diamond cube structure is not beneficial for the deintercalation of lithium ions. Although there is carbon layer formed on the surface of particle after a carbon coating treatment, large particle size and without support by RGO would extremely decrease the electronic conductivity of the LMP particles. Fig. 4 represents a scheme of the morphology change of LMP according to the different content of graphene oxide. It can be seen clearly that graphene oxide has a great influence on grain growth and morphology of LMP samples prepared with liquid phase system. The cocoon-like structure has excellent ion and electron conduction ability. 3.3. Raman spectroscopy To further identify the degree of graphitization of carbon in the samples, Raman spectra were acquired; they are shown in Fig. 5. It can be seen that sample LMP, with no GO, did not show any of the graphite characteristic peaks; this indicated that the glucose added in the liquid phase synthesis system had not been carbonized and was not loaded on the surface of LMP sample particle, but it only acted as a surfactant. The sample C-LMP, which underwent carbon coating treatment at 650 C, shows two strong peaks at 1350 cm1 and 1600 cm1, corresponding to the defect mode of the disordered carbon
Fig. 4. A schematic diagram of the reaction process and the changes of the morphology.
278
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
Fig. 5. Raman spectra of LMP, C-LMP, LMP/RGO (3 wt.%), C-LMP/G(-3).
(D-band) and the E2g vibration mode of the ordered graphitic carbon (G-band). The ID/IG value is 0.803. This may be because some glucose pyrolysis formed small, flat shaped, irregular arrangement of two dimensional carbon materials. The sample LMP/RGO (3 wt.%) with added GO have distinct peaks corresponding to D-band and G-band; the ID/IG values is 0.872. These data indicated that the added graphene oxide can be reduced to graphene by glucose in GO/Glucose/DMSO/H2O system, and that, combined with LMP particles, can form LMP/RGO. After this sample was coated with carbon at 650 C, its ID/IG values decreased to 0.806. The increase of graphitization indicates that the amorphous carbon coming from the glucose pyrolysis further enhanced the GO reduction during the high temperature carbon treatment processing. The ID/IG values of samples LMP, C-LMP, LMP/GO (3 wt.%), C-LMP/G(-3) are summarized in Table 1. 3.4. Electrochemical properties Fig. 6 shows the discharge curves of the samples with different contents of graphene, at different rates of 0.05C- 1C. It can be seen the samples have a discharge voltage platform around 4.1 V vs. Li/ Li+, which is typical for the Mn(II) $ Mn(III) redox for olivine manganese phosphates. The specific discharge capacities of the different samples are listed in Table 2. Fig. 6(a) shows the discharge curves of the samples at 0.05C. According to the data in Table 2, the first discharge capacities of CLMP, C-LMP/G(-1)and C-LMP/G(-3) were 161 mAh g1,162.5 mAh g1,160.8 mAh g1, respectively, while for samples C-LMP/G(-5) and C-LMP/G(-7) much lower values were measured 131.1 mAh g1 and 102.7 mAh g1. It can be seen that the first three samples all have a discharge capacity higher than 160 mAhg1. Fig. 6a indicates that three samples exhibit the different high-voltage performance at 4.0 V. The voltage of the discharge platform for C-LMP/G(-1) and C-LMP/G(-3) samples are higher than that of C-LMP sample without GO. The discharge capacities of these two samples are also significantly higher than that of C-LMP above 4.0 V high voltage discharge. Table S2 listed the electrochemical performance of LMP
Table 1 ID/IG values of different materials. Sample
LMP
C-LMP
LMP/GO (3 wt.%)
C-LMP/G(-3)
ID/IG
0
0.803
0.872
0.806
reported in recent years, it can be seen that our proposed LMP/RGO electrode exhibits the best high-voltage properties. Fig. 6(a-2) gives a comparison of the discharge capacity above 4.0 V high voltage discharge area between C-LMP/G(-3) and C-LMP electrodes. Obviously, the discharge capacity of C-LMP/G(-3) is 115 mAh g1 which is higher than that of C-LMP with 82 mAh g1. This fact implies that adding a small amount of graphene oxides can significantly improve the discharge performance of the samples, in particular at high voltage. The main reason may be that the presence of small amount of GO give C-LMP/G particles a cocoon-like structure. So this morphology has excellent ion and electronic conductivity, and it effectively reduces the polarization in the process of charging and discharging; making it close to the inherent Mn(II)$ Mn (III) oxidation reduction potential (4.1 v vs Li/ Li +) for LMP. Fig. 6(b–d) show, respectively, the discharge curves of samples C-LMP, C-LMP/G(-1), C-LMP/G(-3) at different rates (0.05C-1C). Comparing them, the capacitance of the three materials are very close at 0.05C-0.2C; for higher discharge ratios, however (from 0.5C to 1.0C), the specific capacity of C-LMP/G(-3) is higher than those of LMP/G and C-LMP/G(-1). This is due to the electrode polarization; in fact when the charge and discharge current is very small, the resistance of the electrode materials (i.e. inherent ohmic voltage drop) is also small. For higher charge and discharge currents, the inherent ohmic voltage drop of the electrodes will become larger. Obviously, the intrinsic resistance of C-LMP/G (-3) sample is smallest, and an addition of 3% of GO led to the material with the best properties. The discharge and cycle performance curves of C-LMP and C- LMP/G (-3) at high rate are shown in It was found that for charging and discharging rate of 2C, 5C and 10C, the specific discharge capacity of C-LMP/G(-3) was clearly higher than for C-LMP. In addition, C-LMP/G(-3) sample showed excellent cycle performance at high rate. At 5C, after 1000 cycles, 83.0% of the discharge capacity of the C-LMP/G(-3) electrode was maintained, while the capacity of C-LMP decreased remarkably after 200 cycles at the same rate (Fig. S3a). Fig. 7(d) compares the capacity recovery performance of C-LMP/G(-3) and C-LMP at different charge-discharge rates. It was observed that the discharge capacity and cycle performance of two samples are comparable at 0.05C. For higher discharge rate, however, significant differences can be observed: the C-LMP/G(-3) electrode maintained the same performance after 10 cycles at 5C, while CLMP showed a significant decrease. After different rate cycle, as the cycle returned to 0.05C, the discharge capacity of the C-LMP/G(-3) electrode was comparable to the initial value (160.8 vs 150.8 mAh g1), while for C-LMP a significant decrease was observed (161.0 vs.
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
279
Fig. 6. (a) Discharge curves of C-LMP and C-LMP/G (with different amounts of GO) for the first cycle at 0.05C; (a-2) Comparison of discharge performance of C-LMP and CLMP/G(-3) at 0.05C; Discharge curves of (b) C-LMP, (c) C-LMP/G(-1) and (d) C-LMP/G(-3) for the first cycle at different rates of 0.05C-1.0C.
81 mAh g1). Fig. 7(e,f) indicate the cyclic performance of C-LMP/G (-3) at 1C and 5C rate under 55 C. It can be seen that at 5C rate, the capacity retention still maintain over 82.0% after 800 cycles (Fig. S3b); at 1C rate, the capacity retention can maintain over 93.0% after 250 cycles. The results demonstrate the good cyclic stability even at high temperature. Excellent discharge capacity and cycle performance of the electrode of C-LMP/G(-3) at high rate are due to its special cocoons structure, package and fusion between LMP particle, RGO, and amorphous carbon. On the one hand, the conduction capacity of lithium ions and electrons has been effectively increased, with the inherent resistance reduced; on the other hand, the change of structure and volume itself can be effectively buffered in the process of the lithium ion disembedding, especially in the condition of charge and discharge at high rate.
To further illustrate the lithium ion diffusion ability and inherent action resistance of cocoon-like C-LMP/G(-3), a comparison was performed between the cyclic voltammograms and the electrochemical impedance of C-LMP/G(-3) and of C-LMP. Fig. 8(a) shows the CV curves of nanocrystalline C-LMP and C-LMP/G(-3) at 0.2 mV s1. In both curves a redox peak was observed in the vicinity of 4.1 V for the two samples, which corresponds to the inherent Mn(II)$ Mn (III) oxidation reduction potential for LMnPO4 (4.1 V vs Li/Li+). The redox peaks of C-LMP/G (-3) were observed at 4.25 V/4.05 V. The difference of potential between the oxidation and the reduction peaks is about 0.2 V for sample C-LMP/G(-3), while for sample C-LMP is about 0.5 V. As it can be seen, the potential difference of C-LMP/G(-3) is clearly less than that of C-LMP, and it is close to the equilibrium electrode potential (4.1 V). The redox peak current (Ip) of C-LMP/G(-3) is
Table 2 Specific discharge capacity of the samples at different rates. sample
0.05C /mAh g1
0.1C /mAh g1
0.2C/mAh g1
0.5C/mAh g1
1.0C/mAh g1
C-LMP C-LMP/G(-1) C-LMP/G(-3) C-LMP/G(-5) C-LMP/G(-7)
161 162.5 160.8 131.1 102.7
137.7 140.1 142.6 – –
110.7 114.6 118.8 – –
82.6 88 109 – –
67.8 76 99.6 – –
–: No data obtained.
