3732
J. Phys. Chem. 1995,99, 3732-3735
Electrochemical Insertion of Lithium into Pyrite from Nonaqueous Electrolytes at Room Temperature: An in Situ Fe K-Edge X-ray Absorption Fine Structure Study Donald A. Tryk, Sunghyun Kim, Yinhg Hu, Wenning Xing, and Daniel A. Scherson* Department of Chemistry, Case Western Reserve Universiv, Cleveland, Ohio 44106-7078
Mark R. Antonio Argonne National Laboratory, Chemistry Division, Argonne, Illinois 60439-4831
Violeta Z. Leger and George E. Blomgren Technology Laboratory, Eveready Battery Company, Inc., Westlake, Ohio 44145 Received: November 10, 199P
The effects of lithium ion insertion on the structural and electronic properties of FeSz (pyrite) have been examined in situ by Fe K-edge X-ray absorption fine stntcture (XAFS) using electrodes and electrolytes similar to those found in conventional, ambient-temperature,primary LilFeSz batteries. A substantial reduction in the amplitudes of the Fe-S and Fe-Fe backscattering was observed as the amount of intercalated lithium in the FeSz lattice was increased from 0 to 2 Li+ equivalents, (Li+),. After the insertion of 2 (Li+),, the second-shell Fe-S interaction and the distant Fe-Fe interactions were no longer discernable in the Fourier transform (FT) data. Curve-fitting analysis of the k3x(k) extended X-ray absorption fine structure for this latter material yielded an average Fe-S distance, d(Fe-S) = 2.31 +/- 0.02 A, which is about 0.05 8, longer than d(Fe-S) in crystalline pyrite. In addition, the X-ray absorption near-edge structure revealed a rounding of the otherwise highly structured edge of FeS2 as the amount of inserted lithium was increased. This behavior is consistent with the formation of Fel,S and thus supports the assignment made on the basis of in situ 57Fe Mossbauer effect spectroscopy of the same system.
Introduction
A better understanding of the changes in the structure and electronic properties of pyrite (FeSz) and its corresponding lithiated derivatives (denoted generically as Li,FeSz, 0 5 x 5 2) induced by the chemical and/or electrochemical insertion and removal of lithium ions is important to the further development of Li-FeSz high-energy, secondary batteries. Considerable insight into certain aspects of this phenomenon has been gained on the basis of recent studies involving a variety of spectroscopic and structural techniques, including infrared,l "Fe Mossbauer effect spectroscopy (MES),2-6 X-ray diffracti~n~-~,' and extended X-ray absorption fine structure (EXAFS)? with much of the emphasis being focused on the identification of the reaction products of these reactions. This work will present in situ X-ray absorption fine structure (XAFS) spectra of FeSz before and after electrochemical insertion of 1 and 2 lithium ion equivalents, (Li+),. Unlike other studies reported in the literature, these measurements were conducted in a specially-designed spectraelectrochemical cell using components and electrolytes very similar to those of conventional, nonaqueous-solvent, ambient-temperature,primary Li-FeSz batteries. Hence, the results obtained are directly relevant to processes occurring in practical devices. Experimental Sectinn
Spectroelectdemical Cell. The cell used in these experiments consists of a thin lithium anode, a separator, and a thin FeSz cast electrode arranged in a sandwich-type configuration (see Figure 1). using a solution of lithium trifluoromethanesulfonate (triflate) in a mixture of dimethoxyethaneldioxolane Abstract published in Advance ACS Abstracts, February 15, 1995.
Figure 1. Schematic diagram of the cell used for in situ XAFS experiments of lithium intercalation into pyrite from nonaqueous solvents.
