2050 J. Electrochem. Soc., Vol. 144, No. 6, June 1997 The Electrochemical Society, Inc.

reacted on discharge, the starting material does not form

SCATTERING ANGLE (deg.) again. Sn peaks do not form after the first charge as
20 30 40 50 observed for SnO, Sn02, and SiSnO3. This could be due to
the fact that for Li2SnO3 3 mol of Li20 per mole of Li2 Sn03
form during the discharge (Eq. 5). This higher concentra-
tion of lithia could prevent large regions of tin metal from
forming and as a result we do not observe well defined tin
04
0
1 peaks in our XRD scan.
We believe that the lithia matrix that is formed during
the initial stages of lithium insertion in these materials
acts as a 'glue," which helps to hold the Li-Sn regions
together through the large volume changes in the alloy-
zw
Cl)
ing/dealloying process. This is based on the reversibility of
the processes as demonstrated electrochemically when
I-
z compared to metallic tin (Fig. 2). Table II shows the calcu-
lated volume per mole of Sn atoms in each of the respec-
tive Li-Sn phases. Notice the relatively small volume
increase factor to LiSn, 1.53, and the large volume in-
crease factor to the alloying limit of Li22Sn5, 3.59.
An investigation of the crystal structures of each of the
various phases shows that both Li2Sn, and LiSn are lay-
Fig. 10. In situ XRD results for SiSnO3; (a) end of charge —2.5 V, ered structures, whereas higher concentrations of lithium
(b) end of discharge —0.0 V, (c) calculated peak positions for Li5Sn2, lead to more complicated structures where layers or tun-
Li135n5, Li75n2, and U225n5 (the majority of peaks for all of these nels of Li atoms are not formed. Figure 12 is a diagram of
structures are around 22° and 381, and (d) first scan —2.5 V. the crystal structures of Li2Sn5 and LiSn. The crystal
Logarithmic intensities are plotted for scans in reference to the left-
structure of Li2Sn5 has some prominent features which are
hand vertical axis and linear intensifies are plotted for the calculat-
ed patterns (normalized to 100) in reference to the right-hand ver-
outlined by the solid lines in the right-hand diagram of
tical axis. Fig. 12a. The lithium resides within pentagonal structures
between the layers of Sn atoms. There are also square and
triangular sites which exist. If all the triangular and
rial does not reappear during the charge as can be deter- square sites were filled with Li, a stoichiometry of LiSn
mined by comparing Fig. iOa and d. would result. However there would be three inequivalent
Figure ii shows some of the in situ XED scans for types of Li atoms. A slight skewing to create an overall
Li2SnO3. Figure lid shows the first scan (2.5 V) and the coordination of 8 Sn per Li (square sites) and vice versa
corresponding calculated peaks for Li2SnO3 are shown by results in the crystal structure shown in Fig. i2b, or LiSn.
the broken lines in that figure. When the cell is fully dis- Over certain ranges the Li-Sn alloy structural transitions
charged, the broad features at 22° and 38°, which are are minimally disruptive and so should be quite reversible.
indicative of the lower Li-Sn alloy phases, appear We cycled cells of SnO over various voltage ranges in
(Fig. lib —0.0 V). The original peaks of Li2SnO3 do not order to confirm that the first two Li-Sn phases should
entirely disappear for this cell. The first discharge capaci- demonstrate good reversibility. Figure 13 shows the results
ty for this cell was approximately 1000 mAh/g instead of of two of these cycling experiments. In Fig. 13a we show a
approximately 1200 mAh/g for other cells (Fig. 2e). cell that has been cycled between 1.3 and 0.0 V Over this
Therefore we believe that not all of the starting material voltage range the process is not very reversible. We believe
has reacted in this in situ cell. By the end of charge that this is due to the cracking and crumbling of the mate-
(Fig. ha —2,5 V) the lower Li-Sn phases are no longer pre- rial by the large volume increases that are involved in real-
sent and, with the exception of the Li2SnO3 that has not izing the lower Li-Sn alloy phases. Figure 13b shows the
results for a SnO cell that has been discharged to 0.4 V and
then charged to 1.3 V for ten cycles. We can see that this
process is more reversible, and that the phase definition as
SCATTERING ANGLE (deg.) indicated by the plateaus seems to improve with cycling.
Since the first two Li-Sn phases, that is Li25n5 and LiSn,
20 30 40 50 are layered and related structures, then over this region
100 TT the reaction with Li is more reversible than if we were to
(a) End ot Charge
3 react over the full range of Li-Sn alloys.
Figure 14 is a schematic of the reaction mechanism that
P1
0r 10
100 —I—I i—Ill .(!l ii
{b) End of Discharge - 0.0 V
Ii occurs in the tin oxides and tin oxide composites that we
have studied. In the diagram we assume SnO as the start-
ing material, but this could he easily replaced by any of
3
10
>.
U) I I.i' I—S Jill iiii 4....—
the other starting materials by simply adjusting the stoi-
chiometry of the added lithium. We first introduce lithium
forming a Li2O "matrix-glue" which assists in the reversi-
zLu (C) Jll Li5Sn2, Li13Sn5, bility of the subsequent alloying/dealloying by resisting
1i7Sn2 Li229n5
I—
100 : :;j j L_.4_,; I

Table Il. Calculated volume per mol Sn (cm3/mol) and volume

i1T5u20L.3
3
increase factor (over Sn metal) for each of U-Sn alloy phases.
10
Phase Volume per mol Sn (cm°/mol) Volume increase factor
Fig. 11. In situ XRD results for Li25n03; (a) end of charge —2.5 V,
(b) end of discharge —0.0 V, (c) calculated peak positions for Li5Sn2, Sn 16.18 1.00
Li2Sn, 19.88 1.23
Li13Sn5, Li7Sn2, and LiSn5 (the majority of peaks for a11 of these 24.74 1.53
structures are around 22° and 381, and (d) first scan —2.5 V. LiSo
Li75n3 36.95 2.28
Logarithmic intensities are plotted for scans in reference to the left- 44.64 2.76
Li55n2 2.81
hand vertical axis and linear intensities are plotted for the calculat- Li3Sn, 45.45
ed patterns (normalized to 100) in reference to the right-hand ver- Li75n2 48.31 2.99
Li,25n5 58.14 3.59
tical axis.