280
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
Fig. 7. Discharge curves of (a) C-LMP, (b) C-LMP/G(-3) for the first cycle at different rates of 2C, 5C, 10C; (c) Comparison of cycling performance of the C-LMP and C-LMP/G(-3) electrodes at 5C; (d) Comparison of reversible capacities of the C-LMP and C-LMP/G(-3) electrodes at charge/discharge rates of 0.05C to 5C; (e) cyclic performance of C-LMP/G (-3) at 1C and 5C rate under 55 C; (f) 55 C, charge/discharge curves of C-LMP/G(-3) at 1C rate under 55 C.
larger than that of C-LMP. The smaller redox potential difference and the larger redox peak current (Ip) indicate that the resistance of electrode material is very small, the overpotential coming from polarization is small too, and that electrode reaction has highly reversible [36]
Fig. 8(b) and (c) show the CV curves of nanocrystalline C-LMP and C-LMP/G(-3) at different scan rates of 0.2, 0.5, 0.8 and 1.0 mV s1. Considering the peak current under different scan rates, the diffusion rate of lithium ion in the electrode can be evaluated [37]. Fig. 8(d) shows the graph of the cathodic peak current (Ip) as a
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
281
Fig. 8. (a) CV curves of the C-LMP and C-LMP/G(-3) electrodes at a scan rate of 0.2 mV s1; CV curves of nanocrystalline (b) C-LMP and (c) C-LMP/G(-3) at different scan rates of 0.2, 0.5, 0.8 and 1.0 mV s1; (d) Relation between the cathodic peak current (Ip) and the square root of the scan rate (n1/2) for both C-LMP and C-LMP/G(-3) electrodes.
function of the square root of the scan rate (n1/2) for C-LMP and CLMP/G(-3). Both samples showed a good linear relationship. Assuming that the electrochemical reaction process of material is affected by the diffusion coefficient of the lithium ions, the correlation between Ip and n1/2 can be used to determine the diffusion coefficient DLi according to the Randles-Sevcik equation [37]: Ip = 2.69 105 n3/2 A DLi1/2 C0 n1/2 (25 C). In the formula, Ip is the peak current, n is number of the transferred electrons of the reaction(n 1 for LiMnPO4), A is the surface area of the electrode (1.13 cm2), DLi is the diffusion coefficient of Li+, C0 is the concentration of Li+ in the electrode and n is the scan rate. Applying the Randles-Sevcik equation, the diffusion coefficients (DLi) of the C-LMP and C-LMP/G(-3) electrodes are calculated to be 3.2 1011cm2.s1 and 6.9 1011cm2.s1, respectively. It can be seen that the diffusion coefficient of cocoon-like C-LMP/G(-3) is 2 times higher than that of C-LMP. For LiMnPO4, the main problem is that the lithium ion has poor electron conductivity, resulting in poor electrochemical property, especially for the high voltage platform and the high rate performances. Therefore, an increase of the diffusion coefficient is very important to improve the overall electrochemical performance of LiMnPO4. Ac impedance spectroscopy is a common experimental method to analyze the electrochemical resistance of the electrode at the solid/liquid interface; information on ion migration resistance can also be obtained. To further understand the excellent electrochemical performance of the C-LMP/G(-3) sample, the electrochemical impedance measurements of the two samples were
carried out at 4.0 V discharge state. Fig.S4 displays the EIS curves of the two samples and the corresponding equivalent circuit model. It can be seen that the AC impedance curve has a semicircle shape for high frequencies while it shows a slope in the low frequency area; the high frequency semicircle has an intercept at the Z' axis, with the intercept reflecting the solution resistance, Re. This semicircle corresponds to the lithium ion migration resistance through the SEI film (Rsf) and the lithium ion transfer resistance Cct. The slope in the low frequency range represents the Warburg impedance (Zw), caused by the diffusion of ions in the electrolyte [38,39], related to lithium ion diffusion coefficient. According to Fig.S4 and Table S3(See Supplementary materials), Rsf and Rct of C-LMP/G(-3) are 96.6 V and 0.61 V, respectively, much smaller than that of C-LMP(Rsf and Rct are 188.3 V and 1.25 V, respectively). This further proved that cocoon-like C-LMP/ G(-3) sample wrapped up by graphene and carbon layers as the electrode material do have a very small resistance to ion migration resistance and charge reaction. This is also the main reason why cocoon-like C-LMP/G(-3) composite material has good high voltage and high rate performance. In addition, the diffusion coefficient obtained from Fig. S4, and the results of 2.9 1011cm2/ s and 6.6 1011cm2/s for C-LMP/G(-3) and C-LMP are agreement with the results from CV test. 4. Conclusions Using manganese acetate, lithium hydroxide and phosphoric acid as raw materials, adding at the same time a small amount of glucose and graphene oxide in DMSO-H2O system, pure phase
282
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
LiMnPO4/RGO nanoparticles with a good crystallinity were directly synthesized, via a one-step liquid-phase method under ambient pressure and at 108 C. The content of GO has a great influence to the particle size and the morphology of LiMnPO4/RGO. With 1 wt.% and 3 wt.% GO addition, the particle size of obtained LiMnPO4/RGO decreased significantly; moreover, the surface of the particles was completely wrapped by soft graphene fold layer and amorphous carbon, liking a cocoon. With 5 wt.% and 7 wt.% GO addition, on the other hand, the generated particles were like larger diamond block. Meanwhile, RGO present state of aggregation. Electrochemical tests indicate cocoon-like C-LiMnPO4/G composite material not only has high discharge specific capacity, but also shows improved high voltage platform and cycle performances at high rate. When the addition of GO is 3 wt.%, the discharge specific capacity of the obtained nanomaterial C-LiMnPO4/G(-3) is 160.86 mAh g1. The discharge specific capacity is up to 115 mAh g1 at the high voltage range ( > 4.0 V), accounting for 70% of the total specific capacity. When the charge and discharge rate was further increased to 1C, 2C, 5C, 10C, C-LMP/G (-3) still showed excellent electrochemical performance. The discharge capacity was 99.6 mAh g1 at 1C. At 5C, after 1000 cycles, 83.0% of the discharge capacity of the C-LMP/ G(-3) electrode was maintained. Cyclic voltammograms and electrochemical impedance spectroscopy analysis showed that cocoon-like C-LiMnPO4/G composite material had a larger lithium ion diffusion coefficient, smaller ion migration and charge reaction resistances. Acknowledgments This work was financially supported by the National Natural Science Foundation of China (21303042, 51672071), the Program for Innovative Research Team in University of Henan Province (No. 14IRTSTHN005). 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. electacta.2016.12.161. References [1] A.K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, J. Electrochem. Soc. 144 (1997) 1188–1194. [2] L. Dimesso, C. Forster, W. Jaegermann, J.P. Khanderi, H. Tempel, A. Popp, J. Engstler, J.J. Schneider, A. Sarapulova, D. Mikhailova, L.A. Schmitt, S. Oswald, H. Ehrenberg, Chem. Soc. Rev. 41 (2012) 5068–5080.