as the electrolyte. This geometry provides optimal conditions for achieving a highly uniform current distribution so that the lithium incorporation (or cell discharge) will occur bomogeueously over the entire FeSz electrode. In addition, the thickness of the complete assembly, which includes a thermally-sealed, aluminized, polymer-based casing to isolate the cell components from air, was sufficiently small so as not to attenuate the X-ray flux appreciably. This made it possible to acquire in situ XAFS spectra in the transmission mode in a rather straightforward fashion. However, the thickness of the pyrite cast electrode in terms of absorption lengths (px= 4.6) was much larger than the ideal absorber thickness (& ca. 1) for Fe K-edge XAFS. Consequently, thickness effect artifacts? such as amplitude distortions,' were unavoidable in these initial measurements; hence, caution must be exercised when comparing the data
0022-3654/95/2099-3732$09.00/0 0 1995 American Chemical Society
J. Phys. Chem., Vol. 99, No. 11, 1995 3733
Electrochemical Insertion of Lithium into Pyrite presented herein with that obtained for the same (or similar) materials involving different values of g. Nevertheless, the fact that the absorber thickness (total amount of Fe in the beam) was kept constant throughout the course of experimentation facilitates the direct comparison of metrical parameters of pristine FeSz (in the electrode) with those for the corresponding lithiated materials generated in situ. Electrochemical Measurements. In situ X A F S spectra were collected for cells which had been discharged to depths equivalent to the insertion of 1 and 2 moles of lithium per mole of FeS2. During discharge, the cells were compressed with a C-clamp (which was removed prior to spectral acquisition) to decrease both electronic (by promoting particle-particle contact) and ionic contributions to the IR drop. Throughout this procedure, the electrodes are always in contact with the electrolyte and current collectors; hence, the integrity of the cell, and therefore the in situ character of the measurements, was never compromised. The discharge was effected using a Pine potentiostat working in the galvanostatic mode while monitoring the cell potential. The current densities involved in the discharge were small, on the order of 0.5-2 mA/cm2, to avoid high electrode polarization. As a result, periods of over 2 hours were required to complete each one-electron discharge step (8.89 mA h capacity; 1-4 mA current). All of the cells were assembled a few days before shipment to the synchrotron facilities and discharged immediately prior to the XAFS experiments. XAFS Spectral Acquisition and Analysis. XAFS data were acquired at the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory (beamline X18B, current 110220 mA, 2.5 GeV, monochromator crystal Si(220), step size 0.77 eV ( X A N E S ) , 0.05 A-1 (EXAFS)) and at the Stanford Synchrotron Radiation Laboratory (SSRL) (beamline VI-2, current 20-40 mA, monochromator crystal Si( 11l), step size 0.23 eV (XANES), 0.05 A-1 (EXAFS). Harmonic rejection was achieved by detuning the primary beam by 50% of its original intensity. The data were analyzed using XFF'AKG with phases and amplitudes obtained from FEW Version 3.25.1°
2.2
I
I
I
I
2.0
I
1
EXAFS
Off
0.8
>
1
1
I
1
0
10
20
30
40
Time, h
:::i
'
1.8
i
1.o 0.8
1
1
1
1
0
20
40
60
1 80
Charge, C Figure 2. Cell discharge curves as a function of time (top) and charge (bottom). The precise times at which the XAFS spectra were collected are indicated in the top panel.
Results and Discussion The cell discharge curves, both as a function of time and charge, are shown in the top and bottom panels in Figure 2, respectively. Also indicated in this figure are the precise times at which the XAFS spectra were collected. During the insertion of the 2 (Li+)q (32.2 C/electron), the cell voltage dropped by only -0.4 V, reaching -1.0 V. The in situ k3-weighted EXAFS spectra, obtained after zero, one, and two electron discharge steps (see curves a-c in Figure 3, respectively), reveal a gradual loss in the fine structure as the amount of lithium in the FeSz is increased. These changes can be better visualized in the corresponding phase-shiftuncorrected Fourier transforms (FT) shown in curves a-c of Figure 4. As indicated, the FeS2 electrode before lithium insertion (curve a) displays features characteristic of pyrite, including pronounced peaks at 1.8 and 2.9 A, attributable to the fiist two Fe-S shells (both with coordination number N = 6) and also at ca. 3.4, 5, and 6 A, assigned to the first three Fe-Fe shells in the lattice ( N = 12, 6, and 24, respectively). After the insertion of 1 and 2 (Li+),, the intensities of the shells with f > 2.5 A decreased monotonically compared to that of the major Fe-S peak. A curve-fitting analysis of the Fourier-filtered Fe-S backscattering k3x(k)for the FeS2 electrode before and after insertion of 1 and 2 (Li+)eq was performed with Fe-S phase and amplitude functions calculated using the FEFF formalism
-101 2
"
"
4
6
"
8
"
"
12
10
I
k (A')
Figure 3. In situ k3-weightedEXAFS spectra obtained during the various stages of discharge: 0 (a), 1 (b), and 2 (c) Li+ equivalents. 20 15
a A -
S 10 'P . 'E
t
5-
0I
I 0
.
