Preparation and electrochemical performance of spinel-type compounds Li4Aly,Ti5-yO12 (y = 0, 0.10, 0.15, 0.25)

Article abstract of DOI:10.1149/1.1833315

Li₄Ti₅O₁₂ and Al³⁺ doped Li₄Al_yTi_5-yO₁₂ (y = 0.10, 0.15, 0.25) were synthesized via solid-state reaction using TiO₂-rutile, Li₂CO₃, and Al₂O₃ as starting reagents. The charge-discharge cycling of the cells showed that the electrochemical performance of Li₄Ti₅O₁₂ prepared from rutile-type TiO₂ by our laboratory was as good as that produced by anatase-type TiO₂. However, Al³⁺ doped Li ₄Al_yTi_5-yO₁₂ (y = 0.10, 0.15) exhibited a much better electrochemical performance in comparison with undoped Li₄Ti₅O₁₂. Among the three samples of Li ₄Al_yTi_5-yO₁₂ (y = 0, 0.10, 0.15), Li₄Al_0.15Ti_4.85O₁₂ exhibited the largest reversible capacity and the highest coulombic efficiency. The discharge capacity values in the first and second cyclings were 195.6 and 173.6 mAh/g, respectively. The value remained 166.9 mAh/g after 30 cycles with a capacity loss of 3.86% compared to the second cycle, and the coulombic efficiency was 99.2% at the 30th cycle.

Full text of DOI:10.1149/1.1833315

A186
Journal of The Electrochemical Society, 152 ͑1͒ A186-A190 ͑2005͒
0013-4651/2004/152͑1͒/A186/5/$7.00 © The Electrochemical Society, Inc.
Preparation and Electrochemical Performance of Spinel-Type
Compounds Li4AlyTi5ÀyO12 „y ϭ 0, 0.10, 0.15, 0.25…
,z
*
Shahua Huang, Zhaoyin Wen, Xiujian Zhu, and Zuxiang Lin
Shanghai Institute of Ceramics, Chinese Academy of Sciences, Shanghai 200050, People’s Republic of China
Li4Ti5O12 and Al3ϩ doped Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15, 0.25͒ were synthesized via solid-state reaction using TiO2-rutile,
Li2CO3 , and Al2O3 as starting reagents. The charge-discharge cycling of the cells showed that the electrochemical performance
of Li4Ti5O12 prepared from rutile-type TiO2 by our laboratory was as good as that produced by anatase-type TiO2 . However, Al3ϩ
doped Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15) exhibited a much better electrochemical performance in comparison with undoped
Li4Ti5O12 . Among the three samples of Li4AlyTi5ϪyO12 (y ϭ 0, 0.10, 0.15), Li4Al0.15Ti4.85O12 exhibited the largest reversible
capacity and the highest coulombic efficiency. The discharge capacity values in the first and second cyclings were 195.6 and 173.6
mAh/g, respectively. The value remained 166.9 mAh/g after 30 cycles with a capacity loss of 3.86% compared to the second cycle,
and the coulombic efficiency was 99.2% at the 30th cycle.
© 2004 The Electrochemical Society. ͓DOI: 10.1149/1.1833315͔ All rights reserved.
Manuscript submitted February 10, 2004; revised manuscript received June 7, 2004. Available electronically December 2, 2004.
Electrodes containing Li4Ti5O12 have good Li-ion intercalation
Characterization.—Phase analysis of the Li4AlyTi5ϪyO12 (y
¨
and deintercalation reversibility and exhibit no structural change
͑zero-strain insertion material͒ during charge-discharge cycling.