[3] D. Choi, J. Xiao, Y.J. Choi, J.S. Hardy, M. Vijayakumar, M.S. Bhuvaneswari, J. Liu, W. Xu, W. Wang, Z.G. Yang, G.L. Graff, J.G. Zhang, Energ. Environ. Sci. 4 (2011) 4560–4566. [4] Y. Dong, Y.M. Zhao, H. Duan, Z.Y. Liang, Electrochim. Acta 132 (2014) 244–250. [5] V. Ramar, P. Balaya, Phys. Chem. Chem. Phys. 15 (2013) 17240–17249. [6] A. Yamada, M. Hosoya, S.C. Chung, Y. Kudo, K. Hinokuma, K.Y. Liu, Y. Nishi, J. Power Sources 119–121 (2003) 232–238. [7] Z.X. Nie, C.Y. Ouyang, J.Z. Chen, Z.Y. Zhong, Y.L. Du, D.S. Liu, S.Q. Shi, M.S. Lei, Solid State Commun. 150 (2010) 40–44. [8] D.G. Kellerman, Y.G. Chukalkin, N.I. Medvedeva, M.V. Kuznetsov, N.A. Mukhina, A.S. Semenova, V.S. Gorshkov, Mater. Chem. Phys. 149–150 (2015) 209–215. [9] C. Sronsri, P. Noisong, C. Danvirutai, Spectrochim. Acta. A Mol. Biomol. Spectrosc 153 (2016) 436–444. [10] Y. Gan, C. Chen, J. Liu, P. Bian, H. Hao, A. Yu, J. Alloy. Compd. 620 (2015) 350–357. [11] H.H. Yi, C.L. Hu, X.M. He, H.Y. Xu, Ionics 21 (2015) 667–671. [12] P.R. Kumar, M. Venkateswarlu, M. Misra, A.K. Mohanty, N. Satyanarayana, Ionics 19 (2013) 461–469. [13] Q. Lu, G.S. Hutchings, Y. Zhou, H.L. Xin, H.M. Zheng, F. Jiao, J. Mater. Chem. A 2 (2014) 6368–6373. [14] H.M. Ji, G. Yang, H. Ni, S. Roy, J. Pinto, X.F. Jiang, Electrochim. Acta 56 (2011) 3093–3100. [15] H.S. Fang, H.H. Yi, C.L. Hu, B. Yang, Y.C. Yao, W.H. Ma, Y.N. Dai, Electrochim. Acta 71 (2012) 266–269. [16] P. Nie, L.F. Shen, F. Zhang, L. Chen, H.F. Deng, X.G. Zhang, CrystEngComm 14 (2012) 4284–4288. [17] J.F. Ni, Y. Kawabe, M. Morishita, M. Watada, T. Sakai, Power Sources 196 (2011) 8104–8109. [18] L.F. Zhang, Q.T. Qu, L. Zhang, J. Li, H.H. Zheng, J. Mater. Chem. A 2 (2014) 711–719. [19] L.Q. Kou, F.J. Chen, F. Tao, Y. Dong, L. Chen, Electrochim Acta 173 (2015) 721– 727. [20] F.J. Chen, F. Tao, C.M. Wang, W.L. Zhang, L. Chen, J. Power Sources 285 (2015) 367–373. [21] X.M.Liu H.J.Zhu, X.D.Shen H.Yang, Ceram. Int. 40 (2014) 6699–6704. [22] E.R. Dai, H.S. Fang, B. Yang, W.H. Ma, Y.N. Dai, Ceram. Int. 41 (2015) 8171–8176. [23] H. Guo, C.Y. Wu, J. Xie, S.C. Zhang, G.S. Cao, X.B. Zhao, J. Mater. Chem. A 2 (2014) 10581–10588. [24] C. Su, X. Bu, L.H. Xu, J.L. Liu, C. Zhang, Electrochim. Acta 64 (2012) 190–195. [25] L. Wang, H.B. Wang, Z.H. Liu, C. Xiao, S.M. Dong, P.X. Han, Z.Y. Zhang, X.Y. Zhang, C.F. Bi, G.L. Cui, Solid State Ionics 181 (2010) 1685–1689. [26] L.B. Chen, M. Zhang, W.F. Wei, J. Nanomater. 2013 (2013) 1–8. [27] Y. Jiang, R.Z. Liu, W.W. Xu, Z. Jiao, M.H. Wu, Y.L. Chu, L. Su, H. Cao, M. Hou, B. Zhao, J. Mater. Res. 28 (2013) 2584–2589. [28] J. Zong, X.J. Liu, Electrochim. Acta 116 (2014) 9–18. [29] Z.R. Chang, Y. Liu, H.W. Tang, X.Z. Yuan, H.J. Wang, Electrochem. Solid-State Lett. 14 (2011) A90–A92. [30] Z.R. Chang, H.W. Tang, Y. Liu, X.Z. Yuan, H.J. Wang, S.Y. Gao, J. Electrochem. Soc. 159 (2012) A331–A335. [31] X.N. Fu, Z.R. Chang, K. Chang, B. Li, H.W. Tang, E.B. Shangguan, X.Z. Yuan, H.J. Wang, Electrochim. Acta 178 (2015) 420–428. [32] W.S. Hummers, R.E. Offeman, J. Am. Chem. Soc. 80 (1958) 1339. [33] M. Hu, X.L. Pang, Z. Zhou, J. Power Sources 237 (2013) 229–242. [34] D.Y. Wang, H. Buqa, M. Crouzet, G. Deghenghi, T. Drezen, I. Exnar, N.H. Kwon, J. H. Miners, L. Poletto, M. Grätzel, J. Power Sources 189 (2009) 624–628. [35] D. Choi, D.H. Wang, I.T. Bae, J. Xiao, Z.M. Nie, W. Wang, V.V. Viswanathan, Y.J. Lee, J.G. Zhang, G.L. Graff, Z.G. Yang, J. Liu, Nano Lett. 10 (2010) 2799–2805. [36] J.F. Ni, H.H. Zhou, J.T. Chen, X.X. Zhang, Mater. Lett. 59 (2005) 2361–2365. [37] R.S. Nicholson, Anal. Chem. 37 (1965) 1351–1355. [38] M. Mohamedi, D. Takahashi, T. Uchiyama, T. Itoh, M. Nishizawa, I. Uchida, J. Power Sources 93 (2001) 93–103. [39] W.K. Kim, D.W. Han, W.H. Ryu, S.J. Lim, H.S. Kwon, Electrochim. Acta 71 (2012) 17–21.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Low-temperature synthesis of LiMnPO4/RGO cathode material with excellent voltage platform and cycle performance Xiaoning Fu, Kun Chang* , Bao Li, Hongwei Tang, Enbo Shangguan, Zhaorong Chang* Collaborative Innovation Center of Henan Province for Green Manufacturing of Fine Chemicals, Key Laboratory of Green Chemical Media and Reactions, Ministry of Education, School of Chemistry and Chemical Engineering, Henan Normal University, Xinxiang 453007, PR China
A R T I C L E I N F O
Article history: Received 24 October 2016 Received in revised form 16 December 2016 Accepted 25 December 2016 Available online 26 December 2016 Keywords: Lithium ion battery lithium manganese phosphate low temperature synthesis high voltage platform graphene oxide
A B S T R A C T
Pure and well-crystallized LiMnPO4/reduced graphene oxide (RGO) nanopowders are synthesized by adding a small amount glucose and graphene oxide simultaneously in dimethyl sulfoxide (DMSO)/H2O, under constant atmospheric pressure and at low-temperature (108 C). The scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images show that the addition of different amounts of graphene oxide can significantly affect the particle size and morphology of LiMnPO4/RGO composites. With small amounts of graphene oxide (1 and 3 wt.%), the small LiMnPO4 particles are wrapped in RGO, in a cocoon-like structure. This special morphology can be maintained after a rapid carbon coating treatment at high temperature. Electrochemical studies show that these cocoonlike C-LiMnPO4/G nanocomposites not only have a higher discharge specific capacity, but also show improved high voltage platform and high rate cycle performance. When the added graphene oxide is 3%, the specific capacity of C-LiMnPO4/G nanocomposite is 160.8 mAh g1 at 0.05 C, the discharge capacity in the area of more than 4.0 V is up to 115 mAh g1, accounting for 70% of the total discharge capacity. The proposed C-LiMnPO4/G nanocomposites also exhibit an outstanding high rate capability, where the discharge specific capacity at 1C can reach to 99.6 mAh g1 and after 1000 cycles at 5 C, it still has 83% of capacity retention. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction At present, olivine-type polyanionic compounds such as LiMPO4 (M = Fe, Mn, Ni and Co) have attracted increasing attention as powerful cathode materials for lithium ion batteries, due to their good structure and chemical stability [1,2]. Compared with LiFePO4, LiMnPO4 is a more interesting material because of its redox potential (4.1 V vs. Li/Li+) which shows 0.7 V higher than that of LiFePO4, and is still within the stable electrochemical window of currently used liquid electrolytes. Meanwhile, LiMnPO4 also has a much higher theoretical energy density than LiFePO4, 701 Wh kg1 = 171 mAh g1 4.1 V vs. 586 Wh kg1 = 170 mAh g1 3.45 V; this plays an important role for applications in electric cars. In addition, LiMnPO4 has other important properties; in fact it can be produced using abundant and low cost raw materials, it is environmental friendly, it has good structural stability, wide chemical compatibility, and it is safe to use; because of all this, it is
* Corresponding authors. E-mail addresses: [email protected] (K. Chang), [email protected] (Z. Chang) . http://dx.doi.org/10.1016/j.electacta.2016.12.161 0013-4686/© 2016 Elsevier Ltd. All rights reserved.
