A

mohacud©

K

1

oibi1tImcyictafasd

0d

Available online at www.sciencedirect.com

Electrochimica Acta 53 (2008) 5071–5075

Kinetic behavior of LiFePO4/C cathode material for lithium-ion batteries

Fei Gao ∗, Zhiyuan TangSchool of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, PR China

Received 8 July 2007; received in revised form 19 October 2007; accepted 29 October 2007Available online 7 November 2007

bstract

The composite cathode materials of LiFePO4/C were synthesized by spray-drying and post-annealing method. The crystalline structure andorphology of products were characterized by X-ray diffraction (XRD) and scanning electron microscopy (SEM). The charge–discharge kinetics

f LiFePO4/C electrode was investigated using electrochemical impedance spectroscopy (EIS). The results show that the increment of the resistanceas a close relation to the appearance of the FePO4 phase during charge–discharge course, and that the ohmic resistance, charge transfer resistancend lithium-ion diffusion coefficients of the LiFePO4/C electrode do not change much by extended cycling tests, implying a relatively superior

yclability of the battery. The effect of cell temperature on the electrochemical reaction behaviors of LiFePO4/C electrode was also investigatedsing the EIS. It is confirmed that the effect of the cell temperature on the impedance results mainly from the enhancement of the lithium-ioniffusion at elevated temperatures.

2008 Elsevier Ltd. All rights reserved.

ficien

bctpCc

ptolE

2

F

eywords: LiFePO4; Electrochemical impedance spectroscopy; Diffusion coef

. Introduction

Since the first report on the electrochemical propertiesf LiFePO4 by Goodenough and co-workers [1], LiFePO4s emerging as a promising cathode material for lithium-ionatteries because of low cost and environmental compatibil-ty. In addition, LiFePO4 has a large theoretical capacity of70 mAh g−1, a flat discharge potential of 3.4 V versus Li/Li+,he good cycle stability, and the excellent thermal stability.n spite of these attractive features, LiFePO4 requires furtherodifications to overcome limitations such as poor electronic

onductivity and slow lithium-ion diffusion. In the past fewears, conductivity is usually enhanced appreciably by coat-ng LiFePO4 particles with electrically conductive materials likearbon [2], metal and metal oxides [3–7]. Minimizing the par-icle size of LiFePO4 material to provide more surface area haslso been investigated as a means to enhance lithium-ion dif-usion [8–10]. The prepared LiFePO4 powders by the above

pproaches have achieved the good electrochemical properties toome extend. However, up to now, it has been difficult to clearlyepict the kinetics of LiFePO4 materials used for lithium-ion

∗ Corresponding author. Tel.: +86 22 27401684; fax: +86 22 27401684.E-mail address: gaofei220@hotmail.com (F. Gao).

C(ssatr

013-4686/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.oi:10.1016/j.electacta.2007.10.069

t; Rate performance

atteries, even though lot of efforts related to this field have beenarried out [11–15]. Several factors including particle size, elec-ronic/ionic conductivity, and phase transition kinetics have beenroposed to affect the electrochemical performance of LiFePO4.onsequently, understanding the kinetic behavior of LiFePO4 isertainly fundamental to optimize the properties of LiFePO4.

Electrochemical impedance spectroscopy (EIS) is a veryowerful technology to determine the rate of individual elec-rode kinetic steps. In this paper, we first synthesized compositesf LiFePO4/C, and then investigated dependence of kinetics andithium-ion diffusion coefficient of the products by means of theIS.

. Experimental

The precursors of CH3COOLi·2H2O (99.9% A.R.),eC2O4·2H2O (98% A.R.), (NH4)2HPO4 (99.5% A.R.) and6H12O6 (glucose A.R.) were mixed in stoichiometric amounts

molar ratio of Li:Fe:P:C6H12O6 = 1:1:1:0.1). First, the precur-ors were ground for 8 h by wet ball-milling in the aqueous

olution mixed with polyvinyl alcohol (PVA) (mass ratio of PVAnd water being 3:100) to decrease the particle size of the reac-ants and ensure intimate and homogeneous mixing. Then theesulting stable suspension was dried to form a mixed dry pre-

5 mica Acta 53 (2008) 5071–5075

cssaccowt

s2rust(

ttmc(Lfiaaaoc

pltuwt1

3

3

acgptbou0(F(

sLrpat

3

vcrttfassociated with lithium-ion diffusion in the LiFePO4 particles.

