Electrochimica Acta 54 (2009) 4655–4661
Contents lists available at ScienceDirect
Electrochimica Acta
journa l homepage: www.e lsev ier .com
Compa by
and co-
Wenbin
a Department o
b School of Met a
a r t i c l
Article history:
Received 24 D
Received in re
Accepted 24 M
Available onlin
Keywords:
Aluminum sub
Lithium-ion ba
Aluminum sub
materials
Co-precipitatio
Solid-state synthesis
ere
ith z
-stat
tent
les w
=0.95
s bet
meas
Litera
e ra
samp
thermal stability of delithiated Li[Co1−zAlz]O2 in electrolyte.
© 2009 Elsevier Ltd. All rights reserved.
1. Introdu
The saf
ued succe
are presen
where the
can be ma
often incor
LiFePO4 an
have small
mon layer
research fo
itive electr
only a sma
through su
Li[Ni1/2−zM
The im
ity of lay
For exam
Li[Ni1−zAlz
material w
∗ Correspo
E-mail ad
0013-4686/$
doi:10.1016/j
ction
ety of lithium-ion batteries is central to their contin-
ss in the market place. LiCoO2-based Li-ion batteries
tly dominant in cell-phone and computer applications
thermal instability of delithiated LiCoO2 in electrolyte
naged due to the small size of the cells. Larger cells
porate less expensive and less reactivematerials such as
d LiMn2O4 positive electrode materials. However these
er volumetric energydensity thanLiCoO2 andother com-
ed positive electrode materials. There has recently been
cusing on improving the thermal stability of layeredpos-
ode materials so they are more benign than LiMn2O4 at
ll penalty in energy density. This has been accomplished
bstitutions of Al in Li[Ni1/3Mn1/3Co1/3−zAlz]O2 [1] and
n1/2−zAl2z]O2 [2].
pact of Al substitutions on the thermal stabil-
ered lithium transition metal oxides is well-known.
ple, Ohzuku’s group showed that Al-substitution in
]O2 reduced the reactivity of the charged electrode
ith electrolyte [3]. Al substitutions have been used to
nding author. Tel.: +1 902 494 2991; fax: +1 902 494 5191.
dress: jeff.dahn@dal.ca (J.R. Dahn).
make Li[Ni0.8Co0.15Al0.05]O2 (NCA) with improved thermal sta-
bility compared to the parent material without Al. Ceder et al.
have reported that Al-substituted LiCoO2 has a higher average
potential versus Li than LiCoO2 according to the first principles
calculation and that it has better thermal stability [4,5]. In spite
of these well-known results, the mechanism for the thermal sta-
bility improvement is not understood in detail. In addition, the
impact of aluminum on the kinetics of the reaction between
delithiated positive electrode materials and electrolyte at high
temperature is not known. These reaction kinetics are required
for models of the response of full Li-ion cells to abuse scenarios
[6–8].
MacNeil et al. have made substantial studies of the kinetics of
the reactions between delithiated LiCoO2 and electrolyte at ele-
vated temperature [9,10]. It is our goal to build on this earlier
research by studying the impact of Al on the kinetics of reaction
between delithiated Li[Co1−zAlz]O2 and electrolyte. In such studies,
Li[Co1−zAlz]O2 would serve as amodel systemandhopefully results
learned could be applied to other Al-substituted layered transition
metal oxides. As the first step, a careful characterization of synthe-
sis, structure and electrochemical properties of Li[Co1−zAlz]O2 is
required. Scattered results can be found in the literature already,
and will be referred to as appropriate below, but none is com-
plete to the level we would like. In this paper, the co-precipitation
methodand the traditional solid-statemethodwereused toprepare
Li[Co1−zAlz]O2. We document the structure and electrochemical
– see front matter © 2009 Elsevier Ltd. All rights reserved.
.electacta.2009.03.068
rative study of Li[Co1−zAlz]O2 prepared
precipitation methods
Luoa,b, J.R. Dahna,∗
f Physics and Atmospheric Science, Dalhousie University, Halifax B3H3J5, Canada
allurgical Science and Engineering, Central South University, Changsha 410083, PR Chin
e i n f o
ecember 2008
vised form 22 March 2009
arch 2009
e 5 April 2009
stituted LiCoO2
tteries
stituted positive electrode
n
a b s t r a c t
Li[Co1−zAlz]O2 (0≤ z≤0.5) samples w
lattice constants varied smoothly w
state samples above z=0.2. The solid
distribution when the aluminum con
stoichiometric Lix[Co0.9Al0.1]O2 samp
nominal compositions x= Li/(Co+Al)
ples suggest the solid solution limit i
Li[Co1−zAlz]O2 samples were used to
be about −250±30 (mAh/g)/(z=1).