This active material has a main discharge platform close to 1.55 V
vs. Liϩ/Li, which is very promising for use in electrodes in numer-
ous battery applications.1 So far, the spinel-type Li4Ti5O12 has
mainly been prepared by a solid-state reaction from TiO2 and
Li2CO3 or LiOH. The reaction typically occurs within 12-24 h at
800-1000°C.2-4
ϭ 0, 0.10, 0.15, 0.25) powder was performed on a Guinier-Hagg
camera by a powder diffraction method using Cu K␣1 radiation as
the radiation source (␭ ϭ 1.54056 Å͒ and Si element as an internal
standard. The obtained films were evaluated with a computerized
scanner system. The cell parameters of the Li4AlyTi5ϪyO12 phase
¨
were determined by the PIRUM program based on the Guinier-Hagg
film data.12 The morphology and particle size distribution of the
Li4AlyTi5ϪyO12 powders were obtained by SEM.
Titanium dioxide adopts eight different crystallographic struc-
tures. Relevant data on Liϩ insertion exist only for anatase and
rutile. Anatase is generally considered to be the most active Liϩ
insertion host; the insertion into rutile-type TiO2 (r-TiO2) is usually
reported to be negligible. Thus, anatase-type TiO2 (a-TiO2) has been
used in most studies as the starting material for Li4Ti5O12 .5 Because
r-TiO2 is a natural material, it has low cost and wide source, and
a-TiO2 changes to r-TiO2 at high temperatures. In this paper, we use
r-TiO2 as the starting material and study its performance.
Although Li4Ti5O12 is an insulator, doping the structure with
small amounts of Mg2ϩ has been reported to improve the electronic
conductivity of the spinel by many orders of magnitude.6 Other
doping ions such as Cr3ϩ, V3ϩ, Mn2ϩ, Ba2ϩ, and Sr2ϩ have also
been studied to improve the capacity and cycling property, but no
satisfactory results have been obtained.7-10 To improve the electro-
chemical performance of Li4Ti5O12 , Al3ϩ was chosen as an
alternative dopant for its extreme stability in an octahedral
environment and for its light weight.11 In this paper, we report the
results obtained for Li4AlyTi5ϪyO12 with various Al3ϩ doping con-
tents. The materials have been characterized by X-ray diffraction
͑XRD͒ and scanning electron microscopy ͑SEM͒ observations.
The electrochemical cycling performance of Li4AlyTi5ϪyO12
(y ϭ 0, 0.10, 0.15, 0.25) in lithium cells was also studied.
The electrochemical cycling performance of the Li4AlyTi5ϪyO12
powders was carried out with laboratory cells using lithium metal as
the counter electrode at room temperature ͑20°C͒. The cells were
based on the following configuration: Li metal ͑Ϫ͒/electrolyte/
Li4AlyTi5ϪyO12 ͑ϩ͒ with a liquid electrolyte ͑1 M solution of LiPF6
in ethylene carbonate ͑EC͒: dimethyl carbonate ͑DMC͒ ͑1:1 v/v͒.
Microporous polypropylene film ͑Celgard͒ was used as a separator.
The working electrode was prepared from a ballmilling paste of 80
wt % Li4AlyTi5ϪyO12 with 10 wt % conductive carbon black and 10
wt % polyvinylidene fluoroethylene binder in N-methyl pyrrolidone
solvent. The paste was coated on an aluminum foil by blade, and
then dried under vacuum at 100°C for 8 h before electrochemical
evaluation. The cell was assembled in a glove box ͑NAC AM-2͒
filled with ultrapure argon. The charge-discharge characteristics of
the cells were recorded in a potential range between 0.5-2.5 V using
a LANDCT2001A computer-controlled galvanostat. A Solartron
model 1287 Electrochemical Interface was used for the cyclic vol-
tammetry measurements.
Results and Discussion
Phase analysis.—The XRD patterns of the Al3ϩ doped
Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15, 0.25) in comparison with the
pristine Li4Ti5O12 are shown in Fig. 1. When y р 0.15, the diffrac-
tion patterns of all materials obtained were similar, with all the
peaks indexable in the Fd3m space group with a cubic lattice. When
y was increased, small amounts of LiAlO2 were also detected, as
evidenced in Fig. 1d for Li4Al0.25Ti4.75O12 (y ϭ 0.25).
