In situ X-ray study of the electrochemical reaction of Li with η′-Cu6Sn5

Article abstract of DOI:10.1149/1.1393413

The electrochemical reaction of lithium with crystalline intermetallic η′-Cu₆Sn₅ has been studied by means of in situ X-ray diffraction. The reaction of Li with Cu₆Sn₅ proceeds in two distinct steps. A first discharge plateau at approximately 0.4 V vs. Li is attributed to the coexistence of Cu₆Sn₅ and a ternary Li₂CuSn-type phase. Upon further reaction Li₂CuSn coexists with a Li-rich tin phase (Li_4.4Sn) and metallic Cu at a potential below 0.1 V. During the charge, the initial intermetallic is reformed in a multistep reaction. First, some Li can be removed from Li_4.4Sn making Li_4.4-xSn with the same structure. As Li is removed from Li_4.4-xSn it reacts with the available Cu to make Li₂CuSn. Then, lithium can be removed from Li₂CuSn, making vacancies in the structure to form Li_2-xCuSn. When x in Li_2-xCuSn reaches about 1, the removal of further lithium results in the formation of a second phase, Cu₆Sn₅. The 0.8 V plateau during the charge of Li/Cu₆Sn₅ cells is associated with this coexistence of Li_2-xCuSn and Cu₆Sn₅. The in situ experiments show that both the Li_4.4Sn and the Li₂CuSn phases can exhibit a range of Li stoichiometries, especially as Li is removed from their structures. When Li/Cu₆Sn₅ cells are charged and discharged above the lower plateau, to prevent the continual expulsion and reincorporation of Cu from the cathode grains, better cycling is obtained than when cycling includes the lower plateau where metallic copper is formed.

Full text of DOI:10.1149/1.1393413

1658
Journal of The Electrochemical Society, 147 (5) 1658-1662 (2000)
S0013-4651(99)09-104-1 CCC: $7.00 © The Electrochemical Society, Inc.
In Situ X-Ray Study of the Electrochemical Reaction of
Li with ␩Ј-Cu6Sn5
D. Larcher,a,** L. Y. Beaulieu,a,* D. D. MacNeil,b and J. R. Dahna,b,**,z
aDepartment of Physics and bDepartment of Chemistry, Dalhousie University, Halifax, Canada B3H 3J5
The electrochemical reaction of lithium with crystalline intermetallic ␩Ј-Cu6Sn5 has been studied by means of in situ X-ray dif-
fraction. The reaction of Li with Cu6Sn5 proceeds in two distinct steps. A first discharge plateau at approximately 0.4 V vs. Li is
attributed to the coexistence of Cu6Sn5 and a ternary Li2CuSn-type phase. Upon further reaction Li2CuSn coexists with a Li-rich
tin phase (Li4.4Sn) and metallic Cu at a potential below 0.1 V. During the charge, the initial intermetallic is reformed in a multistep
reaction. First, some Li can be removed from Li4.4Sn making Li4.4ϪxSn with the same structure. As Li is removed from Li4.4ϪxSn
it reacts with the available Cu to make Li2CuSn. Then, lithium can be removed from Li2CuSn, making vacancies in the structure
to form Li2ϪxCuSn. When x in Li2ϪxCuSn reaches about 1, the removal of further lithium results in the formation of a second
phase, Cu6Sn5. The 0.8 V plateau during the charge of Li/Cu6Sn5 cells is associated with this coexistence of Li2ϪxCuSn and
Cu6Sn5. The in situ experiments show that both the Li4.4Sn and the Li2CuSn phases can exhibit a range of Li stoichiometries, espe-
cially as Li is removed from their structures. When Li/Cu6Sn5 cells are charged and discharged above the lower plateau, to prevent
the continual expulsion and reincorporation of Cu from the cathode grains, better cycling is obtained than when cycling includes
the lower plateau where metallic copper is formed.
© 2000 The Electrochemical Society. S0013-4651(99)09-104-1. All rights reserved.
Manuscript submitted September 28, 1999; revised manuscript received January 4, 2000.
