Study on dynamics of structural transformation during charge/discharge of LiFePO4 cathode

(1)

Study on dynamics of structural transformation during

charge/discharge of LiFePO

4

cathode

Hao-Hsun Chang

a

, Chun-Chih Chang

a

, Hung-Chun Wu

b

, Mo-Hua Yang

b

,

Hwo-Shuenn Sheu

c

, Nae-Lih Wu

a,*

a_{Department of Chemical Engineering, National Taiwan University, Taipei 106, Taiwan, ROC}

b_{Materials Research Laboratories, Industrial Technology Research Institute, Chutung, Hsinchu 310, Taiwan, ROC} c_{National Synchrotron Radiation Research Center, Hsinchu 30077, Taiwan, ROC}

Received 24 October 2007; received in revised form 14 November 2007; accepted 15 December 2007 Available online 25 December 2007

Abstract

Structural transformation taking place during charge/discharge of the LiFePO4electrode in an organic Li-ion electrolyte has been studied by in situ synchrotron X-ray diffraction (XRD) concurrently with electrochemical analysis. The data reveal complex structural transformation patterns which result from significantly delayed structural transformation, even at low to moderate current rates (C/ 10 1 C). The extent of the deviation is affected by charge/discharge conditions as well as history, and strong deviation appears detri-mental to electrode cycle life. Furthermore, the two-phase characteristic of the Li+insertion/extraction reactions is found to persist even within one metastable crystal structure. The result suggests the presence of domain structures that are probably caused by the strong electron/ion site coulombic interaction as previously suggested.

Keywords: LiFePO4; Structural transformation; Synchrotron X-ray diﬀraction; Li-ion secondary battery; Cathode

1. Introduction

LiFePO4has drawn intensive attention in recent years

for its potential application as a cathode material for

high-power rechargeable Li-ion batteries [1–6]. Several

reports have demonstrated high-rate capability (>10 C) with high coulombic eﬃciency (>80%) and satisfactory

cycle life [7–9]. To date, the crystallographic properties

and phase–stoichiometry relation of the Li1DxFePO4

sys-tem have been thoroughly studied [10,11]. The crystal

structure of LiFePO4 can be described by space group

Pmna and is composed of corner-sharing FeO6octahedra

forming two-dimensional square lattice perpendicular to

the a-axis, edge-sharing LiO6 aligned along the b-axis,

and PO4groups connecting neighboring planes. The

struc-ture of the delithiated phase, FePO4, is described by the

same space group but with distinctly diﬀerent lattice

parameters. Yamada et al. [12] studied on the

room-tem-perature phase behavior of Li1DxFePO4 system with

0 6 Dx 6 1 by progressive chemical extraction of Li-ions

from LiFePO4, and their data showed that there does not

exist stable intermediate compound except for two small regions (<5.0 mol.% in Li content) of solid solutions toward the end compounds. (The small solid-solution ranges will hereafter be neglected for simplicity in

discus-sion.) Study by Delacourt et al.[13]on the phase behavior

by thermal annealing basically conﬁrmed the same phase

relation at temperatures below 50°C. Solid-solution over

entire Li stoichiometry range (0 6 Dx 6 1) was, on the

other hand, found to exist above 350°C.

It is noted that the phase–structure–stoichiometry rela-tions have so far been established mainly by non-electro-chemical methods. It is intriguing to know how well the structural transformation during electrochemical lithiation/

delithiation of LiFePO4 actually follows the established

*

Corresponding author. Tel.: +886 2 23627158; fax: +886 2 23623040. E-mail address:[email protected](N.-L. Wu).

www.elsevier.com/locate/elecom Electrochemistry Communications 10 (2008) 335–339

(2)

relations and how the deviation, if any, could aﬀect the per-formance of the LiFePO4electrode in practical applications.

To answer these questions, structural transformation during electrochemical lithiation/delithiation of the LiFe-PO4electrode was studied in this work by using in situ

syn-chrotron X-ray diﬀraction (XRD). The results revealed complex structural transformation dynamics and unique characteristics of the two-phase reaction nature of this elec-trochemical system.

2. Experimental

Commercial LiFePO4 powder (Phostech) is used as

received. The powder has a particle size of d50= 3.0 lm.