Experimental
Sample preparation.—The Al3ϩ substituted spinels were synthe-
sized according to the stoichiometric quantities of Li4.15AlyTi5ϪyO12
(y ϭ 0,0.10,0.15,0.25) by a solid-state reaction. Excess Li was pro-
vided to compensate for the loss of Li at the synthesis temperature.
Given amounts of r-TiO2 , Li2CO3 ͑Aldrich͒, and Al2O3 ͑Aldrich͒
were thoroughly ground in an agate mortar and pelletized, fired in a
muffle furnace at 950°C for 24 h in air, and then naturally cooled to
room temperature. The pellets were then ground again.
The cell parameters of the Li4AlyTi5ϪyO12 phase determined us-
¨
ing the PIRUM program based on the Guinier-Hagg film data are
shown in Table I. The a value of the undoped Li4Ti5O12 , 8.357 Å, is
in accordance with literature data and ICSD reference number
015787, ␭ ϭ 8.357 Å.13 For the Al3ϩ substituted systems, the lat-
tice parameters were smaller than that of the undoped Li4Ti5O12
,
and the lattice parameter gradually decreased with increasing Al3ϩ
content, which is probably due to the smaller ionic size of Al3ϩ
͑0.053 nm͒ compared to that of Ti3ϩ ͑0.067 nm͒.
The SEM images of samples Li4Ti5O12 , Li4Al0.10Ti4.90O12 , and
Li4Al0.15Ti4.85O12 are shown in Fig. 2. The agglomeration in the
* Electrochemical Society Active Member.
z E-mail: zywen@mail.sic.ac.cn
Downloaded on 2015-04-05 to IP 169.233.248.124 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑1͒ A186-A190 ͑2005͒
A187
Figure 1. XRD patterns of ͑a͒ Li4Ti5O12
Li4Al0.15Ti4.85O12 , and ͑d͒ Li4Al0.25Ti4.75O12
, ͑b͒ Li4Al0.10Ti4.90O12 , ͑c͒
.
Al3ϩ doped samples was clearly alleviated. But the particle sizes of
the Al3ϩ doped Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15) samples were
comparable to that of Li4Ti5O12 , both were about 1 ␮m. We suggest
that doping Al3ϩ improved the dispersion of Li4Ti5O12 powders but
did not change the particle size.
Electrochemical characteristics.—Cyclic voltammetry.—The
electrochemical behaviors of Li/Li4AlyTi5ϪyO12 prepared in this
work were characterized by cyclic voltammograms ͑CVs͒ with coin
cells. As shown in Fig. 3, the voltage was scanned from 0.5 to 2.7 V
and then back to 0.5 V at a scan rate of 5 mV/s. Note that all cycling
tests were made under the same conditions. The figures clearly dem-
onstrated the differences in redox behavior among the cells with
various cathodes made from Li4AlyTi5ϪyO12 (y ϭ 0, 0.10, 0.15).
In all three systems, the oxidation peaks were located at 2.1, 2.1,
and 1.9, and the corresponding reduction peaks around 1.33,
1.25, and 1.38 for the Li4Ti5O12
,
Li4Al0.1Ti4.9O12
,
and
Li4Al0.15Ti4.85O12 , respectively. For Li4Ti5O12 with the starting
material r-TiO2 prepared in this work, the peak was much broader,
reflecting
a profile typical of a disordered electrochemical
system.14,15 Compared to Li4Ti5O12 , the peaks of oxidation and
reduction of Al3ϩ doped Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15) became
clearly sharp, and the gap between the oxidation and reduction po-
tentials was small. These results indicate that lithium extraction and
insertion become easier with cathodes made from Al3ϩ doped
Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15) because the stable Al-O octahe-
dron gives rise to a more reversible Liϩ transfer in the three-
dimensional framework of Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15). That
only one electrochemical peak appeared for both the oxidation and
reduction processes indicates that both the insertion and extraction
processes of lithium ions occurred in one stage. All these results
demonstrate that Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15) is a promising
anodic electrode host.
