High rate capability by sulfur-doping into LiFePO4 matrix

Enhanced electrochemical performance of LiFePO4 for Li-ion batteries has been anticipated by anion doping at the O-site rather than cation doping at the Fe-site. We report on the electrochemical performance of S-doped LiFePO4 nanoparticles synthesized by a solvothermal method using thioacetamide as a sulfur source. S-doping into the LiFePO4 matrix expands the lattice due to the larger ionic radius of S2− than that of O2−. The lattice parameters a and b increase by around 0.2% with sulfur content, while that of c remains almost unchanged with only 0.03% increase. The S-doping also contributes to the suppression of antisite defects (Fe occupying Li sites), which facilitates the easy migration of Li in the diffusion channels without blockage. Owing to these effects of S-doping, the S-doped LiFePO4 nanoparticles show enhanced electrochemical properties with a high discharge capacity of ∼113 mA h g−1 even at a high rate of 10C.


Introduction
Rechargeable lithium-ion batteries (LIBs) are widely used for portable electronics and are also considered as promising energy storage devices for electric vehicles due to their long cycle life and high energy density. 1,2 Ensuring safety of the devices and the use of environmentally-friend and cheap materials are important requirements to deal with further use of LIBs in the near future. Also, the current requirement for LIBs is to achieve both high charging/discharging rates and high capacity. 3 Development of cathode materials plays an important role in satisfying these requirements because anode materials (mainly carbon materials) already exhibit high capacity at high charging/discharging rates. 4,5 Olivine-structured LiFePO 4 is considered as one of the most promising cathode materials for LIBs due to its high thermal stability, good cycling property, high energy density, theoretical capacity, and environmental friendliness as well as low cost. 6,7 Although the theoretical capacity (170 mA h g À1 ) of LiFePO 4 is a bit lower than that of conventional cathode materials (e.g., LiCoO 2 : 274 mA h g À1 ), its benets such as low toxicity, high safety, low-cost and good cycling property will make LiFePO 4 replaced to conventional cathode materials as now being produced commercially. 8,9 However, its poor electronic conductivity (10 À8 to 10 À10 S cm À1 ) and low lithium ion diffusion coefficient (10 À10 to 10 À16 cm 2 s À1 ) lead to a decreased capacity at a high charging/discharging rate, which has inhibited the use in high-rate applications. [10][11][12][13] Many efforts have been made to solve the problems by mainly three strategies: (1) carbon coating, 14,15 (2) reducing particle size, 16,17 (3) doping with certain element. [18][19][20] Excellent electrochemical properties have been reported by combining these three strategies. The carbon coating can increase the surface electronic conductivity, leading to an improvement of rate performance and cycling life. The method of carbon coating has been developed and established already as a thermal treatment of the mixture of LiFePO 4 and carbon sources under inert gas. Reducing particle size can shorten the Li + diffusion pathway length, that is an effective way to improve rate capability. Hydro/ solvo-thermal synthesis approaches have been applied to obtain the LiFePO 4 nanoparticles. As for the doping strategy, doping proper ions into the LiFePO 4 can increase the electronic and ionic conductivity of the LiFePO 4 and also expand lithium ions diffusion channels in the structure, leading to a signicant improvement of the rate capability and cyclic performance. Various elements such as La, Cu, Y, V, Mn, Mg, Bi, Co, Pt, Pd, Ni, Zn, Mo and Cr have been previously used as dopant elements in cation-site (Fe-site) of LiFePO 4 . Certain cation doping elements like Ni and Zn can remarkably enhance the electrochemical performance of LiFePO 4 . [21][22][23][24] There are much less reports on anion doping at O-site for LiFePO 4 . However, the O-site doping is expected to signicantly enhance the conductivity of LiFePO 4 rather than Fe-site doping as indicated by rst-principle calculations. 25,26 Indeed, signicantly enhanced high rate capability and cycling performance have been reported for anion-doped (such as F and Cl) LiFePO 4 . 27-29 Among considerable anion species, sulfur doping would be efficient to improve the electrochemical performance (rate capability and cyclic performance) because easy Li + intercalation and extraction are expected by an expansion of lattice arising from larger anion doping and the lower bond dissociation energy of Li-S (312.5 kJ mol À1 ) than that of Li-O (340.5 kJ mol À1 ). 30 Sulfurdoping has been only reported by Park et al. where S-doped LiFePO 4 nanoparticles exhibit enhanced electrochemical properties compared to non-doped and N-doped LiFePO 4 . 31 However, only surface substitution of LiFePO 4 nanoparticles was achieved by exposing solvothermally-synthesized LiFePO 4 nanoparticles to sulfur vapor at high temperature. The further enhancement of electrochemical properties can be expected for the sulfur doping into the LiFePO 4 matrix.
