Combined effects of Ni and Li doping on the phase transitions in LixCoO2 electrochemical and 7Li nuclear magnetic resonance studies

Article abstract of DOI:10.1149/1.1516219

High temperature Li_x0Co_1-yNi_yO₂ (x₀ = 1.0, 1.10; y = 0.0, 0.03, 0.06, and 0.10) phases were synthesized by solid-state chemistry. Their characterization of X-ray diffraction and galvanostatic measurements shows that 3% of Ni ions substituted for Co in the LiCoO₂ lattice suppress the two-phase domain, related to the semiconductor-to-metal transition, that is observed at the beginning of the charge process in Li_xCoO₂. These ions, trapped in the lattice, prevent the phase separation. On the other hand, more than 10% of Ni ions need to be substituted for Co in order to inhibit the monoclinic distortion due to a lithium/vacancy ordering in the interslab for Li_0.50Co_1-yNi_yO₂. Besides, a Li/(Ni + Co) ratio (x₀) strictly higher than one in Li_x0Co_0.97Ni_0.03O₂ leads, as in the case of the unsubstituted Li_1.10CoO₂ phase, to the disappearance of all the phase transitions upon deintercalation. ⁷Li magic angle spinning nuclear magnetic resonance measurements show that Ni^III ions are the only paramagnetic species in the LiCo_1-yNi_yO₂ phases while in the overlithiated Li_x0Co_1-yNi_yO₂ (x₀ > 1.0) phases, Ni^III and intermediate spin Co^3+(IS) are present. This suggests the existence of structural defects associated with O vacancies which are responsible for the suppression of the electronic delocalization and of the lithium/vacancy ordering upon lithium deintercalation.

Full text of DOI:10.1149/1.1516219

Journal of The Electrochemical Society, 149 ͑12͒ A1533-A1540 ͑2002͒
A1533
0013-4651/2002/149͑12͒/A1533/8/$7.00 © The Electrochemical Society, Inc.
Combined Effects of Ni and Li Doping on the Phase
Transitions in LixCoO2
7
Electrochemical and Li Nuclear Magnetic Resonance Studies
,z
´ ´
S. Levasseur, M. Menetrier, and C. Delmas
*
`
´
´
Institut de Chimie de la Matiere Condensee de Bordeaux-CNRS and Ecole Nationale Superieure de Chimie et
Physique de Bordeaux, 33608 Pessac cedex, France
High temperature Lix Co1ϪyNiyO2 (x0 ϭ 1.0, 1.10; y ϭ 0.0, 0.03, 0.06, and 0.10͒ phases were synthesized by solid-state chem-
istry. Their characteri0zation by X-ray diffraction and galvanostatic measurements shows that 3% of Ni ions substituted for Co in
the LiCoO2 lattice suppress the two-phase domain, related to the semiconductor-to-metal transition, that is observed at the
beginning of the charge process in LixCoO2 . These ions, trapped in the lattice, prevent the phase separation. On the other hand,
more than 10% of Ni ions need to be substituted for Co in order to inhibit the monoclinic distortion due to a lithium/vacancy
ordering in the interslab for Li0.50Co1ϪyNiyO2 . Besides, a Li/(Ni ϩ Co) ratio (x0) strictly higher than one in Lix Co0.97Ni0.03O2
0
leads, as in the case of the unsubstituted Li1.10CoO2 phase, to the disappearance of all the phase transitions upon deintercalation.
7Li magic angle spinning nuclear magnetic resonance measurements show that NiIII ions are the only paramagnetic species in the
III
1.0) phases, Ni and intermediate spin Co3ϩ(IS) are
Ͼ
LiCo1ϪyNiyO2 phases while in the overlithiated Lix Co1ϪyNiyO2 (x0
0
present. This suggests the existence of structural defects associated with O vacancies which are responsible for the suppression of
the electronic delocalization and of the lithium/vacancy ordering upon lithium deintercalation.
© 2002 The Electrochemical Society. ͓DOI: 10.1149/1.1516219͔ All rights reserved.
Manuscript submitted December 31, 2001; revised manuscript received May 20, 2002. Available electronically October 18, 2002.
Experimental
LiCoO2 is the most widely used positive electrode material in
commercial Li-ion batteries.1,2 The high-temperature form of
The Lix Co1ϪyNiyO2 (x0 ϭ 1.0, 1.10; y ϭ 0.0, 0.03, 0.06, and
0
¯
LiCoO2 crystallizes in the trigonal system ͑space group: R3m) with
0.10͒ materials were prepared by calcination at 600°C under O2 for
ˆ
the ideal layered ␣-NaFeO2 structure in which LiO6 and CoO6 oc-
tahedra share their edges and are stacked alternately along the c axis
direction with an AB CA BC oxygen packing.3 Numerous investi-
gations have reported the phase changes upon lithium deintercala-
tion in LixCoO2 , with the existence of a two-phase domain for
0.75 р x р 0.94 due to a macroscopic semiconductor-to-metal
transition and the presence of a monoclinic distortion for x ϭ 0.50
due to the establishment of an interslab Li/vacancy ordering.1,4-7
However, since the earliest work, in an effort to improve its electro-
chemical performances, substitution of numerous metallic elements
for cobalt has been attempted ͑Cr,8 Al,9,10 Mg,11,12 Fe,13 or Mn14,15͒.