Table I. Variations in lattice parameters as a function of the dop-
ant concentration in Li4AlyTi5ÀyO12 „y ϭ 0, 0.10, 0.15, 0.25….
Figure 2. SEM images of samples ͑a͒ Li4Ti5O12 , ͑b͒ Li4Al0.10Ti4.90O12 , and
͑c͒ Li4Al0.15Ti4.85O12
.
Sample
Lattice parameter a ͑Å͒
8.357
8.355
8.353
8.341
Li4Ti5O12
Constant current charge-discharge of Li/Li4AlyTi5ϪyO12 cells.—
The specific capacity of various Li4AlyTi5ϪyO12 samples was deter-
mined by charge-discharge cycling at a constant current density of
Li4Al0.10Ti4.90O12
Li4Al0.15Ti4.85O12
Li4Al0.25Ti4.75O12
Downloaded on 2015-04-05 to IP 169.233.248.124 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A188
Journal of The Electrochemical Society, 152 ͑1͒ A186-A190 ͑2005͒
Figure 4. Discharge curves of the first cycle of Li/Li4AlyTi5ϪyO12 (y
ϭ 0, 0.10, 0.15) cells employing 1 M solution of LiPF6 in EC:DMC ͑1:1
v/v͒ as the electrolyte and Li4AlyTi5ϪyO12 (y ϭ 0, 0.10, 0.15) samples as
the cathode. Current density: 0.15 mA/cm2.
discharge capacities were 164.5, 169.0, and 195.6 mAh/g with
columbic efficiencies of 99.8, 99.9, and 99.8% for Li4Ti5O12
,
Li4Al0.1Ti4.9O12 , and Li4Al0.15Ti4.85O12 , respectively. In addition,
the discharge capacities corresponding to the 1.53 V platform were
122.5, 130.9, and 126 mAh gϪ1, about 74.5, 77.5, and 64.4% of the
total discharge capacities for Li4Ti5O12 , Li4Al0.1Ti4.9O12 , and
Li4Al0.15Ti4.85O12 , respectively. As the quantity of Al3ϩ increased,
the capacity of the cell increased at the first discharge. But at y
ϭ 0.25, the capacities in the initial several cycles decreased
quickly, which resulted from the impure phases. Moreover, in addi-
tion to the 1.5 V platform, another one at 0.68 V was observed in all
samples, and its capacity increased with increasing amounts of dop-
ing Al3ϩ. However, the reverse of this platform occurs around 1.7
16
VLi in conjunction with the oxidation of Ti3ϩ to Ti4ϩ
,
which sug-
gests a multistep restore of Ti4ϩ during the discharge period. As
described, clearly, Al3ϩ doped Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15)
improved the first discharge capacity of the spinel Li4Ti5O12 . We
explain this effect as follows: aluminum is lightweight, and there-
fore improves the specific capacity. Doping Al3ϩ changed the
chemical environment of the ions in the spinel structure and in-
creased the amount of Ti4ϩ content available for redox, thus increas-
ing the specific capacity.
The charge and discharge curves of ͑a͒ Li4Ti5O12 ͑b͒
Li4Al0.10Ti4.90O12 , and ͑c͒ Li4Al0.15Ti4.85O12 for every 10 cycles are
shown in Fig. 5. Each sample has two platforms corresponding to
the 1.53 and 0.68 V platforms. With an increasing doping Al3ϩ
content, the discharge capacity corresponding to the 1.53 V platform
decreases, but the 0.68 V platform capacity increases. The 10th dis-
charge capacity was 119, 111, and 104 mAh/g, and the 50th dis-
charge capacity was 109, 105, and 92 mAh/g at 1.53 V for
Li4Ti5O12 , Li4Al0.10Ti4.90O12 , and Li4Al0.15Ti4.85O12 , respectively.