Herein, we report on the synthesis of S-doped LiFePO 4 nanoparticles by a single-step solution approach and its improved electrochemical performance compared to non-doped LiFePO 4 . Thioacetamide was used as a sulfur source, which was the conventional way for the synthesis of S-doped TiO 2 and metal sulde such as CdS, ZnS and Bi 2 S 3 by the solution approach. [32][33][34][35] The successful S-doping into LiFePO 4 matrix was conrmed by Xray diffraction (XRD), energy dispersive X-ray spectrometry (EDS) analysis and X-ray photoelectron spectrometry (XPS). The formation of LiFePO 4 nanoparticles with $100 nm in diameter was observed by scanning electron microscopy (SEM). The S-doped LiFePO 4 nanoparticles showed a high discharge capacity of $113 mA h g À1 even at a high rate of 10C.

Preparation of non-doped and S-doped LiFePO 4 nanoparticles
Non-doped and S-doped LiFePO 4 nanoparticles were prepared by a solvothermal method. The non-doped LiFePO 4 nanoparticles were prepared as following the literature. 36 At rst, 6 mmol of FeSO 4 $7H 2 O and 6 mmol of H 3 PO 4 (85%) were completely dissolved in 20 mL of ethylene glycol. At same time, 16.2 mmol of LiOH$H 2 O was also dissolved in 15 mL of ethylene glycol. Then, the solution containing a Li source was added dropwise to the solution containing Fe and P sources. The nal molar ratio of Li : Fe : P was 2.7 : 1 : 1. The reaction was carried out in an autoclave at 180 C for 10 h. Aer the reaction, the resultant gray precipitate was washed with deionized water and ethanol for several times and then dried at 70 C overnight under vacuum. S-doped LiFePO 4 nanoparticles were prepared by adding thioacetamide with different amount (0.13, 0.40, 0.67, 1.33, 2.66 and 3.99 mmol) to the reaction solution containing Li, Fe and P sources before solvothermal treatment (the molar ratio; Li : Fe : P : S ¼ 2.7 : 1 : 1 : 0.02, 0.07, 0.11, 0.22, 0.44 or 0.67). In the manuscript, the non-doped and S-doped LiFePO 4 nanoparticles prepared with different amount of thioacetamide were denoted as LFP and LFP-S-x (where x ¼ 0.02, 0.07, 0.11, 0.22, 0.44 or 0.67), respectively. For the electrochemical measurements, the dried samples with a certain amount of sucrose (10 wt%) were mixed in ethanol and calcined at 650 C for 6 h under N 2 atmosphere in order to obtain the carbon-coated samples.

Characterization
The crystalline phases of samples were identied by X-ray diffraction (XRD; Rigaku, RINT-2200) using a Cu Ka radiation.
The unit cell parameters were investigated by Rietveld renement (using the RIETAN-FP program). 37 Surface morphologies of samples were observed using a scanning electron microscope (SEM, Hitachi Corp., S-3000HXS); the energy dispersive X-ray analysis has been performed using 15 keV. X-ray photoelectron spectrometry (XPS) measurements were carried out using an electron spectrometer (Shimadzu Corp./Kratos, AXIS-165Â) with an Al Ka X-ray source. The position of each peaks was collected by using C spectra. Fourier transform infrared spectroscopy (FTIR) measurements were carried out using a FTIR 4700 spectrometer (JASCO).
