The impact of carbon additives on lithium ion diffusion kinetic of LiFePO4/C composites

Introduction: LiFePO4/C composites were synthesized via physical mixing assisted solvothermal process. Different kinds of carbon materials were investigated including 0D (carbon Ketjen black), 1D (carbon nanotubes) and 2D (graphene) materials. X-rays diffraction patterns of carbon coated LiFePO4 synthesized by solvothermalwas indexed to pure crystalline phasewithout the emergence of second phase. LiFePO4 platelets and rods were in range size of 80-200 nm and dispersed well in carbonmatrix. The lithium ion diffusion kinetics was evaluated through the calculated diffusion coefficients to explore the impact of carbonmixing. Methods: In this work, we studied the structure, morphologies and the lithium ion diffusion kinetic of LiFePO4/C composites for Li-ion batteries. Different characterization methods were used including powder X-rays (for crystalline structure); Transmission Electron Microscopy (for particle and morphologies observation) and Cyclic voltammetry (for electrochemical kinetic study). Results: The study indicated LiFePO4/C compositeswere successfully obtained bymixing process and the electrochemical performance throughout the calculated diffusion coefficient was significantly improved by adding the carbon types. Conclusion: The excellent iondiffusionwasobtained for composites LiFePO4/Ketjenblack (KB) and LiFePO4/CNT compared to LiFePO4/Graphene. KB could be a potential candidate for large-scale production due to low-cost, stable and high electrochemical performance.


INTRODUCTION
With the rapid growth of technology, Li-ion batteries (LIBs) have gradually become one of the most innovative and potential energy storage devices for a variety of energy applications. Regarding this, materials development plays a crucial role to provide LIBs outstanding characteristics compared to conventional batteries [1][2][3] . Following LiCoO 2, LiMn 2 O 4 as cathode materials, olivine LiFePO 4 (LFP) has been considered the most compatible candidate for portable devices or larger scale energy storage nowadays. This material possesses various desired properties, such as a long flat plateau of 3.45 V over a large lithium solid solution, straightforward fabrication, environmental benignity, and safety in handling and operation 4 . Unfortunately, there remain two major obstacles including low lithium-ion diffusion coefficient (10 −14 -10 −16 cm 2 .s −1 ) and poor electronic conductivity (< 10 −9 S.cm −1 ) which struggle the electrochemical performance and the commercialization of LFP 5,6 . A variety of solutions have been proposed to overcome these problems. There most effective strategies consist of nano-sizing grains, conductive carbon utilization, controlled off-stoichiometry, and transition metals doping [7][8][9] . By using nanoscale LFP particles, the diffusion pathway of lithium ions into octahedral vacancies of LFP crystalline becomes shorter, and then the electrochemical performance could be enhanced. Till now, top-down (solidstate reactions) and bottom-up (co-precipitation, hydro/solvothermal, sol-gel) methods are two main routes to synthesize nanoscale LFP and its composites. Unlike the former, the bottom-up or solutionbased methods are much interested due to energysaving and due to the ease control of size and morphology. Meanwhile, in the second strategy, carbon materials play the role of forming a conductive network linking the LFP particles to improve its electron transport 10 . Additionally, carbon deposition or coating on the surface of LFP particles in a reducing atmosphere also helps to limit the growth or agglomeration of LFP particles 11 . Since the outstanding revolution of carbon materials, the unique characteristics come up with their own shapes such as spheres [12][13][14][15] , nanotubes [16][17][18] , nanofibers 19,20 , or sheets (graphene) [21][22][23] . These advantages could be potentially exploited for different purposes of materials improvement. However, the morphology of carbon, if not properly integrated, sometimes restrains lithium ion movements in olivine structure and decreases the energy density in LFP/C composite. Apart from the above approaches, multivalent cations are usually used as suitable dopants for LFP to enhance high current rate performance as well as reduce polarization 14 . However, this increase of conductivity is doubted whether dopants could penetrate easily into LFP lattice and conduct the formation of surface conductive phases 8 .
With an aim to increase the ionic conductivity of LFP, LFP/C composites with different carbon matrix including Ketjen black (KB), CNT and graphene (Gr) were prepared. By using the solvothermal technique, LFP particles are the nanoscale size, and hence the lithium ion diffusion pathway could be expectedly shortening. To evaluate the impact of carbon coating, Cyclic Voltammetry method was used to understand the electrochemical kinetic and calculate the lithium ion diffusion coefficient as well.