A simplified equivalent circuit model (Fig. 4) was constructedto analyze the impedance spectra. A constant phase elementCPE was placed to represent the double layer capacitance and

072 F. Gao, Z. Tang / Electrochi

ursor via a spray-dryer. The suspension was atomized via aprinkler at an air pressure of 0.2 MPa, and was dried in thepray-dryer by hot air. The inlet air temperature was 220 ◦C,nd the exit air temperature was 110 ◦C. The as-prepared pre-ursor powders with pellets ranging from 50 to 100 �m, werealcined at 300 ◦C for 5 h, and then to the elevated temperaturef 600 ◦C for 10 h in nitrogen atmosphere. Finally, the samplesere reground for 8 h by the ball-milling after cooled to room

emperature and the final products of LiFePO4/C were obtained.The structures and phases of the LiFePO4 and LiFePO4/C

amples were identified by X-ray diffraction (XRD, RU-00B/RINT, Rigaku, Rotaflex using monochromatic Cu K�adiation). The morphology of the samples was recorded bysing a scanning electron microscopy (SEM, Hitachi S800). Thepecific surface area of the powder particles was measured byhe low temperature nitrogen adsorption and desorption methodGemini V-2380).

Thick film electrodes were manufactured for electrochemicalesting of the samples by casting on an aluminum current collec-or a N-methylpyrrolidone (NMP) slurry of the LiFePO4 active

aterial (82 wt.%) mixed with a carbon black (Super P-Timcal)onductive additive (10 wt.%) and a polyvinylidene fluorideKyner2801-Eif Atochem) binder (8 wt.%). These film-typeiFePO4 electrodes of 120 �m were assembled in a nitrogen-lled glove box using mesocarbon microbeads (MCMB) anodes counter electrode. The electrolyte was 1 mol L−1 LiPF6 inmixture of ethylene carbonate (EC) and dimethyl carbon-

te (DMC) (1:1, v/v). The charge and discharge characteristicsf the batteries were examined between 2.0 and 4.2 V using aylindrical battery (18650 type).

The electrochemical impedance spectroscopy (EIS) waserformed using a voltalab system with VM4 (Radiometer Ana-ytical SAS. PGZ301). These tests were performed using ahree-electrode cell. The thin film electrodes of LiFePO4 weresed as working electrodes. The counter and reference electrodesere lithium foil. The sinusoidal excitation voltage applied to

he cells was 10 mV rms with a frequency range of between00 kHz and 10 mHz.

. Results and discussion

.1. Physical properties

The XRD pattern (Fig. 1) of the sample LiFePO4/C showsn absence of parasitic peaks, and all the diffraction peaksan be indexed on the orthorhombic structure with the spaceroup Pnmb, being in agreement with a well-crystalline singlehase LiFePO4 according to the PDF2 (83-2092). The lat-ice parameters of well-crystallized LiFePO4 are a = 1.0313 nm,= 0.6001 nm and c = 0.4691 nm. The crystallite size (D) basedn the (1 3 1) diffraction peaks for the sample was calculatedsing the Scherer formula: D = kλ/β cos θ, where k is 0.9, λ is

.154 nm and β is the full width at the half maximum lengthFWHM) of the diffraction peak on a 2θ scale, respectively.rom the Scherer equation, the D1 3 1 value, calculated from the1 3 1) peak width, is 65 nm.

Fig. 1. XRD patterns of the carbon-coated LiFePO4.

Fig. 2 is the SEM image of LiFePO4/C composites. Near-pherical particles with minimal agglomeration are obtained foriFePO4/C. The particle size (estimated from SEM analysis)

anges from 100 to 300 nm. The specific surface area of theowder particles was measured by the low temperature nitrogendsorption and desorption method. The specific surface area ofhe sample is 15.8 m2 g−1.

.2. Electrochemical impedance spectroscopy

Fig. 3a shows the Nyquist plot in the open circuit state witharying Li contents, Li1 − δFePO4 at room temperature. An inter-ept at the Z′ axis in high frequency corresponded to the ohmicesistance (Re), which represented the resistance of the elec-rolyte. The semicircle in the middle frequency range indicatedhe charge transfer resistance (Rct). The inclined line in the lowrequency represented the Warburg impedance (Zw), which was

Fig. 2. The SEM image of the carbon-coated LiFePO4.

F. Gao, Z. Tang / Electrochimica

Fig. 3. (a) Impedance spectra for Li1 − δFePO4 at various lithium content and(b) the relationship between Z′′ and square root of frequency (ω−1/2) in thelow-frequency region for Li1 − δFePO4 at various lithium content.

pcterFt

TEm

L

δ

δ

δ

δ

0e

D

wttσ

Z

of(a5ta

Lmdischarged state were carried out at room temperature. Fig. 5bshows the relationship between Z′′ and square root of frequency(ω−1/2) in the low-frequency region. Table 2 shows the param-eters of the equivalent circuit for the carbon-coated LiFePO4.