However, only the derivatives of the mixed LiCo1ϪyNiyO2 oxide,
for which a complete solid solution exists between the two end
members LiCoO2 and LiNiO2 , seem to be an interesting candidate
from the electrochemical point of view.5,16,17
12 h of Li2Co3 ͑Rhone Poulenc Rectapur % min 99͒, Co3O4 ͓calci-
nation at 450°C for 12 h under O2 of Co(NO3)2•6H2O Carlo Erba
% min 99͔, and NiO ͑calcination at 450°C for 12 h under O2 of
Ni(NO3)2•6H2O Carlo Erba % min 99͒; x0 denotes the nominal
Li/(Co ϩ Ni) ratio of the mixture; two compositions were consid-
ered in this study, x0 ϭ 1.0 and 1.10. Two successive heat-
treatments at 900°C for 24 h with an intermediate grinding were
done after the calcination.
Electrochemical measurements were carried out at room
temperature ͑22°C͒ with Li/LiClO4-propylene carbonate
(PC)/LixCo1ϪyNiyO2 cells. The positive electrode consisted of a
mixture of 88% by weight active material, 2% polytetrafluorethyl-
ene ͑PTFE͒, and 10% carbon black. The first series of cells, as-
sembled in an argon-filled dry box, was charged at 100 ␮A cmϪ2
(mLiCoO ϭ 30 mg); the second series, for long-range cycling, was
cycled at 400 ␮A cmϪ2 (mLiCoO ϭ 15 mg, C/20 rate͒. The
2
2
In this context, in 1993 Reimers et al. studied the effect of Ni
doping in LiCoO2 . They showed that 2% of the nickel ions substi-
tuted for cobalt suppress the occurrence of the monoclinic distortion
in Li0.50Co0.98Ni0.02O2 concluding that phase transitions are very
sensitive to impurity levels and could be used as a quality control
check on the material.18 More recently, we observed that both phase
transitions upon deintercalation can be suppressed as well in the
unsubstituted lithium cobalt oxide by using nominal Li/Co ratios
strictly higher than 1.0 in a high-temperature synthesis.19
LixCo0.97Ni0.03O2 (x0 ϭ 1.0) deintercalated materials for XRD
characterization were recovered in an argon-filled dry box, washed
in dimethyl carbonate ͑DMC͒ and dried under vacuum.
The XRD patterns of the Lix Co1ϪyNiyO2 starting materials were
0
recorded using a Siemens D5000 powder diffractometer using the
Cu K␣ radiation and a graphite diffracted beam monochromator.
Rietveld refinements were performed using the Fullprof program
with a pseudo-Voigt fitting function.20 XRD patterns of the
LixCo0.97Ni0.03O2 (x0 ϭ 1.0) deintercalated materials were recorded
on a Philips PW1820 powder diffractometer using the Cu K␣ radia-
tion, in a special airtight holder under argon atmosphere in order to
prevent any reaction with air moisture.
In this general context, we performed a study on the combined
effects of Ni doping and Li overstoichiometry in LiCoO2 . We syn-
thesized Lix Co1ϪyNiyO2 (x0 ϭ 1.0, 1.10; y ϭ 0.0, 0.03, 0.06, and
7Li MAS NMR spectra were recorded on a Bruker MSL200
spectrometer at 77.7 MHz, with a standard 4 mm Bruker MAS
probe. The samples were mixed with dry silica ͑typically 50% by
weight͒, in order to facilitate the spinning and improve the field
homogeneity, since they may exhibit metallic or paramagnetic prop-
erties. The mixture was placed into a 4 mm diam zirconia rotor in
0
0.10͒ materials and characterized them using X-ray diffraction
͑XRD͒, galvanostatic intercalation/deintercalation tests, and 7Li
magic angle spinning nuclear magnetic resonance ͑MAS NMR͒
spectroscopy.
the dry box. Spinning speeds ( ) of 10 and 15 kHz were used. For
vr
all phases, a Hahn echo sequence
lized in order to facilitate the phasing of all the signals and of their
t
Ϫ ␶ Ϫ t Ϫ ␶ was uti-
͔
␲/2 1 ␲ 2
͓
* Electrochemical Society Active Member.
z E-mail: delmas@icmcb.u-bordeaux.fr
Downloaded on 2015-03-28 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1534
Journal of The Electrochemical Society, 149 ͑12͒ A1533-A1540 ͑2002͒
Figure 1. Observed and calculated
XRD profiles for Li1.0Co0.94Ni0.06O2 ,
filled circles: observed; solid line, calcu-
lated; lower trace, difference plot; bars,
reflections.