C 2
.
8
, 4
) 6
.
' 8
I
r (4
Figure 4. Fourier transforms for the stages of discharge specified in the caption for Figure 3.
(Table 1). The scale factor of the material in the one- and twoelectrode discharged states was determined from the EXAFS of the undischarged in situ electrode on the basis of the actual bond distances and coordination numbers for pyrite." On the
Tryk et al.
3734 J. Phys. Chem., Vol. 99, No. 11, 1995
TABLE 1: Curve-Fitting Results for the Fourier-Filteredk3&) Fe K-Edge in Situ EXAFS of Pyrite before and after Electrochemical Insertion of 1 (1 e Discharge) and 2 (2 e Discharge) Li+ Equivalents into Pyrite on the Basis of Theoretical Phase Shifts and Scattering Amplitudes Obtained from FEFF'O sample shell Ak,b A-1 Ar: %, CNd r: A U2f A 2 SFa A&> eV F 1.2-2.5 6.0 2.26 0.0021 Fe-S 4-12 0.56 9.9 undischarged 0.43 1.2-2.5 3.2 2.27 Fe-S 4- 12 0.0022 0.56 1 e discharged 10.3 0.41 4- 12 1.2-2.5 2.5 2.31 0.0053 2 e discharged Fe-S 0.56 0.5 0.35 '
Best fits were achieved by nonlinear least squares refinements using the routines available in the XFPAKG package. Ak, range of k values over which the curve fitting was performed. Ar, window for inverse Fourier transform. CN, coordination number. r, distance in A. f u,DebyeWaller factor in A. g SF, scale factor. * AEo, energy threshold difference in eV. F, goodness of fit. basis of this analysis, the Fe-S interatomic distances, d(FeS ) , for FeS2 after lithium insertion, yielded 2.27 +/- 0.02 and 2.31 +/- 0.02 A, for 1 and 2 (Li+)es respectively. Essentially identical results were obtained using phases and amplitudes extracted from the first-shell Fe-S backscattering in pyrite. These d(Fe-S) values are significantly shorter than those reported for LizFeSz? namely, 2.3737(7) and 2.4126(15) A. For bulk powder specimens of Li,FeSz deintercalates (0.21 x 2.0), Brec et al. report Fe-S interactions based on EXAFS measurements in the range 2.28-2.33 A, which are consistent with those found in this work.3 Further compelling evidence that the material produced by the insertion of 2 (Li+),q into FeS2 is not crystalline LizFeSz has been obtained from in situ room-temperature 57FeMossbauer effect spectroscopy (MES) measurements using a spectroelectrochemical cell of similar design to that described here.12 According to these in situ data, the isomer shift (6) of the single doublet spectrum, 6 = 0.30 mm/s vs a-Fe at room temperature, for FeS2 after insertion of 2 (Li+), is substantially different from those of the two doublets observed for crystalline LizFeS2 in this (61 = 0.47, AI = 1.59; 6 2 = 0.47, A2 = 0.62 mm/ s)12 and other 1ab0ratories.l~ It is interesting to note that the presence of two doublets in the MES spectra of this material appears to be associated with occupational disorder between iron and lithium at the tetrahedral sites.3 The reduction in the S and Fe backscattering amplitudes upon incorporating Li+ into FeS2 can be explained on the basis of at least three different models: (i) Loss of single and multiple scattering (or net decrease in connectivity between the absorbing iron atoms and the backscattering atoms) due to a nonuniform distribution of Li (a very weak scatterer, Z = 3) in the coordination environment of iron. If lithium ions were distributed between iron (or sulfur) atoms to form linear .Fe-LiFee. (or .*Fe-Li-S- .) pathways, the resulting photoelectron focusing would increase the magnitude of the iron (or sulfur) peaks in the FT data. Although not observed here, such focusing effects in K-edge EXAFS have been demonstrated for the insertion of hydrogen (Z = 1, which has no measurable backscattering of its own) directly between metal atoms.14 (ii) Loss of coherency of the EXAFS signal due to the presence of a broad distribution of d(Fe-S) and d(Fe-Fe) values in the lithiated phase. This is to be contrasted with the narrow distribution of d(Fe-S) and d(Fe-Fe) for highly crystalline pyrite. (iii) Phase cancellation effects between Fe-S and FeFe interactions, similar to the behavior observed for Fe-Fe and Fe-Mo interactions in MoFe& ~1usters.l~ On the basis of these arguments, the most probable structure of the electrochemically formed material involves a highly disordered (possibly amorphous) form of pyrrhotite, Fel-,S (with Li+ counterbalancing the charge). Support of this view is provided by the similarity between the values of 6 and A of crystalline FeS2 (6 = 0.313 +/- 0.008 mm/s; A = 0.611 +/0.003 mm/s)16 and ultrafine (or superparamagnetic) particles of Fel-,S (6 = 0.27-0.30 mm/s; A = 0.54-0.77 mm/s in one and 6 = 0.36 +/- 0.03 d s ; A = 0.65 +/- 0.03
-
Energy, eV
Figure 5. XANES for the stages of discharge specified in the caption for Figure 3.
