Abstract

LiMnPO4 is anticipated to be a promising cathode material for next generation lithium battery. A reduction of particle size is recognized as a good strategy to improve its performance and it can be achieved by ball milling. However, the ball milling including carbon addition forms small LiMnPO4 particles with large carbon content, which leads to low volumetric energy density of electrode. In this study, carbon-coated LiMnPO4 prepared by hydrothermal route was applied to the ball milling without carbon addition. The reduction of particle size of carbon-coated LiMnPO4 was achieved by the ball milling without destroying the surface carbon layer. The ball-milled LiMnPO4 particle revealed better cathodic performance than non-milled sample. This was attributed to shortening Li ion diffusion path, improvement of structural flexibility, and large surface area of electrode due to reduction of particle size. The ball milling is attested to be a promising method to improve cathodic performance of carbon-coated LiMnPO4.

1. Introduction

LiMPO4 compounds (M = Fe, Mn, Co, and Ni) with olivine structure have been attracted as alternative cathode materials for lithium ion batteries owing to their low cost, low toxicity, chemical and thermal stabilities compared with currently used LiCoO2 [1]. Among these compounds, LiMnPO4 has been recognized the most attractive compound due to its high operation voltage determined by Mn3+/4+ redox couple at 4.1 V versus Li/Li+ and is compatible with the system presently used in lithium ion battery [2].

However, the most serious problem of this cathode material is low intrinsic electronic and Li ion conductivities [3, 4]. Various experimental reports have indicated that carbon-coating can provide high electronic conductivity [57]. In fact, a large charge-discharge capacity of carbon-coated LiFePO4 synthesized by a hydrothermal synthesis method has been reported [8].

As for improvement of Li ion conductivity, much effort has been paid for particle size reduction to shorten Li ion conduction path [9]. Drezen et al. reported reversible capacities for 140 and 270 nm diameter LiMnPO4 particles prepared by sol-gel method were 81 and 7 mA h g−1, respectively [10]. Some groups reported a good performance of ball-milled small LiMnPO4 [11] and LiCoPO4 particles [12]. However, their processes included conductive carbon addition (≤20 wt.%) before the ball milling to obtain a carbon composite. This large amount of carbon causes surely improvement of the electronic conductivity, however, such heavy carbon coating decreases volumetric capacity. Additionally, commercial battery foils typically only contain 2.5 weight percent (or lower) carbon blacks with close to 95 weight percent active materiel. Therefore, it may not appeal for practical application.

The carbon coated olivine type cathode materials prepared by hydrothermal route have been reported [13, 14]. The method allows us to prepare carbon-coated particles readily by only addition of a carbon source to start materials under appropriate condition. Consequently, it is guessed that if carbon-coated LiMnPO4 (LiMnPO4/C) prepared by hydrothermal condition was supplied to the ball milling, carbon amount could be suppressed by omission of further carbon addition. On the other hand, harmful influence by the ball-milling of LiMnPO4/C is also surmised, such as destroying the carbon coating layer. Hence, research on effect of the ball-milling to LiMnPO4/C on electrochemical properties is a worthwhile work.

Herein, we applied to the ball-milling technique to hydrothermally synthesized LiMnPO4/C and its electrochemical property was compared with non-milled LiMnPO4/C to clarify influence of the ball-milling technique on electrochemical property of LiMnPO4/C.

2. Experimental

LiMnPO4 was synthesized by a modification of hydrothermal process in previous reports for LiFePO4 [1517]. MnSO4·5H2O (43.84 g) and Li3PO4 (20.84 g) were dissolved into purified water (44 mL) under N2 atmosphere. A molar ratio of Li : Mn : P in a precursor solution was 3 : 1 : 1. In order to prepare carbon-coated LiMnPO4, 6.00 g of carboxy methyl cellulose (CMC) was added into the precursor solution [1]. The precursor solution was put into a glass-lined Parr reactor with N2 gas sealed in a stainless steel autoclave, and then heated at 200°C for 3 h with stirring at 680 rpm. A precipitation was produced in the reactor under hydrothermal conditions. The precipitation was separated centrifugally at first, and then by a mean of freeze-drying at −50°C for 12 h. Yield was always higher than 95%. The obtained sample was dispersed into ethanol and ball milled at 400 rpm for 12 h with ZrO2 ball using a planetary ball mill equipment (Pulverissette P-6, Fritsch GmbH). The milled and non-milled samples were treated at 700°C under 3% H2/Ar flow for 1 h to obtain electroconductive graphite carbon [1, 14].