spinning side bands and to ensure the observation of possibly very
wide signals which would be lost during the receiver dead time in
single pulse experiments. The 90° pulse duration used (t␲/2) was
equal to 3.05 ␮s. In order to synchronize the spin echo with the first
Results and Discussion
X-ray diffraction study. Rietveld refinements.—All the starting
Lix Co1ϪyNiyO2 materials were characterized by XRD. In all cases,
pure phases are obtained without any trace of synthesis reactants or
parasitic phases. All materials exhibit narrow diffraction lines what-
ever the percentage of nickel ions in the structure. They crystallize
0
rotational echo, ␶ was fixed to the rotor period T ϭ 1/ r . With
v
1
r
such a Hahn echo sequence, phasing can be done easily, with only
zero order correction if required, and no baseline correction is
needed. A 200 kHz spectral width was used, and the recycle time
D0 ϭ 1 s is long enough to avoid T1 saturation effects. The isotro-
pic shifts reported in parts per million are relative to an external
sample of 1 M LiCl solution in water.
¯
in the trigonal system ͑space group: R3m), with the ideal
␣-NaFeO2 layered structure of HT-LiCoO2 . Since the XRD pat-
terns are very similar and close to those reported in the literature for
the LiCo1ϪyNiyO2 system,5,16,17,21 only that of the LiCo0.94Ni0.06O2
material is shown in Fig. 1.
Since the purpose of this study is to emphasize the combined
effect of Ni doping and Li overstoichiometry on the properties of
Lix(Co,Ni)O2 phases, all the results reported in this work were ob-
All the Lix Co1ϪyNiyO2 samples were characterized by Rietveld
0
refinement of powder XRD data. As the morphology of the crystal-
lites could influence the line intensity ratios on diffraction patterns,
all the Rietveld samples were prepared in such a way that preferred
orientation was minimized.22 The refinement was led, in a first step,
tained for the same batch for each Lix Co1ϪyNiyO2 material. Fur-
0
thermore, in order to control the reproducibility of the experimental
results, two different batches of each material were characterized.
using
a strictly 2D model. Global occupancies for the 3a
(1 Ϫ y)Co ϩ yNi , 3b ͑Li͒, and 6c ͑O͒ sites were fixed to 1.0.
͓
͔
Due to Co absorption, the release of the occupancy values of the 3a
and 3b sites always leads to a decrease of the Co/Ni ͑3a͒ site occu-
pancy, no significant change in the occupancy value for the Li 3b
Table I. Rietveld refinement of the Li1.0Co0.94Ni0.06O2 XRD spec-
trum.
¯
Li1.0Co0.94Ni0.06O2 S.G.: R3m ahex.ϭ2.8176(2) Å, chex.ϭ14.062(1) Å
Table II. Hexagonal cell parameters and oxygen z position from
Wyckoff positions
refinement of XRD patterns for the various Lix Co1ÀyNiyO2
Atom
Site
Occupancy
x
y
z
B (Å2)
0
phases.
Li͑1͒
Co͑1͒
Ni͑1͒
O
3b
3a
3a
6c
0
0
0
0
0
0
0
0
1/2
0
0
1.0
0.51(40)
Ϫ0.04(8)
Ϫ0.04(8)
0.05͑16͒
a
0.94
0.06
1.0
Sample
ahex. ͑Å͒
chex. ͑Å͒
zox
c/a
b
2.81506͑16͒
2.8177͑1͒
2.8176͑2͒
2.8182͑2͒
2.81605͑18͒
2.8172͑1͒
14.0516͑8͒
14.056͑1͒
14.062͑1͒
14.067͑1͒
14.0511͑8͒
14.058͑1͒
0.2603͑4͒
0.2582͑9͒
0.2597͑5͒
0.2584͑9͒
0.2593͑6͒
0.2597͑7͒
4.99
4.99
4.99
4.99
4.99
4.99
Li1.0CoO2
0.2597(5)
Li1.0Co0.97Ni0.03O2
Li1.0Co0.94Ni0.06O2
Li1.0Co0.90Ni0.10O2
Profile function: pseudo-Voigt PV ϭ ␩L ϩ (1 Ϫ ␩)G; ␩ ϭ ␩
0
0
ϩ X(2␪) U ϭ 0.025(8), V ϭ 0.013(4), W ϭ 0.005(1);
␩
b
Li1.10CoO2
ϭ 0.23(4), X ϭ 0.004(1).
Conventional Rietveld factors: Rwp ϭ 13.4, RB ϭ 2.85, ␹ ϭ 4.78,
14 refined parameters.
2
Li1.10Co0.97Ni0.03O2
a 6c oxygen z position.