Electrochemical measurements. The electrochemical performances were measured by using an assembly cell (Hohsen, Japan) with lithium foil as counter and reference electrodes. For the preparation of working electrodes, at rst, the slurries of carbon-coated non-doped and S-doped LiFePO 4 were prepared by mixing active materials, acetylene black, and polyvinylidene diuoride (PVDF) binder (a mass ratio of 6.5 : 2.5 : 1) in N-methyl-2-pyrrolidone (NMP) solution. Then, the slurry was coated on an Al foil by using a coater. Aer being dried under vacuum at 90 C for 10 h, working electrodes were punched and weighed. Then, the cells containing 1.0 mol L À1 LiPF 6 in ethylene carbonate (EC)-diethyl carbonate (DEC) (1 : 1, v/v) as electrolyte were assembled in a glove box under a dry and high purity argon atmosphere. Charge/discharge tests were performed between 2.0 and 4.3 V by using a battery tester (Kikusui, PFX2011, Japan) at room temperature at around 20 C.

Results and discussion
Non-doped and S-doped LiFePO 4 nanoparticles were prepared by a solvothermal method. The non-doped LiFePO 4 nanoparticles were prepared from lithium, iron and phosphorus sources dissolved in ethylene glycol at 180 C for 10 h as following the literature. 36 For the sulfur doping, thioacetamide was added to the reaction solution. Hereaer, the non-doped and S-doped LiFePO 4 nanoparticles prepared with different amount of thioacetamide were denoted as LFP and LFP-S-x (where x ¼ 0.02, 0.07, 0.11, 0.22, 0.44 or 0.67), respectively. Fig. 1(a) shows XRD patterns of the obtained samples (where silicon was used as a standard). The peaks of LFP and LFP-S-x (x # 0.22) were exclusively indexed to the LiFePO 4 phase (JCPDS card number: 81-1173). While, the formation of FeS 2 (JCPDS card number: 42-1340) was detected as an impurity for the samples prepared with an excess of thioacetamide (x $ 0.44). The unit cell parameters were investigated by Rietveld renement (using the RIETAN-FP program). 37 An increase of lattice parameters with sulfur doping amount was conrmed as shown in Fig. 1(b) and Table S1, † indicating the successful doping of sulfur in LiFePO 4 matrix because the ionic radius of S 2À (1.84 A) is larger than O 2À (1.40 A). 38 Interestingly, the lattice parameters of a and b were found to increase by around 0.2% with sulfur doping, while that of c remained almost unchanged as only 0.03% increase. Such large changes of a and b lattice parameters were similarly reported for the F-doping into LiFePO 4 . 39 In the case of cation doping at Fe-site, the most cationic elements (except vanadium) are found to change the lattice parameters uniformly in literature. [40][41][42][43][44][45] Thus, it should be noted that the anion doping at O-site can specially change the a and b lattice parameters and remain c lattice parameter unchanged, which is not observed for cation doping. That might be explained by the position of oxygen atoms in the LiFePO 4 lattice. Oxygen atoms are located parallel to the a and b axes of LiFePO 4 lattice while irregularly for the c axis. 46 Thus, it can be considered that the sulfur doping at O-site simply expands the lattice along the both a and b axes and the stress of c axis induced by the sulfur doping are released toward a and b axes by distortion, which would induce the specic changes for a and b axes of LiFePO 4 lattice. The expansion of LiFePO 4 lattice by sulfur doping was maintained aer carbon coating at 650 C under inert atmosphere (Fig. S1 †). A further increase in a and b lattice parameters was observed aer the carbon coating process. The thermal treatment at 650 C might make the crystal structure more stable by displacing the S atoms into favored positions and minimizing ion disorder. 47 The expansion of LiFePO 4 lattice by a sulfur doping would give us the benet of high capacity at high charging/discharging rate because of easy Li + intercalation and extraction by expanded lithium ions diffusion channels. Fig. 2 shows the SEM images of LFP, LFP-S-0.22 and LFP-S-0.67. Uniform nanoparticles with $100 nm in diameter were observed for LFP ( Fig. 2(a)). Similar nanoparticles formed in LFP-S-0.22 (Fig. 2(b)), indicating S-doping did not affect the morphology. In the LFP-S-0.67, larger and crystalline particles over 1 mm were observed with the presence of LiFePO 4 nanoparticles (Fig. 2(c) and (d)). The large particles were concluded as an impurity FeS 2 by the XRD result and EDS analysis (Fig. S2 †). EDS mapping and analysis for S-doped LiFePO 4 (LFP-S-0.22) conrmed that sulfur was uniformly dispersed as an atomic percent of $1.4% (Fig. 3 and S3 †). Based on this result, the composition of LFP-S-0.22 was estimated as LiFePO 3.9 S 0.1 .