Synthesis process
Carbon-coated LFP particles were prepared via solvothermal route. All the reagents are analytical grade and readily used without further purification. Precursors are lithium hydroxide monohydrate LiOH.H 2 O (Fisher, USA), ferrous sulfate heptahydrate FeSO 4 .7H 2 O (Fisher, USA), phosphoric acid H 3 PO 4 85 wt.% (Fisher, USA). Carbon sources are Ketjen black EC-600JD (Azko Nobel, Dutch), multiwalled carbon nanotubes (Sigma-Aldrich, USA) and graphene (Fisher, USA). 10 wt.% of different kind of carbons was well-dispersed by ultrasonication in 10 mL ethanol for 1 h. After that, 1.2 g of LiOH.H 2 O was dissolved in 50 mL of ethylene glycol (EG, 98.5%, Merck, Germany) and distilled water until the mixture was homogeneous. Subsequently, carbon was added and magnetically stirred for 10 min to obtain mixture 1. Likewise, mixture 2 was prepared by dissolving 2.7 g FeSO 4 .7H 2 O with 1 mL H 3 PO 4 (99 %, Merck, Germany) in 50 mL EG/H 2 O solvent under the inert atmosphere. Mixture 1 was slowly added to mixture 2, continuously magnetically stirred for several minutes and transferred into a Teflon-lined stainless-steel autoclave. The solvothermal reaction occurred at 180 O C for 5 h. The obtained mixture was cooled down in ambient condition followed by filtration and drying. Finally, LFP/C powder was annealed in the tube furnace at 650 O C for 3 h to obtain the composite.

Materials characterization
The synthesized materials were characterized by various advanced techniques. Powder X-ray diffraction (XRD) data was collected on Bruker D8 Advance (France) with Cu-K α 1 radiation (λ = 1.54056 Å) at 2θ range of 15 o to 60 o . Transmission electron microscopy (TEM) for particle size study and morphology evaluation were performed using JEOL JEM 1400 microscope (120 kV).

Electrochemical measurements
For electrochemical characterization, the prepared composites were laminated to thin film cathode for half-cell testing. LFP/C composite was grounded using mortar for approximately 2 h to obtain fine particles. Then, 5 wt.% poly(vinylidene fluoride-cohexafluoropro-pylene) (PVdF-HFP) as a binder in the solution of N-methyl-pyrrolidone (NMP) was added to LFP/C composite and mixed until a homogeneous slurry was obtained. Then, the slurry was coated on an Al foil and dried in a vacuum chamber at 80 O C for 14 h. The cathode foil was punched into 10 mm-diameter discs and used as working electrodes. Lithium metal foil was used as the anode while commercial Whatman glass-fiber membrane and the 1 M solution of LiPF 6 in EC: DMC (1:1) were used as separators and electrolyte, respectively. Finally, Swagelok cells were assembled in an argon-filled glove box. Cyclic voltammetry tests were performed on Bio-Logic MPG-2 battery tester (France) in the voltage range of 3.0-4.0 V (vs. Li + /Li) at room temperature. For each of composite material, the measurement was performed in three times repeated in two different cells.

Structural analysis
Crystalline structure and chemical composition of LiFePO 4 were evaluated, as shown in Figure 1. Comparing to XRD reference patterns, olivine LFP phase was successfully prepared with the index of Pnmb space group in orthorhombic system (ICDD no. 40-1499) without emerged impurities. Additionally, from wt.% by analyzing the XRD diagram; which might penalize the expected effect of graphene on the electrochemical performance. LFP/KB composite contains a fraction of amorphous carbon nanoparticles causing a noisy baseline and some extremely weak peaks in the XRD diagram. According to diffraction data, the crystallite size of LFP was determined based on the Scherrer equation.
Where τ is the mean size of crystalline domains, which is smaller or equal to grain size, λ is the X-ray wavelength, β is the full width at half the maximum (FWHM) peak intensity and θ is the Bragg or diffraction angle. In

Particle size and morphology of LFP/C composites
The morphology and microstructure of three LFP/C composites were analyzed by TEM. As shown in Figure 2a, b, and c, LFP particles distributed regularly in the carbon matrix. The conductive bridges were built to connect LFP particles. Hence, it could effectively enhance the electronic conductivity of LFP/C composites. Additionally, the structure of KB and CNT based LFP composite is highly porous-structure than LFP/graphene which is helpful for electrolyte penetration and facilitate the fast ion exchange or even fast discharge process. In-depth examination (Figure 2 c), Gr sheets are folded together and supposed to form layered graphite structure as evident analyzing from XRD diagram. Consequently, the outstanding electronic conductivity of Gr was completely failed, and it's considered to be a value of graphite. Additionally, the graphite formation also diminishes the surface area of LFP preventing lithium ions moving into the bulk material under charge-discharge conditions, thus resulting in high polarization and low capacity. Through TEM images, the platelets LFP particles (Figure 2d, e) on the rods shape (Figure 2 f) could be seen. Compared to the values of crystallite size from Scherrer equation (Equation (1)), the observed LFP grain size was larger and in range of 80-200 nm by using the scale-bar of TEM instrument. The larger size of LFP composites was explained that the carbon coated on LFP particle to prevent the agglomeration was not entirely induced, thus, leading to a wide size distribution of LFP.