considered the most promising cathode material for Li batteries [3–5]. Nevertheless, LiMnPO4 has a very poor electronic conductivity and Li+ diffusion coefficient, which make it an insulator material [6], with poor electrochemical activity [7]. To improve LiFePO4 performance, different methods were developed which can increase the ion diffusion coefficient and the electron conductivity; these include carbon coating on the surface of nanoparticles, bulk phase element doping, nanocrystallization of material particles [8– 13]. These improvements are mostly focused on the enhancement of the discharge capacity at low rates. Less investigation was performed, however, on the performances of the discharge of high voltage platform and of the high rate discharges; consequently some materials have a higher specific capacity at low rate, but the platform performance at high voltage is poor [14–19] .For examples, Chen et al. [20] synthesized CeO2-modified LiMnPO4/ C composites by solvothermal method at 190 C for 24 h, and then calcined at 550 C for 3 h under dynamic argon atmosphere, the discharge capacity of the prepared composites was about 60 mAh g1 above 4.0 V high voltage. Zhu et al. [21] used Mn (CH3COO)24H2O and LiH2PO4 as the raw material, synthesized LiMnPO4 in diethylene glycol solution keeping 3 h in the
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
temperature of 160 C. LiMnPO4/C composites were prepared using planetary ball-milling of the LiMnPO4 powders with 20 wt.% carbon black; they can deliver a discharge capacity of 129 mAh g1 at 0.05 C, but at more than 4.0 V high voltage discharge area, the discharge capacity of the material was very low, about less 30 mAh g1. Dai et al. [22] studied the impact of the rate properties of LMP, when doped with different proportions of vanadium. A series of LiMn1xVxPO4/C (0 < x < 0.075) samples were synthesized via an improved solid-state method. Results showed that the LiMn0.075V0.025PO4/C exhibited the highest discharge capacity of 108 mAh g1at 0.2 C for the first cycle; for over 4.0 V high voltage discharge area, however, the discharge capacity of the material was still too low, about less than 40 mAh g1. Guo et al. [23] synthesized the plate-like LiMnPO4 material by a solvothermal method. After a high temperature carbon treatment, at over 4.0 V high voltage discharge area, the discharge platform of C- LiMnPO4 was relatively good. Coating with too much carbon (weight ratio of LiMnPO4/glucose is 2:1) would greatly reduce the amount of active material and, hence, reduce the volume ratio of energy of the electrode. Meanwhile, the solvent thermal conditions at high temperature and high pressure were quite harsh. As it is well-known, graphene has good electrical conductivity and chemical and mechanical stability; literature showed that it can improve the electrochemical performance of LiMPO4 cathode materials [24–27]. Zong et al. [28] synthesized LiMnPO4-C/ Acetylene black (AB) and LiMnPO4-C/graphenes by solvothermal method. The samples had a capacity of 134 mAh g1 and 139 mAh g1 at 0.05 C, respectively. The retention of discharge capacity maintained 90% and 94% of the initial values after 40 cycles. When the discharge ratio was increased to 0.5 C and 1.0 C, the discharge capacities of LiMnPO4-C/Acetylene black were 92 mAh g1and 60 mAh g1, but that of LiMnPO4-C/graphenes still could be as high as 125 mAh g1and 119 mAh g1. These data confirm that the electric capacity and cycle performance of LiMnPO4 samples improved substantially with the addition of graphene. Despite this, however, the discharge voltage platform was still not ideal; in fact at over 4.0 V high voltage discharge area, the discharge capacity of the material was about less than 40 mAh g1, accounting for only 30% of the total capacity. To take advantage of the 4.1 V vs. Li/Li+ redox potential of LiMnPO4, further improvements in the performance of the materials are needed. These can be achieved by optimizing the size of the nanoparticle and their structural morphology, and by optimizing the preparation of LiMnPO4/ graphene composites. An effective improvement of the electrical conductivity of the particle will enhance the material’s discharge platform, taking full advantage of its high energy density. In our previous work, our team had synthesized sphere-like LiFePO4/C nanocomposites with particle sizes of about 100– 300 nm via a solution-phase method. The synthesized LiFePO4/C sample showed excellent rate and cycle performances [29,30]. Based on our previous works, we added glucose as additive in dimethyl sulfoxide solvent; the glycoside formed during the reaction acted as a surfactant and led to the successful preparation of hollow spindle LiMnPO4 particles [31]. The discharge capacity was 161.8 mAh g1mAh g1 at 0.05 C. To further improve the high voltage platform performance, the high rate performance and the electrical conductivity between the particles, we also added different mass fraction of graphene oxide in the solution, together with the glucose. The results showed that nano LiMnPO4/RGO was synthesized after the reaction. With the graphene oxide ratios of 1%, 3%, the materials have a cocoon-like morphology, with the surface of the particles completely wrapped up by soft graphene fold layer. Nano C-LiMnPO4/G composite was then produced after a rapid high temperature carbon treatment. This kind of composite with special morphological structure had excellent conductive network
273
structure, produced by the close contact of LiMnPO4 particles with amorphous carbon and graphene layers; this will greatly improve the high voltage platform performance and the high rate cycle performance of the material. This proposed method is based on a simple and facile procedure, using cheap raw materials and under mild reaction conditions (no high temperatures or pressure need to be employed). This method can also be applied in the preparation of other olivine structure for lithium ion battery materials, it has therefore great application prospects. 2. Experimental section 2.1. Synthesis of graphene oxide Graphene oxide (GO) was prepared by an improved Hummers method [32]: 1.0 g graphite were added to 100 ml concentrated H2SO4 (98%). 4.5 g of KMnO4 and 0.5 g of NaNO3 were slowly added into the above solution, which was magnetically stirred in an ice bath for 10 min. The mixture continued to be stirred for 3 days at room temperature. Subsequently, water was added dropwise to the mixture under vigorous stirring, and the stirring was continued until the solution was cooled to room temperature. 15 ml of H2O2 solution (30 wt.%) and 200 ml of water were added dropwise under vigorous stirring for another 2 h. Then, the resulting graphite oxide suspension was washed 3 times with HCl (0.5 M), and washed repeatedly with distilled water, until the solution pH keep a constant value of about 5.0. The sol was scattered to H2O again. After ultrasound treatment, the complete delamination of graphite oxide was achieved. 2.2. Synthesis of samples The synthesizing procedure of sample LiMnPO4/RGO is shown in Fig.S1 (See Supplementary materials); the process is based on our previous work to prepare LiMnPO4 [31]. Deionized water and DMSO (VH2O = VDMSO = 25 ml) were mixed in a three neck round bottom flask. 5% glucose was slowly added to the above solution. The DMSO-H2O solution with glucose was vigorously stirred and heated to 108 C for 2 h, then cooled to room temperature. This is named Solution A. Solution B was obtained by dissolving 0.05 mol of manganese acetate tetrahydrate (Mn(CH3COO)24H2O, 99%) and 3.42 ml of phosphoric acid (H3PO4, 85%) into the solution A under constant stirring, for 1 h at 50 C. Then, 6.3545 g of lithium hydroxide (LiOHH2O, 99%) were slowly added into the Solution B. The final concentration of lithium hydroxide was 1.5 mol L1. The reaction solution was heated at 108 C for 2 h under nitrogen atmosphere, cooled to room temperature, and stored overnight. The LiMnPO4 material (named LMP) was obtained by precipitation; the precipitate was washed several times with deionized water and ethanol and dried overnight in an oven at 60 C. To facilitate the measurement and guarantee the repeatability of experiments, experiments according to the theoretical production of LiMnPO4 were performed adding different amounts of GO to the reaction system; more specifically, solution with GO concentrations of 1 wt.%, 3 wt.%, 5 wt.% and 7 wt.% were prepared. The corresponding samples obtained by liquid phase reaction were denoted as LMP/RGO (1 wt.%), LMP/RGO (3 wt.%), LMP/RGO (5 wt. %), and LMP/RGO (7 wt.%). The C-LiMnPO4/G was synthesized via the following steps; samples LMP, LMP/RGO(1 wt.%), LMP/RGO(3 wt.%), LMP/RGO(5 wt. %) and LMP/RGO(7 wt.%) were milled with 10% of glucose. The milled powder was collected in boat shape alumina crucible and then heated at 650 C for 4 h in an atmosphere of (N2:H2 = 95:5) in a tube furnace, and then cooled to room temperature. The resulting product was named as C-LMP, C-LMP/G(-1), C-LMP/G(-3), C-LMP/G