This observation is consistent with the fact that the in situ room-temperature MES spectra of FeS2 obtained for batteries of the type used in this work appear to be independent of the extent of Li+ insertion up to the insertion of 2 (Li+),.12 Additional support for Fel-,S as the most probable reaction product is provided by the XANES of the three specimens examined, shown in Figure 5. Despite the much decreased resolution of the otherwise highly structured near-edge region of crystalline pyrite,lgcaused by the thick nature of the specimen (vide supra), the insertion of Li+ into the lattice leads, in addition to a much steeper edge jump, to a rounding of the edge peak region. Such lack of edge structure has been found for the socalled NiAs-type Fel-,S pyrrhotite20,21and ascribed to a shift in the position of the fine structure toward lower energies induced by an increase in the metal-ligand distance. The same phenomenon has been observed for C U S and ~ ~ explained in terms of the complexity of the crystal structure, where Cu atoms often occupy various types of low-symmetry sites. Also noteworthy is the overall shift of the XANES toward lower energies as the amount of Li+ in the lattice is increased, an effect that may be consistent with the gradual electrochemical reduction of the iron centers. Such effects on the X-ray absorption edge position and shape have been observed in V K-edge XAFS of Lil-,V30g as a function of x for 0.00 x 3.74.23 The incorporation of Li+ into pyrite could, in principle, yield a sulfur-deficient material with a formal composition FeS2-, for x 0.15, without affecting significantly the short- and longrange order about iron.24 Although neither 57FeMES nor XAFS data have been reported for these nonstoichiometric pyrite specimens, the lattice perfection and crystallinity of FeS2-, are identical to those of stoichiometric FeS2. Although it is conceivable that such a structure may be formed during the initial stages of Li+ intercalation, the XAFS data presented herein suggest that long-range order is not preserved following Li+ insertion into the pyrite lattice. It is expected that an extension of these studies to include in situ XAFS measurements of the S K-edge may provide further insight into the Li+ insertion and deintercalation in iron sulfide mm/s in anotheP).
Electrochemical Insertion of Lithium into Pyrite materials and shed light on the processes responsible for the lack of reversibility of ambient-temperature, nonaqueouselectrolyte Li-FeS2 cells.
Summary In situ Fe K-edge XAFS spectra of FeSz cathodes have shown that the electrochemical insertion of lithium in nonaqueous electrolytes into FeS2 brings about a marked decrease in the amplitude of the EXAFS oscillations, particularly for shells associated with distant atoms, a rounding of the XANES region, and a value of d(Fe-S) (based on the fitting of the EXAFS data) of 2.27 and 2.31 +/- 0.02 A, following the insertion of 1 and 2 Li+ equivalents, respectively. On this basis and additional in situ room-temperature 57Fe Mossbauer effect spectroscopy data for the same system, it is proposed that the electrochemically-formed material involves a highly disordered (possibly amorphous) form of pyrrhotite, Fel-,S (with Li+ counterbalancing the charge). Acknowledgment. This work was supported in part by the Department of Energy, Basic Energy Sciences, and by a subcontract from the Lawrence Berkeley Laboratory. Additional funding was provided by Eveready Battery Co., Westlake, OH. The research was carried out at the National Synchrotron Light Source, Brookhaven National Laboratory, which is supported by the U.S. Department of Energy, Division of Materials Sciences and Division of Chemical Sciences, and at the Stanford Synchrotron Radiation Laboratory, which is operated by the Department of Energy, Office of Basic Energy Sciences.