Although doping Al3ϩ improved the total discharge capacity, it re-
duced the discharge capacity of the 1.53 V platform. Comparing the
first discharge, we found that with an increased number of cyclings,
the 1.53 V platform discharge capacity decreased more quickly in
the Al3ϩ doped samples. However, with an increased doping Al3ϩ
amount, the fade of the 1.53 V platform discharge increased.
Cycling stability is an important and desirable feature in most
battery applications. Figure 6 shows the variation of the discharge
capacities for Li/Li4AlyTi5ϪyO12 (y ϭ 0, 0.10, 0.15) cells with the
number of cycles. The electrochemical performance of Li4Ti5O12
prepared from the starting material r-TiO2 in this work was as
Figure 3. CVs of cells employing 1 M solution of LiPF6 in EC:DMC ͑1:1
v/v͒ as the electrolyte and ͑a͒ Li4Ti5O12 , ͑b͒ Li4Al0.10Ti4.90O12 , and ͑c͒
Li4Al0.15Ti4.85O12 samples as the cathode. Scan rate: 5 mV/s.
0.15 mA/cm2 between the cutoff voltages of 2.5 and
0.5 V. The first discharge characteristics of cells employing 1 M
solution of LiPF6 in EC:DMC ͑1:1 v/v͒ as the electrolyte and
Li4AlyTi5ϪyO12 (y ϭ 0, 0.10, 0.15) samples as the cathode are
shown in Fig. 4. From the figure, we can see that the voltage
dropped quickly to below 2 V and decreased as the reaction pro-
ceeded until the voltage reached about 1.53 V, after which it was
almost independent of the degree of reduction.2,16,17 The first
Downloaded on 2015-04-05 to IP 169.233.248.124 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑1͒ A186-A190 ͑2005͒
A189
Figure 6. Variation of the discharge capacities for Li/Li4AlyTi5ϪyO12 (y
ϭ 0, 0.10, 0.15) cells with the number of cycles between 0.5 to 2.5 V at a
current density of 0.15 mA/cm2.
Li4Al0.1Ti4.9O12 , and Li4Al0.15Ti4.85O12 were 154.1, 159.5, and
166.9 mAh/g with capacity losses of 2.65, 3.1, and 3.86%, respec-
tively, compared to the corresponding second cycle. The 30th charge
and discharge capacities and coulombic efficiencies of Li4Ti5O12
Li4Al0.1Ti4.9O12 , and Li4Al0.15Ti4.85O12 are listed in Table II.
,
Clearly, Li4Al0.15Ti4.85O12 exhibited the largest reversible capac-
ity and the highest coulombic efficiency among the three samples.
A large irreversible loss between the first charge and discharge
capacity was observed for all the samples, probably due to the in-
complete recovery of the fully intercalated phase after the first
charge or some possible decomposition of the electrolyte solution.
Conclusion
The charge and discharge cycling of the cells showed that the
electrochemical performance of Li4Ti5O12 with r-TiO2 as the start-
ing material prepared in this work was as good as that produced
from a-TiO2 . The electrochemical performance of Li4Ti5O12 did not
change at high temperature, although the insertion into rutile re-
quired a larger distortion of the crystal lattice. Because r-TiO2 is a
natural material, the cost of materials can be greatly reduced. Al3ϩ
doped Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15) exhibited much better
electrochemical performance in comparison with the undoped
Li4Ti5O12
. Among the three samples of Li4AlyTi5ϪyO12 (y
ϭ 0, 0.10, 0.15), Li4Al0.15Ti4.85O12 exhibited the largest reversible
capacity and the highest coulombic efficiency, with a discharge ca-
pacity in the first and second cycles of 195.6 and 173.6 mAh/g,
respectively. It remained 166.9 mAh/g after 30 cycles with a capac-
ity loss of 3.86% compared to the second cycle, and its coulombic
efficiency in the 30th cycle was 99.2%.
Acknowledgment
This work was financially supported by the key project of the
Natural Science Foundation of China ͑NSFC͒ no. 20333040.
Figure 5. The charge and discharge curves of ͑a͒ Li4Ti5O12
Li4Al0.10Ti4.90O12 , and ͑c͒ Li4Al0.15Ti4.85O12 for every 10 cycles.