The chemical states of sulfur, phosphorous and oxygen ions were investigated by XPS (Fig. 4). Fig. 4(a) shows the XPS spectra for the S 2p electrons of LFP, LFP-S-0.22 and LFP-S-0.67. A broad peak at around 160-163 eV was conrmed in the LFP-S-0.22 although no peaks was detected in LFP. The peak was tted as the core levels of S 2p 1/2 (at 162.4 eV) and S 2p 3/2 (at 161.4 eV) separated by a spin-orbit splitting of 1.0 eV, demonstrating that the sulfur atoms are doped in the state of S 2À . 48,49 On the other hand, a broad peak at $1.0 eV higher than that of LFP-S-0.22 was detected in LFP-S-0.67. The peak at higher energy could be assigned to that of disulde derived from FeS 2 . 50,51 The presence of FeS 2 in LFP-S-0.67 was consisted with the XRD results. The XPS spectra for the P 2p electrons of LFP, LFP-S-0.22 and LFP-S-0.67 are shown in Fig. 4(b). As shown in Fig. 4(b), the sulfur doping shied the P 2p XPS peak to lower binding energy. Similar chemical shi of binding energy has been reported for the nitrogen doping at Li 3 PO 4 and Na 3 PO 4 where a decrease in the P 2p binding energy was explained by simple charged-shell models 52,53 as reducing the average ionic charge on the phosphorus ions due to the replacement of P-O bonds with less ionic P-N bonds. 54,55 The binding energy chemical shi by sulfur doping to LiFePO 4 can be explained by the similar effect; reducing the average ionic charge on the phosphorus ions due   to the replacement of P-O bonds with less ionic P-S bonds results in the lower P 2p binding energy. O 1s XPS peaks in LFP-S-0.22 also showed lower binding energy compared to LFP, which is similar to N-doped Li 3 PO 4 and Na 3 PO 4 (ref. 54) (Fig. S4 †). These XPS results for the S 2p, P 2p and O 1s electrons clearly indicate the successful S 2À doping at O-site of LiFePO 4 matrix.