The crystalline phases of the synthesized sample were identified with X-ray diffraction (XRD, RINT-Ultima, Rigaku) with Cu Kα radiation. Raman measurement (NRS-1000, JASCO) was carried out at room temperature by a laser radiation of 532 nm. The morphology of the synthesized particle was observed by scanning electron microscope (SEM, JEOL). Coated carbon amount of the samples was estimated by Thermogravimetry (TG) analysis (DTG-60, Shimazu) and BET surface areas of milled and non-milled particles were measured using BELSORP-mini (BEL JAPAN Inc.).

Performances of milled and non-milled LiMnPO4/C as a positive electrode were tested by a galvanostatic charge/discharge test. A composite LiMnPO4/C electrode was prepared by mixing LiMnPO4/C, Ketjen black, and Polyvinylidene difluoride (PVdF) in a weight ratio of 90 : 5 : 5 in 1.2 mL of NMP (N-methyl pyrrolidone). The LiMnPO4/C electrode was painted onto a thin Al sheet and dried overnight at 85°C under reduced pressure for 12 h. The LiMnPO4 electrode (14 mm diameter) was set in a coin cell 2032 with a mixture of ethylene carbonate (EC) and dimethyl carbonate (DEC) (volume ratio = 1 : 1) containing 1 mol·dm−3 LiPF6 as an electrolyte and with lithium metal as a negative and a reference electrodes. The galvanostatic charge/discharge tests of coin cell with LiMnPO4 positive electrode were performed by using HJ1001SM8A (HOKUTODENKO). In the test, charging process was done under CC-CV (constant current-constant voltage) mode, that is, charge in a constant current condition of 0.1 C until 4.5 V and then constant voltage charging was performed at 4.5 V until current dropped to 0.01 C. The discharge process was carried out at constant current condition of 0.1 C. Cutoff voltages were 2.0 V and 4.5 V for discharge and charge processes, respectively.

3. Results

Figure 1 displays SEM images of non-milled and milled samples. In both cases, particles aggregated each other and formed secondary particles. The sizes of primary particles of non-milled and milled samples were estimated to be about 500 and 50 nm, respectively. The particle shapes of both samples were irregular. It is concluded that much smaller particles can be obtained by ball milling. BET surface area of non-milled and milled samples were 8.9 and 49.0 m2 g−1, respectively. This result is well consistent with SEM observation.

XRD patterns of non-milled and milled samples are depicted in Figure 2. In non-milled sample, all diffraction peaks were attributed to LiMnPO4 with olivine structure and no impurity phase was observed. In milled sample, the diffraction peaks were much broader due to small particle and low crystallinity given by the ball milling. An appearance of new phase was not confirmed. Therefore, it is inferred that the ball milling can crash the LiMnPO4 particles, but did not produce any impurity phase.

The Raman spectroscopy is sensitive to surface of the materials comparing with XRD. Dokko et al. have succeeded in detection of even small amount of impurity on the LiFePO4 synthesized by the hydrothermal process [17] and carbon coating layer on the surface of the LiFePO4/C [8], LiMnPO4/C [1], and LiCoPO4/C [14]. Thus, the Raman measurement was carried out to detect carbon layer on the samples that could not be detected by the XRD measurement. Figure 3 reveals the Raman spectra of non-milled and milled LiMnPO4/C. In non-milled sample, clear peaks appeared at 948, 1361, and 1610 cm−1. The peak at 948 cm−1 is assigned to the symmetric vibration of the PO4 group [18]. As mentioned above, clear diffraction peaks of the LiMnPO4 were observed in the XRD patterns. The peak at 948 cm−1 was consistent with the result of XRD and it was clear that the LiMnPO4 has been synthesized successfully. Peaks at 1610 cm−1 and 1361 cm−1 were attributed to graphite and disorder carbon (G and D bands), respectively [19], indicating that conductive carbon layer was formed on the surface. In milled sample, the peak at 948 cm−1 was not observed because of small particles and low crystallinity of the milled-sample. A high noise level of the spectrum is also possible reason. As for the G and D bands, they clearly appeared in the spectrum although their intensities became weaker, indicating that surface carbon layer still existed even after crashing the particles by the ball milling. The ratio of G band to D band, that is, quality of the carbon layer, was 1.4 and 1.3 for non-milled and milled samples, respectively. Therefore, the qualities of surface carbon layer of both samples are considered to be same. TG analysis was performed to estimate surface carbon amount. Estimated carbon amounts of non-milled and milled samples were 3.4 and 2.2 wt%, respectively. It seems that some carbon coatings were destroyed and peeled off during the ball milling.