Li[Mn2−yAly]O4 demonstrates the sam
serve as baseline characterization of
/ locate /e lec tac ta
solid-state
prepared by co-precipitation and solid-state methods. The
for the co-precipitated samples but deviated for the solid-
e method may not produce materials with a uniform cation
is large or when the duration of heating is too brief. Non-
ere synthesized by the co-precipitation method at various
, 1.0, 1.1, 1.2, 1.3. XRD patterns of the Lix[Co0.9Al0.1]O2 sam-
ween Li/(Co+Al) = 1.1 and 1.2. Electrochemical studies of the
ure the rate of capacity reduction with Al content, found to
ture work on Li[Ni1/3Mn1/3Co1/3−zAlz]O2, Li[Ni1−zAlz]O2 and
te of capacity reduction with Al/(Al +M) ratio. These studies
les to be used to determine the impact of Al content on the
4656 W. Luo, J.R. Dahn / Electrochimica Acta 54 (2009) 4655–4661
performance of Li[Co1−zAlz]O2 prepared by twomethods and com-
pare to literature results.
2. Experimental
2.1. Material preparation
2.1.1. Li[Co1−zAlz]O2 samples prepared by the co-precipitation
method
A LiOH·H2O (Sigma Aldrich, 98%) solution and a mixed solution
of Co(NO3)2·6H2O (Sigma Aldrich, 98%) and Al(NO3)3·9H2O (Sigma
Aldrich, 98%) were simultaneously added over the course of about
30min to a stirredflask using a two-channel peristaltic pump (Mas-
terflex C/L pump, Barnant Co.). The concentrations of the solutions
were adjusted to set the Al:(Al +Co) ratio, z. The precipitate was fil-
tered out and washed with distilled water several times to remove
any dissolved salts and dried at 80 ◦C overnight. The variation in
structural properties and chemical composition of the precipitate
with Al content, z, have been reported in another publication [11].
The dried precipitate was mixed with an appropriate amount
of Li2CO3 (Alfa Aesar, 99%) and ground. The precursors were
heated in air. Samples were heated at selected temperatures for
selected times to be described below. Samples with 0≤ z≤0.5
in Li[Co1−zAlz]O2 were prepared. In addition, non-stoichiometric
Lix[Co0.9Al0.1]O2 samples were synthesized at various nominal
compositions x= Li/(Co+Al) = 0.95, 1.0, 1.1, 1.2, 1.3.
2.1.2. Li[Co1−zAlz]O2 samples prepared by the solid-state method
The starting materials for synthesis were Li2CO3 (Alfa Aesar,
99%), Co3O4 (Alfa Aesar, 99.7%) and Al(OH)3 (Sigma Aldrich). Sto-
ichiometric amounts of Co- and Al-containing starting materials
along with an appropriate amount Li2CO3 were mixed and ground
togetherusing anautomatic grinder (RetschRM-0). Finally the sam-
ples were heated at a selected temperature for a selected time in
air.
2.2. Material characterization
2.2.1. X-ray diffraction
XRD patterns were collected with a Siemens D5000 diffrac-
tometer equipped with a Cu target x-ray tube and a diffracted
beammonochromator. Li[Co1−zAlz]O2 samplesweremeasuredover
a scattering angle range between 10◦ and 90◦ using 0.05◦ steps and
a 10 s counting time.
2.2.2. Electrochemical testing
Coin cells (23mm diameter and 2.5mm thick) were used for
testing the electrochemical performance of the samples. Positive
electrodes were prepared by mixing the active material, Super S
Carbon Black, and PVDF in a weight ratio of 90:5:5. An appropriate
amount of NMP was added to the mixture to form a slurry, which
was then thoroughly mixed. The slurry was coated on Al foil and
dried overnight in an oven. The electrochemical cells used a single
lithium metal foil as the counter electrode, Celgard 2320 microp-
orous film as the separator and 1MLiPF6 in ethylene carbonate (EC)
and diethyl carbonate (DEC) (1:2, v/v) as the electrolyte. Cells were
assembled in an argon-filled glove box. The cells were removed
from the glove box and connected to a computer-controlled charg-
ing system (E-One/Moli Energy). The cells were initially charged
and discharged at a C/20 rate between 3.3 and 4.3V versus Li metal
for two cycles. After the first two cycles, the charge–discharge rate
was increased to C/5.