Electrochemical measurements were carried out to test the electrical properties of LFP and LFP-S-0.22 aer carbon coating by following a reported method. 56 Fig. 5(a) shows the charge/ discharge curves of carbon-coated LFP and LFP-S-0.22 measured at 5C (850 mA g À1 ) between 2.0 and 4.3 V vs. Li + /Li. Both samples showed a charging and discharging plateau region at $3.4 V which is correspond to the redox reaction between FePO 4 and LiFePO 4 . 57 A smaller voltage difference between the charge and discharging plateaus was conrmed in LFP-S-0.22 compared to LFP, implying that LFP-S-0.22 possess the better electronic conductivity and higher reaction reversibility. 58 The non-doped LFP presented a discharge capacity of 99.0 mA h g À1 at 5C. In contrast, S-doped LFP-S-0.22 led to a higher discharge capacity of 121.6 mA h g À1 at 5C. The discharge capacities of LFP and LFP-S-0.22 were investigated at different rates from 0.5C to 10C in order to examine the rate capabilities ( Fig. 5(b)). The LFP-S-0.22 showed much higher discharge capacities (131.7, 128.5, 121.6 and 112.7 mA h g À1 at 0.5, 1, 5 and 10C, respectively) at each rates compared to LFP (120.6, 116.4, 99.0 and 81.1 mA h g À1 at 0.5, 1, 5 and 10C, respectively). In addition, the discharge capacity of LFP-S-0.22 varied much smaller from 128.5 to 112.7 mA h g À1 at rates of 1C to 10C, retaining 87.7% of discharge capacity. There was no difference between LFP and LFP-S-0.22 for coulombic efficiency at each C rate (Fig. S5 †). These electrochemical investigations exhibited that sulfur doping into LiFePO 4 effectively improved the rate capability. It has been considered that high rate capability can be achieved by an increase in electric conductivity of LiFePO 4 . The signicantly enhanced high rate capability achieved in LFP-S-0.22 was considered due to an increase in electric conductivity by sulfur doping as expected by rst-principle calculations. 31 The differential capacity (dQ/dV) studies showed smaller peak voltage separation between the anodic and cathodic peaks in LFP-S-0.22 compared to LFP, indicating the smaller electrochemical polarization and better electrochemical kinetics of the S-doped LiFePO 4 (Fig. S6 †). 59,60 Also, the LFP-S-0.22 exhibited higher internal electronic conductivity and lower resistance than LFP (Fig. S7 †). In addition, the sulfur doping results in the suppression of Fe c Li antisite defects (Fe occupying Li sites) in LiFePO 4 which is important as it facilitates the easy migration of Li in the diffusion channels without blockage. 61 The antisite defect is considered to be present in the one-dimensional channel of LiFePO 4 and effectively block the Li + pathways, 62-64 which reduces the capacity, rate capability and cycle life of a battery. The suppression of Fe c Li antisite defects can be investigated by Fourier transform infrared (FTIR) spectra. 39,65 Fig. 6 shows the FTIR spectra of LFP and LFP-S-0.22. The Infrared absorption spectra in the range 800-1200 cm À1 and 400-700 cm À1 were assigned to the symmetric (n 1 ) and asymmetric (n 3 ) P-O stretching, symmetric (n 2 ) and asymmetric (n 4 ) O-P-O bending in (PO 4 ) 3À tetrahedral, respectively. 66 As shown in Fig. 6, the sulfur doping shied a peak at $975 cm À1 to lower wavenumber, while there were no signicant changes for the other peaks. This result is clearly consistent with the Fe c   Li antisite defects by S-doping would also contribute to improvement in the capacity and the rate capability compared to non-doped LiFePO 4 . It should be also mentioned that the discharge capacity of 112.7 mA h g À1 at 10C achieved in LFP-S-0.22 is much higher than that of sulfur-doped LiFePO 4 only on the surface (86.4 mA h g À1 at 10C) reported in the literature. 31 Thus, it can be concluded that anion doping into the LiFePO 4 matrix is more important than anion doping only on the surface of LiFePO 4 particle in order to enhance the electrochemical performance. The synthesis of cation-doped LiFePO 4 has been reported by similar solvothermal approaches. 67,68 Thus, our system would be simply applied to the cation-doping approach. The further enhancement of electrochemical performance is expected by multi-elements codoped LiFePO 4 with certain elements.

Conclusions
In this work, we successfully prepared the S-doped LiFePO 4 nanoparticles with $100 nm in diameter by a solvothermal method using thioacetamide as a sulfur source. The LiFePO 4 lattice is expanded with increasing the sulfur amount, but the excess amount of sulfur results in the formation of FeS 2 as an impurity. The lattice parameters of a and b increase by around 0.2% with the sulfur content, while that of c remains almost unchanged as only 0.03% increase. The sulfur doping at O-site also results in the suppression of Fe c Li antisite defects in lithium ions diffusion channels. Both the lattice expansion and the suppression of Fe c Li antisite defects by S-doping contribute to the enhancement of electrochemical properties due to the easy migration of Li in the diffusion channels without blockage.

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