The standard deviations were multiplied by the Scor parameter (ϭ4)
to correct for local correlations.
b Ref. 19.
Downloaded on 2015-03-28 to IP 138.251.14.35 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, 149 ͑12͒ A1533-A1540 ͑2002͒
A1535
Figure 2. First galvanostatic charge of Li//LixCo1ϪyNiyO2 (x0 ϭ 1.0,
y ϭ 0.0, 0.03, 0.06, and 0.10͒ electrochemical cells (J ϭ 100 ␮A cmϪ2
;
mLi(CoNi)O ϭ 30 mg).
2
Figure 4. Expansion of the ͑003͒ Bragg reflection of the XRD patterns of
the electrochemically deintercalated LixCo0.97Ni0.03O2 (x0 ϭ 1.0) samples
(0.60 р x р 1.0).
site being observed. The results are presented in Table I for the
LiCo0.94Ni0.06O2 phase and are in agreement with the structural data
reported for pure 2D-LiCo1ϪyNiyO2 (y р 0.70) phases by Rougier
et al.21 In our Lix Co1ϪyNiyO2 phases, the hypothesis of a fraction
0
of Co and/or Ni ions in the lithium 3b site is rejected since the fits
invariably lead to a negative occupancy. However, due to the impos-
sibility of quantifying Co absorption in the Bragg-Brentano geom-
Figure 5. First galvanostatic charge of Li//LixCo1ϪyNiyO2 (x0 ϭ 1.10,
y ϭ 0.0, 0.03͒ electrochemical cells (J ϭ 100 ␮A cmϪ2
ϭ 30 mg).
; mLi(Co,Ni)O
Figure 3. XRD patterns of the electrochemically deintercalated
LixCo0.97Ni0.03O2 (x0 ϭ 1.0) samples (0.60 р x р 1.0).
2
Downloaded on 2015-03-28 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1536
Journal of The Electrochemical Society, 149 ͑12͒ A1533-A1540 ͑2002͒
Figure 6. Cycling behavior of the various Li//LixCo1ϪyNiyO2 (x0 ϭ 1.0, 1.10; y ϭ 0.0, 0.03, 0.06, and 0.10͒ electrochemical cells (J ϭ 400 ␮A cmϪ2
;
mLi(Co,Ni)O ϭ 15 mg, C/20 rate͒ between 2.7 and 4.15 V.
2
etry, no conclusion could be drawn about the presence of Li ions in
the 3a crystallographic site of Co/Ni. Nevertheless, in agreement
with the previous neutron diffraction studies on pure LiCoO2 , one
can consider the amount of Liϩ ions in the Co/Ni 3a site as
negligible.3,19 The very good fitting of the profile of the XRD lines
and the small intensity of the line in the difference pattern empha-
size the quality of the Rietveld refinement.
due to a semiconductor-to-metal transition and the particular feature
associated to the monoclinic distortion expected for Li0.50CoO2 due
to the interslab lithium/vacancy ordering.4,7
A quite different behavior is observed for the Ni-doped materials.
Ͼ
In the first part of the oxidation of the LiCo1ϪyNiyO2 (y
0)
phases, the preferential oxidation of Ni ions compared to Co ones
leads to the appearance of a lower potential step, all the more pro-
nounced when the nickel amount increases. This difference in po-
tential is due to the lower oxidizing power of the NiIII/NiIV redox
couple compared to the CoIII/CoIV one.17,26 Moreover, the voltage
plateau at the beginning of the charge process is less marked for the
Ni-doped materials, even with only 3% of the doping element in the
LiCoO2 lattice, suggesting that the two-phase domain does not exist
anymore in these materials. This is fully confirmed by the XRD
patterns of electrochemically deintercalated LixCo0.97Ni0.03O2 mate-
rials; as shown in Fig. 3 and 4, a solid solution is obtained for
0.60 р x р 1.0 in the LixCo0.97Ni0.03O2 (x0 ϭ 1.0) system. One
can see in Fig. 2 that the monoclinic distortion for x ϭ 0.50 exists
in all the LixCo1ϪyNiyO2 (y ϭ 0.0, 0.03, 0.06, and 0.10͒ systems
studied here. This result conflicts with the conclusion by Reimers
et al. who showed in a former study that 2% of the Ni ions substi-
tuted for cobalt are sufficient to suppress the monoclinic distortion at
room temperature.18 Besides, one can see on the inset of Fig. 2 that
this order/disorder transition tends to disappear for Ni contents
higher than 10%.