2.2.3. SEM testing
A Hitachi S4700 field-emission scanning electron microscope
(SEM) was used to image the materials.
Fig. 1. XRD pa he co-
samples are in lumn
top panels.
tterns of LiCo1−zAlzO2 (0≤ z≤0.5) synthesized by both the solid-state method and t
the left column and the patterns of the co-precipitation samples are in the right co
precipitation method at 900 ◦C for 3h. The patterns of the solid-state
. Three Bragg peaks are indexed based on the LiCoO2 structure in the
W. Luo, J.R. Dahn / Electrochimica Acta 54 (2009) 4655–4661 4657
Fig. 2. SEM images of LiCo0.7Al0.3O2 synthesized by the solid-state method A, B and the co-precipitation method C, D.
Fig. 3. XRD patterns of LiCo0.8Al0.2O2 synthesized by both the solid-state method
and the co-precipitation method. The XRD pattern shown in the bottom panel is
for a sample synthesized by the co-precipitation method and then heated at 750 ◦C
for 3h. The remaining XRD patterns are for samples synthesized by the solid-state
method at 750 ◦C with different heating times.
3. Results and discussion
Fig. 1 s
LiCo1−zAlzO
state and co
single phas
broader Bra
show the s
gesting the
panels may
Bragg peak
imageswer
Fig. 2 sh
LiCo0.7Al0.3
state samp
Fig. 4. Lattice
for samples pr
and by the sol
hows the XRD patterns (between 60◦ and 75◦) of
2 (0≤ z≤0.5) synthesizedat900 ◦C for3hbyboth solid-
-precipitationmethods. All samples appear at first to be
e, however the solid-state samples show substantially
gg peaks for z>0.2. The right panels of Fig. 1 clearly
mooth shift of the Bragg peak positions versus z, sug-
broad Bragg peaks for the solid-state samples in the left
be caused by non-uniform Al distribution. Increased
width can also be caused by smaller grain size, so SEM
eused to ensure that theparticle sizeswere comparable.
ows scanning electron micrographs of two samples of
O2. Fig. 2A and B show the morphology of the solid-
le. Fig. 2C and D show the morphology of the sample
constants, a [panel A] and c [panel B] of Li[Co1−zAlz]O2 versus z. Data
epared by the co-precipitation method (750 ◦C, 20h and 900 ◦C, 3h)
id-state method (900 ◦C, 3h) are included.
4658 W. Luo, J.R. Dahn / Electrochimica Acta 54 (2009) 4655–4661
Fig. 5. Lattice constants versus z for Li[Co1−zAlz]O2 samples prepared by the co-
precipitation method (900 ◦C, 3h) compared to previous literature reports. —co-
precipitation method (900 ◦C; 3h) are compared to the results from �—ref. [12],
�—ref. [13], �—ref. [14], �—ref. [15], +—ref. [16], —ref. [17].
made by the co-precipitation route. The sample synthesized by
the solid-state method shows crystallites with flat facets, having
primary particle sizes around 2�m. The sample synthesized by co-
precipitation route shows a similar morphology, but has primary
particles that are around 1�m. The solid-state sample has larger
particle size, but broader Bragg peaks, suggesting that non-uniform
cation distr
Fig. 3 sh
both the so
sample in b
method at 7
ing panels
method at
are observe
the solid-st
Fig. 7. Lattice constants of Lix[Co0.9Al0.1]O2 (x=0.95, 1.0, 1.1, 1.2 and 1.3) synthesized
by the co-precipitation method (900 ◦C, 3h).
16h. A pure phase of LiCo0.8Al0.2O2 is only obtained by heating
20h at 750 ◦C for the solid-state method. Based on the results in
Fig. 3, it can be concluded that shorter heating times are required
for samples prepared by the co-precipitation method, as expected.
Fig. 4 shows the lattice constants (a and c) as a function of z in
LiCo1−zAlzO2 for samples prepared by the solid-state (900 ◦C, 3h)
and co-precipitation methods (both 900 ◦C, 3h and 750 ◦C, 20h).