274
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
(-5), C-LMP/G(-7). According to the TGA analysis in Fig. S2, it can be seen that the carbon weight loss of sample C-LMP/G(-3), C-LMP, and LMP/RGO (3 wt.%) is 4.8%, 3.4%, and 1.4%, respectively. It is therefore calculated that the carbon content is 0.47% for LMP/RGO (1 wt.%), 2.35% for LMP/RGO (5 wt.%), and 3.29% for LMP/RGO (7 wt. %); after coating carbon, the carbon content is 3.87% for C-LMP/ RGO (-1), 5.75% for C-LMP/RGO (-5), and 6.69% for C-LMP/RGO (-7), respectively. 2.3. Characterization of samples Phase analysis and cell parameter determination of all samples were performed by X-ray diffraction (XRD) using a D8 X diffractometer (Germany, Bruker) with Cu-Ka radiation. The particle morphology of all samples was analyzed by the Scanning electron microscopy (SEM, S4800, Japan) and transmission electron microscopy (TEM, JEM-2010, JEOL). Raman spectra of CLFP/G cathodes were recorded on a Raman spectrometer (HR800) using laser excitation energy of 532 nm. 2.4. Electrochemical measurements The electrodes were prepared using the following procedure: the LMP-based sample, acetylene black and polyvinylidene fluoride (PVDF) were mixed at a mass ratio of 7:2:1. The suspensions were prepared after adding a certain amount of
analytical grade N-methyl-2-pyrrolidinone (NMP) (Sinopharm Chemical Reagent Co., Ltd.) to the mixture, and grinding for 60 min with a pulp refiner. Then, the suspensions were uniformly coated on an aluminum foil with an applicator. The solvents were then evaporated by drying the fabricated electrodes at 80 C overnight in a vacuum oven. The dried electrodes were pressed using a twin roller to ensure a close contact between the electrode materials and the current collector. The pressed electrodes were cut into circular discs of 12 mm diameter and then dried again at 120 C overnight in a vacuum oven. The LMP-based material loading on the electrode was about 2.5 mg. For half cells, lithium metal was used as negative electrode, polypropylene microporous membrane as the separator, while 1 mol L1 LiPF6 (EC: DEC: DMC is 1: 1: 1, V/V/V) was used as the electrolyte. The coin cells (CR2025) were assembled in an Ar-filled box (MBraun, Germany). The cells were tested using a land CT2001A battery tester (Wuhan Jinnuo Electronics Co. Ltd., China) at room temperature, at rates of 0.05 C– 10 C. In the electrochemical stability experiments, the cells were cycled between 2.50 V and 4.50 V vs. Li/Li+. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were performed using an electrochemical work-station (Shanghai Chenhua Instrument Company, China). The CV test scan rates were 0.2 mV s1, 0.5 mV s1, 0.8 mV s1 and 1.0 mV s1 over a cell voltage from 2.5 V to 4.5 V. EIS was performed over a frequency range of 0.05 Hz–10 kHz.
Fig. 1. (a, b) XRD patterns of LiMnPO4/RGO and C-LMP/G; (c, d) XRD spectrums of GO before and after reaction.
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
3. Results and discussion 3.1. Structure and morphology characterization Fig. 1(a, b) shows the XRD patterns of the LMP/RGO and C-LMP/ G materials obtained by adding different content of graphene oxide. Comparing with the standard LiMnPO4 pattern (PDF reference number 33-0804), it can be seen that all samples are well crystallized, regardless of the graphene oxide content. All diffraction peaks could readily be indexed to the pure phase of LiMnPO4 with an orthorhombic olivine-type structure, without other impurity peaks. These data confirmed that the pure phase LiMnPO4 can be directly prepared using the liquid phase method; the addition of graphene oxide does not change the crystal structure of the sample When a certain amount of glucose is blended with samples, followed by a rapid high temperature carbon treatment, the obtained samples C-LMP/G(-1), C-LMP/G(-3), C-LMP/G(-5), C-LMP/ G(-7) still showed to be pure phases, with an orthorhombic olivinetype structure. As no other impurity peaks were detected, it is likely that carbon originated from the glucose thermolysis in the composite is amorphous.
275
The crystal cell parameters, crystal cell volume, the grain size, FWHM of samples were calculated using the Scherrer equation, and they listed in Table S1(See Supplementary materials). It can be seen that the cell volumes of C-LMP/G(-1), C-LMP/G(-3), C-LMP/G (-5), C-LMP/G(-7) were larger than that of C-LMP. It can be seen that the introduction of graphene oxide could increase the cell volume of LiMnPO4 particles, which is beneficial for the deintercalation of lithium ions. [33]. From the values of FWHM of the (111), (020) and (311) crystal faces it can be seen that, with the increase of graphene oxide, the FWHM of the characteristic peaks first decreased and then increased again; for the grain size, on the contrary, the opposite behavior was observed, i.e. an increase followed by a decrease. The FWHM of the main crystal plane diffraction peaks (111, 020 and 311) reached a minimum for sample C-LMP/G(-3), while the corresponding grain size is the largest. This not only shows that CLMP/G (-3) had the highest crystallinity, but also that the grain orientation growth was preferential along the vertical crystal direction, especially along the vertical direction of the (020) crystal face. This lattice orientation can favor the deintercalation of lithium ions, beneficial to the improvement of the performance [34,35]. To examine the possible changes of the graphene oxide due to the reaction in liquid phase, we took a certain quantity of graphene
Fig. 2. SEM micrographs of (a) LMP; (b) LMP/GO (1 wt.%); (c) LMP/GO (3 wt.%); (d) LMP/GO (5 wt.%);(e) LMP/GO (7 wt.%); (f) C-LMP/G(-1); (g) C-LMP/G(-3).
276
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
oxide, and made it react for 2 hours in the same liquid phase system. XRD spectrums of GO before and after the reaction are shown in Fig. 1(c, d). It can be seen that after the liquid phase reaction, the characteristic peak of graphene oxide for 2u = 11 disappeared; different strong diffraction peaks, however, appeared for 2u = 26 , 43 and 46 , which correspond to the characteristic peaks of graphite (PDF#26-1079). This shows that graphene oxide can be reduced to RGO by glucose in DMSO-glucose-GO-H2O reaction system. Therefore, in the above reaction medium, GO was gradually reduced to RGO and wrapped on the surface of LiMnPO4 particles, generating LMP/RGO composites with olivine phase. 3.2. Study of the morphology Fig. 2 shows SEM images of LMP/RGO samples with different content of graphene oxide and of samples C-LMP/G(-1), C-LMP/G (-3). It can be seen that the different amounts of graphene oxide affected both the particle size and the morphology of the material. As seen in Fig. 2(a), the morphology of the LMP sample without graphene oxide is spindle-like, with a mean length of 700–900 nm and width of 200–300 nm. The particle size distribution was very uniform, with no agglomeration occurring. Under the lower contents of graphene oxide (1 wt.% and 3 wt.%), the LMP particles
maintained the spindle morphology and wrapped with soft graphene oxide fold layers. During the stirring solution phased system, the composites resembled to form a silkworm cocoon-like structure, where the graphene layers wrapped LMP particles with 300–500 nm, as show in Fig. 2(b, c-1, c-2). During the formation process of the final products, the Mn2+ ions can be adsorbed on the surface of graphene oxide due to the containing large amount of oxygen functional groups, and then reacted with Li+ ions and PO43 ions in the solution. The LiMnPO4 nanocrystals were in-situ anchored on the surface of graphene oxide layers and finally grown into nanoparticles. Due to the less amount of GO, the particle morphology of LMP/RGO (1 wt. %, 3 wt.%) sample did not change but maintained a spindle shape, with the particles just becoming smaller. The surface of the particles is wrapped up by reduced graphene oxide layer, it is similar to cocoons. Fig. 2(d-1–d-3) are SEM Fig. of LMP/RGO(5 wt.%) and at different magnifications. As it can be seen, when the amount of graphene oxide was increased to 5 wt.%, the particle of LMP/RGO(5 wt.%) has a clear change in morphology, gradually becoming diamond cube-like, and it is mixed with RGO. In the images with higher magnification of Fig. 2(d-2 and d-3), it can be seen clearly that, for the diamond cubic particles, the local surface is smooth, and clearly not wrapped by RGO. This phenomenon shows that when the amount of graphene oxide is
Fig. 3. The TEM of (a) LMP/RGO (1 wt.%); (b) LMP/RGO(3 wt.%); (c) LMP/RGO(5 wt.%) and (a-1) C-LMP/G(-1); (b-1) C-LMP/G(-3); (c-1) C-LMP/G(-5) and) (a-2, b-2, c-2 the local amplification Fig. 3(a-1, b-1, c-1).