Fig. 4. Equivalent circuit used for fitting the experimental EIS data.

assivation film capacitance. The parameters of the equivalentircuit by computer simulations are shown in Table 1. Obviously,he charge transfer resistance (Rct) increases with lithium ions

xtracting from the structure. It seems that the increment of theesistance would have a close relation to the appearance of theePO4 phase. The charge transfer resistance is the highest when

he FePO4 phase is the major phase at charged state.

able 1lectrode kinetic parameters obtained from equivalent circuit fitting of experi-ental data for Li1 − δFePO4 at various lithium content

i1 − δFePO4 Re (�) Rct (�) D (m2 s−1)

= 0 7.08 87.54 1.27 × 10−16

= 0.3 7.63 109.72 5.35 × 10−17

= 0.6 7.44 135.36 3.65 × 10−17

= 0.9 8.10 161.96 8.82 × 10−18

Fci

Acta 53 (2008) 5071–5075 5073

The lithium-ion diffusion coefficients (D) at δ = 0, 0.3, 0.6 and.9 in Li1 − δFePO4, were calculated according to the followingquation:

= R2T 2

2n4F4C2σ2 (1)

here R is the gas constant, T is the absolute temperature, n ishe number of electrons per molecule during oxidization, F ishe Faraday constant, C is the concentration of lithium ion, and

is the Warburg factor which has relationship with Z′′:′′ = σω1/2 (2)

Fig. 3b shows the relationship between Z′′ and square rootf frequency (ω−1/2) in the low-frequency region. The dif-usion coefficient of lithium ion is calculated based on Eqs.1) and (2). The calculated lithium-ion diffusion coefficientst δ = 0, 0.3, 0.6 and 0.9 in Li1 − δFePO4 are 1.27 × 10−16,.35 × 10−17, 3.65 × 10−17, and 8.82 × 10−18 m2 s−1, respec-ively. The lithium-ion diffusion coefficients decrease markedlylong with the formation of FePO4.

Fig. 5a shows Nyquist plots obtained from the carbon-coatediFePO4 electrodes after 5th, 20th, and 50th cycle tests. The EISeasurements of the carbon-coated LiFePO4 electrodes at fully

ig. 5. (a) Impedance spectra for LiFePO4/C electrodes after 5th, 20th, and 50thycle and (b) the relationship between Z′′ and square root of frequency (ω−1/2)n the low-frequency region.

5074 F. Gao, Z. Tang / Electrochimica Acta 53 (2008) 5071–5075

Table 2Electrode kinetic parameters obtained from equivalent circuit fitting of experi-mental data for LiFePO4/C electrodes after 5th, 20th, and 50th cycle

Cycles Re (�) Rct (�) D (×10−17 m2 s−1)

5th 7.56 75.54 5.9525

ItsilOwTstc

otttLtt

FL(

Table 3Electrode kinetic parameters obtained from equivalent circuit fitting of experi-mental data for LiFePO4/C, measured at −20, 0, 20, 40 and 60 ◦C

Temperature (◦C) Re (�) Rct (�) D (m2 s−1)

−20 23.84 318.24 1.01 × 10−18

0 16.73 154.56 2.92 × 10−18

−17

e

i

wAtiit2toLp

0th 7.84 81.25 6.980th 8.12 90.09 5.44

n particular, Re, Rct and D of the carbon-coated LiFePO4 elec-rode do not change much by extended cycling tests. This resultuggests that the kinetic behavior of the electrode is stable andmplies a relatively superior cyclability of the battery. The simi-ar EIS of the carbon-coated LiFePO4 appears due to two factors.ne is the reduced size of carbon-coated LiFePO4 particles,hich shortens the distance of the diffusion path for lithium ion.he other is the improved electrical conductivity of the cathodeince the carbon in the LiFePO4 provides good electrical con-acts between the LiFePO4 particles and the conductor in theathode composite.

The influence of the electrode temperature on the impedancef the LiFePO4 redox reaction was also investigated withhe EIS. Fig. 6a shows the Nyquist plot of LiFePO4 on theemperature of −20, 0, 20, 40 and 60 ◦C. Table 3 showshe parameters of the equivalent circuit for the carbon-coated

iFePO4. Obviously, the ohmic resistance (Re) and the charge

ransfer resistance (Rct) decreases with the elevated electrodeemperature.

ig. 6. (a) The influence of the cell temperature on the Nyquist plot ofiFePO4/C and (b) the relationship between Z′′ and square root of frequencyω−1/2) in the low-frequency region, measured at −20, 0, 20, 40 and 60 ◦C.

o

3

Ldcawdtccic

F

20 8.74 106.78 4.31 × 1040 5.08 70.93 7.61 × 10−17

60 3.65 51.22 2.70 × 10−16

The exchange current i0 was obtained from Rct by usingquation:

0 = RTA

nFRct(3)

here A is the real surface area of the LiFePO4 electrode.lthough the value of A is unobtainable, it is invariable for

he same electrode. Therefore, the temperature dependence of0 was evaluated by using the value of Rct. The values of i0ncrease about 6 times with the elevated temperature from −20o 60 ◦C, whereas the values of D increase substantially about70 times. These results indicate that the influence of the cellemperature on the lithium diffusion process is larger than thatf the charge transfer for the lithium extraction reaction of theiFePO4 electrode. It is confirmed that the effect of the cell tem-erature on the impedance results mainly from the enhancementf the lithium-ion diffusion at elevated temperatures.