Lithium-ion cells normally use graphite negative electrodes and
lithium transition metal oxide positive electrodes. In order to in-
crease the energy density and specific energy of such cells, new elec-
trode materials must be found. Lithium alloys offer larger specific
and volumetric capacities than graphite, and these have been the sub-
ject of recent research. In particular, Mao et al. have shown that in-
Experimental
Samples of ␩Ј-Cu6Sn5 powder were prepared by heating stoi-
chiometric mixtures of elemental Cu and Sn powders (purity
>99.9%) under argon flow at 400ЊC for a 10 h period.3 The heating
rate was 1ЊC/min. To ensure good homogeneity and contact of the
reactants, the Cu/Sn mixture was intensively ground in a mortar and
pressed into a pellet at a pressure of 200 bar.
The HT-Li2CuSn reference phase was prepared from powdered
Sn and Cu, and Li foil in stoichiometric proportions. Reagents were
poured in a stainless steel tube under purified argon in a glove box.
The glove box was equipped with a welder. The ends of the stain-
less steel tube were TIG welded closed within the glove box. The
sealed tube was heated at 700ЊC under argon flow for a 10 h peri-
od.7 The tube was opened in a He-filled glove box, and the recov-
ered purple-colored powder was finely ground in an agate mortar. A
weight comparison showed that no leakage occurred during the
annealing treatment.
X-ray diffraction was performed using a Siemens D5000 diffrac-
tometer equipped with a Cu target and diffracted-beam monochro-
mator. Air-sensitive samples were characterized in a specially de-
signed airtight X-ray holder as described in Ref. 8. In situ X-ray
experiments were conducted using an in situ cell as reported in
Ref. 9. For all in situ measurements the counting time was set to be
15 s for a step of 0.05Њ. The ranges between 20 and 58Њ (experiment
no. 1) and between 36 and 42Њ (experiment no. 2) in scattering angle
were measured repeatedly while the cell was discharged and charged
once. The cell was cycled using a constant current corresponding to
C/100 (experiment no. 1) or C/30 (experiment no. 2), at room tem-
perature. To improve counting statistics, two electrode disks were
stacked in the cell to effectively make an electrode having twice the
thickness. The C-rate is calculated assuming a full reaction of Sn in
Cu6Sn5 with Li (i.e., formation of Li4.4Sn). (The C-rate is the current
needed to discharge the battery in 1 h.)
Electrochemical tests were made in 2325 coin cell hardware con-
structed as detailed in Ref. 10. The electrolyte used was 1 M LiPF6
dissolved in a 33/67 v/v mixture of ethylene carbonate (EC) and
diethyl carbonate (DEC) solvents (Mitsubishi Chemical). The nega-
tive electrode was a 250 ␮m thick Li foil. For the preparation of the
positive electrode, the powder was dispersed in a plastic matrix ac-
cording to Bellcore technology.11 The plastic layer was made by mix-
ing 56 wt % active material and 6 wt % Super-S carbon black, and by
intimately dispersing this mixture in 15 wt % copolymer (Kynar
2801, 88:12 VDF:hexafluoro-propylene copolymer, Elf-Atochem)
1
termetallic compounds like FeSn2 and active/inactive nanocom-
2
posites like FeSn2/SnFe3C function well as anodes although their
cycle life is limited. The limited cycle life of such alloys is probably
caused by relatively large volume changes associated with the re-
versible reaction of Li.
In recent papers Kepler et al. and Thackeray et al.3,4 have report-
ed electrochemical data on Li/Cu6Sn5 cells. It was found that Cu6Sn5
reacts reversibly with lithium to deliver an initial discharge capacity
of approximately 350 mAh/g. Based on structural similarity, it was
proposed that this reaction could be described as an insertion of lithi-
um in the Cu6Sn5 structure to reversibly form Li2.17CuSn0.83 (or
Li13Cu6Sn5) that has the same structure as Li2CuSn. In this paper we
do not distinguish between Li2.17CuSn0.83 and Li2CuSn. In Ref. 3
and 4, it is suspected that further reaction of Li with Li2CuSn could
result in the formation of Li-Sn alloys and the simultaneous expul-
sion of Cu. Surprisingly, even when discharged to 0 V, this reaction
does not occur under the conditions used in Ref. 3 and 4. It is also
shown that better cycling life is achieved when the discharge cut-off
voltage is raised to 200 mV from 0 V.