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
more than 5 wt.%, the ability of GO to wrap around crystal grain decreases. According to Fig. 2(e), when the amount of graphene oxide is to 7 wt.% (sample LMP/GO (7 wt.%)) the particles gradually became a larger rhombus block structure, with a very smooth surface. It also can be seen that the RGO aggregates together, not package crystal particles. These images show that when the amount of GO is to 7 wt.%, the wrapping ability of LiMnPO4 grains almost disappears. With the increase of GO content, the solution became acidic because of many oxygen functional group in GO. Thus, with the increase of GO content, the pH value in reaction system would be decreased, leading to the aggregation the GO, which decreased the ability of coating LiMnPO4 particles. In addition, when GO content increased, partial of added glucose as reductant and surfactant would take part into the GO reduction reaction, decreasing the effect of its surfactant. The interfacial tension of the LiMnPO4 particles would also affected during the growing process, resulting in the final structure and morphology. Based on these results, samples LMP/RGO(1 wt.%) and LMP/RGO (3 wt.%) underwent a short high temperature carbon treatment; the morphology of the obtained samples C-LMP/G(-1) and C-LMP/ G(-3) is shown in Fig. 2(f, g). The images show that the morphology of the composites did not change significantly; indeed particles still maintained the original morphology of silkworm cocoon. The size of the particles also did not show significant changes. The close contact of LiMnPO4 particles with amorphous carbon and the reduced graphene layers formed an excellent conductive network, which led to the deintercalation of lithium ions and enhanced electron transportation. Fig. 3 shows the TEM images of LMP/RGO (1 wt.%), LMP/RGO (3 wt.%), LMP/RGO (5 wt.%) before and after high temperature carbon treatment. It can be seen that LMP/RGO (1 wt.%) and LMP/ RGO (3 wt.%) have a spindle crystal structure, wrapped up by RGO, similar to the shape of silkworm cocoon. After a rapid high
277
temperature carbon treatment, the composites maintain the original morphology of silkworm cocoon. Graphene and amorphous carbon form a dense cladding layer on the surface of LiMnPO4 particles. The carbon layer thickness is about 10 nm. LMP/RGO (5 wt.%) is like a diamond cube with a mean length of 600–800 nm. Some RGO layers are distributed around the LMP/ RGO (5 wt.%) particles. The originally smooth surface was coated by the amorphous carbon after a carbon treatment, the carbon layer thickness being approximate 10 nm. Obviously, the larger size diamond cube structure is not beneficial for the deintercalation of lithium ions. Although there is carbon layer formed on the surface of particle after a carbon coating treatment, large particle size and without support by RGO would extremely decrease the electronic conductivity of the LMP particles. Fig. 4 represents a scheme of the morphology change of LMP according to the different content of graphene oxide. It can be seen clearly that graphene oxide has a great influence on grain growth and morphology of LMP samples prepared with liquid phase system. The cocoon-like structure has excellent ion and electron conduction ability. 3.3. Raman spectroscopy To further identify the degree of graphitization of carbon in the samples, Raman spectra were acquired; they are shown in Fig. 5. It can be seen that sample LMP, with no GO, did not show any of the graphite characteristic peaks; this indicated that the glucose added in the liquid phase synthesis system had not been carbonized and was not loaded on the surface of LMP sample particle, but it only acted as a surfactant. The sample C-LMP, which underwent carbon coating treatment at 650 C, shows two strong peaks at 1350 cm1 and 1600 cm1, corresponding to the defect mode of the disordered carbon
Fig. 4. A schematic diagram of the reaction process and the changes of the morphology.
278
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
Fig. 5. Raman spectra of LMP, C-LMP, LMP/RGO (3 wt.%), C-LMP/G(-3).
(D-band) and the E2g vibration mode of the ordered graphitic carbon (G-band). The ID/IG value is 0.803. This may be because some glucose pyrolysis formed small, flat shaped, irregular arrangement of two dimensional carbon materials. The sample LMP/RGO (3 wt.%) with added GO have distinct peaks corresponding to D-band and G-band; the ID/IG values is 0.872. These data indicated that the added graphene oxide can be reduced to graphene by glucose in GO/Glucose/DMSO/H2O system, and that, combined with LMP particles, can form LMP/RGO. After this sample was coated with carbon at 650 C, its ID/IG values decreased to 0.806. The increase of graphitization indicates that the amorphous carbon coming from the glucose pyrolysis further enhanced the GO reduction during the high temperature carbon treatment processing. The ID/IG values of samples LMP, C-LMP, LMP/GO (3 wt.%), C-LMP/G(-3) are summarized in Table 1. 3.4. Electrochemical properties Fig. 6 shows the discharge curves of the samples with different contents of graphene, at different rates of 0.05C- 1C. It can be seen the samples have a discharge voltage platform around 4.1 V vs. Li/ Li+, which is typical for the Mn(II) $ Mn(III) redox for olivine manganese phosphates. The specific discharge capacities of the different samples are listed in Table 2. Fig. 6(a) shows the discharge curves of the samples at 0.05C. According to the data in Table 2, the first discharge capacities of CLMP, C-LMP/G(-1)and C-LMP/G(-3) were 161 mAh g1,162.5 mAh g1,160.8 mAh g1, respectively, while for samples C-LMP/G(-5) and C-LMP/G(-7) much lower values were measured 131.1 mAh g1 and 102.7 mAh g1. It can be seen that the first three samples all have a discharge capacity higher than 160 mAhg1. Fig. 6a indicates that three samples exhibit the different high-voltage performance at 4.0 V. The voltage of the discharge platform for C-LMP/G(-1) and C-LMP/G(-3) samples are higher than that of C-LMP sample without GO. The discharge capacities of these two samples are also significantly higher than that of C-LMP above 4.0 V high voltage discharge. Table S2 listed the electrochemical performance of LMP
Table 1 ID/IG values of different materials. Sample
LMP
C-LMP
LMP/GO (3 wt.%)
C-LMP/G(-3)
ID/IG
0
0.803
0.872
0.806
reported in recent years, it can be seen that our proposed LMP/RGO electrode exhibits the best high-voltage properties. Fig. 6(a-2) gives a comparison of the discharge capacity above 4.0 V high voltage discharge area between C-LMP/G(-3) and C-LMP electrodes. Obviously, the discharge capacity of C-LMP/G(-3) is 115 mAh g1 which is higher than that of C-LMP with 82 mAh g1. This fact implies that adding a small amount of graphene oxides can significantly improve the discharge performance of the samples, in particular at high voltage. The main reason may be that the presence of small amount of GO give C-LMP/G particles a cocoon-like structure. So this morphology has excellent ion and electronic conductivity, and it effectively reduces the polarization in the process of charging and discharging; making it close to the inherent Mn(II)$ Mn (III) oxidation reduction potential (4.1 v vs Li/ Li +) for LMP. Fig. 6(b–d) show, respectively, the discharge curves of samples C-LMP, C-LMP/G(-1), C-LMP/G(-3) at different rates (0.05C-1C). Comparing them, the capacitance of the three materials are very close at 0.05C-0.2C; for higher discharge ratios, however (from 0.5C to 1.0C), the specific capacity of C-LMP/G(-3) is higher than those of LMP/G and C-LMP/G(-1). This is due to the electrode polarization; in fact when the charge and discharge current is very small, the resistance of the electrode materials (i.e. inherent ohmic voltage drop) is also small. For higher charge and discharge currents, the inherent ohmic voltage drop of the electrodes will become larger. Obviously, the intrinsic resistance of C-LMP/G (-3) sample is smallest, and an addition of 3% of GO led to the material with the best properties. The discharge and cycle performance curves of C-LMP and C- LMP/G (-3) at high rate are shown in It was found that for charging and discharging rate of 2C, 5C and 10C, the specific discharge capacity of C-LMP/G(-3) was clearly higher than for C-LMP. In addition, C-LMP/G(-3) sample showed excellent cycle performance at high rate. At 5C, after 1000 cycles, 83.0% of the discharge capacity of the C-LMP/G(-3) electrode was maintained, while the capacity of C-LMP decreased remarkably after 200 cycles at the same rate (Fig. S3a). Fig. 7(d) compares the capacity recovery performance of C-LMP/G(-3) and C-LMP at different charge-discharge rates. It was observed that the discharge capacity and cycle performance of two samples are comparable at 0.05C. For higher discharge rate, however, significant differences can be observed: the C-LMP/G(-3) electrode maintained the same performance after 10 cycles at 5C, while CLMP showed a significant decrease. After different rate cycle, as the cycle returned to 0.05C, the discharge capacity of the C-LMP/G(-3) electrode was comparable to the initial value (160.8 vs 150.8 mAh g1), while for C-LMP a significant decrease was observed (161.0 vs.