.3. Electrochemical properties

The electrochemical rate performance and cyclability ofiFePO4 are shown in Figs. 7 and 8. With increasing currentensity, the discharge capacity is rapidly reduced. The initial dis-harge capacities are 134.6, 129.5 and 120.1 mAh g−1 at C/5, 1Cnd 5C rate, respectively. A decrease in initial discharge capacityith higher current density results from the intrinsic lithium-ioniffusion limitations of the material. Moreover, these capaci-ies seem to be stable upon cycling. After 50 cycles at different

urrent density, the capacity decrease is slight, even at the highurrent density of 5C, a discharge capacity of 114.8 mAh g−1

s obtained (95.6% of its initial value at 5C rate) with only aapacity fading of 0.08% per cycle.

ig. 7. Discharge curves of LiFePO4/C at (a) C/5, (b) 1C and (c) 5C rates.

F. Gao, Z. Tang / Electrochimica

Fc

4

pokTactLtpo

A

us2

R

[

[[

ig. 8. Cycle performance of lithium cells at room temperature with LiFePO4/Cathodes at (a) C/5, (b) 1C and (c) 5C rates.

. Conclusions

Pure and well-crystallized LiFePO4/C composites with smallarticle sizes, and homogeneous particle size distribution werebtained by spray-drying and post-annealing method. Theinetic behaviors of LiFePO4/C were measured by the EIS.he analysis shows that the increment of the resistance hasclose relation to the appearance of the FePO4 phase during

harge–discharge course. And that the ohmic resistance, chargeransfer resistance and lithium-ion diffusion coefficients of the

iFePO4/C electrode do not change much by extended cycling

ests. In addition, it is confirmed that the effect of the cell tem-erature on the impedance results mainly from the enhancementf the lithium-ion diffusion at elevated temperatures.

[[

[

Acta 53 (2008) 5071–5075 5075

cknowledgements

We acknowledge Great Power Battery Co. Ltd., for supplyings experimental equipments. This work was supported by thecience and technology research item of Guangzhou, China (no.007Z3-D0021).

eferences

[1] A.K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, J. Electrochem. Soc.144 (1997) 1188.

[2] C.S. Ho, I.C. Won, J. Ho, J. Power Sources 159 (2006) 1383.[3] G.X. Wang, S. Needham, J. Yao, J.Z. Wang, R.S. Liu, H.K. Liu, J. Power

Sources 159 (2006) 282.[4] T. Tsung-Hsien, Y. Mu-Rong, W. She-hung, C. Yi-Ping, Solid State Com-

mun. 142 (2007) 389.[5] K.S. Park, J.T. Son, H.T. Chung, S.J. Kim, Solid State Commun. 129 (2004)

311.[6] H. Liu, Q. Cao, L.J. Fu, C. Li, Y.P. Wu, Electrochem. Commun. 5 (2006)

1553.[7] C. Sung-Yoon, J.T. Bloking, C. Yet-Ming, Nat. Mater. 1 (2002) 123.[8] C. Daiwon, P.N. Kumta, J. Power Sources 163 (2007) 1064.[9] A. Singhal, G. Skandan, G. Amatucci, F. Badway, N. Ye, J. Power Sources

129 (2004) 38.10] A. Ait Salah, A. Mauger, C.M. Julien, F. Gendron, Mater. Sci. Eng. B 129

(2006) 232.11] J. Hong, C.S. Wang, U. Kasavajjula, J. Power Sources 162 (2006) 1289.12] M. Seung-Taek, K. Shinichi, H. Norimitsu, Y. Hitoshi, K. Naoaki, Elec-

trochim. Acta 49 (2004) 4213.

13] T. Masaya, T. Shin-ichi, T. Koji, S. Yoji, Solid State Ionics 148 (2002) 283.14] H. Liu, C. Li, H.P. Zhang, L.J. Fu, Y.P. Wu, H.Q. Wu, J. Power Sources

159 (2006) 717.15] P. Pier Paolo, L. Marida, Z. Daniela, P. Mauro, Solid State Ionics 148 (2002)

45.