The structural parameters of the various Lix Co1ϪyNiyO2 mate-
0
rials are presented in Table II. Both the ahex. and the chex. parameters
increase smoothly with Ni doping due to the difference in size be-
tween the trivalent nickel and cobalt ions (rNiIII ϭ 0.56 Å, rCoIII
ϭ 0.545 Å23͒. Furthermore, in all materials, the chex. /ahex. ratio
close to five (Þ4.90 for a cubic lattice͒ confirms the 2D character of
the structure of the Ni-doped LiCoO2 phases,21,24 in good agreement
with the high and constant value of the oxygen z position (zox.) and
the consequent difference in the slab (S) and interslab space (I)
thicknesses (S ϭ 2.06 Å Þ I ϭ 2.63 Å for Lix Co1ϪyNiyO2
0
phases͒.25
Electrochemical Study
Lithium-stoichiometric LiCo1ϪyNiyO2 phases (y ϭ 0.0, 0.03,
0.06, and 0.10).—Figure 2 shows the first galvanostatic charge
curve of Li//LixCo1ϪyNiyO2 cells (x0 ϭ 1.0; y ϭ 0.0, 0.03, 0.06,
and 0.10͒ at low rate J ϭ 100 ␮A cmϪ2, mLi(Co,Ni)O ϭ 30 mg;
͓
2
C/160 rate ͑160 h are required to exchange one equivalent of
electron͔͒.
Lithium overstoichiometric Li1.10Co1ϪyNiyO2 phases (y ϭ 0.0
and 0.03).—Contrary to the case of the Li stoichiometric materials,
as shown in Fig. 5, the first galvanostatic charge curves of
Li//LixCo1ϪyNiyO2 cells (x0 ϭ 1.10; y ϭ 0.0 and 0.03͒ at a low
As already reported, the charge curve obtained with the pure
LixCoO2 (x0 ϭ 1.0) material exhibits both the voltage plateau at ca.
3.93 V corresponding to the biphasic domain for 0.75 р x р 0.94
Downloaded on 2015-03-28 to IP 138.251.14.35 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, 149 ͑12͒ A1533-A1540 ͑2002͒
A1537
Figure 7. Ϫdx/
͉dV͉ ϭ f(V) derivatives curves of the first ͑solid line͒ and 10th ͑dashed line͒ cycles of the various Li//LixCo1ϪyNiyO2 (x0 ϭ 1.0, 1.10; y
ϭ 0.0, 0.03, 0.06, and 0.10͒ electrochemical cells (J ϭ 400 ␮A cmϪ2; mLi(CoNi)O ϭ 15 mg) between 2.7 and 4.15 V.
2
rate (J ϭ 100 ␮A cmϪ2, mLi(Co,Ni)O ϭ 30 mg, C/180 charging
and that the irreversible capacity at the first cycle is especially low
compared to that of the LiNiO2-type compounds.27
The derivative curves Ϫdx/ dV ϭ f(V) of the first and tenth
2
rate͒ have a quite similar profile, whatever the value of y. The
feature corresponding to the monoclinic distortion for x ϭ 0.50 and
the voltage plateau at the beginning of the charge have disappeared
both for the Ni-doped and for the pure Li1.10CoO2 phases. The uni-
directional increase of the potential during lithium deintercalation
suggests the occurrence of a monophasic reaction throughout the
charge process for the Li1.10Co1ϪyNiyO2 materials. An XRD study
of electrochemically deintercalated LixCoO2 (x0 ϭ 1.10) materials
carried out in a former work had confirmed such an observation.19
The same behavior is expected for the Li1.10Co0.97Ni0.03O2 system,
although no XRD characterization was performed on deintercalated
materials.
͉
͉
cycles of the Li//Lix Co1ϪyNiyO2 cells presented just above are re-
0
ported in Fig. 7. One can note the peak at the lower potential due to
the preferential oxidation of NiIII ions compared to CoIII ones at the
beginning of the charging process. Furthermore, as already evi-
denced from the slow rate charging curves ͑Fig. 2͒, the trace of the
monoclinic distortion around 4.15 V is easily observable in all the
cycling curves from the Li stoichiometric LiCo1ϪyNiyO2 phases and
still exists after ten cycles, confirming the persistence of an order/
disorder transition for x ϭ 0.50 in these materials. On the contrary,
it does not appear for lithium overstoichiometric materials.
7Li MAS NMR Study
Cycling tests.—Li/LiClO4-PC/Lix Co1ϪyNiyO2 (x0 ϭ 1.0, 1.10;
0
y ϭ 0.0, 0.03, 0.06, and 0.10͒ cells were cycled between 2.7 and
4.15 V at 400 ␮A cmϪ2 ͑C/20 rate͒ as shown in Fig. 6. The revers-
ible capacity for the first and tenth cycle is also indicated on the
figure. The purpose of this study is to characterize the structural
transitions and not to obtain large reversible capacities, a relatively
low high-voltage cutoff was used.
7Li MAS NMR spectroscopy measurements were carried out on
the various Lix Co1ϪyNiyO2 starting materials. Figures 8 and 9 show
0
the 15 kHz 7Li MAS NMR spectra of the lithium-stoichiometric
LiCo1ϪyNiyO2 materials (y ϭ 0.0, 0.03, and 0.06͒ and of the Li
overstoichiometric Li1.10Co0.97Ni0.03O2 phase.