The lattice constantswere determined by least squares refinements
meas
e lat
tice
s bu
ugge
h a u
, con
5
zAlz
Fig. 6. XRD pa
shown. The lef
for all samples
ibution is the cause as was suggested above.
ows the XRD patterns of LiCo0.8Al0.2O2 synthesized by
lid-state method and the co-precipitation method. The
ottom panel was synthesized by the co-precipitation
50 ◦C for 3h. The sample is a single phase. The remain-
in Fig. 3 show samples synthesized by the solid-state
750 ◦C for different times (3–28h). Some Li2CO3 peaks
d in the XRD patterns of LiCo0.8Al0.2O2 synthesized by
ate method and heated at 750 ◦C for 3h, 6h, 12h and
to the
ple. Th
The lat
sample
again s
als wit
is large
Fig.
Li[Co1−
tterns of Lix[Co0.9Al0.1]O2 (x=0.95, 1.0, 1.1, 1.2 and 1.3) synthesized by the co-precipitat
t panels show the presence of Bragg peaks from Li2CO3 in the x=1.2 and 1.3 patterns. The r
.
ured positions of at least 10 Bragg peaks for each sam-
tice constant a decreases and c increases as z increases.
constants vary smoothly with z for the co-precipitated
t deviate for the solid-state samples above z=0.15. This
sts that the solid-statemethodmaynot producemateri-
niform cation distribution when the aluminum content
sistent with the results in Fig. 1.
shows the variation of the lattice constants of
]O2 (co-precipitation method, 900 ◦C, 3h) compared to
ion method (900 ◦C, 3h) Two regions of the diffraction patterns are
ight panels show that the 018, 110 and 111 Bragg peaks are crystalline
W. Luo, J.R. Dahn / Electrochimica Acta 54 (2009) 4655–4661 4659
Fig. 8. The potential (V) versus specific capacity (mAh/g) of Li/Li[Co1−zAlzO2]cells cycled between 3.3 and 4.3V at a rate of C/20. Samples prepared by the solid-state method
(900 ◦C, 3h) are shown in the left columns and samples prepared by the co-precipitation method (900 ◦C, 3h) are shown in the right panels.
literature results [12–17]. The literature results are for samples
prepared by a variety of methods, including the sol–gel method,
the emulsion drying method, the co-precipitation method, the
citrate prec
methods (th
precipitationmethodand the citrate precursormethod) that ensure
good Co-Al cation mixing in the precursor yield results in good
agreement with our work. By contrast, the data set ( ) in Fig. 5
in p
ds, w
Fig. 9. The dif
method (900 ◦
ursor method and the solid-state method. In general,
e sol–gel method, the emulsion dryingmethod, the co-
that is
metho
ferential capacity (dQ/dV) versus potential (V) of Li/Li[Co1−zAlz]O2 cells cycled between
C, 3h) are shown in the left columns and samples prepared by the co-precipitation meth
oor agreement is for samples prepared by solid-state
hich may not lead to good cation mixing.
3.3 and 4.3V at a rate of C/20. Samples prepared by the solid-state
od (900 ◦C, 3h) are shown in the right panels.
4660 W. Luo, J.R. Dahn / Electrochimica Acta 54 (2009) 4655–4661
Fig. 6 shows the XRD patterns of Lix[Co0.9Al0.1]O2 (x=0.95, 1.0,
1.1, 1.2, 1.3) synthesizedat900 ◦C for3hbyco-precipitationmethod.
The value x in the figure shows the nominal Li/(Co+Al) molar ratio.
The left panels in Fig. 6 show the portion of the XRD pattern where
the Li2CO3
right panel
resolved K�
single phas
0.95≤ x≤1.