, ͑b͒
Table II. The 30th charge and discharge capacity and coulombic
efficiency of Li4Ti5O12 , Li4Al0.1Ti4.9O12 , and Li4Al0.15Ti4.85O12
.
Charge capacity Discharge capacity
Reversible
Sample
͑mAh/g͒
͑mAh/g͒
efficiency ͑%͒
good as that produced by a-TiO2 . Furthermore, Al3ϩ doped
Li4AlyTi5ϪyO12 (y ϭ 0.10, 0.15) exhibited a higher reversible ca-
pacity but a slightly decreased cycling stability compared to
155.5
161.9
168.2
154.1
159.5
166.9
99.1
98.5
99.2
Li4Ti5O12
Li4Al0.10Ti4.90O12
Li4Al0.15Ti4.85O12
Li4Ti5O12 . The reversible capacities after 30 cycles for Li4Ti5O12
,
Downloaded on 2015-04-05 to IP 169.233.248.124 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A190
Journal of The Electrochemical Society, 152 ͑1͒ A186-A190 ͑2005͒
The Chinese Academy of Sciences assisted in meeting the publication
costs of this article.
8. T. Ohzuku, K. Tatsumi, N. Matoba, and K. Sawai, J. Electrochem. Soc., 147, 3592
͑2000͒.
9. P. Reale, S. Panero, F. Ronci, V. Rossi Albertini, and B. Scrosati, Chem. Mater., 15,
3437 ͑2003͒.
10. I. Belharouak and K. Amine, Electrochem. Commun., 5, 435 ͑2003͒.
11. W. Xiaomei, Z. Xiangfu, Y. Qinghe, J. Zhongkao, and W. Haoqing, J. Fluorine
Chem., 107, 39 ͑2001͒.
References
1. A. Guerfi, S. Sevigny, M. Lagace, P. Hovington, K. Kinoshita, and K. Zaghib, J.
Power Sources, 119-121, 88 ͑2003͒.
2. D. Peramunage and K. M. Abraham, J. Electrochem. Soc., 145, 2609 ͑1998͒.
3. K. Nakahara, R. Nakajima, T. Matsushima, and H. Majima, J. Power Sources, 117,
131 ͑2003͒.
4. K. Zaghib, M. Simoneau, M. Armand, and M. Gauthier, J. Power Sources, 81-82,
300 ͑1999͒.
5. L. Kavan, D. Fattakhova, and P. Krtil, J. Electrochem. Soc., 146, 1375 ͑1999͒.
6. C. H. Chen, J. T. Vaughey, A. N. Jansen, D. W. Dees, A. J. Kagaian, T. Goacher,
and M. M. Thackeray, J. Electrochem. Soc., 148, A102 ͑2001͒.
7. P. Kubiak, A. Garcia, M. Womes, L. Aldon, J. Olivier-Fourcade, P. Lippens, and J.
Jumas, J. Power Sources, 119-121, 626 ͑2003͒.
12. P. E. Werner, Arkiv Fur Kemi, 31, 513 ͑1964͒.
13. A. Deschanvres, B. Raveau, and Z. Sekkal, Mater. Res. Bull., 6, 699 ͑1971͒.
14. A. Stashans, S. Lunell, R. Bergstrom, A. Hagfeldt, and S. E. Lindquist, Phys. Rev.
B, 53, 159 ͑1996͒.
15. W. Liu, G. C. Farrington, F. Chaput, and B. Dunn, J. Electrochem. Soc., 143, 879
͑1996͒.
16. A. D. Robertson, H. Tukamoto, and J. T. S. Irvine, J. Electrochem. Soc., 146, 3958
͑1999͒.
17. E. Ferg, R. J. Gummow, A. De Kock, and M. M. Thackeray, J. Electrochem. Soc.,
141, L147 ͑1994͒.
Downloaded on 2015-04-05 to IP 169.233.248.124 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).