As already reported, LiCoO2 exhibits a single narrow signal
close to 0 ppm due to the exclusive presence of diamagnetic low
spin CoIII ions in the structure, so that no hyperfine interaction is
expected.28,29 For lithium stoichiometric Ni-doped materials, three
new signals at 110, Ϫ15, and Ϫ30 ppm are observed. These signals
have been attributed by Marichal et al. to the presence of paramag-
netic low spin NiIII ions with one single electron in the eg orbital
(t62ge1g). The transfer of hyperfine interaction between the single eg
electron toward the surrounding lithium ions can be done via the
As already reported elsewhere, the cycling behavior of the Co-
rich LixCo1ϪyNiyO2 phases is very satisfactory and close to that of
pure LiCoO2 .16,24 The cycling curves, on Fig. 6, show that the re-
versibility of the deintercalation/intercalation process is good, as the
reversible capacities for the first and tenth cycles, shown in Fig. 6,
are quite high in the voltage window used. Note that the slight
decrease in potential caused by 3% doping of Ni leads to a particu-
larly good cycling capacity for the cutoff voltage used ͑4.15 V͒.
Furthermore, one can note that polarization is very small in all cases
Downloaded on 2015-03-28 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1538
Journal of The Electrochemical Society, 149 ͑12͒ A1533-A1540 ͑2002͒
Figure 8. 15 kHz 7Li MAS NMR spectra for the various Lix Co1ϪyNiyO2
0
Figure 9. Expansion of the central signal of 15 kHz 7Li MAS NMR spectra
*
phases (x0 ϭ 1.0, 1.10; y ϭ 0.0, 0.03, 0.06͒, ( ϭ spinning side bands͒.
for the various Lix Co1ϪyNiyO2 phases (x0 ϭ 1.0, 1.10; y ϭ 0.0, 0.03,
0
0.06͒.
overlap with the oxygen orbitals with a 90° ͑first neighbor͒ or 180°
geometry ͑second neighbor͒. The 110 ppm signal is assigned to Li
ions with one NiIII ions as the first neighbor while the Ϫ15 and
Ϫ30 ppm signals result from the presence of Li ions with one and
two NiIII ions as the second neighbor, respectively.26,29 As expected,
the magnitude of the NiIII ions signals is all the more pronounced
since the percentage of Ni ions in the material is large.
close to the oxygen vacancy. These shoulders are clearly apparent in
the spectra of the Li1.10Co0.97Ni0.03O2 .19 However, since the Ϫ16
and Ϫ40 ppm Co3ϩ(IS) signals must get mixed up with the Ϫ15 and
Ϫ30 ppm NiIII signals and because the first spinning side band must
hide the possible presence of a signal at 185 ppm, one cannot clearly
observe these signals on the 15 kHz MAS spectra.
As concerns the first-neighbor effect of Co3ϩ(IS) ions in
Li1.10Co0.97Ni0.03O2 , the first spinning side band at Ϯ190 ppm is
clearly superposed on another band in the case of the lithium over-
stoichiometric material ͑Fig. 8͒. Therefore, in order to better observe
the 150/250 ppm range, we have performed MAS NMR measure-
ments using a 10 kHz rotation speed to shift the spinning side band
without modifying the isotropic shifts. The NMR spectra, shown
in Fig. 10 and compared to that of the Li1.10CoO2 phase clearly
show the signal associated with Co3ϩ(IS) at 185 ppm for the
Li1.10Co0.97Ni0.03O2 material. The expansion of the central signal
of the various compounds ͑Fig. 11͒ confirms the presence of
both NiIII and Co3ϩ(IS) ions in the lithium overstoichiometric
Li1.10Co0.97Ni0.03O2 phase, the contribution of each type of paramag-
netic ion adding in the Ϫ40/10 ppm range and leading to a loss in
the central signal definition.