region was
Thus, the ad
non-stoichi
Fig. 7 sh
molar ratio
with x. This
LixCoO2 sam
Fig. 8 sh
having Li[Co
from solid-
those from
column. Th
the increasi
cannot be o
ing. Therefo
content of a
Fig. 8 sh
the smalles
versible ca
irreversible
strongly for
x-ray diffra
the irrevers
be more cha
origin of th
Fig. 9 s
described b
are describ
are describe
associated w
ened and sh
solid-state
to z=0.25,
samples is n
agreement
dQ/dV for t
the second
the reasons
Fig. 10a
(4.3–3.3V,
for the soli
samples (90
sample typ
solid-state
precipitated
in the solid
average am
dashed line
samples in
samples sho
as the sam
als designer
with the inc
The imp
ties of Li[N
[20] and Li
Li[Co1−zAlz]
Panel a) First discharge capacities as a function of z in Li[Co1−zAlz]O2 (z=0,
, 0.15, 0.2, 0.25,0.3, 0.4, and 0.5). Samples were prepared by co-precipitation
state methods as indicated in the legend. The discharge capacities for sam-
ared in thiswork:� solid-state (900 ◦C, 3h), co-precipitation (900 ◦C, 3h),
o-precipitation (750 ◦C, 20h) are compared to previous literature work—�
� ref. [17], + ref. [19]. Panel b) First discharge capacities versus z=Al/(Al +M)
amples prepared in this work (same as Panel a) compared to different Al-
ositive electrodematerials in the literature: LiNi1/3Mn1/3Co(1/3−z)AlzO2 ref.
Ni1−zAlzO2 ref.[20]; LiMn2−yAlyO4 ref.[21].
s been renormalized and plotted as z= y/2 along the x-axis so
lmaterials canbe comparedon the samegraph. The impact of
tions on cell capacity for all these systems, both layered-type
inel-type is basically identical. If each addedAl atom reduced
acitybyone lithiumatom, then the slopeof the lines through
data in Fig. 10a and b should be about −280 (mAh/g)/(z=1)
approximately matches the data. This “universality” is pre-
ly a result of the fact that Al3+ cannot be oxidized while
er cations in the samples described in Fig. 10 undergo an
e oxidation state change of +1 when one Li per metal is
ed.
clusions
o1−zAlz]O2 samples were prepared by co-precipitation and
tatemethods. Cationuniformityasmeasuredbysharpnessof
eaks, smooth variation of lattice constantswith z and differ-
capacity versus potential was better for the co-precipitated
s. Lattice constant versus z behaviour for samples prepared
co-precipitation routewas found toagreewellwith literature
ses where uniform cation mixing was ensured.
-stoichiometric Lix[Co0.9Al0.1]O2 samples with x=0.95, 1.0,
and 1.3were prepared by the co-precipitationmethod. Sam-
ith x<1.2 were found to be single phase.
impurity appears for the x=1.2 and x=1.3 samples. The
s show that the XRD peaks remain sharp with well-
doublets in all cases. The results in Fig. 6 suggest that
e samples can be prepared under these conditions for
1. Delmas et al. [18] showed that a similar single phase
possible for LixCoO2 samples synthesized at 900 ◦C.
dition of Al does not strongly impact the single phase
ometric range.
ows the lattice constants, a and c, plotted versus x, the
Li/(Co+Al). The lattice constants basically do not vary
is in agreement with the findings of Delmas et al. for
ples [See Table 1 in reference [18]].
ows the first charge/discharge cycling curves for the cells
1−zAlz]O2 electrodes. The results from electrodesmade
state samples (900 ◦C, 3h) are in the left column while
co-precipitated samples (900 ◦C, 3h) are in the right
e reversible capacity of Li[Co1−zAlz]O2 decreases with
ng Al content. As is well-known [5] this is because Al3+
xidizedwhile Co3+ can be oxidized to Co4+ during charg-
re, substitution of Al3+ for Co3+ in LiCoO2 decreases the
ctive species.
ows that the co-precipitated sample with z=0 has
t charge–discharge polarization and the smallest irre-
pacity of all the samples. Fig. 8 also shows that the
capacity increases with z and that it increases more
the samples prepared by co-precipitation. Based on the
ction studies in Figs. 1 and 4, we believe the variation of
ible capacity with z for the co-precipitated samples to
racteristic of samples with uniform cation mixing. The
is irreversible capacity needs to be understood.
hows the differential capacity of the same samples
y Fig. 8 plotted versus potential. The solid-state samples
ed by the left column and the co-precipitated samples
d by the right column. The sharp peak in dQ/dV at z=0,
ith the insulator–metal transition [5] is rapidly broad-
ifted to higher potential as z increases. However, in the
samples, a small remnant peak exists, even all the way
again suggesting that cation mixing in the solid-state
ot complete. The average potential increases with z, in
with previous reports [5]. One curious feature is that
he first charge becomes quite different from dQ/dV for
charge as z increases. Presumably this is associatedwith
for the irreversible capacity and deserves more study.
shows the variation of