In other work, Thackeray’s group studied the material discharged
to zero volts and the reaction below 400 mV was attributed to the ex-
trusion of Cu and the formation of Li-Sn alloy.5 The extraction of
lithium from Li2CuSn was studied in Ref. 6. However the results in
Ref. 3-6 are not self-consistent, nor are they of the highest possible
quality. Thus, we decided to undertake a careful study of the reaction
of lithium with Cu6Sn5.
Here, we show that the reaction of Li with Cu6Sn5 results first in
the formation of Li2CuSn as an intermediate phase through a two-
phase reaction, and finally leads to a disordered Li4.4Sn alloy and
nanoscopic grains of Cu. These reactions are found to be basically
reversible although some aspects of the mechanism are different dur-
ing discharge and charge. The changes we observe in the charge-dis-
charge cycle life depending on the selected voltage window are easi-
ly explained by the structural changes involved in these two reactions.
** Electrochemical Society Student Member.
** Electrochemical Society Active Member.
z
*
E-mail: jeff.dahn@dal.ca

Journal of The Electrochemical Society, 147 (5) 1658-1662 (2000)
1659
S0013-4651(99)09-104-1 CCC: $7.00 © The Electrochemical Society, Inc.
Figure 3. Voltage as a function of X-ray scan number and capacity measured
for the first cycle of the Li/Cu6Sn5 in situ X-ray cell. (C/100, 0.0-2.0 V.)
The first discharge of a Li/Cu6Sn5 cell at C/100 (Fig. 2a) revealed
two plateaus on discharge, one at about 0.4 V and another below
0.1 V. Figure 2a shows that the calculated capacity for the formation
of Li2CuSn (330 mAh/g) is much less than the length of the first dis-
charge (630 mAh/g). The lengths of the two plateaus match the the-
oretical capacity for the consecutive formation of Li2CuSn and
Li4.4Sn ϩ Cu, respectively. Based on these preliminary observations,
we believe that the end product of the full lithiation of Cu6Sn5 is not
Li2CuSn, but instead is Li4.4Sn ϩ Cu. The differential capacity vs.
voltage (Fig. 2b) clearly shows these two discharge plateaus. Fig-
ure 2b shows a broad and symmetric peak at 0.45 V followed by a
sharp one at 0.8 V during the charge. Note that the anisotropy and
shape of this latter peak is characteristic of a two-phase reaction.
In order to shed some light on these reactions, we performed an
in situ X-ray experiment. The voltage vs. scan number plot is shown
in Figure 3. The evolution of the X-ray patterns during scans 0-20
(Fig. 4) clearly indicates that the first discharge plateau corresponds
to the formation of Li2CuSn phase. In this region, the Cu6Sn5 peaks
(strongest peak at 30.4Њ) vanish progressively while broader peaks of
the ternary phase grow. Figure 5 compares in situ scan no. 20 (end
of first plateau) to the XRD pattern of HT-Li2CuSn prepared by
solid-state reaction. Only a small difference in peak positions can be
observed, and that could probably be attributed to the difference in
Cu and Sn stoichiometries.
Figure 1. (a) Solid dots show the XRD of ␩Ј-Cu6Sn5 along with the fit as
determined by Rietveld refinement shown by the solid line. (b) Difference
plot between the experimental data and the refinement.
binder matrix plasticized with 23 wt % dibutylphthalate (DBP). After
addition of anhydrous acetone and stirring for 24 h, the slurry was
coated on a glass substrate, the acetone evaporated and the resulting
dried plastic layer was cut into disks. Before use in coin cells, the
DBP was extracted by three successive rinses in anhydrous ether.
Electrochemical experiments were carried out using a current
corresponding to C/30, at 30 Ϯ 0.1ЊC. Different cutoff potentials
were used as indicated later.