Furthermore, one can observe in Fig. 9 that the central signal of
the Li1.10Co0.97Ni0.03O2 material is quite different from those of the
lithium stoichiometric Ni-doped phases, showing characteristic
shoulders centered approximately at 3 and Ϫ6 ppm. These features
are similar to those observed in the case of the Li1.10CoO2 phase ͑see
Fig. 11͒, resulting from the presence of a structural defect leading to
oxygen vacancies in the material. In Li1.10CoO2 , these signals, ini-
tially attributed to the presence of divalent cobalt ions, are in fact
due to Liϩ ions in the vicinity of paramagnetic trivalent cobalt ions
in a quite unusual spin state.19,30 Indeed, this structural defect, in-
duced by a nominal Li excess during the synthesis, leads to the
presence of intermediate spinning Co3ϩ ions ͑labeled Co3ϩ(IS)
)
which are responsible for three characteristic first/second neighbor
effect NMR signals at 185 ppm ͑one Co3ϩ(IS) first neighbor͒ and
Ϫ16 and Ϫ40 ppm ͑one and two Co3ϩ(IS) second neighbor, respec-
tively͒. Furthermore, the two shoulders on the 0 ppm signal ͑3 and
Ϫ6 ppm) are considered to be due to Liϩ ions in an oxygen envi-
ronment perturbed by the oxygen vacancy, leading to various
mechanisms of transfer of the hyperfine interaction from the Liϩ ion
Conclusions
The galvanostatic measurements and XRD results allow us to
point out some conclusions on the influence of Ni doping and Li
Downloaded on 2015-03-28 to IP 138.251.14.35 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, 149 ͑12͒ A1533-A1540 ͑2002͒
A1539
7
Figure 10. 10 kHz Li MAS NMR spectra for the various Lix Co1ϪyNiyO2
0
*
Figure 11. Expansion of the central signal of 10 kHz 7Li MAS NMR spectra
phases (x0 ϭ 1.0, 1.10; y ϭ 0.0, 0.03͒, ( ϭ spinning side bands͒.
for the various Lix Co1ϪyNiyO2 phases (x0 ϭ 1.0, 1.10; y ϭ 0.0, 0.03͒.
0
nonstoichiometry in HT-LiCoO2 . This study clearly allows us to
separate the relative effects of Ni-doping and lithium overstoichiom-
etry which was not considered by Reimers et al.18 We show here
that 3% of Ni ions substituted for Co do not make the monoclinic
distortion disappear.
close ionic radii of the NiIV and CoIII ions.23 When 10% of Ni are
substituted for Co, the effect becomes significant and the trace of the
monoclinic distortion becomes weak. As a possible explanation of
the discrepancy between our and Reimers’s results on that point, we
suggest that the Ni-substituted samples studied by this author may
have been slightly Li-overstoichiometric.
Second, from the electronic properties point of view, the Ni ions
are trapped in the lattice and are an obstacle to electron delocaliza-
tion. Therefore, the expected phase separation associated with the
semiconductor-to-metal transition cannot occur; a solid solution do-
main is observed upon lithium deintercalation. The behavior result-
ing from the presence of the nickel ions is very similar to that due to
Before any conclusion, we have to consider what are the driving
forces for the phase transitions in the pure LixCoO2 (x0 ϭ 1.0)
system. On the one hand, it is now totally admitted that the biphasic
domain for 0.75 р x р 0.94 results from a brutal change in the
electronic properties of the LixCoO2 system as it undergoes a
semiconductor-to-metal transition at the macroscopic scale.7 On the
other hand, the monoclinic distortion for x ϭ 0.50 is considered to
have a structural origin related to the establishment of a Li/vacancy
ordering in the interslab layer.4 Indeed, a recent TEM study clearly
showed, for the first time, the Li/vacancy ordering in the monoclinic
Ͼ
the structural defects associated with lithium overstoichiometry in
19,30
Ͼ
the Lix CoO2(x0
1.0) phases.
0
Li0.50CoO2 phase.31 In the Lix CoO2 system (x0
1.0), the pres-
When both type of defects are simultaneously present ͑Ni ions
and defects related to lithium overstoichiometry͒ as shown by NMR
measurements conducted on the Li1.10Co0.97Ni0.03O2 phase, all the
phase transitions disappear.
Finally, the NMR measurements show that the doping effects of
nickel and lithium are cumulative and independent. Indeed, in the
Li1.10Co0.97Ni0.03O2 phase, both Co3ϩ(IS) ions ͑associated to an oxy-
gen vacancy͒ and NiIII ions coexist in the diamagnetic CoIII lattice.
0
ence of structural defects ͑oxygen vacancy associated to Co3ϩ(IS)
ions͒ induces enough perturbations in the slab to prevent any Li/
vacancy ordering in the interslab space.30 From the electronic prop-
erties point of view, even if the semiconductor-to-metal transition is
observed upon lithium deintercalation, this transition is not associ-
ated with a phase separation since the defects are trapped in the
lattice and cannot move at room temperature.
In the case of Ni-doped materials synthesized without any
lithium excess, the experimental results show first that 6% of Ni ions
substituted for Co do not create sufficient perturbations in the
(Co,Ni)O2 slab ordering to prevent the lithium/vacancy ordering in
the insterlab and the resulting monoclinic distortion because of the
Acknowledgments
The authors thank I. Saadoune, A. Audemer, L. Gautier, and J.
Scoyer for fruitful discussions, and Umicore, ANRT, and Region
Aquitaine for financial support.
´
Downloaded on 2015-03-28 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1540
Journal of The Electrochemical Society, 149 ͑12͒ A1533-A1540 ͑2002͒
`
´
Institut de Chimie de la Matiere Condensee de Bordeaux assisted in
meeting the publication costs of this article.