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
279
Fig. 6. (a) Discharge curves of C-LMP and C-LMP/G (with different amounts of GO) for the first cycle at 0.05C; (a-2) Comparison of discharge performance of C-LMP and CLMP/G(-3) at 0.05C; Discharge curves of (b) C-LMP, (c) C-LMP/G(-1) and (d) C-LMP/G(-3) for the first cycle at different rates of 0.05C-1.0C.
81 mAh g1). Fig. 7(e,f) indicate the cyclic performance of C-LMP/G (-3) at 1C and 5C rate under 55 C. It can be seen that at 5C rate, the capacity retention still maintain over 82.0% after 800 cycles (Fig. S3b); at 1C rate, the capacity retention can maintain over 93.0% after 250 cycles. The results demonstrate the good cyclic stability even at high temperature. Excellent discharge capacity and cycle performance of the electrode of C-LMP/G(-3) at high rate are due to its special cocoons structure, package and fusion between LMP particle, RGO, and amorphous carbon. On the one hand, the conduction capacity of lithium ions and electrons has been effectively increased, with the inherent resistance reduced; on the other hand, the change of structure and volume itself can be effectively buffered in the process of the lithium ion disembedding, especially in the condition of charge and discharge at high rate.
To further illustrate the lithium ion diffusion ability and inherent action resistance of cocoon-like C-LMP/G(-3), a comparison was performed between the cyclic voltammograms and the electrochemical impedance of C-LMP/G(-3) and of C-LMP. Fig. 8(a) shows the CV curves of nanocrystalline C-LMP and C-LMP/G(-3) at 0.2 mV s1. In both curves a redox peak was observed in the vicinity of 4.1 V for the two samples, which corresponds to the inherent Mn(II)$ Mn (III) oxidation reduction potential for LMnPO4 (4.1 V vs Li/Li+). The redox peaks of C-LMP/G (-3) were observed at 4.25 V/4.05 V. The difference of potential between the oxidation and the reduction peaks is about 0.2 V for sample C-LMP/G(-3), while for sample C-LMP is about 0.5 V. As it can be seen, the potential difference of C-LMP/G(-3) is clearly less than that of C-LMP, and it is close to the equilibrium electrode potential (4.1 V). The redox peak current (Ip) of C-LMP/G(-3) is
Table 2 Specific discharge capacity of the samples at different rates. sample
0.05C /mAh g1
0.1C /mAh g1
0.2C/mAh g1
0.5C/mAh g1
1.0C/mAh g1
C-LMP C-LMP/G(-1) C-LMP/G(-3) C-LMP/G(-5) C-LMP/G(-7)
161 162.5 160.8 131.1 102.7
137.7 140.1 142.6 – –
110.7 114.6 118.8 – –
82.6 88 109 – –
67.8 76 99.6 – –
–: No data obtained.
280
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
Fig. 7. Discharge curves of (a) C-LMP, (b) C-LMP/G(-3) for the first cycle at different rates of 2C, 5C, 10C; (c) Comparison of cycling performance of the C-LMP and C-LMP/G(-3) electrodes at 5C; (d) Comparison of reversible capacities of the C-LMP and C-LMP/G(-3) electrodes at charge/discharge rates of 0.05C to 5C; (e) cyclic performance of C-LMP/G (-3) at 1C and 5C rate under 55 C; (f) 55 C, charge/discharge curves of C-LMP/G(-3) at 1C rate under 55 C.
larger than that of C-LMP. The smaller redox potential difference and the larger redox peak current (Ip) indicate that the resistance of electrode material is very small, the overpotential coming from polarization is small too, and that electrode reaction has highly reversible [36]
Fig. 8(b) and (c) show the CV curves of nanocrystalline C-LMP and C-LMP/G(-3) at different scan rates of 0.2, 0.5, 0.8 and 1.0 mV s1. Considering the peak current under different scan rates, the diffusion rate of lithium ion in the electrode can be evaluated [37]. Fig. 8(d) shows the graph of the cathodic peak current (Ip) as a
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
281
Fig. 8. (a) CV curves of the C-LMP and C-LMP/G(-3) electrodes at a scan rate of 0.2 mV s1; CV curves of nanocrystalline (b) C-LMP and (c) C-LMP/G(-3) at different scan rates of 0.2, 0.5, 0.8 and 1.0 mV s1; (d) Relation between the cathodic peak current (Ip) and the square root of the scan rate (n1/2) for both C-LMP and C-LMP/G(-3) electrodes.
function of the square root of the scan rate (n1/2) for C-LMP and CLMP/G(-3). Both samples showed a good linear relationship. Assuming that the electrochemical reaction process of material is affected by the diffusion coefficient of the lithium ions, the correlation between Ip and n1/2 can be used to determine the diffusion coefficient DLi according to the Randles-Sevcik equation [37]: Ip = 2.69 105 n3/2 A DLi1/2 C0 n1/2 (25 C). In the formula, Ip is the peak current, n is number of the transferred electrons of the reaction(n 1 for LiMnPO4), A is the surface area of the electrode (1.13 cm2), DLi is the diffusion coefficient of Li+, C0 is the concentration of Li+ in the electrode and n is the scan rate. Applying the Randles-Sevcik equation, the diffusion coefficients (DLi) of the C-LMP and C-LMP/G(-3) electrodes are calculated to be 3.2 1011cm2.s1 and 6.9 1011cm2.s1, respectively. It can be seen that the diffusion coefficient of cocoon-like C-LMP/G(-3) is 2 times higher than that of C-LMP. For LiMnPO4, the main problem is that the lithium ion has poor electron conductivity, resulting in poor electrochemical property, especially for the high voltage platform and the high rate performances. Therefore, an increase of the diffusion coefficient is very important to improve the overall electrochemical performance of LiMnPO4. Ac impedance spectroscopy is a common experimental method to analyze the electrochemical resistance of the electrode at the solid/liquid interface; information on ion migration resistance can also be obtained. To further understand the excellent electrochemical performance of the C-LMP/G(-3) sample, the electrochemical impedance measurements of the two samples were
carried out at 4.0 V discharge state. Fig.S4 displays the EIS curves of the two samples and the corresponding equivalent circuit model. It can be seen that the AC impedance curve has a semicircle shape for high frequencies while it shows a slope in the low frequency area; the high frequency semicircle has an intercept at the Z' axis, with the intercept reflecting the solution resistance, Re. This semicircle corresponds to the lithium ion migration resistance through the SEI film (Rsf) and the lithium ion transfer resistance Cct. The slope in the low frequency range represents the Warburg impedance (Zw), caused by the diffusion of ions in the electrolyte [38,39], related to lithium ion diffusion coefficient. According to Fig.S4 and Table S3(See Supplementary materials), Rsf and Rct of C-LMP/G(-3) are 96.6 V and 0.61 V, respectively, much smaller than that of C-LMP(Rsf and Rct are 188.3 V and 1.25 V, respectively). This further proved that cocoon-like C-LMP/ G(-3) sample wrapped up by graphene and carbon layers as the electrode material do have a very small resistance to ion migration resistance and charge reaction. This is also the main reason why cocoon-like C-LMP/G(-3) composite material has good high voltage and high rate performance. In addition, the diffusion coefficient obtained from Fig. S4, and the results of 2.9 1011cm2/ s and 6.6 1011cm2/s for C-LMP/G(-3) and C-LMP are agreement with the results from CV test. 4. Conclusions Using manganese acetate, lithium hydroxide and phosphoric acid as raw materials, adding at the same time a small amount of glucose and graphene oxide in DMSO-H2O system, pure phase
282
X. Fu et al. / Electrochimica Acta 225 (2017) 272–282
LiMnPO4/RGO nanoparticles with a good crystallinity were directly synthesized, via a one-step liquid-phase method under ambient pressure and at 108 C. The content of GO has a great influence to the particle size and the morphology of LiMnPO4/RGO. With 1 wt.% and 3 wt.% GO addition, the particle size of obtained LiMnPO4/RGO decreased significantly; moreover, the surface of the particles was completely wrapped by soft graphene fold layer and amorphous carbon, liking a cocoon. With 5 wt.% and 7 wt.% GO addition, on the other hand, the generated particles were like larger diamond block. Meanwhile, RGO present state of aggregation. Electrochemical tests indicate cocoon-like C-LiMnPO4/G composite material not only has high discharge specific capacity, but also shows improved high voltage platform and cycle performances at high rate. When the addition of GO is 3 wt.%, the discharge specific capacity of the obtained nanomaterial C-LiMnPO4/G(-3) is 160.86 mAh g1. The discharge specific capacity is up to 115 mAh g1 at the high voltage range ( > 4.0 V), accounting for 70% of the total specific capacity. When the charge and discharge rate was further increased to 1C, 2C, 5C, 10C, C-LMP/G (-3) still showed excellent electrochemical performance. The discharge capacity was 99.6 mAh g1 at 1C. At 5C, after 1000 cycles, 83.0% of the discharge capacity of the C-LMP/ G(-3) electrode was maintained. Cyclic voltammograms and electrochemical impedance spectroscopy analysis showed that cocoon-like C-LiMnPO4/G composite material had a larger lithium ion diffusion coefficient, smaller ion migration and charge reaction resistances. Acknowledgments This work was financially supported by the National Natural Science Foundation of China (21303042, 51672071), the Program for Innovative Research Team in University of Henan Province (No. 14IRTSTHN005). 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. electacta.2016.12.161. References [1] A.K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, J. Electrochem. Soc. 144 (1997) 1188–1194. [2] L. Dimesso, C. Forster, W. Jaegermann, J.P. Khanderi, H. Tempel, A. Popp, J. Engstler, J.J. Schneider, A. Sarapulova, D. Mikhailova, L.A. Schmitt, S. Oswald, H. Ehrenberg, Chem. Soc. Rev. 41 (2012) 5068–5080.