LiFePO4 electrodes were made of 85 wt.% LiFePO4,

5 wt.% graphitic ﬂakes (KS6, 3 lm, TIMCAL), 2% nano-sized carbon black (Super P, 40 nm, TIMCAL) and 8 wt.% organic binder (polyvinylidene diﬂuoride; Aldrich) with Al foil as the current collector. The resulting LiFePO4

electrode was assembled together with a Li-foil counter electrode to make either coﬀee-bag cells or CR2032 coin cells, which use electrolyte of 1 M LiPF6in a 1:2 v/v

mix-ture of ethylene carbonate (EC) and ethyl methyl carbonate (EMC; Mitsubishi Chemical). All the cells were assembled in a dry room where the dew point was maintained at

between40 and 45 °C.

In situ synchrotron XRD was conducted by using beam-line 01-C2 in National Synchrotron Radiation Research

Center, Taiwan, ROC, and an X-ray source of

0.10223 nm in wavelength was employed. XRD spectra were acquired during the charge/discharge tests carried out at 55°C.

3. Results and discussion

Fig. 1a shows the synchrotron XRD patterns of selected 2h range acquired during a discharge/charge cycle under 1 C rate with an intermittent rest period, while the

corre-sponding voltage curve is shown inFig. 1b. The structure

transformation can be monitored by, for instance, follow-ing the evolution of the (2 1 1) and (0 2 0) reﬂection peaks

of the end compounds, as marked in Fig. 1a. As shown,

the XRD pattern show only the FePO4 structure before

the 8th measurement during discharge, and then

trans-formed rapidly into that of the LiFePO4-structure during

the period between the 9th and 11th measurements, when

the test was switched to rest. The inset shown in Fig. 1a

plots the intensities of the (0 2 0) peaks of the LiFePO4

-and FePO4-structures, , conﬁrming the delayed FePO4

-to-LiFePO4transformation until near the end of discharge.

Similarly, during charge, the crystal structure of the elec-trode remained as the LiFePO4-structure until near the end

of charge, and then transformed abruptly to the FePO4

-structure upon reaching another rest period (Fig. 1a and

b). It is important to note that, both the discharge and charge voltage curves (Fig. 1b) exhibit a ﬂat-plateau region, where the XRD patterns show essentially only one

struc-ture. A plateau typically observed over the middle capacity region of an either charge or discharge voltage curve has been considered as a signature of the two-phase reaction nature, namely

LiFePO4$ e+ Liþ + FePO4

of the present electrochemical system[1,12,14].

Delayed structural transformation was also detected even at C/10 rate at room temperature (not shown). Under

such a condition, transformation from the LiFePO4

-struc-ture to the FePO4-structure did not commence until70%

Fig. 1. XRD and electrochemical data for charge/discharge test of Cell #1 at 1 C rate at 55°C with an intermittent rest period. (a) Synchrotron XRD patterns. The subscripts F and L indicate, respectively, reﬂections from the FePO4- and LiFePO4-structures. The bold curves mark the onset of

diﬀerent stages (C for charge; D, discharge; R, rest) within a test cycle. The inset shows the XRD peak intensities of LiFePO4(0 2 0) (j) and FePO4

(0 2 0) (h) versus the measurement number. (b) Voltage curve. The numbers shown along the right-axis in (a) and within the ﬁgure in (b) index the XRD measurement number.

(3)

of Li-ion was extracted. Again, the characteristic of the voltage ﬂat-plateau was also observed during the ‘‘struc-ture-frozen” period. It is worth noting that the ‘‘frozen” structures are diﬀerent from structures of solid-solutions. For the latter, the lattice parameters change progressively with the stoichiometry, while the lattice parameters simply remain unchanged for the former.

Fig. 2shows the result of the ﬁrst cycle of another cell (Cell #2) which was tested without intermittent rest. There

was essentially no structure change during charge (Fig. 2a),

similar to the charge period shown inFig. 1a. However, as

there was no intermittent rest in this case, structure trans-formation from the LiFePO4-structure to the FePO4

-struc-ture actually took place during discharge (lithiation)! The

XRD peak intensities of the FePO4-structure reach the

maximum sometime during the discharge period. In spite of such a complex structural transformation process, the voltage curves showed the same ﬂat-plateau characteristic as seen inFig. 1b.