15. S. Waki, K. Dokko, T. Itoh, M. Nishizawa, T. Abe, and I. Uchida, J. Solid State
Electrochem., 4, 205 ͑2000͒.
16. C. Delmas, I. Saadoune, and A. Rougier, J. Power Sources, 43-44, 595 ͑1993͒.
17. I. Saadoune and C. Delmas, J. Mater. Chem., 6, 193 ͑1996͒.
18. J. N. Reimers, J. R. Dahn, and U. Von Sacken, J. Electrochem. Soc., 140, 2752
͑1993͒.
References
1. K. Mizushima, P. C. Jones, P. J. Wiseman, and J. B. Goodenough, Mater. Res. Bull.,
15, 783 ͑1980͒.
2. H. Abe, K. Zaghib, K. Tatsumi, and S. Higuchi, J. Power Sources, 54, 236 ͑1995͒.
3. H. J. Orman and P. J. Wiseman, Acta Crystallogr., Sect. C: Cryst. Struct. Commun.,
1984, 12.
4. J. N. Reimers and J. R. Dahn, J. Electrochem. Soc., 139, 2091 ͑1992͒.
5. T. Ohzuku and A. Ueda, J. Electrochem. Soc., 141, 2972 ͑1994͒.
6. G. G. Amatucci, J. M. Tarascon, and L. C. Klein, J. Electrochem. Soc., 143, 1114
͑1996͒.
7. M. Menetrier, I. Saadoune, S. Levasseur, and C. Delmas, J. Mater. Chem., 9, 1135
͑1999͒.
8. C. D. W. Jones, E. Rossen, and J. R. Dahn, Solid State Ionics, 68, 65 ͑1994͒.
9. R. Alcantara, P. Lavela, P. L. Relano, J. L. Tirado, E. Zhecheva, and R. Stoyanova,
Inorg. Chem., 37, 264 ͑1998͒.
10. Y. I. Jang, B. Huang, H. Wang, G. R. Maslkaly, G. Ceder, D. R. Sadoway, Y. M.
Chiang, H. Liu, and H. Tamura, J. Power Sources, 81-82, 589 ͑1999͒.
11. H. Tukamoto and A. R. West, J. Electrochem. Soc., 144, 3164 ͑1997͒.
12. E. Antolini, L. Giorgi, and M. Carewska, J. Mater. Sci. Lett., 18, 325 ͑1999͒.
13. H. Kobayachi, H. Shigemura, M. Tabuchi, H. Sakaebe, K. Ado, H. Kageyama, A.
Hirano, R. Kanno, M. Wakita, S. Morimoto, and S. Nasu, J. Electrochem. Soc.,
147, 960 ͑2000͒.
´
´
19. S. Levasseur, M. Menetrier, E. Suard, and C. Delmas, Solid State Ionics, 128, 11
͑2000͒.
20. J. Rodriguez-Carvajal, in Fullprof, Laboratoire L. Brillouin ͑CEA-CNRS͒, http://
www-llb.cea.fr/fullweb/powder.htm ͑1990͒.
21. A. Rougier, I. Saadoune, P. Gravereau, P. Willmann, and C. Delmas, Solid State
Ionics, 90, 83 ͑1996͒.
22. L. B. Mccusker, R. B. Von Dreele, D. E. Cox, D. Louer, and P. Scardi, J. Appl.
Crystallogr., 32, 36 ͑1999͒.
23. R. D. Shannon and C. T. Prewitt, J. Appl. Crystallogr., 25, 925 ͑1969͒.
´
´
´
´
24. I. Saadoune, M. Menetrier, and C. Delmas, J. Mater. Chem., 7, 2505 ͑1997͒.
25. A. Rougier, P. Gravereau, and C. Delmas, J. Electrochem. Soc., 143, 1168 ͑1996͒.
´
´
26. D. Carlier, M. Menetrier, and C. Delmas, J. Mater. Chem., 11, 594 ͑2001͒.
´ `
27. J. P. Peres, C. Delmas, A. Rougier, M. Broussely, F. Perton, P. Biensan, and P.
Willmann, J. Phys. Chem. Solids, 57, 1057 ͑1996͒.
´
´
28. M. Menetrier, A. Rougier, and C. Delmas, Solid State Commun., 90, 439 ͑1994͒.
´
´
29. C. Marichal, J. Hirschinger, P. Granger, M. Menetrier, A. Rougier, and C. Delmas,
Inorg. Chem., 34, 1773 ͑1995͒.
´
´
30. S. Levasseur, M. Menetrier, L. Gautier, A. Audemer, G. Demazeau, A. Largeteau,
and C. Delmas, J. Mater. Chem., Submitted ͑2001͒.
14. R. Stoyanova, E. Zhecheva, and L. Zarkova, Solid State Ionics, 73, 233 ͑1994͒.
31. Y. Shao Horn, F. Weill, S. Levasseur, and C. Delmas, To be published.
Downloaded on 2015-03-28 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).