Results and Discussion
Figure 1 shows the X-ray diffraction (XRD) pattern of the start-
ing material along with a calculated pattern determined by Rietveld
refinement. The pattern was fitted successfully to a hexagonal struc-
ture with cell parameters a ϭ 4.2012 Ϯ 0.0002 Å and c ϭ 5.0928 Ϯ
0.0003 Å in excellent agreement with earlier reported data.12 The re-
fined site occupation in space group P63/mmc (no. 194) leads to the
formula Cu0.545Sn0.455 that perfectly matches the expected Cu6Sn5
composition. More recent structure determinations of Cu6Sn5 sug-
gest that it has monoclinic symmetry,13 but we found better agree-
ment when refining to the hexagonal structure.
Figure 2. (a) Voltage vs. capacity of the first discharge/charge cycle of a
Li/Cu6Sn5 cell cycled between 2.0 and 0.0 V at C/100. (b) Differential capac-
ity vs. voltage of the cell shown in (a).
Figure 4. First in situ X-ray experiment for the Li/Cu6Sn5 cell. Scans col-
lected along the first discharge plateau.

1660
Journal of The Electrochemical Society, 147 (5) 1658-1662 (2000)
S0013-4651(99)09-104-1 CCC: $7.00 © The Electrochemical Society, Inc.
disordered Li4.4Sn phase. At the end of the reaction of Li with
FeSn2, 15 Li4.4Sn was found to coexist with nanoscopic grains of Fe.
This Fe could not be detected by XRD after the first discharge, but it
could be observed by Mössbauer spectroscopy. We suspect the copper
formed here is similarly small, and hence difficult to observe by XRD.
Figure 7 shows the in situ X-ray results during part of the subse-
quent charge. First there is a transformation of Li4.4Sn into the
Li2CuSn phase. However, this transformation appears quite unusual.
There is a progressive shift of the Li4.4Sn peak located at about 38.5Њ
toward the Li2CuSn peak located at about 40.5Њ (Fig. 7 and 8). At the
same time, the Li2CuSn(111) peak (24.5Њ) grows continuously with-
out any noticeable shift in position, and only a small shift is observed
for the (113) reflection (48Њ). It is our opinion that the transition pro-
ceeds as follows
Li4.4Sn r Li4.4ϪxSn ϩ xLi
and simultaneously
Figure 5. Comparison of the X-ray patterns of HT-Li2CuSn and in situ scan
no. 20. This scan corresponds to the end of the first discharge plateau. Aster-
isks indicate Li-Sn and Cu-Sn impurities.
Li4.4ϪxSn ϩ Cu r Li2CuSn ϩ (2.4 Ϫ x) Li
Figure 6 shows the appearance of another series of peaks located
at scattering angles of 23, 38.5, and 43.5Њ while the Li2CuSn peaks
disappear during the discharge on the second plateau. These new
peaks have been previously attributed to a disordered Li-rich Sn
alloy.14 Once the cell reaches 0.0 V, the pattern shows mainly the
Li4.4Sn phase. The complete transformation into Li4.4Sn confirms the
observed value of the first discharge capacity (630 mAh/g) compared
to the calculation for the full reaction of tin (600 mAh/g). The
strongest diffraction peak of copper (the 111 peak) is found at a scat-
tering angle of about 43.3Њ. The top panel of Fig. 6 shows a peak con-
sistent with the presence of copper, but this peak is also present in the
Figure 7. First in situ X-ray experiment for the Li/Cu6Sn5 cell. Scans col-
lected during the first part of the recharge and showing the transformation of
Li4.4Sn into Li4.4ϪxSn and Li2CuSn.
Figure 6. First in situ X-ray experiment for the Li/Cu6Sn5 cell. Scans col-
lected along the second discharge plateau until 0 V. The end product is iden-
tified as disordered Li4.4Sn.

Journal of The Electrochemical Society, 147 (5) 1658-1662 (2000)
1661
S0013-4651(99)09-104-1 CCC: $7.00 © The Electrochemical Society, Inc.
(Notice that we have not balanced the second reaction with respect
to tin content. This is because we do not know the exact stoichiom-
etry of the phase we are calling Li2CuSn, which could be
Li2CuSn0.83, as discussed in the introduction.) The small amount of
nontransformed Li2ϪxCuSn contributes to the irreversible capacity
evaluated to about 120 mAh/g on the first cycle.