It was, nevertheless, found that, after the non-interrupt-ing ﬁrst cycle, the structural transformation started to pro-ceed ‘‘normally” in the subsequent cycles either with or without intermittent resting between charge and discharge (Fig. 2b). For a ‘‘normal” structure transformation pro-cess, the extent of the transformation, as indicated by the XRD peak intensities, is correlated almost linearly with the extent of Li-ion extraction or insertion.

The results described above point out two important conclusions. Firstly, structural transformation during

charge/discharge of LiFePO4more than often proceeds in

ways that deviate far from the equilibrium phase–struc-ture–stoichiometry relations. The extent of the deviation depends not only on the test condition but also on the charge/discharge history that the electrode has previously experienced. Secondly, the voltage curves consistently show the characteristics of the two-phase mechanism, irrespec-tive of the structural transformation pattern and crystal structure.

In the cases of the delayed structural transformation, it is noted that transformation has been greatly

acceler-ated when there is strong disturbance during the

charge/discharge process, such as the sudden stop for

the intermittent rest period in Fig. 1 and the lithiation/

delithiation reversal in Fig. 2a. This may explain the fact that the frozen-structure phenomenon has never been detected in the previous ex situ XRD or TEM studies. From the viewpoint of nucleation theory, these observa-tions may suggest that the delay in the transformation result from slow nucleation kinetics of the resulting phase, and the process disturbances tend to stimulate the formation of such nuclei. This assumption may also explain why the transformation process is history-depen-dent, as diﬀerent test protocols could result in particular microstructures that serve as favorable nucleation sites.

An example is when the LiFePO4 electrode experienced

the overlapped structural transformation processes during

discharge shown in Fig. 2a. The overlapped

transforma-tion process might have introduced microstructures that facilitate nucleation in the subsequent cycles, which show

exclusively ‘‘normal” transformation process (Fig. 2b).

The results also suggest that a few initial non-stopping cycles at moderate to high rates may be a good exercise in practical applications to avoid delayed structure trans-formation, which, as shown below, could otherwise have adverse eﬀects on the cycling performance of the electrode.

Fig. 2. XRD data for Cell #2 at 1 C rate at 55°C: (a) the ﬁrst cycle without intermittent rest; (b) the second and third cycles with intermittent rest. The subscripts F and L indicate, respectively, reﬂections from the FePO4- and LiFePO4-structures. The bold curves mark the onset of

(4)

The eﬀect of the non-equilibrium structural

transfor-mation on the performance of LiFePO4electrode was

pre-liminarily assessed by conducting prolonged cycling tests on two coin cells which have the same electrode composi-tions as in the coﬀee-bag cells for the synchrotron studies.

Cell #3 was cycled at 1 C rate at 55°C always with an

intermittent rest period between charge and discharge, while Cell #4 was cycled under the same condition but without any intermittent rest period. According to the XRD data shown above, it is anticipated that structural transformation occurs abruptly only near the end of every charge or discharge period for Cell #3. On the other hand, transformation following closely the equilibrium phase relation is expected from the second cycle up for

Cell #4.Fig. 3summarizes the discharge capacities versus

cycle number for these two cells. Cell #3 was found to exhibit much worse performance, including lower dis-charge capacities during initial cycles and faster capacity fading, than Cell #4. The results suggest that structural transformation that strongly deviates from the equilib-rium phase–stoichiometry relation has profound adverse eﬀects on electrode performance. Systematic research is currently underway for understanding the interplay among the structural transformation process, initial dis-charge capacity and the fading mechanism.

The presence of the voltage plateau suggest that some sort of two-‘‘phase” relation always be maintained within the crystallite, irrespective of the crystal structure. It may be possible to understand this phenomenon based on the theoretical study by Zhou et al.[15]on the phase stability of LiFePO4/FePO4. By using ﬁrst-principle calculation, it

was demonstrated in their study that electron localization and charge ordering at Fe-ion sites play a very important role in the phase relation of LiFePO4–FePO4; strong

elec-tron/site coulombic interaction becomes a key factor that forbids the formation of continuous solid solution in the binary system. To explain the present observations, it is

proposed that, owing to the strong localized charge interac-tion, Li-ion extraction do not occur randomly even within the frozen singled-structured crystallites, but rather two-‘‘phase” interface be maintained at the boundaries between ‘‘Li-rich” and ‘‘Li-lean domains which are, respectively, stabilized and energetically diﬀerentiated mainly by the coulombic interaction energy diﬀerence.