References and Notes (1) Gard. P.: Sourisseau.. C.:. Ouvrard,. G.: . Brec, R. Solid State Ionics 1988,io,23i. (2) Blandeau, L.; Ouvrard, G.:Calage, - Y.; Brec, R.; Rowel, J. J . Phys. C: Soiid State Phys. 1987,20,4271. (3) Brec, R.; Rouzet, E.; Ouvrard, G. J . Power Sources 1989,26,325.
J. Phys. Chem., Vol. 99, No. 11, 1995 3735 (4) Fong, R.; Jones, C. H. W.; Dahn, J. R. J . Power Sources 1989,26, 333. ( 5 ) Fong, R.; Dahn, J. R.; Jones, C. H. W.J. Electrochem. Soc. 1989, 136, 3206. (6) Jones, C. H.W.; Kovacs, P. E.; Sharma, R. D.; McMilIan, R. S. J . Phys. Chem. 1990,94, 832. (7) Batchelor, R. J.; Einstein, F. W . B.; Jones, C. H. W.; Fong, R.; Dahn, J. R. Phys. Rev. B 1988,37, 3699. (8) (a) Stem, E.A.; Kim, K. Phys. Rev. B. 1981,23,3781. (b) Goulon, J.; Goulon-Ginet, C.; Cortes, R.; Dubois, J. M. J. Phys. (Paris) 1982,43, 539. (c) Lytle, F. W.In Applications of Synchrotron Radiation; Winick, H., Xian, D., Ye, M. H., Huang, T., Eds.; Gordon and Breach: New York, 1989;Vol. 4,pp 135. (9)Afrer-rhe-fact thickness effect corrections for strongly-absorbing materials, such as those studied in this work, cannot always eliminate amplitude distortions. See for example: (a) Teo, B. K.; Antonio, M. R.; Averill, B. A. J . Am. Chem. SOC. 1983, 105, 3751. (b) Antonio, M. R.; Song, I.; Yamada, H. J. Solid State Chem. 1991,93, 183. (10) (a) Rehr, J.J.; Mustre de Leon, J.;Zabinsky, S. I.; Albers, R. C. J. Am. Chem. Soc. 1991, 113, 5135. (b) Mustre de Leon, J.; Rehr, J. J.; Zabinsky, S. I.; Albers, R. C. Phys. Rev. B 1991,44, 4146. (1 1) Finklea, S., III.; LeConte, C. Acta Crystallogr. 1976,A32, 529. (12) Fierro, C.; Leger, V.; Akridge, J.; Scherson, D. To be submitted. (13) See for example: Melendres, C. J . Phys. Chem. 1978,82,2850. (14)Lengeler, B. Phys. Rev.Lett. 1984,53, 74. (15) Antonio, M. R.; Teo, B. K.; Cleland, W . E.; Averill, B. A. J . Am. Chem. SOC. 1983,105, 3477. (16)Evans, B. J.; Johnson, R. G.;Senftle, F. E.; Cecil, C. B.; Dulong, F. Geochim. Cosmochim. Acta 1982,46,761. (17)Hidaka, S.; Iino, A.; Nita, K.; Morinaga, K.; Yamazoe, N. Bull. Chem. SOC.Jpn. 1988,61,3169. (1 8) Vaishnava, P.P.; Montano, P. A.; Tischler, R. E.; Pollack, S. S. J. Catal. 1982,78,454. (19)Drager, G.;Frahm, R.; Materlik, G.;Bmmmer, 0. Phys. Status Solidi 1988,146,287. (20)Iida,A.; Noma,T.; Hayakawa, S.; Takahashi, M.; Gohshi, Y. Jpn. J. Appl. Phys., Suppl. 32 1993,2, 160. (21) Petiau, J.; Sainctavit, Ph.; Calas, G.Mater. Sci. Eng. 1988,BI, 237. (22) Sugiura, C. J. Chem. Phys. 1984,80,1047. (23) Tossici, R.; Marassi, R.; Berrettoni,M.; Stizza, S.; Pistoia, G.Solid State Ionics 1992,57, 227. (24)Fiechter, S.; Birkholz, M.; Hartmann, A.; Dulski, P.; Giersig, M.; Tributsch, H.;Tilley, R. J. D. J . Mater. Res. 1992, 7, 1829. JP943040W