Lithium-ion diffusion kinetics
The CV measurement of three LFP/C composites was examined at various scan rates in the range of 0.01 -0.20 mV.s −1 (Figure 3 a-c). From the CV data obtained with a scanning rate of 0.01 mV/s, the difference between the anodic and cathodic peak voltages (hysteresis) has been found to be~0.15 V for sample LFP/CNT, whereas slightly higher values of~0.20 V and~0.30 V for samples LFP/KB and LFP/Gr. These results are consistent with slower kinetics and larger over-potentials exhibited by LFP/Gr. As the scan rate increased, the charge transfer rate became faster following the significant increase of lithium ions concentration on the surface of electrode material. Thus the intensity of redox peak significantly increased. For small scan rates, the anodic and cathodic peak currents vary linearly with the square root of the scan rate, indicating that the Li-ion insertion/extraction in LiFePO4 is a diffuse controlled process 25 . Figure 3d shows such plots for anodic currents in composites samples. According to Randles-Sevcik equation (Equation (1) ), i p versus v 1/2 is linear and the diffusion coefficient can be estimated from the slope of this line.
Where i p is the peak current (A), n is number of electrons involved in the reaction of redox couple (for Li + it is 1), ν is the rate at which potential is swept (V.s −1 ), A is the effective working electrode area (cm 2 ), D is the diffusion coefficient of electroactive species Li + (cm 2 .s −1 ) and C * is initial concentration of Li in LiFePO 4 material (defined as the ratio bulk density to molar mass, for which the corresponding Li concentration C* should be 0.0228 mol/cm 3 ). The lithium ion diffusion coefficients of LFP/KB and LFP/CNT samples were estimated to be 3.79×10 −11 and 2.16×10 −10 cm 2 .s −1 , respectively, which is 4-6 times higher order of magnitude than that of pristine LFP (10 −14 -10 −16 cm 2 .s −1 ) as early reported by Prosini et al. 5 . In contrast, lithium ion diffusion coefficient of LFP/Gr composite is 1.96×10 −16 cm 2 .s −1 , which is quite like the original LFP.

DISCUSSION
As expected, coating LFP particles with conductive carbon are the predominant way to enhance electrical conduction within an insulating LFP cathode.
In the solvothermal process, the nanosized particles could be effectively controlled by the solvothermal method in modifying different parameters (pH, chemical agents…). While using the carbon matrix, the LFP particles were physically dispersed into carbon matrix and well-connected with carbon particles or carbon tubes, as obviously seen in case of LFP/KB or LFP/CNT (Figure 2 d-e). Nevertheless, the synthesis pathway isn't effective to disperse LFP carbon into the graphene sheets due to high Val-der-Waals force between graphene layers even with the vigorous stirring. As a result, the graphene layers were stacked together and destroyed the superior electronic conductor (Figure 2

f).
Cyclic Voltammetry (CV) is a common technique for studying the properties of an electrochemical system. Within the scanning potential range, a current peak occurs at a certain potential indicating an occurrence of an electrode reaction. In the case of LiFePO 4 electrode, the two peaks are expectedly observed around 3.45 V vs. Li + /Li, correspond to the two-phase charge-discharge reaction of Fe 3+ /Fe 2+ redox couple. By analyzing the resultant current versus potential profiles, information on the kinetics and thermodynamics of the electrode reaction can be obtained 24 . CV measurements confirmed the significant enhancement of electrochemical kinetic of LFP/KB and LFP/CNT composite electrodes through the calculation of lithium ion diffusion. Due to expected role of tailored carbon materials (carbon KB, carbon nanotubes), during the reduction reaction, electrons transfer rapidly into Fe d-orbitals of LFP compound and simultaneously formed electrostatic forces attracting lithium ions to diffuse along the electrode surface and then go into host matrix for charge neutralization. The composite LFP/Gr didn't get any benefit from the superior electronic conductivity of graphene due to stacked layers; indeed, the electrochemical performance was nearly penalized. Therefore, pre-treatment or modification of graphene surface should be considered to eliminate restacking phenomenon of Gr sheets before mixing them into desired composites.

CONCLUSIONS
LFP/KB, LFP/CNT and LFP/Gr composites were successfully obtained by physical mixing assisted solvothermal process. LFP particles distributed regularly in carbon matrix with the grain size ranging from 80-200 nm. Regarding the electrochemical performance, the synthesized LFP composites showed excellent ion diffusion coefficient compared to pristine LFP, except for LFP/Gr. Additionally, based on the calculated values, KB and CNT based LFP composite almost had a good impact on the electrochemical kinetics of lithium diffusion. Concerning the commercialization, KB could be a potential candidate for large-scale production due to low-cost, stable and high electrochemical performance.