Figure 8. Evolution of the position of the Li4.4Sn (110 of bcc substructure)
and Li2CuSn (022) peaks observed along the first cycle (in situ experiment
no. 1). A smooth evolution of the peak position is observed during the first
stage of the recharge. The peak positions for Li2CuSn (homemade) and
Li4.4Sn (JCPDS 18-753) are given as guidelines.
During such a process, the Bragg peaks of Li4.4ϪxSn would shift
in position as x changes and the Li2CuSn phase would appear. Such
a reaction scenario occurs apparently because of the slow diffusion
of Cu to be reincorporated into the Li2CuSn phase.
The top two panels of Fig. 7 and all of Fig. 9 show the in situ
results collected during the second charge plateau. The voltage
varies during the region of scans 65-73 (see Fig. 2), and this corre-
sponds to a region where the (022) peak of Li2CuSn shifts apprecia-
bly in angle. Most likely this is caused by a region where Li2ϪxCuSn
can have a variable stoichiometry. During scans 73-80, the Bragg
peaks of Cu6Sn5 reappear in a two-phase region during a corre-
sponding plateau in Fig. 3. The reaction during charge above 0.5 V
is best described by
Li2CuSn r Li2ϪxCuSn ϩ xLi, for about 0 < x < 1
followed by
Li2ϪxCuSn r 1/6 Cu6Sn5 ϩ (2 Ϫ x) Li
Figure 10. Second in situ experiment for the Li/Cu6Sn5 cell. The scans are
collected along a charge-discharge sweep (C/30) over the Li2CuSn } Li4.4Sn
plateau.
Figure 9. First in situ X-ray experiment for the Li/Cu6Sn5 cell. Scans col-
lected during the final part of the recharge.

1662
Journal of The Electrochemical Society, 147 (5) 1658-1662 (2000)
S0013-4651(99)09-104-1 CCC: $7.00 © The Electrochemical Society, Inc.
Thackeray’s group.3,4 Unlike3,4 we attribute the improved cycling to
the fact that the Cu atoms are not expelled into a second distinct phase
in this voltage region. That is, the Cu atoms are always found next to
tin atoms in both Cu6Sn5 and in Li2CuSn. They are not expelled to
form copper. Kepler3 and Thackeray4 argue that a topotactic reaction
is responsible for the improved cycling in this region. We do not
agree. The volume change associated with the Cu6Sn5 to Li2CuSn
transition is about 45%, and structural integrity most likely will not be
maintained. It is interesting to note that the same group invokes much
smaller lattice changes in two-phase regions as the cause of capacity
loss in the LiMn2O4 Ϫ Li2Mn2O4 reaction.17 Finally, the cycling per-
formance in the high-voltage region, although best, is still poor com-
pared to that obtained for topotactic reactions, such as Li insertion in
graphite, where hundreds of fully reversible cycles can be obtained.
Conclusion
A careful survey of the structural modifications occurring during
the electrochemical reaction of Li with crystallized intermetallic ␩Ј-
Cu6Sn5 has been performed. During discharge, this reaction proceeds
Figure 11. Capacity vs. cycle number for Li/Cu6Sn5 cells measured at C/30.
The different voltage ranges and the corresponding reactions are indicated.
in two distinct single first-order transitions, namely, Cu6Sn5
}
Li2CuSn and Li2CuSn } Li4.4Sn ϩ Cu. The reverse reactions occur
during charge, but both Li4.4ϪxSn and Li2ϪxCuSn appear to show
some range of variable Li stoichiometry during charge. Although the
full understanding of all of the mechanisms is not yet totally achieved,
we found that large volume expansion and copper expulsion leads to
poor capacity retention for extended cycling tests. Finally, this study
clearly shows that situations where alloys have similar structure (but
very different volumes!) do not ensure long cycle life for the electro-
chemical insertion/alloying reaction.
Figure 10 shows in situ X-ray diffraction results recorded near the
composition Li4.4Sn. The first 15 scans correspond to the removal of
Li, making Li4.4ϪxSn (probably with some parallel back reaction to
Li2CuSn) and scans 16-30 to the reverse reaction with the removed
lithium. The Bragg peak near 38Њ (corresponding to the 110 peak of
the body-centered cubic (bcc) substructure of Li4.4Sn 14) shifts re-
versibly during this experiment, suggesting a single-phase reaction.