Recently studies by Chen et al. [16] and Laﬀont et al.

[17]using high-resolution electron microscopy and energy

loss spectroscopy provide insight into the LiFePO4–

FePO4phase transition at the microscopic level. The

lith-iation/delithiation was carried out either with chemical means or with electrochemical charge/discharge at very low rate (C/50). The data in both studies suggested that, during lithiation/delithiation, Li-ions diﬀuse in a direction parallel to the b-axis of the olivine crystallites, while the phase boundary progresses parallel to the a-axis. The data of Laﬀont et al.[17]further indicate the presence of sharp

boundaries between LiFePO4 and FePO4-structure

domains within the crystallites. Our presumption of the two-phase equilibrium maintained at the boundaries between the compositional domains is to some extent in line with the interface model of Laﬀont et al., only that the interface within one lattice structure in the presence study is merely metastable.

In summary, in situ synchrotron XRD study has revealed the complex nature of the transformation

dynam-ics during charge/discharge of LiFePO4 electrode, which

cannot be predicted from current understanding of the equilibrium phase relations. The transformation dynamics, in turn, could have a signiﬁcant impact on the cycling per-formance of the electrode. Comparison between the XRD and electrochemical data have indicated that the two-phase reaction nature of the electrochemical lithiation/delithia-tion processes is not governed by crystallographic consider-ation but more likely by localized electron/ion site coulombic interaction.

Acknowledgments

This work is partially supported by Industrial Technol-ogy Research Institute and by National Taiwan University (contract number 95R0066-BE04-01).

References

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

[2] J. Chen, M.S. Whittingham, Electrochem. Commun. 8 (2006) 855. [3] H. Huang, S.-C. Yin, L.F. Nazar, Electrochem. Solid-State Lett. 4

(2001) A170.

[4] S.-Y. Chung, J.T. Bloking, Y.-M. Chiang, Nature Mater. 1 (2002) 123.

[5] H.C. Wu, C.Y. Su, D.T. Shieh, M.H. Yang, N.L. Wu, Electrochem. Solid-State Lett. 9 (2006) A537.

[6] H.H. Chang, C.C. Chang, H.C. Wu, Z.Z. Guo, M.H. Yang, Y.P. Chiang, H.S. Sheu, N.L. Wu, J. Power Sources 158 (2006) 550. [7] A.S. Gozdz, A.C. Chu, Y.M. Chiang, G.N. Riley, Jr., US patent

application #2005/0233219A1. Fig. 3. Discharge capacity versus cycle number. Cell #3: coin-cell cycled

with intermittent rest between charge and discharge; Cell #4: coin-cell cycled without intermittent rest (1 C rate, 55°C).

(5)

[8] H.C. Wu, C.Y. Su, P.H. Hsieh, B.M. Lin, M.Y. Yang, N.L. Wu, Taiwan patent #I270994; US patent application #US2007/ 0154809A1.

[9] B. Kang, G. Cedar, International Meeting on Lithium Batteries, Biarritz, France, 18–23 June, 2006, paper #162.

[10] A.S. Andersson, J.O. Thomas, J. Power Sources 97 (2001) 498. [11] C. Delacourt, J. Rodriguez-Carvajal, B. Schmitt, J.M. Tarascon, C.

Masquelier, Solid State Sci. 7 (2005) 1506.

[12] A. Yamada, H. Koizumi, N. Sonoyama, R. Kanno, Electrochem. Solid-State Lett. 8 (2005) A409.

[13] C. Delacourt, P. Poizot, J.M. Tarascon, C. Masquelier, Nature Mater. 4 (2005) 254.

[14] N. Meethong, H.Y.S. Huang, W.C. Carter, Y.M. Chiang, Electro-chem. Solid-State Lett. 10 (2007) A134.

[15] F. Zhou, C.A. Marianetti, M. Cococcioni, D. Morgan, G. Ceder, Phys. Rev. B 69 (2004) 201101.

[16] G. Chen, X. Song, T.J. Richardson, Electrochem. Solid-State Lett. 9 (2006) A295.

[17] L. Laﬀont, C. Delacourt, P. Gibot, M.Y. Wu, P. Kooyman, C. Masquelier, J.M. Tarascon, Chem. Mater. 18 (2006) 5520.