[3] D. Choi, J. Xiao, Y.J. Choi, J.S. Hardy, M. Vijayakumar, M.S. Bhuvaneswari, J. Liu, W. Xu, W. Wang, Z.G. Yang, G.L. Graff, J.G. Zhang, Energ. Environ. Sci. 4 (2011) 4560–4566. [4] Y. Dong, Y.M. Zhao, H. Duan, Z.Y. Liang, Electrochim. Acta 132 (2014) 244–250. [5] V. Ramar, P. Balaya, Phys. Chem. Chem. Phys. 15 (2013) 17240–17249. [6] A. Yamada, M. Hosoya, S.C. Chung, Y. Kudo, K. Hinokuma, K.Y. Liu, Y. Nishi, J. Power Sources 119–121 (2003) 232–238. [7] Z.X. Nie, C.Y. Ouyang, J.Z. Chen, Z.Y. Zhong, Y.L. Du, D.S. Liu, S.Q. Shi, M.S. Lei, Solid State Commun. 150 (2010) 40–44. [8] D.G. Kellerman, Y.G. Chukalkin, N.I. Medvedeva, M.V. Kuznetsov, N.A. Mukhina, A.S. Semenova, V.S. Gorshkov, Mater. Chem. Phys. 149–150 (2015) 209–215. [9] C. Sronsri, P. Noisong, C. Danvirutai, Spectrochim. Acta. A Mol. Biomol. Spectrosc 153 (2016) 436–444. [10] Y. Gan, C. Chen, J. Liu, P. Bian, H. Hao, A. Yu, J. Alloy. Compd. 620 (2015) 350–357. [11] H.H. Yi, C.L. Hu, X.M. He, H.Y. Xu, Ionics 21 (2015) 667–671. [12] P.R. Kumar, M. Venkateswarlu, M. Misra, A.K. Mohanty, N. Satyanarayana, Ionics 19 (2013) 461–469. [13] Q. Lu, G.S. Hutchings, Y. Zhou, H.L. Xin, H.M. Zheng, F. Jiao, J. Mater. Chem. A 2 (2014) 6368–6373. [14] H.M. Ji, G. Yang, H. Ni, S. Roy, J. Pinto, X.F. Jiang, Electrochim. Acta 56 (2011) 3093–3100. [15] H.S. Fang, H.H. Yi, C.L. Hu, B. Yang, Y.C. Yao, W.H. Ma, Y.N. Dai, Electrochim. Acta 71 (2012) 266–269. [16] P. Nie, L.F. Shen, F. Zhang, L. Chen, H.F. Deng, X.G. Zhang, CrystEngComm 14 (2012) 4284–4288. [17] J.F. Ni, Y. Kawabe, M. Morishita, M. Watada, T. Sakai, Power Sources 196 (2011) 8104–8109. [18] L.F. Zhang, Q.T. Qu, L. Zhang, J. Li, H.H. Zheng, J. Mater. Chem. A 2 (2014) 711–719. [19] L.Q. Kou, F.J. Chen, F. Tao, Y. Dong, L. Chen, Electrochim Acta 173 (2015) 721– 727. [20] F.J. Chen, F. Tao, C.M. Wang, W.L. Zhang, L. Chen, J. Power Sources 285 (2015) 367–373. [21] X.M.Liu H.J.Zhu, X.D.Shen H.Yang, Ceram. Int. 40 (2014) 6699–6704. [22] E.R. Dai, H.S. Fang, B. Yang, W.H. Ma, Y.N. Dai, Ceram. Int. 41 (2015) 8171–8176. [23] H. Guo, C.Y. Wu, J. Xie, S.C. Zhang, G.S. Cao, X.B. Zhao, J. Mater. Chem. A 2 (2014) 10581–10588. [24] C. Su, X. Bu, L.H. Xu, J.L. Liu, C. Zhang, Electrochim. Acta 64 (2012) 190–195. [25] L. Wang, H.B. Wang, Z.H. Liu, C. Xiao, S.M. Dong, P.X. Han, Z.Y. Zhang, X.Y. Zhang, C.F. Bi, G.L. Cui, Solid State Ionics 181 (2010) 1685–1689. [26] L.B. Chen, M. Zhang, W.F. Wei, J. Nanomater. 2013 (2013) 1–8. [27] Y. Jiang, R.Z. Liu, W.W. Xu, Z. Jiao, M.H. Wu, Y.L. Chu, L. Su, H. Cao, M. Hou, B. Zhao, J. Mater. Res. 28 (2013) 2584–2589. [28] J. Zong, X.J. Liu, Electrochim. Acta 116 (2014) 9–18. [29] Z.R. Chang, Y. Liu, H.W. Tang, X.Z. Yuan, H.J. Wang, Electrochem. Solid-State Lett. 14 (2011) A90–A92. [30] Z.R. Chang, H.W. Tang, Y. Liu, X.Z. Yuan, H.J. Wang, S.Y. Gao, J. Electrochem. Soc. 159 (2012) A331–A335. [31] X.N. Fu, Z.R. Chang, K. Chang, B. Li, H.W. Tang, E.B. Shangguan, X.Z. Yuan, H.J. Wang, Electrochim. Acta 178 (2015) 420–428. [32] W.S. Hummers, R.E. Offeman, J. Am. Chem. Soc. 80 (1958) 1339. [33] M. Hu, X.L. Pang, Z. Zhou, J. Power Sources 237 (2013) 229–242. [34] D.Y. Wang, H. Buqa, M. Crouzet, G. Deghenghi, T. Drezen, I. Exnar, N.H. Kwon, J. H. Miners, L. Poletto, M. Grätzel, J. Power Sources 189 (2009) 624–628. [35] D. Choi, D.H. Wang, I.T. Bae, J. Xiao, Z.M. Nie, W. Wang, V.V. Viswanathan, Y.J. Lee, J.G. Zhang, G.L. Graff, Z.G. Yang, J. Liu, Nano Lett. 10 (2010) 2799–2805. [36] J.F. Ni, H.H. Zhou, J.T. Chen, X.X. Zhang, Mater. Lett. 59 (2005) 2361–2365. [37] R.S. Nicholson, Anal. Chem. 37 (1965) 1351–1355. [38] M. Mohamedi, D. Takahashi, T. Uchiyama, T. Itoh, M. Nishizawa, I. Uchida, J. Power Sources 93 (2001) 93–103. [39] W.K. Kim, D.W. Han, W.H. Ryu, S.J. Lim, H.S. Kwon, Electrochim. Acta 71 (2012) 17–21.