The peak does broaden near scan 15, suggesting the formation of the
Li2CuSn phase. In addition, the voltage profile shows significant hys-
teresis, not expected for a single-phase process. Thus, it is possible
that some formation of Li2CuSn has occurred during charge to 0.5 V.
If no Li2CuSn forms during charge to 0.5 V, then the hysteresis
in the voltage profile must have another origin. The equilibrium Li-
Sn phase diagram does not show single-phase regions, and this hys-
teresis may therefore result from a tendency of the small (several
nanometers) Li4.4ϪxSn grains to attempt to phase separate into re-
gions of significantly different Li concentration, as the bulk phase
diagram would predict.
Dalhousie University assisted in meeting the publication costs of this
article.
References
1. O. Mao, R. A. Dunlap, and J. R. Dahn, J. Electrochem. Soc., 146, 405 (1999).
2. O. Mao, B. L. Turner, D. W. McClure, I. A. Courtney, L. J. Krause, and J. R. Dahn,
Electrochem. Solid-State Lett., 2, 3 (1999).
3. K. D. Kepler, J. T. Vaughey, and M. M. Thackeray, Electrochem. Solid-State Lett.,
2, 307 (1999).
4. M. M. Thackeray, J. T. Vaughey, A. J. Kahaian, K. D. Kepler, and R. Benedek,
Electrochem. Commun., 1, 111 (1999).
5. M. M. Thackeray, C. S. Johnson, A. J. Kahaian, K. D. Kepler, and J. T. Vaughey,
ITE Battery Lett., 1, 26 (1999).
6. J. T. Vaughey, K. D. Kepler, R. Benedek, and M. M. Thackeray, Electrochem. Com-
mun., 1, 517 (1999).
7. H. Pauly, A. Weiss, and H. Witte, Z. Metallkde, 59, 47 (1969).
8. R. A. Dunlap, D. A. Small, D. D. McNeil, M. N. Obrovac, and J. R. Dahn, J. Alloys
Compd., 289, 135 (1999).
9. M. N. Richard, I. Koetschau, and J. R. Dahn, J. Electrochem. Soc., 144, 554 (1997).
10. A. M. Wilson and J. R. Dahn, J. Electrochem. Soc., 142, 326 (1995).
11. D. Guyomard and J-M. Tarascon, U.S. Pat. 5,192,629 (1993).
12. A. Westgren and G. Phragmen, Z. Anorg. Chem., 175, 80 (1928).
13. A. K. Larsson, Acta Crystallogr., B50, 636 (1994).
Figure 11 compares the charge-discharge cycling of Li/Cu6Sn5
cells constrained to different upper and lower voltage cutoffs. Three
ranges were tested, 0.0 Ϫ 2.0 V, 0.0 Ϫ 0.6 V (restricted to the range
Li2CuSn } Li4.4Sn ϩ Cu), and 0.2 Ϫ 2.0 V (restricted to Cu6Sn5 }
Li2CuSn). The capacity vs. cycle number declines rapidly when the
entire voltage range is accessed. There is an improvement when re-
stricted ranges are used. In the low-voltage range, Cu is repeatedly
expelled and reincorporated, and we feel that this is the cause for the
relatively poor reversibility observed here. If the back reaction of
copper could be hindered, as it was for Fe in Li/FeSn2 cells charged
only to 0.55 V, 16 then we expect good cycling would ensue.
Figure 11 shows that the best cycling results are obtained for cells
that are never discharged below 0.2 V. This was also observed by
14. J. R. Dahn, I. A. Courtney, and O. Mao, Solid State Ionics, 111, 289 (1998).
15. O. Mao, R. A. Dunlap, and J. R. Dahn, J. Electrochem. Soc., 145, 4195 (1998).
16. O. Mao, R. A. Dunlap, and J. R. Dahn, J. Electrochem. Soc., 146, 405 (1999).
17. M. M. Thackeray, J. Electrochem. Soc., 142, 2558 (1995).