The charge/discharge test of non-milled and milled LiMnPO4/C was carried out (Figure 4). In both cases, charge and discharge plateaus at around 4.1 V versus Li/Li+, which are attributed to intrinsic redox of LiMnPO4, were observed. In non-milled LiMnPO4, the plateau was much shorter than milled sample and the discharge capacity was 109 mA h g−1. This was 64% of theoretical one (171 mA h g−1) [20]. In milled LiMnPO4/C, a potential difference of the plateau between charge and discharge curves was small owing to small internal resistance of the milled LiMnPO4/C electrode. The discharge capacity improved, 145 mA h g−1, 85% of theoretical capacity. A rate capability test also revealed favorable performance of milled sample (Figure 5). In this experiment, the charge process was performed at constant current condition of 0.1 C. The discharge capacities of milled and non-milled samples at 0.1 C were 127 and 98 mA h g−1, respectively. Then, the capacity of non-milled sample decreased more quickly, only 6 mA h g−1 at 5.0 C. On the other hand, the milled sample still maintained high capacity, 64 mA h g−1 at same C rate. This corresponded to 50% of capacity retention.

4. Discussion

In this study, we applied the ball milling to hydrothermally synthesized LiMnPO4/C and its electrochemical property was compared with non-milled LiMnPO4/C. By the ball milling, small particle of LiMnPO4/C was obtained, although a little loss of surface carbon layer was confirmed. The ball-milled LiMnPO4/C demonstrated larger discharge capacity than non-milled LiMnPO4/C. Fey et al. studied on a dependency of discharge capacity on surface carbon amount of LiFePO4 [21]. They reported the discharge capacities of LiFePO4/C with coated carbon from 1.0 to 6.0 wt% were almost identical. In this study, carbon amounts of non-milled and milled samples were 3.4 and 2.2 wt.%, respectively. Consequently, it can be said that the carbon layer on the LiMnPO4 particle surface was destroyed a little by the ball milling; however, the deconstruction would hardly influence on electrochemical property. The difference of cathodic performance between non-milled and milled samples seems to be attributed to difference in length of Li ion transport paths by diverse particle size. Also, it is fairly well known that the charged PO4 phase undergoes a severe lattice deformation due to the asymmetric electronic configuration of Mn3+ ions (3d4( )) [22, 23]. The Jahn-Teller deformation as well as the large lattice misfit between LiMnPO4 and MnPO4 phases destroys the integrity of lattice, thus leading to a low electrochemical activity. The small size of LiMnPO4 enhances structural flexibility against lattice deformation [24]. Thus, improvement of integrity of the lattice by structural flexibility of small particles would be also another reason for superior performance of the milled sample. Moreover, the surface area of electrode increased with reduction of LiMnPO4 particle size as indicated by BET surface area measurement, leading to large contact area between electrolyte and electrode. This enlargement of electrode area would also help improved performance of milled sample.

The reduction of LiMnPO4/C particle size by the ball-milling was achieved and improved performance of ball-milled LiMnPO4/C was confirmed. This is attributed to short Li ion diffusion paths, improvement of structure flexibility, and large surface area of electrode by reduction of LiMnPO4/C particle size. The ball milling is attested to be a promising method to improve cathodic performance of carbon-coated LiMnPO4.

5. Conclusion

The ball milling was applied to hydrothermally synthesized LiMnPO4/C. The LiMnPO4/C particle size was successfully reduced by the ball milling. Although a little loss of surface carbon layer by the ball milling was observed, this loss did not influence on electrochemical properties. The ball-milled LiMnPO4/C demonstrated higher cathodic performance than non-milled sample. This would be attributed to short Li ion diffusion paths, improvement of structure flexibility, and large surface area provided by reduction of LiMnPO4/C particle size. The ball milling is proven to be a promising method to improve cathodic performance of carbon-coated LiMnPO4.